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, get-mark, ls-tree, and cat-blob requests may
 138      # appear anywhere in the input, except within a data command. Any
 139      # form of the data command always escapes the related input from
 140      # 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   get_mark ::= 'get-mark' sp idnum lf;
 148   cat_blob ::= 'cat-blob' sp (hexsha1 | idnum) lf;
 149   ls_tree  ::= 'ls' sp (hexsha1 | idnum) sp path_str lf;
 150
 151   comment ::= '#' not_lf* lf;
 152   not_lf  ::= # Any byte that is not ASCII newline (LF);
 153 */
 154
 155 #include "builtin.h"
 156 #include "cache.h"
 157 #include "config.h"
 158 #include "lockfile.h"
 159 #include "object.h"
 160 #include "blob.h"
 161 #include "tree.h"
 162 #include "commit.h"
 163 #include "delta.h"
 164 #include "pack.h"
 165 #include "refs.h"
 166 #include "csum-file.h"
 167 #include "quote.h"
 168 #include "dir.h"
 169 #include "run-command.h"
 170
 171 #define PACK_ID_BITS 16
 172 #define MAX_PACK_ID ((1<<PACK_ID_BITS)-1)
 173 #define DEPTH_BITS 13
 174 #define MAX_DEPTH ((1<<DEPTH_BITS)-1)
 175
 176 /*
 177  * We abuse the setuid bit on directories to mean "do not delta".
 178  */
 179 #define NO_DELTA S_ISUID
 180
 181 struct object_entry {
 182         struct pack_idx_entry idx;
 183         struct object_entry *next;
 184         uint32_t type : TYPE_BITS,
 185                 pack_id : PACK_ID_BITS,
 186                 depth : DEPTH_BITS;
 187 };
 188
 189 struct object_entry_pool {
 190         struct object_entry_pool *next_pool;
 191         struct object_entry *next_free;
 192         struct object_entry *end;
 193         struct object_entry entries[FLEX_ARRAY]; /* more */
 194 };
 195
 196 struct mark_set {
 197         union {
 198                 struct object_entry *marked[1024];
 199                 struct mark_set *sets[1024];
 200         } data;
 201         unsigned int shift;
 202 };
 203
 204 struct last_object {
 205         struct strbuf data;
 206         off_t offset;
 207         unsigned int depth;
 208         unsigned no_swap : 1;
 209 };
 210
 211 struct mem_pool {
 212         struct mem_pool *next_pool;
 213         char *next_free;
 214         char *end;
 215         uintmax_t space[FLEX_ARRAY]; /* more */
 216 };
 217
 218 struct atom_str {
 219         struct atom_str *next_atom;
 220         unsigned short str_len;
 221         char str_dat[FLEX_ARRAY]; /* more */
 222 };
 223
 224 struct tree_content;
 225 struct tree_entry {
 226         struct tree_content *tree;
 227         struct atom_str *name;
 228         struct tree_entry_ms {
 229                 uint16_t mode;
 230                 struct object_id oid;
 231         } versions[2];
 232 };
 233
 234 struct tree_content {
 235         unsigned int entry_capacity; /* must match avail_tree_content */
 236         unsigned int entry_count;
 237         unsigned int delta_depth;
 238         struct tree_entry *entries[FLEX_ARRAY]; /* more */
 239 };
 240
 241 struct avail_tree_content {
 242         unsigned int entry_capacity; /* must match tree_content */
 243         struct avail_tree_content *next_avail;
 244 };
 245
 246 struct branch {
 247         struct branch *table_next_branch;
 248         struct branch *active_next_branch;
 249         const char *name;
 250         struct tree_entry branch_tree;
 251         uintmax_t last_commit;
 252         uintmax_t num_notes;
 253         unsigned active : 1;
 254         unsigned delete : 1;
 255         unsigned pack_id : PACK_ID_BITS;
 256         struct object_id oid;
 257 };
 258
 259 struct tag {
 260         struct tag *next_tag;
 261         const char *name;
 262         unsigned int pack_id;
 263         struct object_id oid;
 264 };
 265
 266 struct hash_list {
 267         struct hash_list *next;
 268         struct object_id oid;
 269 };
 270
 271 typedef enum {
 272         WHENSPEC_RAW = 1,
 273         WHENSPEC_RFC2822,
 274         WHENSPEC_NOW
 275 } whenspec_type;
 276
 277 struct recent_command {
 278         struct recent_command *prev;
 279         struct recent_command *next;
 280         char *buf;
 281 };
 282
 283 /* Configured limits on output */
 284 static unsigned long max_depth = 50;
 285 static off_t max_packsize;
 286 static int unpack_limit = 100;
 287 static int force_update;
 288
 289 /* Stats and misc. counters */
 290 static uintmax_t alloc_count;
 291 static uintmax_t marks_set_count;
 292 static uintmax_t object_count_by_type[1 << TYPE_BITS];
 293 static uintmax_t duplicate_count_by_type[1 << TYPE_BITS];
 294 static uintmax_t delta_count_by_type[1 << TYPE_BITS];
 295 static uintmax_t delta_count_attempts_by_type[1 << TYPE_BITS];
 296 static unsigned long object_count;
 297 static unsigned long branch_count;
 298 static unsigned long branch_load_count;
 299 static int failure;
 300 static FILE *pack_edges;
 301 static unsigned int show_stats = 1;
 302 static int global_argc;
 303 static const char **global_argv;
 304
 305 /* Memory pools */
 306 static size_t mem_pool_alloc = 2*1024*1024 - sizeof(struct mem_pool);
 307 static size_t total_allocd;
 308 static struct mem_pool *mem_pool;
 309
 310 /* Atom management */
 311 static unsigned int atom_table_sz = 4451;
 312 static unsigned int atom_cnt;
 313 static struct atom_str **atom_table;
 314
 315 /* The .pack file being generated */
 316 static struct pack_idx_option pack_idx_opts;
 317 static unsigned int pack_id;
 318 static struct sha1file *pack_file;
 319 static struct packed_git *pack_data;
 320 static struct packed_git **all_packs;
 321 static off_t pack_size;
 322
 323 /* Table of objects we've written. */
 324 static unsigned int object_entry_alloc = 5000;
 325 static struct object_entry_pool *blocks;
 326 static struct object_entry *object_table[1 << 16];
 327 static struct mark_set *marks;
 328 static const char *export_marks_file;
 329 static const char *import_marks_file;
 330 static int import_marks_file_from_stream;
 331 static int import_marks_file_ignore_missing;
 332 static int import_marks_file_done;
 333 static int relative_marks_paths;
 334
 335 /* Our last blob */
 336 static struct last_object last_blob = { STRBUF_INIT, 0, 0, 0 };
 337
 338 /* Tree management */
 339 static unsigned int tree_entry_alloc = 1000;
 340 static void *avail_tree_entry;
 341 static unsigned int avail_tree_table_sz = 100;
 342 static struct avail_tree_content **avail_tree_table;
 343 static struct strbuf old_tree = STRBUF_INIT;
 344 static struct strbuf new_tree = STRBUF_INIT;
 345
 346 /* Branch data */
 347 static unsigned long max_active_branches = 5;
 348 static unsigned long cur_active_branches;
 349 static unsigned long branch_table_sz = 1039;
 350 static struct branch **branch_table;
 351 static struct branch *active_branches;
 352
 353 /* Tag data */
 354 static struct tag *first_tag;
 355 static struct tag *last_tag;
 356
 357 /* Input stream parsing */
 358 static whenspec_type whenspec = WHENSPEC_RAW;
 359 static struct strbuf command_buf = STRBUF_INIT;
 360 static int unread_command_buf;
 361 static struct recent_command cmd_hist = {&cmd_hist, &cmd_hist, NULL};
 362 static struct recent_command *cmd_tail = &cmd_hist;
 363 static struct recent_command *rc_free;
 364 static unsigned int cmd_save = 100;
 365 static uintmax_t next_mark;
 366 static struct strbuf new_data = STRBUF_INIT;
 367 static int seen_data_command;
 368 static int require_explicit_termination;
 369 static int allow_unsafe_features;
 370
 371 /* Signal handling */
 372 static volatile sig_atomic_t checkpoint_requested;
 373
 374 /* Where to write output of cat-blob commands */
 375 static int cat_blob_fd = STDOUT_FILENO;
 376
 377 static void parse_argv(void);
 378 static void parse_get_mark(const char *p);
 379 static void parse_cat_blob(const char *p);
 380 static void parse_ls(const char *p, struct branch *b);
 381
 382 static void write_branch_report(FILE *rpt, struct branch *b)
 383 {
 384         fprintf(rpt, "%s:\n", b->name);
 385
 386         fprintf(rpt, "  status      :");
 387         if (b->active)
 388                 fputs(" active", rpt);
 389         if (b->branch_tree.tree)
 390                 fputs(" loaded", rpt);
 391         if (is_null_oid(&b->branch_tree.versions[1].oid))
 392                 fputs(" dirty", rpt);
 393         fputc('\n', rpt);
 394
 395         fprintf(rpt, "  tip commit  : %s\n", oid_to_hex(&b->oid));
 396         fprintf(rpt, "  old tree    : %s\n",
 397                 oid_to_hex(&b->branch_tree.versions[0].oid));
 398         fprintf(rpt, "  cur tree    : %s\n",
 399                 oid_to_hex(&b->branch_tree.versions[1].oid));
 400         fprintf(rpt, "  commit clock: %" PRIuMAX "\n", b->last_commit);
 401
 402         fputs("  last pack   : ", rpt);
 403         if (b->pack_id < MAX_PACK_ID)
 404                 fprintf(rpt, "%u", b->pack_id);
 405         fputc('\n', rpt);
 406
 407         fputc('\n', rpt);
 408 }
 409
 410 static void dump_marks_helper(FILE *, uintmax_t, struct mark_set *);
 411
 412 static void write_crash_report(const char *err)
 413 {
 414         char *loc = git_pathdup("fast_import_crash_%"PRIuMAX, (uintmax_t) getpid());
 415         FILE *rpt = fopen(loc, "w");
 416         struct branch *b;
 417         unsigned long lu;
 418         struct recent_command *rc;
 419
 420         if (!rpt) {
 421                 error_errno("can't write crash report %s", loc);
 422                 free(loc);
 423                 return;
 424         }
 425
 426         fprintf(stderr, "fast-import: dumping crash report to %s\n", loc);
 427
 428         fprintf(rpt, "fast-import crash report:\n");
 429         fprintf(rpt, "    fast-import process: %"PRIuMAX"\n", (uintmax_t) getpid());
 430         fprintf(rpt, "    parent process     : %"PRIuMAX"\n", (uintmax_t) getppid());
 431         fprintf(rpt, "    at %s\n", show_date(time(NULL), 0, DATE_MODE(ISO8601)));
 432         fputc('\n', rpt);
 433
 434         fputs("fatal: ", rpt);
 435         fputs(err, rpt);
 436         fputc('\n', rpt);
 437
 438         fputc('\n', rpt);
 439         fputs("Most Recent Commands Before Crash\n", rpt);
 440         fputs("---------------------------------\n", rpt);
 441         for (rc = cmd_hist.next; rc != &cmd_hist; rc = rc->next) {
 442                 if (rc->next == &cmd_hist)
 443                         fputs("* ", rpt);
 444                 else
 445                         fputs("  ", rpt);
 446                 fputs(rc->buf, rpt);
 447                 fputc('\n', rpt);
 448         }
 449
 450         fputc('\n', rpt);
 451         fputs("Active Branch LRU\n", rpt);
 452         fputs("-----------------\n", rpt);
 453         fprintf(rpt, "    active_branches = %lu cur, %lu max\n",
 454                 cur_active_branches,
 455                 max_active_branches);
 456         fputc('\n', rpt);
 457         fputs("  pos  clock name\n", rpt);
 458         fputs("  ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~\n", rpt);
 459         for (b = active_branches, lu = 0; b; b = b->active_next_branch)
 460                 fprintf(rpt, "  %2lu) %6" PRIuMAX" %s\n",
 461                         ++lu, b->last_commit, b->name);
 462
 463         fputc('\n', rpt);
 464         fputs("Inactive Branches\n", rpt);
 465         fputs("-----------------\n", rpt);
 466         for (lu = 0; lu < branch_table_sz; lu++) {
 467                 for (b = branch_table[lu]; b; b = b->table_next_branch)
 468                         write_branch_report(rpt, b);
 469         }
 470
 471         if (first_tag) {
 472                 struct tag *tg;
 473                 fputc('\n', rpt);
 474                 fputs("Annotated Tags\n", rpt);
 475                 fputs("--------------\n", rpt);
 476                 for (tg = first_tag; tg; tg = tg->next_tag) {
 477                         fputs(oid_to_hex(&tg->oid), rpt);
 478                         fputc(' ', rpt);
 479                         fputs(tg->name, rpt);
 480                         fputc('\n', rpt);
 481                 }
 482         }
 483
 484         fputc('\n', rpt);
 485         fputs("Marks\n", rpt);
 486         fputs("-----\n", rpt);
 487         if (export_marks_file)
 488                 fprintf(rpt, "  exported to %s\n", export_marks_file);
 489         else
 490                 dump_marks_helper(rpt, 0, marks);
 491
 492         fputc('\n', rpt);
 493         fputs("-------------------\n", rpt);
 494         fputs("END OF CRASH REPORT\n", rpt);
 495         fclose(rpt);
 496         free(loc);
 497 }
 498
 499 static void end_packfile(void);
 500 static void unkeep_all_packs(void);
 501 static void dump_marks(void);
 502
 503 static NORETURN void die_nicely(const char *err, va_list params)
 504 {
 505         static int zombie;
 506         char message[2 * PATH_MAX];
 507
 508         vsnprintf(message, sizeof(message), err, params);
 509         fputs("fatal: ", stderr);
 510         fputs(message, stderr);
 511         fputc('\n', stderr);
 512
 513         if (!zombie) {
 514                 zombie = 1;
 515                 write_crash_report(message);
 516                 end_packfile();
 517                 unkeep_all_packs();
 518                 dump_marks();
 519         }
 520         exit(128);
 521 }
 522
 523 #ifndef SIGUSR1 /* Windows, for example */
 524
 525 static void set_checkpoint_signal(void)
 526 {
 527 }
 528
 529 #else
 530
 531 static void checkpoint_signal(int signo)
 532 {
 533         checkpoint_requested = 1;
 534 }
 535
 536 static void set_checkpoint_signal(void)
 537 {
 538         struct sigaction sa;
 539
 540         memset(&sa, 0, sizeof(sa));
 541         sa.sa_handler = checkpoint_signal;
 542         sigemptyset(&sa.sa_mask);
 543         sa.sa_flags = SA_RESTART;
 544         sigaction(SIGUSR1, &sa, NULL);
 545 }
 546
 547 #endif
 548
 549 static void alloc_objects(unsigned int cnt)
 550 {
 551         struct object_entry_pool *b;
 552
 553         b = xmalloc(sizeof(struct object_entry_pool)
 554                 + cnt * sizeof(struct object_entry));
 555         b->next_pool = blocks;
 556         b->next_free = b->entries;
 557         b->end = b->entries + cnt;
 558         blocks = b;
 559         alloc_count += cnt;
 560 }
 561
 562 static struct object_entry *new_object(struct object_id *oid)
 563 {
 564         struct object_entry *e;
 565
 566         if (blocks->next_free == blocks->end)
 567                 alloc_objects(object_entry_alloc);
 568
 569         e = blocks->next_free++;
 570         oidcpy(&e->idx.oid, oid);
 571         return e;
 572 }
 573
 574 static struct object_entry *find_object(struct object_id *oid)
 575 {
 576         unsigned int h = oid->hash[0] << 8 | oid->hash[1];
 577         struct object_entry *e;
 578         for (e = object_table[h]; e; e = e->next)
 579                 if (!oidcmp(oid, &e->idx.oid))
 580                         return e;
 581         return NULL;
 582 }
 583
 584 static struct object_entry *insert_object(struct object_id *oid)
 585 {
 586         unsigned int h = oid->hash[0] << 8 | oid->hash[1];
 587         struct object_entry *e = object_table[h];
 588
 589         while (e) {
 590                 if (!oidcmp(oid, &e->idx.oid))
 591                         return e;
 592                 e = e->next;
 593         }
 594
 595         e = new_object(oid);
 596         e->next = object_table[h];
 597         e->idx.offset = 0;
 598         object_table[h] = e;
 599         return e;
 600 }
 601
 602 static void invalidate_pack_id(unsigned int id)
 603 {
 604         unsigned int h;
 605         unsigned long lu;
 606         struct tag *t;
 607
 608         for (h = 0; h < ARRAY_SIZE(object_table); h++) {
 609                 struct object_entry *e;
 610
 611                 for (e = object_table[h]; e; e = e->next)
 612                         if (e->pack_id == id)
 613                                 e->pack_id = MAX_PACK_ID;
 614         }
 615
 616         for (lu = 0; lu < branch_table_sz; lu++) {
 617                 struct branch *b;
 618
 619                 for (b = branch_table[lu]; b; b = b->table_next_branch)
 620                         if (b->pack_id == id)
 621                                 b->pack_id = MAX_PACK_ID;
 622         }
 623
 624         for (t = first_tag; t; t = t->next_tag)
 625                 if (t->pack_id == id)
 626                         t->pack_id = MAX_PACK_ID;
 627 }
 628
 629 static unsigned int hc_str(const char *s, size_t len)
 630 {
 631         unsigned int r = 0;
 632         while (len-- > 0)
 633                 r = r * 31 + *s++;
 634         return r;
 635 }
 636
 637 static void *pool_alloc(size_t len)
 638 {
 639         struct mem_pool *p;
 640         void *r;
 641
 642         /* round up to a 'uintmax_t' alignment */
 643         if (len & (sizeof(uintmax_t) - 1))
 644                 len += sizeof(uintmax_t) - (len & (sizeof(uintmax_t) - 1));
 645
 646         for (p = mem_pool; p; p = p->next_pool)
 647                 if ((p->end - p->next_free >= len))
 648                         break;
 649
 650         if (!p) {
 651                 if (len >= (mem_pool_alloc/2)) {
 652                         total_allocd += len;
 653                         return xmalloc(len);
 654                 }
 655                 total_allocd += sizeof(struct mem_pool) + mem_pool_alloc;
 656                 p = xmalloc(st_add(sizeof(struct mem_pool), mem_pool_alloc));
 657                 p->next_pool = mem_pool;
 658                 p->next_free = (char *) p->space;
 659                 p->end = p->next_free + mem_pool_alloc;
 660                 mem_pool = p;
 661         }
 662
 663         r = p->next_free;
 664         p->next_free += len;
 665         return r;
 666 }
 667
 668 static void *pool_calloc(size_t count, size_t size)
 669 {
 670         size_t len = count * size;
 671         void *r = pool_alloc(len);
 672         memset(r, 0, len);
 673         return r;
 674 }
 675
 676 static char *pool_strdup(const char *s)
 677 {
 678         size_t len = strlen(s) + 1;
 679         char *r = pool_alloc(len);
 680         memcpy(r, s, len);
 681         return r;
 682 }
 683
 684 static void insert_mark(uintmax_t idnum, struct object_entry *oe)
 685 {
 686         struct mark_set *s = marks;
 687         while ((idnum >> s->shift) >= 1024) {
 688                 s = pool_calloc(1, sizeof(struct mark_set));
 689                 s->shift = marks->shift + 10;
 690                 s->data.sets[0] = marks;
 691                 marks = s;
 692         }
 693         while (s->shift) {
 694                 uintmax_t i = idnum >> s->shift;
 695                 idnum -= i << s->shift;
 696                 if (!s->data.sets[i]) {
 697                         s->data.sets[i] = pool_calloc(1, sizeof(struct mark_set));
 698                         s->data.sets[i]->shift = s->shift - 10;
 699                 }
 700                 s = s->data.sets[i];
 701         }
 702         if (!s->data.marked[idnum])
 703                 marks_set_count++;
 704         s->data.marked[idnum] = oe;
 705 }
 706
 707 static struct object_entry *find_mark(uintmax_t idnum)
 708 {
 709         uintmax_t orig_idnum = idnum;
 710         struct mark_set *s = marks;
 711         struct object_entry *oe = NULL;
 712         if ((idnum >> s->shift) < 1024) {
 713                 while (s && s->shift) {
 714                         uintmax_t i = idnum >> s->shift;
 715                         idnum -= i << s->shift;
 716                         s = s->data.sets[i];
 717                 }
 718                 if (s)
 719                         oe = s->data.marked[idnum];
 720         }
 721         if (!oe)
 722                 die("mark :%" PRIuMAX " not declared", orig_idnum);
 723         return oe;
 724 }
 725
 726 static struct atom_str *to_atom(const char *s, unsigned short len)
 727 {
 728         unsigned int hc = hc_str(s, len) % atom_table_sz;
 729         struct atom_str *c;
 730
 731         for (c = atom_table[hc]; c; c = c->next_atom)
 732                 if (c->str_len == len && !strncmp(s, c->str_dat, len))
 733                         return c;
 734
 735         c = pool_alloc(sizeof(struct atom_str) + len + 1);
 736         c->str_len = len;
 737         memcpy(c->str_dat, s, len);
 738         c->str_dat[len] = 0;
 739         c->next_atom = atom_table[hc];
 740         atom_table[hc] = c;
 741         atom_cnt++;
 742         return c;
 743 }
 744
 745 static struct branch *lookup_branch(const char *name)
 746 {
 747         unsigned int hc = hc_str(name, strlen(name)) % branch_table_sz;
 748         struct branch *b;
 749
 750         for (b = branch_table[hc]; b; b = b->table_next_branch)
 751                 if (!strcmp(name, b->name))
 752                         return b;
 753         return NULL;
 754 }
 755
 756 static struct branch *new_branch(const char *name)
 757 {
 758         unsigned int hc = hc_str(name, strlen(name)) % branch_table_sz;
 759         struct branch *b = lookup_branch(name);
 760
 761         if (b)
 762                 die("Invalid attempt to create duplicate branch: %s", name);
 763         if (check_refname_format(name, REFNAME_ALLOW_ONELEVEL))
 764                 die("Branch name doesn't conform to GIT standards: %s", name);
 765
 766         b = pool_calloc(1, sizeof(struct branch));
 767         b->name = pool_strdup(name);
 768         b->table_next_branch = branch_table[hc];
 769         b->branch_tree.versions[0].mode = S_IFDIR;
 770         b->branch_tree.versions[1].mode = S_IFDIR;
 771         b->num_notes = 0;
 772         b->active = 0;
 773         b->pack_id = MAX_PACK_ID;
 774         branch_table[hc] = b;
 775         branch_count++;
 776         return b;
 777 }
 778
 779 static unsigned int hc_entries(unsigned int cnt)
 780 {
 781         cnt = cnt & 7 ? (cnt / 8) + 1 : cnt / 8;
 782         return cnt < avail_tree_table_sz ? cnt : avail_tree_table_sz - 1;
 783 }
 784
 785 static struct tree_content *new_tree_content(unsigned int cnt)
 786 {
 787         struct avail_tree_content *f, *l = NULL;
 788         struct tree_content *t;
 789         unsigned int hc = hc_entries(cnt);
 790
 791         for (f = avail_tree_table[hc]; f; l = f, f = f->next_avail)
 792                 if (f->entry_capacity >= cnt)
 793                         break;
 794
 795         if (f) {
 796                 if (l)
 797                         l->next_avail = f->next_avail;
 798                 else
 799                         avail_tree_table[hc] = f->next_avail;
 800         } else {
 801                 cnt = cnt & 7 ? ((cnt / 8) + 1) * 8 : cnt;
 802                 f = pool_alloc(sizeof(*t) + sizeof(t->entries[0]) * cnt);
 803                 f->entry_capacity = cnt;
 804         }
 805
 806         t = (struct tree_content*)f;
 807         t->entry_count = 0;
 808         t->delta_depth = 0;
 809         return t;
 810 }
 811
 812 static void release_tree_entry(struct tree_entry *e);
 813 static void release_tree_content(struct tree_content *t)
 814 {
 815         struct avail_tree_content *f = (struct avail_tree_content*)t;
 816         unsigned int hc = hc_entries(f->entry_capacity);
 817         f->next_avail = avail_tree_table[hc];
 818         avail_tree_table[hc] = f;
 819 }
 820
 821 static void release_tree_content_recursive(struct tree_content *t)
 822 {
 823         unsigned int i;
 824         for (i = 0; i < t->entry_count; i++)
 825                 release_tree_entry(t->entries[i]);
 826         release_tree_content(t);
 827 }
 828
 829 static struct tree_content *grow_tree_content(
 830         struct tree_content *t,
 831         int amt)
 832 {
 833         struct tree_content *r = new_tree_content(t->entry_count + amt);
 834         r->entry_count = t->entry_count;
 835         r->delta_depth = t->delta_depth;
 836         memcpy(r->entries,t->entries,t->entry_count*sizeof(t->entries[0]));
 837         release_tree_content(t);
 838         return r;
 839 }
 840
 841 static struct tree_entry *new_tree_entry(void)
 842 {
 843         struct tree_entry *e;
 844
 845         if (!avail_tree_entry) {
 846                 unsigned int n = tree_entry_alloc;
 847                 total_allocd += n * sizeof(struct tree_entry);
 848                 ALLOC_ARRAY(e, n);
 849                 avail_tree_entry = e;
 850                 while (n-- > 1) {
 851                         *((void**)e) = e + 1;
 852                         e++;
 853                 }
 854                 *((void**)e) = NULL;
 855         }
 856
 857         e = avail_tree_entry;
 858         avail_tree_entry = *((void**)e);
 859         return e;
 860 }
 861
 862 static void release_tree_entry(struct tree_entry *e)
 863 {
 864         if (e->tree)
 865                 release_tree_content_recursive(e->tree);
 866         *((void**)e) = avail_tree_entry;
 867         avail_tree_entry = e;
 868 }
 869
 870 static struct tree_content *dup_tree_content(struct tree_content *s)
 871 {
 872         struct tree_content *d;
 873         struct tree_entry *a, *b;
 874         unsigned int i;
 875
 876         if (!s)
 877                 return NULL;
 878         d = new_tree_content(s->entry_count);
 879         for (i = 0; i < s->entry_count; i++) {
 880                 a = s->entries[i];
 881                 b = new_tree_entry();
 882                 memcpy(b, a, sizeof(*a));
 883                 if (a->tree && is_null_oid(&b->versions[1].oid))
 884                         b->tree = dup_tree_content(a->tree);
 885                 else
 886                         b->tree = NULL;
 887                 d->entries[i] = b;
 888         }
 889         d->entry_count = s->entry_count;
 890         d->delta_depth = s->delta_depth;
 891
 892         return d;
 893 }
 894
 895 static void start_packfile(void)
 896 {
 897         struct strbuf tmp_file = STRBUF_INIT;
 898         struct packed_git *p;
 899         struct pack_header hdr;
 900         int pack_fd;
 901
 902         pack_fd = odb_mkstemp(&tmp_file, "pack/tmp_pack_XXXXXX");
 903         FLEX_ALLOC_STR(p, pack_name, tmp_file.buf);
 904         strbuf_release(&tmp_file);
 905
 906         p->pack_fd = pack_fd;
 907         p->do_not_close = 1;
 908         pack_file = sha1fd(pack_fd, p->pack_name);
 909
 910         hdr.hdr_signature = htonl(PACK_SIGNATURE);
 911         hdr.hdr_version = htonl(2);
 912         hdr.hdr_entries = 0;
 913         sha1write(pack_file, &hdr, sizeof(hdr));
 914
 915         pack_data = p;
 916         pack_size = sizeof(hdr);
 917         object_count = 0;
 918
 919         REALLOC_ARRAY(all_packs, pack_id + 1);
 920         all_packs[pack_id] = p;
 921 }
 922
 923 static const char *create_index(void)
 924 {
 925         const char *tmpfile;
 926         struct pack_idx_entry **idx, **c, **last;
 927         struct object_entry *e;
 928         struct object_entry_pool *o;
 929
 930         /* Build the table of object IDs. */
 931         ALLOC_ARRAY(idx, object_count);
 932         c = idx;
 933         for (o = blocks; o; o = o->next_pool)
 934                 for (e = o->next_free; e-- != o->entries;)
 935                         if (pack_id == e->pack_id)
 936                                 *c++ = &e->idx;
 937         last = idx + object_count;
 938         if (c != last)
 939                 die("internal consistency error creating the index");
 940
 941         tmpfile = write_idx_file(NULL, idx, object_count, &pack_idx_opts, pack_data->sha1);
 942         free(idx);
 943         return tmpfile;
 944 }
 945
 946 static char *keep_pack(const char *curr_index_name)
 947 {
 948         static const char *keep_msg = "fast-import";
 949         struct strbuf name = STRBUF_INIT;
 950         int keep_fd;
 951
 952         odb_pack_name(&name, pack_data->sha1, "keep");
 953         keep_fd = odb_pack_keep(name.buf);
 954         if (keep_fd < 0)
 955                 die_errno("cannot create keep file");
 956         write_or_die(keep_fd, keep_msg, strlen(keep_msg));
 957         if (close(keep_fd))
 958                 die_errno("failed to write keep file");
 959
 960         odb_pack_name(&name, pack_data->sha1, "pack");
 961         if (finalize_object_file(pack_data->pack_name, name.buf))
 962                 die("cannot store pack file");
 963
 964         odb_pack_name(&name, pack_data->sha1, "idx");
 965         if (finalize_object_file(curr_index_name, name.buf))
 966                 die("cannot store index file");
 967         free((void *)curr_index_name);
 968         return strbuf_detach(&name, NULL);
 969 }
 970
 971 static void unkeep_all_packs(void)
 972 {
 973         struct strbuf name = STRBUF_INIT;
 974         int k;
 975
 976         for (k = 0; k < pack_id; k++) {
 977                 struct packed_git *p = all_packs[k];
 978                 odb_pack_name(&name, p->sha1, "keep");
 979                 unlink_or_warn(name.buf);
 980         }
 981         strbuf_release(&name);
 982 }
 983
 984 static int loosen_small_pack(const struct packed_git *p)
 985 {
 986         struct child_process unpack = CHILD_PROCESS_INIT;
 987
 988         if (lseek(p->pack_fd, 0, SEEK_SET) < 0)
 989                 die_errno("Failed seeking to start of '%s'", p->pack_name);
 990
 991         unpack.in = p->pack_fd;
 992         unpack.git_cmd = 1;
 993         unpack.stdout_to_stderr = 1;
 994         argv_array_push(&unpack.args, "unpack-objects");
 995         if (!show_stats)
 996                 argv_array_push(&unpack.args, "-q");
 997
 998         return run_command(&unpack);
 999 }
1000
1001 static void end_packfile(void)
1002 {
1003         static int running;
1004
1005         if (running || !pack_data)
1006                 return;
1007
1008         running = 1;
1009         clear_delta_base_cache();
1010         if (object_count) {
1011                 struct packed_git *new_p;
1012                 struct object_id cur_pack_oid;
1013                 char *idx_name;
1014                 int i;
1015                 struct branch *b;
1016                 struct tag *t;
1017
1018                 close_pack_windows(pack_data);
1019                 sha1close(pack_file, cur_pack_oid.hash, 0);
1020                 fixup_pack_header_footer(pack_data->pack_fd, pack_data->sha1,
1021                                     pack_data->pack_name, object_count,
1022                                     cur_pack_oid.hash, pack_size);
1023
1024                 if (object_count <= unpack_limit) {
1025                         if (!loosen_small_pack(pack_data)) {
1026                                 invalidate_pack_id(pack_id);
1027                                 goto discard_pack;
1028                         }
1029                 }
1030
1031                 close(pack_data->pack_fd);
1032                 idx_name = keep_pack(create_index());
1033
1034                 /* Register the packfile with core git's machinery. */
1035                 new_p = add_packed_git(idx_name, strlen(idx_name), 1);
1036                 if (!new_p)
1037                         die("core git rejected index %s", idx_name);
1038                 all_packs[pack_id] = new_p;
1039                 install_packed_git(new_p);
1040                 free(idx_name);
1041
1042                 /* Print the boundary */
1043                 if (pack_edges) {
1044                         fprintf(pack_edges, "%s:", new_p->pack_name);
1045                         for (i = 0; i < branch_table_sz; i++) {
1046                                 for (b = branch_table[i]; b; b = b->table_next_branch) {
1047                                         if (b->pack_id == pack_id)
1048                                                 fprintf(pack_edges, " %s",
1049                                                         oid_to_hex(&b->oid));
1050                                 }
1051                         }
1052                         for (t = first_tag; t; t = t->next_tag) {
1053                                 if (t->pack_id == pack_id)
1054                                         fprintf(pack_edges, " %s",
1055                                                 oid_to_hex(&t->oid));
1056                         }
1057                         fputc('\n', pack_edges);
1058                         fflush(pack_edges);
1059                 }
1060
1061                 pack_id++;
1062         }
1063         else {
1064 discard_pack:
1065                 close(pack_data->pack_fd);
1066                 unlink_or_warn(pack_data->pack_name);
1067         }
1068         FREE_AND_NULL(pack_data);
1069         running = 0;
1070
1071         /* We can't carry a delta across packfiles. */
1072         strbuf_release(&last_blob.data);
1073         last_blob.offset = 0;
1074         last_blob.depth = 0;
1075 }
1076
1077 static void cycle_packfile(void)
1078 {
1079         end_packfile();
1080         start_packfile();
1081 }
1082
1083 static int store_object(
1084         enum object_type type,
1085         struct strbuf *dat,
1086         struct last_object *last,
1087         struct object_id *oidout,
1088         uintmax_t mark)
1089 {
1090         void *out, *delta;
1091         struct object_entry *e;
1092         unsigned char hdr[96];
1093         struct object_id oid;
1094         unsigned long hdrlen, deltalen;
1095         git_SHA_CTX c;
1096         git_zstream s;
1097
1098         hdrlen = xsnprintf((char *)hdr, sizeof(hdr), "%s %lu",
1099                            typename(type), (unsigned long)dat->len) + 1;
1100         git_SHA1_Init(&c);
1101         git_SHA1_Update(&c, hdr, hdrlen);
1102         git_SHA1_Update(&c, dat->buf, dat->len);
1103         git_SHA1_Final(oid.hash, &c);
1104         if (oidout)
1105                 oidcpy(oidout, &oid);
1106
1107         e = insert_object(&oid);
1108         if (mark)
1109                 insert_mark(mark, e);
1110         if (e->idx.offset) {
1111                 duplicate_count_by_type[type]++;
1112                 return 1;
1113         } else if (find_sha1_pack(oid.hash, packed_git)) {
1114                 e->type = type;
1115                 e->pack_id = MAX_PACK_ID;
1116                 e->idx.offset = 1; /* just not zero! */
1117                 duplicate_count_by_type[type]++;
1118                 return 1;
1119         }
1120
1121         if (last && last->data.buf && last->depth < max_depth && dat->len > 20) {
1122                 delta_count_attempts_by_type[type]++;
1123                 delta = diff_delta(last->data.buf, last->data.len,
1124                         dat->buf, dat->len,
1125                         &deltalen, dat->len - 20);
1126         } else
1127                 delta = NULL;
1128
1129         git_deflate_init(&s, pack_compression_level);
1130         if (delta) {
1131                 s.next_in = delta;
1132                 s.avail_in = deltalen;
1133         } else {
1134                 s.next_in = (void *)dat->buf;
1135                 s.avail_in = dat->len;
1136         }
1137         s.avail_out = git_deflate_bound(&s, s.avail_in);
1138         s.next_out = out = xmalloc(s.avail_out);
1139         while (git_deflate(&s, Z_FINISH) == Z_OK)
1140                 ; /* nothing */
1141         git_deflate_end(&s);
1142
1143         /* Determine if we should auto-checkpoint. */
1144         if ((max_packsize && (pack_size + 60 + s.total_out) > max_packsize)
1145                 || (pack_size + 60 + s.total_out) < pack_size) {
1146
1147                 /* This new object needs to *not* have the current pack_id. */
1148                 e->pack_id = pack_id + 1;
1149                 cycle_packfile();
1150
1151                 /* We cannot carry a delta into the new pack. */
1152                 if (delta) {
1153                         FREE_AND_NULL(delta);
1154
1155                         git_deflate_init(&s, pack_compression_level);
1156                         s.next_in = (void *)dat->buf;
1157                         s.avail_in = dat->len;
1158                         s.avail_out = git_deflate_bound(&s, s.avail_in);
1159                         s.next_out = out = xrealloc(out, s.avail_out);
1160                         while (git_deflate(&s, Z_FINISH) == Z_OK)
1161                                 ; /* nothing */
1162                         git_deflate_end(&s);
1163                 }
1164         }
1165
1166         e->type = type;
1167         e->pack_id = pack_id;
1168         e->idx.offset = pack_size;
1169         object_count++;
1170         object_count_by_type[type]++;
1171
1172         crc32_begin(pack_file);
1173
1174         if (delta) {
1175                 off_t ofs = e->idx.offset - last->offset;
1176                 unsigned pos = sizeof(hdr) - 1;
1177
1178                 delta_count_by_type[type]++;
1179                 e->depth = last->depth + 1;
1180
1181                 hdrlen = encode_in_pack_object_header(hdr, sizeof(hdr),
1182                                                       OBJ_OFS_DELTA, deltalen);
1183                 sha1write(pack_file, hdr, hdrlen);
1184                 pack_size += hdrlen;
1185
1186                 hdr[pos] = ofs & 127;
1187                 while (ofs >>= 7)
1188                         hdr[--pos] = 128 | (--ofs & 127);
1189                 sha1write(pack_file, hdr + pos, sizeof(hdr) - pos);
1190                 pack_size += sizeof(hdr) - pos;
1191         } else {
1192                 e->depth = 0;
1193                 hdrlen = encode_in_pack_object_header(hdr, sizeof(hdr),
1194                                                       type, dat->len);
1195                 sha1write(pack_file, hdr, hdrlen);
1196                 pack_size += hdrlen;
1197         }
1198
1199         sha1write(pack_file, out, s.total_out);
1200         pack_size += s.total_out;
1201
1202         e->idx.crc32 = crc32_end(pack_file);
1203
1204         free(out);
1205         free(delta);
1206         if (last) {
1207                 if (last->no_swap) {
1208                         last->data = *dat;
1209                 } else {
1210                         strbuf_swap(&last->data, dat);
1211                 }
1212                 last->offset = e->idx.offset;
1213                 last->depth = e->depth;
1214         }
1215         return 0;
1216 }
1217
1218 static void truncate_pack(struct sha1file_checkpoint *checkpoint)
1219 {
1220         if (sha1file_truncate(pack_file, checkpoint))
1221                 die_errno("cannot truncate pack to skip duplicate");
1222         pack_size = checkpoint->offset;
1223 }
1224
1225 static void stream_blob(uintmax_t len, struct object_id *oidout, uintmax_t mark)
1226 {
1227         size_t in_sz = 64 * 1024, out_sz = 64 * 1024;
1228         unsigned char *in_buf = xmalloc(in_sz);
1229         unsigned char *out_buf = xmalloc(out_sz);
1230         struct object_entry *e;
1231         struct object_id oid;
1232         unsigned long hdrlen;
1233         off_t offset;
1234         git_SHA_CTX c;
1235         git_zstream s;
1236         struct sha1file_checkpoint checkpoint;
1237         int status = Z_OK;
1238
1239         /* Determine if we should auto-checkpoint. */
1240         if ((max_packsize && (pack_size + 60 + len) > max_packsize)
1241                 || (pack_size + 60 + len) < pack_size)
1242                 cycle_packfile();
1243
1244         sha1file_checkpoint(pack_file, &checkpoint);
1245         offset = checkpoint.offset;
1246
1247         hdrlen = xsnprintf((char *)out_buf, out_sz, "blob %" PRIuMAX, len) + 1;
1248
1249         git_SHA1_Init(&c);
1250         git_SHA1_Update(&c, out_buf, hdrlen);
1251
1252         crc32_begin(pack_file);
1253
1254         git_deflate_init(&s, pack_compression_level);
1255
1256         hdrlen = encode_in_pack_object_header(out_buf, out_sz, OBJ_BLOB, len);
1257
1258         s.next_out = out_buf + hdrlen;
1259         s.avail_out = out_sz - hdrlen;
1260
1261         while (status != Z_STREAM_END) {
1262                 if (0 < len && !s.avail_in) {
1263                         size_t cnt = in_sz < len ? in_sz : (size_t)len;
1264                         size_t n = fread(in_buf, 1, cnt, stdin);
1265                         if (!n && feof(stdin))
1266                                 die("EOF in data (%" PRIuMAX " bytes remaining)", len);
1267
1268                         git_SHA1_Update(&c, in_buf, n);
1269                         s.next_in = in_buf;
1270                         s.avail_in = n;
1271                         len -= n;
1272                 }
1273
1274                 status = git_deflate(&s, len ? 0 : Z_FINISH);
1275
1276                 if (!s.avail_out || status == Z_STREAM_END) {
1277                         size_t n = s.next_out - out_buf;
1278                         sha1write(pack_file, out_buf, n);
1279                         pack_size += n;
1280                         s.next_out = out_buf;
1281                         s.avail_out = out_sz;
1282                 }
1283
1284                 switch (status) {
1285                 case Z_OK:
1286                 case Z_BUF_ERROR:
1287                 case Z_STREAM_END:
1288                         continue;
1289                 default:
1290                         die("unexpected deflate failure: %d", status);
1291                 }
1292         }
1293         git_deflate_end(&s);
1294         git_SHA1_Final(oid.hash, &c);
1295
1296         if (oidout)
1297                 oidcpy(oidout, &oid);
1298
1299         e = insert_object(&oid);
1300
1301         if (mark)
1302                 insert_mark(mark, e);
1303
1304         if (e->idx.offset) {
1305                 duplicate_count_by_type[OBJ_BLOB]++;
1306                 truncate_pack(&checkpoint);
1307
1308         } else if (find_sha1_pack(oid.hash, packed_git)) {
1309                 e->type = OBJ_BLOB;
1310                 e->pack_id = MAX_PACK_ID;
1311                 e->idx.offset = 1; /* just not zero! */
1312                 duplicate_count_by_type[OBJ_BLOB]++;
1313                 truncate_pack(&checkpoint);
1314
1315         } else {
1316                 e->depth = 0;
1317                 e->type = OBJ_BLOB;
1318                 e->pack_id = pack_id;
1319                 e->idx.offset = offset;
1320                 e->idx.crc32 = crc32_end(pack_file);
1321                 object_count++;
1322                 object_count_by_type[OBJ_BLOB]++;
1323         }
1324
1325         free(in_buf);
1326         free(out_buf);
1327 }
1328
1329 /* All calls must be guarded by find_object() or find_mark() to
1330  * ensure the 'struct object_entry' passed was written by this
1331  * process instance.  We unpack the entry by the offset, avoiding
1332  * the need for the corresponding .idx file.  This unpacking rule
1333  * works because we only use OBJ_REF_DELTA within the packfiles
1334  * created by fast-import.
1335  *
1336  * oe must not be NULL.  Such an oe usually comes from giving
1337  * an unknown SHA-1 to find_object() or an undefined mark to
1338  * find_mark().  Callers must test for this condition and use
1339  * the standard read_sha1_file() when it happens.
1340  *
1341  * oe->pack_id must not be MAX_PACK_ID.  Such an oe is usually from
1342  * find_mark(), where the mark was reloaded from an existing marks
1343  * file and is referencing an object that this fast-import process
1344  * instance did not write out to a packfile.  Callers must test for
1345  * this condition and use read_sha1_file() instead.
1346  */
1347 static void *gfi_unpack_entry(
1348         struct object_entry *oe,
1349         unsigned long *sizep)
1350 {
1351         enum object_type type;
1352         struct packed_git *p = all_packs[oe->pack_id];
1353         if (p == pack_data && p->pack_size < (pack_size + 20)) {
1354                 /* The object is stored in the packfile we are writing to
1355                  * and we have modified it since the last time we scanned
1356                  * back to read a previously written object.  If an old
1357                  * window covered [p->pack_size, p->pack_size + 20) its
1358                  * data is stale and is not valid.  Closing all windows
1359                  * and updating the packfile length ensures we can read
1360                  * the newly written data.
1361                  */
1362                 close_pack_windows(p);
1363                 sha1flush(pack_file);
1364
1365                 /* We have to offer 20 bytes additional on the end of
1366                  * the packfile as the core unpacker code assumes the
1367                  * footer is present at the file end and must promise
1368                  * at least 20 bytes within any window it maps.  But
1369                  * we don't actually create the footer here.
1370                  */
1371                 p->pack_size = pack_size + 20;
1372         }
1373         return unpack_entry(p, oe->idx.offset, &type, sizep);
1374 }
1375
1376 static const char *get_mode(const char *str, uint16_t *modep)
1377 {
1378         unsigned char c;
1379         uint16_t mode = 0;
1380
1381         while ((c = *str++) != ' ') {
1382                 if (c < '0' || c > '7')
1383                         return NULL;
1384                 mode = (mode << 3) + (c - '0');
1385         }
1386         *modep = mode;
1387         return str;
1388 }
1389
1390 static void load_tree(struct tree_entry *root)
1391 {
1392         struct object_id *oid = &root->versions[1].oid;
1393         struct object_entry *myoe;
1394         struct tree_content *t;
1395         unsigned long size;
1396         char *buf;
1397         const char *c;
1398
1399         root->tree = t = new_tree_content(8);
1400         if (is_null_oid(oid))
1401                 return;
1402
1403         myoe = find_object(oid);
1404         if (myoe && myoe->pack_id != MAX_PACK_ID) {
1405                 if (myoe->type != OBJ_TREE)
1406                         die("Not a tree: %s", oid_to_hex(oid));
1407                 t->delta_depth = myoe->depth;
1408                 buf = gfi_unpack_entry(myoe, &size);
1409                 if (!buf)
1410                         die("Can't load tree %s", oid_to_hex(oid));
1411         } else {
1412                 enum object_type type;
1413                 buf = read_sha1_file(oid->hash, &type, &size);
1414                 if (!buf || type != OBJ_TREE)
1415                         die("Can't load tree %s", oid_to_hex(oid));
1416         }
1417
1418         c = buf;
1419         while (c != (buf + size)) {
1420                 struct tree_entry *e = new_tree_entry();
1421
1422                 if (t->entry_count == t->entry_capacity)
1423                         root->tree = t = grow_tree_content(t, t->entry_count);
1424                 t->entries[t->entry_count++] = e;
1425
1426                 e->tree = NULL;
1427                 c = get_mode(c, &e->versions[1].mode);
1428                 if (!c)
1429                         die("Corrupt mode in %s", oid_to_hex(oid));
1430                 e->versions[0].mode = e->versions[1].mode;
1431                 e->name = to_atom(c, strlen(c));
1432                 c += e->name->str_len + 1;
1433                 hashcpy(e->versions[0].oid.hash, (unsigned char *)c);
1434                 hashcpy(e->versions[1].oid.hash, (unsigned char *)c);
1435                 c += GIT_SHA1_RAWSZ;
1436         }
1437         free(buf);
1438 }
1439
1440 static int tecmp0 (const void *_a, const void *_b)
1441 {
1442         struct tree_entry *a = *((struct tree_entry**)_a);
1443         struct tree_entry *b = *((struct tree_entry**)_b);
1444         return base_name_compare(
1445                 a->name->str_dat, a->name->str_len, a->versions[0].mode,
1446                 b->name->str_dat, b->name->str_len, b->versions[0].mode);
1447 }
1448
1449 static int tecmp1 (const void *_a, const void *_b)
1450 {
1451         struct tree_entry *a = *((struct tree_entry**)_a);
1452         struct tree_entry *b = *((struct tree_entry**)_b);
1453         return base_name_compare(
1454                 a->name->str_dat, a->name->str_len, a->versions[1].mode,
1455                 b->name->str_dat, b->name->str_len, b->versions[1].mode);
1456 }
1457
1458 static void mktree(struct tree_content *t, int v, struct strbuf *b)
1459 {
1460         size_t maxlen = 0;
1461         unsigned int i;
1462
1463         if (!v)
1464                 QSORT(t->entries, t->entry_count, tecmp0);
1465         else
1466                 QSORT(t->entries, t->entry_count, tecmp1);
1467
1468         for (i = 0; i < t->entry_count; i++) {
1469                 if (t->entries[i]->versions[v].mode)
1470                         maxlen += t->entries[i]->name->str_len + 34;
1471         }
1472
1473         strbuf_reset(b);
1474         strbuf_grow(b, maxlen);
1475         for (i = 0; i < t->entry_count; i++) {
1476                 struct tree_entry *e = t->entries[i];
1477                 if (!e->versions[v].mode)
1478                         continue;
1479                 strbuf_addf(b, "%o %s%c",
1480                         (unsigned int)(e->versions[v].mode & ~NO_DELTA),
1481                         e->name->str_dat, '\0');
1482                 strbuf_add(b, e->versions[v].oid.hash, GIT_SHA1_RAWSZ);
1483         }
1484 }
1485
1486 static void store_tree(struct tree_entry *root)
1487 {
1488         struct tree_content *t;
1489         unsigned int i, j, del;
1490         struct last_object lo = { STRBUF_INIT, 0, 0, /* no_swap */ 1 };
1491         struct object_entry *le = NULL;
1492
1493         if (!is_null_oid(&root->versions[1].oid))
1494                 return;
1495
1496         if (!root->tree)
1497                 load_tree(root);
1498         t = root->tree;
1499
1500         for (i = 0; i < t->entry_count; i++) {
1501                 if (t->entries[i]->tree)
1502                         store_tree(t->entries[i]);
1503         }
1504
1505         if (!(root->versions[0].mode & NO_DELTA))
1506                 le = find_object(&root->versions[0].oid);
1507         if (S_ISDIR(root->versions[0].mode) && le && le->pack_id == pack_id) {
1508                 mktree(t, 0, &old_tree);
1509                 lo.data = old_tree;
1510                 lo.offset = le->idx.offset;
1511                 lo.depth = t->delta_depth;
1512         }
1513
1514         mktree(t, 1, &new_tree);
1515         store_object(OBJ_TREE, &new_tree, &lo, &root->versions[1].oid, 0);
1516
1517         t->delta_depth = lo.depth;
1518         for (i = 0, j = 0, del = 0; i < t->entry_count; i++) {
1519                 struct tree_entry *e = t->entries[i];
1520                 if (e->versions[1].mode) {
1521                         e->versions[0].mode = e->versions[1].mode;
1522                         oidcpy(&e->versions[0].oid, &e->versions[1].oid);
1523                         t->entries[j++] = e;
1524                 } else {
1525                         release_tree_entry(e);
1526                         del++;
1527                 }
1528         }
1529         t->entry_count -= del;
1530 }
1531
1532 static void tree_content_replace(
1533         struct tree_entry *root,
1534         const struct object_id *oid,
1535         const uint16_t mode,
1536         struct tree_content *newtree)
1537 {
1538         if (!S_ISDIR(mode))
1539                 die("Root cannot be a non-directory");
1540         oidclr(&root->versions[0].oid);
1541         oidcpy(&root->versions[1].oid, oid);
1542         if (root->tree)
1543                 release_tree_content_recursive(root->tree);
1544         root->tree = newtree;
1545 }
1546
1547 static int tree_content_set(
1548         struct tree_entry *root,
1549         const char *p,
1550         const struct object_id *oid,
1551         const uint16_t mode,
1552         struct tree_content *subtree)
1553 {
1554         struct tree_content *t;
1555         const char *slash1;
1556         unsigned int i, n;
1557         struct tree_entry *e;
1558
1559         slash1 = strchrnul(p, '/');
1560         n = slash1 - p;
1561         if (!n)
1562                 die("Empty path component found in input");
1563         if (!*slash1 && !S_ISDIR(mode) && subtree)
1564                 die("Non-directories cannot have subtrees");
1565
1566         if (!root->tree)
1567                 load_tree(root);
1568         t = root->tree;
1569         for (i = 0; i < t->entry_count; i++) {
1570                 e = t->entries[i];
1571                 if (e->name->str_len == n && !fspathncmp(p, e->name->str_dat, n)) {
1572                         if (!*slash1) {
1573                                 if (!S_ISDIR(mode)
1574                                                 && e->versions[1].mode == mode
1575                                                 && !oidcmp(&e->versions[1].oid, oid))
1576                                         return 0;
1577                                 e->versions[1].mode = mode;
1578                                 oidcpy(&e->versions[1].oid, oid);
1579                                 if (e->tree)
1580                                         release_tree_content_recursive(e->tree);
1581                                 e->tree = subtree;
1582
1583                                 /*
1584                                  * We need to leave e->versions[0].sha1 alone
1585                                  * to avoid modifying the preimage tree used
1586                                  * when writing out the parent directory.
1587                                  * But after replacing the subdir with a
1588                                  * completely different one, it's not a good
1589                                  * delta base any more, and besides, we've
1590                                  * thrown away the tree entries needed to
1591                                  * make a delta against it.
1592                                  *
1593                                  * So let's just explicitly disable deltas
1594                                  * for the subtree.
1595                                  */
1596                                 if (S_ISDIR(e->versions[0].mode))
1597                                         e->versions[0].mode |= NO_DELTA;
1598
1599                                 oidclr(&root->versions[1].oid);
1600                                 return 1;
1601                         }
1602                         if (!S_ISDIR(e->versions[1].mode)) {
1603                                 e->tree = new_tree_content(8);
1604                                 e->versions[1].mode = S_IFDIR;
1605                         }
1606                         if (!e->tree)
1607                                 load_tree(e);
1608                         if (tree_content_set(e, slash1 + 1, oid, mode, subtree)) {
1609                                 oidclr(&root->versions[1].oid);
1610                                 return 1;
1611                         }
1612                         return 0;
1613                 }
1614         }
1615
1616         if (t->entry_count == t->entry_capacity)
1617                 root->tree = t = grow_tree_content(t, t->entry_count);
1618         e = new_tree_entry();
1619         e->name = to_atom(p, n);
1620         e->versions[0].mode = 0;
1621         oidclr(&e->versions[0].oid);
1622         t->entries[t->entry_count++] = e;
1623         if (*slash1) {
1624                 e->tree = new_tree_content(8);
1625                 e->versions[1].mode = S_IFDIR;
1626                 tree_content_set(e, slash1 + 1, oid, mode, subtree);
1627         } else {
1628                 e->tree = subtree;
1629                 e->versions[1].mode = mode;
1630                 oidcpy(&e->versions[1].oid, oid);
1631         }
1632         oidclr(&root->versions[1].oid);
1633         return 1;
1634 }
1635
1636 static int tree_content_remove(
1637         struct tree_entry *root,
1638         const char *p,
1639         struct tree_entry *backup_leaf,
1640         int allow_root)
1641 {
1642         struct tree_content *t;
1643         const char *slash1;
1644         unsigned int i, n;
1645         struct tree_entry *e;
1646
1647         slash1 = strchrnul(p, '/');
1648         n = slash1 - p;
1649
1650         if (!root->tree)
1651                 load_tree(root);
1652
1653         if (!*p && allow_root) {
1654                 e = root;
1655                 goto del_entry;
1656         }
1657
1658         t = root->tree;
1659         for (i = 0; i < t->entry_count; i++) {
1660                 e = t->entries[i];
1661                 if (e->name->str_len == n && !fspathncmp(p, e->name->str_dat, n)) {
1662                         if (*slash1 && !S_ISDIR(e->versions[1].mode))
1663                                 /*
1664                                  * If p names a file in some subdirectory, and a
1665                                  * file or symlink matching the name of the
1666                                  * parent directory of p exists, then p cannot
1667                                  * exist and need not be deleted.
1668                                  */
1669                                 return 1;
1670                         if (!*slash1 || !S_ISDIR(e->versions[1].mode))
1671                                 goto del_entry;
1672                         if (!e->tree)
1673                                 load_tree(e);
1674                         if (tree_content_remove(e, slash1 + 1, backup_leaf, 0)) {
1675                                 for (n = 0; n < e->tree->entry_count; n++) {
1676                                         if (e->tree->entries[n]->versions[1].mode) {
1677                                                 oidclr(&root->versions[1].oid);
1678                                                 return 1;
1679                                         }
1680                                 }
1681                                 backup_leaf = NULL;
1682                                 goto del_entry;
1683                         }
1684                         return 0;
1685                 }
1686         }
1687         return 0;
1688
1689 del_entry:
1690         if (backup_leaf)
1691                 memcpy(backup_leaf, e, sizeof(*backup_leaf));
1692         else if (e->tree)
1693                 release_tree_content_recursive(e->tree);
1694         e->tree = NULL;
1695         e->versions[1].mode = 0;
1696         oidclr(&e->versions[1].oid);
1697         oidclr(&root->versions[1].oid);
1698         return 1;
1699 }
1700
1701 static int tree_content_get(
1702         struct tree_entry *root,
1703         const char *p,
1704         struct tree_entry *leaf,
1705         int allow_root)
1706 {
1707         struct tree_content *t;
1708         const char *slash1;
1709         unsigned int i, n;
1710         struct tree_entry *e;
1711
1712         slash1 = strchrnul(p, '/');
1713         n = slash1 - p;
1714         if (!n && !allow_root)
1715                 die("Empty path component found in input");
1716
1717         if (!root->tree)
1718                 load_tree(root);
1719
1720         if (!n) {
1721                 e = root;
1722                 goto found_entry;
1723         }
1724
1725         t = root->tree;
1726         for (i = 0; i < t->entry_count; i++) {
1727                 e = t->entries[i];
1728                 if (e->name->str_len == n && !fspathncmp(p, e->name->str_dat, n)) {
1729                         if (!*slash1)
1730                                 goto found_entry;
1731                         if (!S_ISDIR(e->versions[1].mode))
1732                                 return 0;
1733                         if (!e->tree)
1734                                 load_tree(e);
1735                         return tree_content_get(e, slash1 + 1, leaf, 0);
1736                 }
1737         }
1738         return 0;
1739
1740 found_entry:
1741         memcpy(leaf, e, sizeof(*leaf));
1742         if (e->tree && is_null_oid(&e->versions[1].oid))
1743                 leaf->tree = dup_tree_content(e->tree);
1744         else
1745                 leaf->tree = NULL;
1746         return 1;
1747 }
1748
1749 static int update_branch(struct branch *b)
1750 {
1751         static const char *msg = "fast-import";
1752         struct ref_transaction *transaction;
1753         struct object_id old_oid;
1754         struct strbuf err = STRBUF_INIT;
1755
1756         if (is_null_oid(&b->oid)) {
1757                 if (b->delete)
1758                         delete_ref(NULL, b->name, NULL, 0);
1759                 return 0;
1760         }
1761         if (read_ref(b->name, old_oid.hash))
1762                 oidclr(&old_oid);
1763         if (!force_update && !is_null_oid(&old_oid)) {
1764                 struct commit *old_cmit, *new_cmit;
1765
1766                 old_cmit = lookup_commit_reference_gently(&old_oid, 0);
1767                 new_cmit = lookup_commit_reference_gently(&b->oid, 0);
1768                 if (!old_cmit || !new_cmit)
1769                         return error("Branch %s is missing commits.", b->name);
1770
1771                 if (!in_merge_bases(old_cmit, new_cmit)) {
1772                         warning("Not updating %s"
1773                                 " (new tip %s does not contain %s)",
1774                                 b->name, oid_to_hex(&b->oid),
1775                                 oid_to_hex(&old_oid));
1776                         return -1;
1777                 }
1778         }
1779         transaction = ref_transaction_begin(&err);
1780         if (!transaction ||
1781             ref_transaction_update(transaction, b->name, b->oid.hash, old_oid.hash,
1782                                    0, msg, &err) ||
1783             ref_transaction_commit(transaction, &err)) {
1784                 ref_transaction_free(transaction);
1785                 error("%s", err.buf);
1786                 strbuf_release(&err);
1787                 return -1;
1788         }
1789         ref_transaction_free(transaction);
1790         strbuf_release(&err);
1791         return 0;
1792 }
1793
1794 static void dump_branches(void)
1795 {
1796         unsigned int i;
1797         struct branch *b;
1798
1799         for (i = 0; i < branch_table_sz; i++) {
1800                 for (b = branch_table[i]; b; b = b->table_next_branch)
1801                         failure |= update_branch(b);
1802         }
1803 }
1804
1805 static void dump_tags(void)
1806 {
1807         static const char *msg = "fast-import";
1808         struct tag *t;
1809         struct strbuf ref_name = STRBUF_INIT;
1810         struct strbuf err = STRBUF_INIT;
1811         struct ref_transaction *transaction;
1812
1813         transaction = ref_transaction_begin(&err);
1814         if (!transaction) {
1815                 failure |= error("%s", err.buf);
1816                 goto cleanup;
1817         }
1818         for (t = first_tag; t; t = t->next_tag) {
1819                 strbuf_reset(&ref_name);
1820                 strbuf_addf(&ref_name, "refs/tags/%s", t->name);
1821
1822                 if (ref_transaction_update(transaction, ref_name.buf,
1823                                            t->oid.hash, NULL, 0, msg, &err)) {
1824                         failure |= error("%s", err.buf);
1825                         goto cleanup;
1826                 }
1827         }
1828         if (ref_transaction_commit(transaction, &err))
1829                 failure |= error("%s", err.buf);
1830
1831  cleanup:
1832         ref_transaction_free(transaction);
1833         strbuf_release(&ref_name);
1834         strbuf_release(&err);
1835 }
1836
1837 static void dump_marks_helper(FILE *f,
1838         uintmax_t base,
1839         struct mark_set *m)
1840 {
1841         uintmax_t k;
1842         if (m->shift) {
1843                 for (k = 0; k < 1024; k++) {
1844                         if (m->data.sets[k])
1845                                 dump_marks_helper(f, base + (k << m->shift),
1846                                         m->data.sets[k]);
1847                 }
1848         } else {
1849                 for (k = 0; k < 1024; k++) {
1850                         if (m->data.marked[k])
1851                                 fprintf(f, ":%" PRIuMAX " %s\n", base + k,
1852                                         oid_to_hex(&m->data.marked[k]->idx.oid));
1853                 }
1854         }
1855 }
1856
1857 static void dump_marks(void)
1858 {
1859         static struct lock_file mark_lock;
1860         FILE *f;
1861
1862         if (!export_marks_file || (import_marks_file && !import_marks_file_done))
1863                 return;
1864
1865         if (safe_create_leading_directories_const(export_marks_file)) {
1866                 failure |= error_errno("unable to create leading directories of %s",
1867                                        export_marks_file);
1868                 return;
1869         }
1870
1871         if (hold_lock_file_for_update(&mark_lock, export_marks_file, 0) < 0) {
1872                 failure |= error_errno("Unable to write marks file %s",
1873                                        export_marks_file);
1874                 return;
1875         }
1876
1877         f = fdopen_lock_file(&mark_lock, "w");
1878         if (!f) {
1879                 int saved_errno = errno;
1880                 rollback_lock_file(&mark_lock);
1881                 failure |= error("Unable to write marks file %s: %s",
1882                         export_marks_file, strerror(saved_errno));
1883                 return;
1884         }
1885
1886         dump_marks_helper(f, 0, marks);
1887         if (commit_lock_file(&mark_lock)) {
1888                 failure |= error_errno("Unable to write file %s",
1889                                        export_marks_file);
1890                 return;
1891         }
1892 }
1893
1894 static void read_marks(void)
1895 {
1896         char line[512];
1897         FILE *f = fopen(import_marks_file, "r");
1898         if (f)
1899                 ;
1900         else if (import_marks_file_ignore_missing && errno == ENOENT)
1901                 goto done; /* Marks file does not exist */
1902         else
1903                 die_errno("cannot read '%s'", import_marks_file);
1904         while (fgets(line, sizeof(line), f)) {
1905                 uintmax_t mark;
1906                 char *end;
1907                 struct object_id oid;
1908                 struct object_entry *e;
1909
1910                 end = strchr(line, '\n');
1911                 if (line[0] != ':' || !end)
1912                         die("corrupt mark line: %s", line);
1913                 *end = 0;
1914                 mark = strtoumax(line + 1, &end, 10);
1915                 if (!mark || end == line + 1
1916                         || *end != ' ' || get_oid_hex(end + 1, &oid))
1917                         die("corrupt mark line: %s", line);
1918                 e = find_object(&oid);
1919                 if (!e) {
1920                         enum object_type type = sha1_object_info(oid.hash, NULL);
1921                         if (type < 0)
1922                                 die("object not found: %s", oid_to_hex(&oid));
1923                         e = insert_object(&oid);
1924                         e->type = type;
1925                         e->pack_id = MAX_PACK_ID;
1926                         e->idx.offset = 1; /* just not zero! */
1927                 }
1928                 insert_mark(mark, e);
1929         }
1930         fclose(f);
1931 done:
1932         import_marks_file_done = 1;
1933 }
1934
1935
1936 static int read_next_command(void)
1937 {
1938         static int stdin_eof = 0;
1939
1940         if (stdin_eof) {
1941                 unread_command_buf = 0;
1942                 return EOF;
1943         }
1944
1945         for (;;) {
1946                 const char *p;
1947
1948                 if (unread_command_buf) {
1949                         unread_command_buf = 0;
1950                 } else {
1951                         struct recent_command *rc;
1952
1953                         strbuf_detach(&command_buf, NULL);
1954                         stdin_eof = strbuf_getline_lf(&command_buf, stdin);
1955                         if (stdin_eof)
1956                                 return EOF;
1957
1958                         if (!seen_data_command
1959                                 && !starts_with(command_buf.buf, "feature ")
1960                                 && !starts_with(command_buf.buf, "option ")) {
1961                                 parse_argv();
1962                         }
1963
1964                         rc = rc_free;
1965                         if (rc)
1966                                 rc_free = rc->next;
1967                         else {
1968                                 rc = cmd_hist.next;
1969                                 cmd_hist.next = rc->next;
1970                                 cmd_hist.next->prev = &cmd_hist;
1971                                 free(rc->buf);
1972                         }
1973
1974                         rc->buf = command_buf.buf;
1975                         rc->prev = cmd_tail;
1976                         rc->next = cmd_hist.prev;
1977                         rc->prev->next = rc;
1978                         cmd_tail = rc;
1979                 }
1980                 if (skip_prefix(command_buf.buf, "get-mark ", &p)) {
1981                         parse_get_mark(p);
1982                         continue;
1983                 }
1984                 if (skip_prefix(command_buf.buf, "cat-blob ", &p)) {
1985                         parse_cat_blob(p);
1986                         continue;
1987                 }
1988                 if (command_buf.buf[0] == '#')
1989                         continue;
1990                 return 0;
1991         }
1992 }
1993
1994 static void skip_optional_lf(void)
1995 {
1996         int term_char = fgetc(stdin);
1997         if (term_char != '\n' && term_char != EOF)
1998                 ungetc(term_char, stdin);
1999 }
2000
2001 static void parse_mark(void)
2002 {
2003         const char *v;
2004         if (skip_prefix(command_buf.buf, "mark :", &v)) {
2005                 next_mark = strtoumax(v, NULL, 10);
2006                 read_next_command();
2007         }
2008         else
2009                 next_mark = 0;
2010 }
2011
2012 static int parse_data(struct strbuf *sb, uintmax_t limit, uintmax_t *len_res)
2013 {
2014         const char *data;
2015         strbuf_reset(sb);
2016
2017         if (!skip_prefix(command_buf.buf, "data ", &data))
2018                 die("Expected 'data n' command, found: %s", command_buf.buf);
2019
2020         if (skip_prefix(data, "<<", &data)) {
2021                 char *term = xstrdup(data);
2022                 size_t term_len = command_buf.len - (data - command_buf.buf);
2023
2024                 strbuf_detach(&command_buf, NULL);
2025                 for (;;) {
2026                         if (strbuf_getline_lf(&command_buf, stdin) == EOF)
2027                                 die("EOF in data (terminator '%s' not found)", term);
2028                         if (term_len == command_buf.len
2029                                 && !strcmp(term, command_buf.buf))
2030                                 break;
2031                         strbuf_addbuf(sb, &command_buf);
2032                         strbuf_addch(sb, '\n');
2033                 }
2034                 free(term);
2035         }
2036         else {
2037                 uintmax_t len = strtoumax(data, NULL, 10);
2038                 size_t n = 0, length = (size_t)len;
2039
2040                 if (limit && limit < len) {
2041                         *len_res = len;
2042                         return 0;
2043                 }
2044                 if (length < len)
2045                         die("data is too large to use in this context");
2046
2047                 while (n < length) {
2048                         size_t s = strbuf_fread(sb, length - n, stdin);
2049                         if (!s && feof(stdin))
2050                                 die("EOF in data (%lu bytes remaining)",
2051                                         (unsigned long)(length - n));
2052                         n += s;
2053                 }
2054         }
2055
2056         skip_optional_lf();
2057         return 1;
2058 }
2059
2060 static int validate_raw_date(const char *src, struct strbuf *result)
2061 {
2062         const char *orig_src = src;
2063         char *endp;
2064         unsigned long num;
2065
2066         errno = 0;
2067
2068         num = strtoul(src, &endp, 10);
2069         /* NEEDSWORK: perhaps check for reasonable values? */
2070         if (errno || endp == src || *endp != ' ')
2071                 return -1;
2072
2073         src = endp + 1;
2074         if (*src != '-' && *src != '+')
2075                 return -1;
2076
2077         num = strtoul(src + 1, &endp, 10);
2078         if (errno || endp == src + 1 || *endp || 1400 < num)
2079                 return -1;
2080
2081         strbuf_addstr(result, orig_src);
2082         return 0;
2083 }
2084
2085 static char *parse_ident(const char *buf)
2086 {
2087         const char *ltgt;
2088         size_t name_len;
2089         struct strbuf ident = STRBUF_INIT;
2090
2091         /* ensure there is a space delimiter even if there is no name */
2092         if (*buf == '<')
2093                 --buf;
2094
2095         ltgt = buf + strcspn(buf, "<>");
2096         if (*ltgt != '<')
2097                 die("Missing < in ident string: %s", buf);
2098         if (ltgt != buf && ltgt[-1] != ' ')
2099                 die("Missing space before < in ident string: %s", buf);
2100         ltgt = ltgt + 1 + strcspn(ltgt + 1, "<>");
2101         if (*ltgt != '>')
2102                 die("Missing > in ident string: %s", buf);
2103         ltgt++;
2104         if (*ltgt != ' ')
2105                 die("Missing space after > in ident string: %s", buf);
2106         ltgt++;
2107         name_len = ltgt - buf;
2108         strbuf_add(&ident, buf, name_len);
2109
2110         switch (whenspec) {
2111         case WHENSPEC_RAW:
2112                 if (validate_raw_date(ltgt, &ident) < 0)
2113                         die("Invalid raw date \"%s\" in ident: %s", ltgt, buf);
2114                 break;
2115         case WHENSPEC_RFC2822:
2116                 if (parse_date(ltgt, &ident) < 0)
2117                         die("Invalid rfc2822 date \"%s\" in ident: %s", ltgt, buf);
2118                 break;
2119         case WHENSPEC_NOW:
2120                 if (strcmp("now", ltgt))
2121                         die("Date in ident must be 'now': %s", buf);
2122                 datestamp(&ident);
2123                 break;
2124         }
2125
2126         return strbuf_detach(&ident, NULL);
2127 }
2128
2129 static void parse_and_store_blob(
2130         struct last_object *last,
2131         struct object_id *oidout,
2132         uintmax_t mark)
2133 {
2134         static struct strbuf buf = STRBUF_INIT;
2135         uintmax_t len;
2136
2137         if (parse_data(&buf, big_file_threshold, &len))
2138                 store_object(OBJ_BLOB, &buf, last, oidout, mark);
2139         else {
2140                 if (last) {
2141                         strbuf_release(&last->data);
2142                         last->offset = 0;
2143                         last->depth = 0;
2144                 }
2145                 stream_blob(len, oidout, mark);
2146                 skip_optional_lf();
2147         }
2148 }
2149
2150 static void parse_new_blob(void)
2151 {
2152         read_next_command();
2153         parse_mark();
2154         parse_and_store_blob(&last_blob, NULL, next_mark);
2155 }
2156
2157 static void unload_one_branch(void)
2158 {
2159         while (cur_active_branches
2160                 && cur_active_branches >= max_active_branches) {
2161                 uintmax_t min_commit = ULONG_MAX;
2162                 struct branch *e, *l = NULL, *p = NULL;
2163
2164                 for (e = active_branches; e; e = e->active_next_branch) {
2165                         if (e->last_commit < min_commit) {
2166                                 p = l;
2167                                 min_commit = e->last_commit;
2168                         }
2169                         l = e;
2170                 }
2171
2172                 if (p) {
2173                         e = p->active_next_branch;
2174                         p->active_next_branch = e->active_next_branch;
2175                 } else {
2176                         e = active_branches;
2177                         active_branches = e->active_next_branch;
2178                 }
2179                 e->active = 0;
2180                 e->active_next_branch = NULL;
2181                 if (e->branch_tree.tree) {
2182                         release_tree_content_recursive(e->branch_tree.tree);
2183                         e->branch_tree.tree = NULL;
2184                 }
2185                 cur_active_branches--;
2186         }
2187 }
2188
2189 static void load_branch(struct branch *b)
2190 {
2191         load_tree(&b->branch_tree);
2192         if (!b->active) {
2193                 b->active = 1;
2194                 b->active_next_branch = active_branches;
2195                 active_branches = b;
2196                 cur_active_branches++;
2197                 branch_load_count++;
2198         }
2199 }
2200
2201 static unsigned char convert_num_notes_to_fanout(uintmax_t num_notes)
2202 {
2203         unsigned char fanout = 0;
2204         while ((num_notes >>= 8))
2205                 fanout++;
2206         return fanout;
2207 }
2208
2209 static void construct_path_with_fanout(const char *hex_sha1,
2210                 unsigned char fanout, char *path)
2211 {
2212         unsigned int i = 0, j = 0;
2213         if (fanout >= 20)
2214                 die("Too large fanout (%u)", fanout);
2215         while (fanout) {
2216                 path[i++] = hex_sha1[j++];
2217                 path[i++] = hex_sha1[j++];
2218                 path[i++] = '/';
2219                 fanout--;
2220         }
2221         memcpy(path + i, hex_sha1 + j, GIT_SHA1_HEXSZ - j);
2222         path[i + GIT_SHA1_HEXSZ - j] = '\0';
2223 }
2224
2225 static uintmax_t do_change_note_fanout(
2226                 struct tree_entry *orig_root, struct tree_entry *root,
2227                 char *hex_oid, unsigned int hex_oid_len,
2228                 char *fullpath, unsigned int fullpath_len,
2229                 unsigned char fanout)
2230 {
2231         struct tree_content *t;
2232         struct tree_entry *e, leaf;
2233         unsigned int i, tmp_hex_oid_len, tmp_fullpath_len;
2234         uintmax_t num_notes = 0;
2235         struct object_id oid;
2236         char realpath[60];
2237
2238         if (!root->tree)
2239                 load_tree(root);
2240         t = root->tree;
2241
2242         for (i = 0; t && i < t->entry_count; i++) {
2243                 e = t->entries[i];
2244                 tmp_hex_oid_len = hex_oid_len + e->name->str_len;
2245                 tmp_fullpath_len = fullpath_len;
2246
2247                 /*
2248                  * We're interested in EITHER existing note entries (entries
2249                  * with exactly 40 hex chars in path, not including directory
2250                  * separators), OR directory entries that may contain note
2251                  * entries (with < 40 hex chars in path).
2252                  * Also, each path component in a note entry must be a multiple
2253                  * of 2 chars.
2254                  */
2255                 if (!e->versions[1].mode ||
2256                     tmp_hex_oid_len > GIT_SHA1_HEXSZ ||
2257                     e->name->str_len % 2)
2258                         continue;
2259
2260                 /* This _may_ be a note entry, or a subdir containing notes */
2261                 memcpy(hex_oid + hex_oid_len, e->name->str_dat,
2262                        e->name->str_len);
2263                 if (tmp_fullpath_len)
2264                         fullpath[tmp_fullpath_len++] = '/';
2265                 memcpy(fullpath + tmp_fullpath_len, e->name->str_dat,
2266                        e->name->str_len);
2267                 tmp_fullpath_len += e->name->str_len;
2268                 fullpath[tmp_fullpath_len] = '\0';
2269
2270                 if (tmp_hex_oid_len == GIT_SHA1_HEXSZ && !get_oid_hex(hex_oid, &oid)) {
2271                         /* This is a note entry */
2272                         if (fanout == 0xff) {
2273                                 /* Counting mode, no rename */
2274                                 num_notes++;
2275                                 continue;
2276                         }
2277                         construct_path_with_fanout(hex_oid, fanout, realpath);
2278                         if (!strcmp(fullpath, realpath)) {
2279                                 /* Note entry is in correct location */
2280                                 num_notes++;
2281                                 continue;
2282                         }
2283
2284                         /* Rename fullpath to realpath */
2285                         if (!tree_content_remove(orig_root, fullpath, &leaf, 0))
2286                                 die("Failed to remove path %s", fullpath);
2287                         tree_content_set(orig_root, realpath,
2288                                 &leaf.versions[1].oid,
2289                                 leaf.versions[1].mode,
2290                                 leaf.tree);
2291                 } else if (S_ISDIR(e->versions[1].mode)) {
2292                         /* This is a subdir that may contain note entries */
2293                         num_notes += do_change_note_fanout(orig_root, e,
2294                                 hex_oid, tmp_hex_oid_len,
2295                                 fullpath, tmp_fullpath_len, fanout);
2296                 }
2297
2298                 /* The above may have reallocated the current tree_content */
2299                 t = root->tree;
2300         }
2301         return num_notes;
2302 }
2303
2304 static uintmax_t change_note_fanout(struct tree_entry *root,
2305                 unsigned char fanout)
2306 {
2307         /*
2308          * The size of path is due to one slash between every two hex digits,
2309          * plus the terminating NUL.  Note that there is no slash at the end, so
2310          * the number of slashes is one less than half the number of hex
2311          * characters.
2312          */
2313         char hex_oid[GIT_MAX_HEXSZ], path[GIT_MAX_HEXSZ + (GIT_MAX_HEXSZ / 2) - 1 + 1];
2314         return do_change_note_fanout(root, root, hex_oid, 0, path, 0, fanout);
2315 }
2316
2317 /*
2318  * Given a pointer into a string, parse a mark reference:
2319  *
2320  *   idnum ::= ':' bigint;
2321  *
2322  * Return the first character after the value in *endptr.
2323  *
2324  * Complain if the following character is not what is expected,
2325  * either a space or end of the string.
2326  */
2327 static uintmax_t parse_mark_ref(const char *p, char **endptr)
2328 {
2329         uintmax_t mark;
2330
2331         assert(*p == ':');
2332         p++;
2333         mark = strtoumax(p, endptr, 10);
2334         if (*endptr == p)
2335                 die("No value after ':' in mark: %s", command_buf.buf);
2336         return mark;
2337 }
2338
2339 /*
2340  * Parse the mark reference, and complain if this is not the end of
2341  * the string.
2342  */
2343 static uintmax_t parse_mark_ref_eol(const char *p)
2344 {
2345         char *end;
2346         uintmax_t mark;
2347
2348         mark = parse_mark_ref(p, &end);
2349         if (*end != '\0')
2350                 die("Garbage after mark: %s", command_buf.buf);
2351         return mark;
2352 }
2353
2354 /*
2355  * Parse the mark reference, demanding a trailing space.  Return a
2356  * pointer to the space.
2357  */
2358 static uintmax_t parse_mark_ref_space(const char **p)
2359 {
2360         uintmax_t mark;
2361         char *end;
2362
2363         mark = parse_mark_ref(*p, &end);
2364         if (*end++ != ' ')
2365                 die("Missing space after mark: %s", command_buf.buf);
2366         *p = end;
2367         return mark;
2368 }
2369
2370 static void file_change_m(const char *p, struct branch *b)
2371 {
2372         static struct strbuf uq = STRBUF_INIT;
2373         const char *endp;
2374         struct object_entry *oe;
2375         struct object_id oid;
2376         uint16_t mode, inline_data = 0;
2377
2378         p = get_mode(p, &mode);
2379         if (!p)
2380                 die("Corrupt mode: %s", command_buf.buf);
2381         switch (mode) {
2382         case 0644:
2383         case 0755:
2384                 mode |= S_IFREG;
2385         case S_IFREG | 0644:
2386         case S_IFREG | 0755:
2387         case S_IFLNK:
2388         case S_IFDIR:
2389         case S_IFGITLINK:
2390                 /* ok */
2391                 break;
2392         default:
2393                 die("Corrupt mode: %s", command_buf.buf);
2394         }
2395
2396         if (*p == ':') {
2397                 oe = find_mark(parse_mark_ref_space(&p));
2398                 oidcpy(&oid, &oe->idx.oid);
2399         } else if (skip_prefix(p, "inline ", &p)) {
2400                 inline_data = 1;
2401                 oe = NULL; /* not used with inline_data, but makes gcc happy */
2402         } else {
2403                 if (parse_oid_hex(p, &oid, &p))
2404                         die("Invalid dataref: %s", command_buf.buf);
2405                 oe = find_object(&oid);
2406                 if (*p++ != ' ')
2407                         die("Missing space after SHA1: %s", command_buf.buf);
2408         }
2409
2410         strbuf_reset(&uq);
2411         if (!unquote_c_style(&uq, p, &endp)) {
2412                 if (*endp)
2413                         die("Garbage after path in: %s", command_buf.buf);
2414                 p = uq.buf;
2415         }
2416
2417         /* Git does not track empty, non-toplevel directories. */
2418         if (S_ISDIR(mode) && is_empty_tree_oid(&oid) && *p) {
2419                 tree_content_remove(&b->branch_tree, p, NULL, 0);
2420                 return;
2421         }
2422
2423         if (S_ISGITLINK(mode)) {
2424                 if (inline_data)
2425                         die("Git links cannot be specified 'inline': %s",
2426                                 command_buf.buf);
2427                 else if (oe) {
2428                         if (oe->type != OBJ_COMMIT)
2429                                 die("Not a commit (actually a %s): %s",
2430                                         typename(oe->type), command_buf.buf);
2431                 }
2432                 /*
2433                  * Accept the sha1 without checking; it expected to be in
2434                  * another repository.
2435                  */
2436         } else if (inline_data) {
2437                 if (S_ISDIR(mode))
2438                         die("Directories cannot be specified 'inline': %s",
2439                                 command_buf.buf);
2440                 if (p != uq.buf) {
2441                         strbuf_addstr(&uq, p);
2442                         p = uq.buf;
2443                 }
2444                 read_next_command();
2445                 parse_and_store_blob(&last_blob, &oid, 0);
2446         } else {
2447                 enum object_type expected = S_ISDIR(mode) ?
2448                                                 OBJ_TREE: OBJ_BLOB;
2449                 enum object_type type = oe ? oe->type :
2450                                         sha1_object_info(oid.hash, NULL);
2451                 if (type < 0)
2452                         die("%s not found: %s",
2453                                         S_ISDIR(mode) ?  "Tree" : "Blob",
2454                                         command_buf.buf);
2455                 if (type != expected)
2456                         die("Not a %s (actually a %s): %s",
2457                                 typename(expected), typename(type),
2458                                 command_buf.buf);
2459         }
2460
2461         if (!*p) {
2462                 tree_content_replace(&b->branch_tree, &oid, mode, NULL);
2463                 return;
2464         }
2465         tree_content_set(&b->branch_tree, p, &oid, mode, NULL);
2466 }
2467
2468 static void file_change_d(const char *p, struct branch *b)
2469 {
2470         static struct strbuf uq = STRBUF_INIT;
2471         const char *endp;
2472
2473         strbuf_reset(&uq);
2474         if (!unquote_c_style(&uq, p, &endp)) {
2475                 if (*endp)
2476                         die("Garbage after path in: %s", command_buf.buf);
2477                 p = uq.buf;
2478         }
2479         tree_content_remove(&b->branch_tree, p, NULL, 1);
2480 }
2481
2482 static void file_change_cr(const char *s, struct branch *b, int rename)
2483 {
2484         const char *d;
2485         static struct strbuf s_uq = STRBUF_INIT;
2486         static struct strbuf d_uq = STRBUF_INIT;
2487         const char *endp;
2488         struct tree_entry leaf;
2489
2490         strbuf_reset(&s_uq);
2491         if (!unquote_c_style(&s_uq, s, &endp)) {
2492                 if (*endp != ' ')
2493                         die("Missing space after source: %s", command_buf.buf);
2494         } else {
2495                 endp = strchr(s, ' ');
2496                 if (!endp)
2497                         die("Missing space after source: %s", command_buf.buf);
2498                 strbuf_add(&s_uq, s, endp - s);
2499         }
2500         s = s_uq.buf;
2501
2502         endp++;
2503         if (!*endp)
2504                 die("Missing dest: %s", command_buf.buf);
2505
2506         d = endp;
2507         strbuf_reset(&d_uq);
2508         if (!unquote_c_style(&d_uq, d, &endp)) {
2509                 if (*endp)
2510                         die("Garbage after dest in: %s", command_buf.buf);
2511                 d = d_uq.buf;
2512         }
2513
2514         memset(&leaf, 0, sizeof(leaf));
2515         if (rename)
2516                 tree_content_remove(&b->branch_tree, s, &leaf, 1);
2517         else
2518                 tree_content_get(&b->branch_tree, s, &leaf, 1);
2519         if (!leaf.versions[1].mode)
2520                 die("Path %s not in branch", s);
2521         if (!*d) {      /* C "path/to/subdir" "" */
2522                 tree_content_replace(&b->branch_tree,
2523                         &leaf.versions[1].oid,
2524                         leaf.versions[1].mode,
2525                         leaf.tree);
2526                 return;
2527         }
2528         tree_content_set(&b->branch_tree, d,
2529                 &leaf.versions[1].oid,
2530                 leaf.versions[1].mode,
2531                 leaf.tree);
2532 }
2533
2534 static void note_change_n(const char *p, struct branch *b, unsigned char *old_fanout)
2535 {
2536         static struct strbuf uq = STRBUF_INIT;
2537         struct object_entry *oe;
2538         struct branch *s;
2539         struct object_id oid, commit_oid;
2540         char path[60];
2541         uint16_t inline_data = 0;
2542         unsigned char new_fanout;
2543
2544         /*
2545          * When loading a branch, we don't traverse its tree to count the real
2546          * number of notes (too expensive to do this for all non-note refs).
2547          * This means that recently loaded notes refs might incorrectly have
2548          * b->num_notes == 0, and consequently, old_fanout might be wrong.
2549          *
2550          * Fix this by traversing the tree and counting the number of notes
2551          * when b->num_notes == 0. If the notes tree is truly empty, the
2552          * calculation should not take long.
2553          */
2554         if (b->num_notes == 0 && *old_fanout == 0) {
2555                 /* Invoke change_note_fanout() in "counting mode". */
2556                 b->num_notes = change_note_fanout(&b->branch_tree, 0xff);
2557                 *old_fanout = convert_num_notes_to_fanout(b->num_notes);
2558         }
2559
2560         /* Now parse the notemodify command. */
2561         /* <dataref> or 'inline' */
2562         if (*p == ':') {
2563                 oe = find_mark(parse_mark_ref_space(&p));
2564                 oidcpy(&oid, &oe->idx.oid);
2565         } else if (skip_prefix(p, "inline ", &p)) {
2566                 inline_data = 1;
2567                 oe = NULL; /* not used with inline_data, but makes gcc happy */
2568         } else {
2569                 if (parse_oid_hex(p, &oid, &p))
2570                         die("Invalid dataref: %s", command_buf.buf);
2571                 oe = find_object(&oid);
2572                 if (*p++ != ' ')
2573                         die("Missing space after SHA1: %s", command_buf.buf);
2574         }
2575
2576         /* <commit-ish> */
2577         s = lookup_branch(p);
2578         if (s) {
2579                 if (is_null_oid(&s->oid))
2580                         die("Can't add a note on empty branch.");
2581                 oidcpy(&commit_oid, &s->oid);
2582         } else if (*p == ':') {
2583                 uintmax_t commit_mark = parse_mark_ref_eol(p);
2584                 struct object_entry *commit_oe = find_mark(commit_mark);
2585                 if (commit_oe->type != OBJ_COMMIT)
2586                         die("Mark :%" PRIuMAX " not a commit", commit_mark);
2587                 oidcpy(&commit_oid, &commit_oe->idx.oid);
2588         } else if (!get_oid(p, &commit_oid)) {
2589                 unsigned long size;
2590                 char *buf = read_object_with_reference(commit_oid.hash,
2591                         commit_type, &size, commit_oid.hash);
2592                 if (!buf || size < 46)
2593                         die("Not a valid commit: %s", p);
2594                 free(buf);
2595         } else
2596                 die("Invalid ref name or SHA1 expression: %s", p);
2597
2598         if (inline_data) {
2599                 if (p != uq.buf) {
2600                         strbuf_addstr(&uq, p);
2601                         p = uq.buf;
2602                 }
2603                 read_next_command();
2604                 parse_and_store_blob(&last_blob, &oid, 0);
2605         } else if (oe) {
2606                 if (oe->type != OBJ_BLOB)
2607                         die("Not a blob (actually a %s): %s",
2608                                 typename(oe->type), command_buf.buf);
2609         } else if (!is_null_oid(&oid)) {
2610                 enum object_type type = sha1_object_info(oid.hash, NULL);
2611                 if (type < 0)
2612                         die("Blob not found: %s", command_buf.buf);
2613                 if (type != OBJ_BLOB)
2614                         die("Not a blob (actually a %s): %s",
2615                             typename(type), command_buf.buf);
2616         }
2617
2618         construct_path_with_fanout(oid_to_hex(&commit_oid), *old_fanout, path);
2619         if (tree_content_remove(&b->branch_tree, path, NULL, 0))
2620                 b->num_notes--;
2621
2622         if (is_null_oid(&oid))
2623                 return; /* nothing to insert */
2624
2625         b->num_notes++;
2626         new_fanout = convert_num_notes_to_fanout(b->num_notes);
2627         construct_path_with_fanout(oid_to_hex(&commit_oid), new_fanout, path);
2628         tree_content_set(&b->branch_tree, path, &oid, S_IFREG | 0644, NULL);
2629 }
2630
2631 static void file_change_deleteall(struct branch *b)
2632 {
2633         release_tree_content_recursive(b->branch_tree.tree);
2634         oidclr(&b->branch_tree.versions[0].oid);
2635         oidclr(&b->branch_tree.versions[1].oid);
2636         load_tree(&b->branch_tree);
2637         b->num_notes = 0;
2638 }
2639
2640 static void parse_from_commit(struct branch *b, char *buf, unsigned long size)
2641 {
2642         if (!buf || size < GIT_SHA1_HEXSZ + 6)
2643                 die("Not a valid commit: %s", oid_to_hex(&b->oid));
2644         if (memcmp("tree ", buf, 5)
2645                 || get_oid_hex(buf + 5, &b->branch_tree.versions[1].oid))
2646                 die("The commit %s is corrupt", oid_to_hex(&b->oid));
2647         oidcpy(&b->branch_tree.versions[0].oid,
2648                &b->branch_tree.versions[1].oid);
2649 }
2650
2651 static void parse_from_existing(struct branch *b)
2652 {
2653         if (is_null_oid(&b->oid)) {
2654                 oidclr(&b->branch_tree.versions[0].oid);
2655                 oidclr(&b->branch_tree.versions[1].oid);
2656         } else {
2657                 unsigned long size;
2658                 char *buf;
2659
2660                 buf = read_object_with_reference(b->oid.hash,
2661                                                  commit_type, &size,
2662                                                  b->oid.hash);
2663                 parse_from_commit(b, buf, size);
2664                 free(buf);
2665         }
2666 }
2667
2668 static int parse_from(struct branch *b)
2669 {
2670         const char *from;
2671         struct branch *s;
2672         struct object_id oid;
2673
2674         if (!skip_prefix(command_buf.buf, "from ", &from))
2675                 return 0;
2676
2677         oidcpy(&oid, &b->branch_tree.versions[1].oid);
2678
2679         s = lookup_branch(from);
2680         if (b == s)
2681                 die("Can't create a branch from itself: %s", b->name);
2682         else if (s) {
2683                 struct object_id *t = &s->branch_tree.versions[1].oid;
2684                 oidcpy(&b->oid, &s->oid);
2685                 oidcpy(&b->branch_tree.versions[0].oid, t);
2686                 oidcpy(&b->branch_tree.versions[1].oid, t);
2687         } else if (*from == ':') {
2688                 uintmax_t idnum = parse_mark_ref_eol(from);
2689                 struct object_entry *oe = find_mark(idnum);
2690                 if (oe->type != OBJ_COMMIT)
2691                         die("Mark :%" PRIuMAX " not a commit", idnum);
2692                 if (oidcmp(&b->oid, &oe->idx.oid)) {
2693                         oidcpy(&b->oid, &oe->idx.oid);
2694                         if (oe->pack_id != MAX_PACK_ID) {
2695                                 unsigned long size;
2696                                 char *buf = gfi_unpack_entry(oe, &size);
2697                                 parse_from_commit(b, buf, size);
2698                                 free(buf);
2699                         } else
2700                                 parse_from_existing(b);
2701                 }
2702         } else if (!get_oid(from, &b->oid)) {
2703                 parse_from_existing(b);
2704                 if (is_null_oid(&b->oid))
2705                         b->delete = 1;
2706         }
2707         else
2708                 die("Invalid ref name or SHA1 expression: %s", from);
2709
2710         if (b->branch_tree.tree && oidcmp(&oid, &b->branch_tree.versions[1].oid)) {
2711                 release_tree_content_recursive(b->branch_tree.tree);
2712                 b->branch_tree.tree = NULL;
2713         }
2714
2715         read_next_command();
2716         return 1;
2717 }
2718
2719 static struct hash_list *parse_merge(unsigned int *count)
2720 {
2721         struct hash_list *list = NULL, **tail = &list, *n;
2722         const char *from;
2723         struct branch *s;
2724
2725         *count = 0;
2726         while (skip_prefix(command_buf.buf, "merge ", &from)) {
2727                 n = xmalloc(sizeof(*n));
2728                 s = lookup_branch(from);
2729                 if (s)
2730                         oidcpy(&n->oid, &s->oid);
2731                 else if (*from == ':') {
2732                         uintmax_t idnum = parse_mark_ref_eol(from);
2733                         struct object_entry *oe = find_mark(idnum);
2734                         if (oe->type != OBJ_COMMIT)
2735                                 die("Mark :%" PRIuMAX " not a commit", idnum);
2736                         oidcpy(&n->oid, &oe->idx.oid);
2737                 } else if (!get_oid(from, &n->oid)) {
2738                         unsigned long size;
2739                         char *buf = read_object_with_reference(n->oid.hash,
2740                                 commit_type, &size, n->oid.hash);
2741                         if (!buf || size < 46)
2742                                 die("Not a valid commit: %s", from);
2743                         free(buf);
2744                 } else
2745                         die("Invalid ref name or SHA1 expression: %s", from);
2746
2747                 n->next = NULL;
2748                 *tail = n;
2749                 tail = &n->next;
2750
2751                 (*count)++;
2752                 read_next_command();
2753         }
2754         return list;
2755 }
2756
2757 static void parse_new_commit(const char *arg)
2758 {
2759         static struct strbuf msg = STRBUF_INIT;
2760         struct branch *b;
2761         char *author = NULL;
2762         char *committer = NULL;
2763         struct hash_list *merge_list = NULL;
2764         unsigned int merge_count;
2765         unsigned char prev_fanout, new_fanout;
2766         const char *v;
2767
2768         b = lookup_branch(arg);
2769         if (!b)
2770                 b = new_branch(arg);
2771
2772         read_next_command();
2773         parse_mark();
2774         if (skip_prefix(command_buf.buf, "author ", &v)) {
2775                 author = parse_ident(v);
2776                 read_next_command();
2777         }
2778         if (skip_prefix(command_buf.buf, "committer ", &v)) {
2779                 committer = parse_ident(v);
2780                 read_next_command();
2781         }
2782         if (!committer)
2783                 die("Expected committer but didn't get one");
2784         parse_data(&msg, 0, NULL);
2785         read_next_command();
2786         parse_from(b);
2787         merge_list = parse_merge(&merge_count);
2788
2789         /* ensure the branch is active/loaded */
2790         if (!b->branch_tree.tree || !max_active_branches) {
2791                 unload_one_branch();
2792                 load_branch(b);
2793         }
2794
2795         prev_fanout = convert_num_notes_to_fanout(b->num_notes);
2796
2797         /* file_change* */
2798         while (command_buf.len > 0) {
2799                 if (skip_prefix(command_buf.buf, "M ", &v))
2800                         file_change_m(v, b);
2801                 else if (skip_prefix(command_buf.buf, "D ", &v))
2802                         file_change_d(v, b);
2803                 else if (skip_prefix(command_buf.buf, "R ", &v))
2804                         file_change_cr(v, b, 1);
2805                 else if (skip_prefix(command_buf.buf, "C ", &v))
2806                         file_change_cr(v, b, 0);
2807                 else if (skip_prefix(command_buf.buf, "N ", &v))
2808                         note_change_n(v, b, &prev_fanout);
2809                 else if (!strcmp("deleteall", command_buf.buf))
2810                         file_change_deleteall(b);
2811                 else if (skip_prefix(command_buf.buf, "ls ", &v))
2812                         parse_ls(v, b);
2813                 else {
2814                         unread_command_buf = 1;
2815                         break;
2816                 }
2817                 if (read_next_command() == EOF)
2818                         break;
2819         }
2820
2821         new_fanout = convert_num_notes_to_fanout(b->num_notes);
2822         if (new_fanout != prev_fanout)
2823                 b->num_notes = change_note_fanout(&b->branch_tree, new_fanout);
2824
2825         /* build the tree and the commit */
2826         store_tree(&b->branch_tree);
2827         oidcpy(&b->branch_tree.versions[0].oid,
2828                &b->branch_tree.versions[1].oid);
2829
2830         strbuf_reset(&new_data);
2831         strbuf_addf(&new_data, "tree %s\n",
2832                 oid_to_hex(&b->branch_tree.versions[1].oid));
2833         if (!is_null_oid(&b->oid))
2834                 strbuf_addf(&new_data, "parent %s\n",
2835                             oid_to_hex(&b->oid));
2836         while (merge_list) {
2837                 struct hash_list *next = merge_list->next;
2838                 strbuf_addf(&new_data, "parent %s\n",
2839                             oid_to_hex(&merge_list->oid));
2840                 free(merge_list);
2841                 merge_list = next;
2842         }
2843         strbuf_addf(&new_data,
2844                 "author %s\n"
2845                 "committer %s\n"
2846                 "\n",
2847                 author ? author : committer, committer);
2848         strbuf_addbuf(&new_data, &msg);
2849         free(author);
2850         free(committer);
2851
2852         if (!store_object(OBJ_COMMIT, &new_data, NULL, &b->oid, next_mark))
2853                 b->pack_id = pack_id;
2854         b->last_commit = object_count_by_type[OBJ_COMMIT];
2855 }
2856
2857 static void parse_new_tag(const char *arg)
2858 {
2859         static struct strbuf msg = STRBUF_INIT;
2860         const char *from;
2861         char *tagger;
2862         struct branch *s;
2863         struct tag *t;
2864         uintmax_t from_mark = 0;
2865         struct object_id oid;
2866         enum object_type type;
2867         const char *v;
2868
2869         t = pool_alloc(sizeof(struct tag));
2870         memset(t, 0, sizeof(struct tag));
2871         t->name = pool_strdup(arg);
2872         if (last_tag)
2873                 last_tag->next_tag = t;
2874         else
2875                 first_tag = t;
2876         last_tag = t;
2877         read_next_command();
2878
2879         /* from ... */
2880         if (!skip_prefix(command_buf.buf, "from ", &from))
2881                 die("Expected from command, got %s", command_buf.buf);
2882         s = lookup_branch(from);
2883         if (s) {
2884                 if (is_null_oid(&s->oid))
2885                         die("Can't tag an empty branch.");
2886                 oidcpy(&oid, &s->oid);
2887                 type = OBJ_COMMIT;
2888         } else if (*from == ':') {
2889                 struct object_entry *oe;
2890                 from_mark = parse_mark_ref_eol(from);
2891                 oe = find_mark(from_mark);
2892                 type = oe->type;
2893                 oidcpy(&oid, &oe->idx.oid);
2894         } else if (!get_oid(from, &oid)) {
2895                 struct object_entry *oe = find_object(&oid);
2896                 if (!oe) {
2897                         type = sha1_object_info(oid.hash, NULL);
2898                         if (type < 0)
2899                                 die("Not a valid object: %s", from);
2900                 } else
2901                         type = oe->type;
2902         } else
2903                 die("Invalid ref name or SHA1 expression: %s", from);
2904         read_next_command();
2905
2906         /* tagger ... */
2907         if (skip_prefix(command_buf.buf, "tagger ", &v)) {
2908                 tagger = parse_ident(v);
2909                 read_next_command();
2910         } else
2911                 tagger = NULL;
2912
2913         /* tag payload/message */
2914         parse_data(&msg, 0, NULL);
2915
2916         /* build the tag object */
2917         strbuf_reset(&new_data);
2918
2919         strbuf_addf(&new_data,
2920                     "object %s\n"
2921                     "type %s\n"
2922                     "tag %s\n",
2923                     oid_to_hex(&oid), typename(type), t->name);
2924         if (tagger)
2925                 strbuf_addf(&new_data,
2926                             "tagger %s\n", tagger);
2927         strbuf_addch(&new_data, '\n');
2928         strbuf_addbuf(&new_data, &msg);
2929         free(tagger);
2930
2931         if (store_object(OBJ_TAG, &new_data, NULL, &t->oid, 0))
2932                 t->pack_id = MAX_PACK_ID;
2933         else
2934                 t->pack_id = pack_id;
2935 }
2936
2937 static void parse_reset_branch(const char *arg)
2938 {
2939         struct branch *b;
2940
2941         b = lookup_branch(arg);
2942         if (b) {
2943                 oidclr(&b->oid);
2944                 oidclr(&b->branch_tree.versions[0].oid);
2945                 oidclr(&b->branch_tree.versions[1].oid);
2946                 if (b->branch_tree.tree) {
2947                         release_tree_content_recursive(b->branch_tree.tree);
2948                         b->branch_tree.tree = NULL;
2949                 }
2950         }
2951         else
2952                 b = new_branch(arg);
2953         read_next_command();
2954         parse_from(b);
2955         if (command_buf.len > 0)
2956                 unread_command_buf = 1;
2957 }
2958
2959 static void cat_blob_write(const char *buf, unsigned long size)
2960 {
2961         if (write_in_full(cat_blob_fd, buf, size) < 0)
2962                 die_errno("Write to frontend failed");
2963 }
2964
2965 static void cat_blob(struct object_entry *oe, struct object_id *oid)
2966 {
2967         struct strbuf line = STRBUF_INIT;
2968         unsigned long size;
2969         enum object_type type = 0;
2970         char *buf;
2971
2972         if (!oe || oe->pack_id == MAX_PACK_ID) {
2973                 buf = read_sha1_file(oid->hash, &type, &size);
2974         } else {
2975                 type = oe->type;
2976                 buf = gfi_unpack_entry(oe, &size);
2977         }
2978
2979         /*
2980          * Output based on batch_one_object() from cat-file.c.
2981          */
2982         if (type <= 0) {
2983                 strbuf_reset(&line);
2984                 strbuf_addf(&line, "%s missing\n", oid_to_hex(oid));
2985                 cat_blob_write(line.buf, line.len);
2986                 strbuf_release(&line);
2987                 free(buf);
2988                 return;
2989         }
2990         if (!buf)
2991                 die("Can't read object %s", oid_to_hex(oid));
2992         if (type != OBJ_BLOB)
2993                 die("Object %s is a %s but a blob was expected.",
2994                     oid_to_hex(oid), typename(type));
2995         strbuf_reset(&line);
2996         strbuf_addf(&line, "%s %s %lu\n", oid_to_hex(oid),
2997                                                 typename(type), size);
2998         cat_blob_write(line.buf, line.len);
2999         strbuf_release(&line);
3000         cat_blob_write(buf, size);
3001         cat_blob_write("\n", 1);
3002         if (oe && oe->pack_id == pack_id) {
3003                 last_blob.offset = oe->idx.offset;
3004                 strbuf_attach(&last_blob.data, buf, size, size);
3005                 last_blob.depth = oe->depth;
3006         } else
3007                 free(buf);
3008 }
3009
3010 static void parse_get_mark(const char *p)
3011 {
3012         struct object_entry *oe = oe;
3013         char output[GIT_MAX_HEXSZ + 2];
3014
3015         /* get-mark SP <object> LF */
3016         if (*p != ':')
3017                 die("Not a mark: %s", p);
3018
3019         oe = find_mark(parse_mark_ref_eol(p));
3020         if (!oe)
3021                 die("Unknown mark: %s", command_buf.buf);
3022
3023         xsnprintf(output, sizeof(output), "%s\n", oid_to_hex(&oe->idx.oid));
3024         cat_blob_write(output, GIT_SHA1_HEXSZ + 1);
3025 }
3026
3027 static void parse_cat_blob(const char *p)
3028 {
3029         struct object_entry *oe = oe;
3030         struct object_id oid;
3031
3032         /* cat-blob SP <object> LF */
3033         if (*p == ':') {
3034                 oe = find_mark(parse_mark_ref_eol(p));
3035                 if (!oe)
3036                         die("Unknown mark: %s", command_buf.buf);
3037                 oidcpy(&oid, &oe->idx.oid);
3038         } else {
3039                 if (parse_oid_hex(p, &oid, &p))
3040                         die("Invalid dataref: %s", command_buf.buf);
3041                 if (*p)
3042                         die("Garbage after SHA1: %s", command_buf.buf);
3043                 oe = find_object(&oid);
3044         }
3045
3046         cat_blob(oe, &oid);
3047 }
3048
3049 static struct object_entry *dereference(struct object_entry *oe,
3050                                         struct object_id *oid)
3051 {
3052         unsigned long size;
3053         char *buf = NULL;
3054         if (!oe) {
3055                 enum object_type type = sha1_object_info(oid->hash, NULL);
3056                 if (type < 0)
3057                         die("object not found: %s", oid_to_hex(oid));
3058                 /* cache it! */
3059                 oe = insert_object(oid);
3060                 oe->type = type;
3061                 oe->pack_id = MAX_PACK_ID;
3062                 oe->idx.offset = 1;
3063         }
3064         switch (oe->type) {
3065         case OBJ_TREE:  /* easy case. */
3066                 return oe;
3067         case OBJ_COMMIT:
3068         case OBJ_TAG:
3069                 break;
3070         default:
3071                 die("Not a tree-ish: %s", command_buf.buf);
3072         }
3073
3074         if (oe->pack_id != MAX_PACK_ID) {       /* in a pack being written */
3075                 buf = gfi_unpack_entry(oe, &size);
3076         } else {
3077                 enum object_type unused;
3078                 buf = read_sha1_file(oid->hash, &unused, &size);
3079         }
3080         if (!buf)
3081                 die("Can't load object %s", oid_to_hex(oid));
3082
3083         /* Peel one layer. */
3084         switch (oe->type) {
3085         case OBJ_TAG:
3086                 if (size < GIT_SHA1_HEXSZ + strlen("object ") ||
3087                     get_oid_hex(buf + strlen("object "), oid))
3088                         die("Invalid SHA1 in tag: %s", command_buf.buf);
3089                 break;
3090         case OBJ_COMMIT:
3091                 if (size < GIT_SHA1_HEXSZ + strlen("tree ") ||
3092                     get_oid_hex(buf + strlen("tree "), oid))
3093                         die("Invalid SHA1 in commit: %s", command_buf.buf);
3094         }
3095
3096         free(buf);
3097         return find_object(oid);
3098 }
3099
3100 static struct object_entry *parse_treeish_dataref(const char **p)
3101 {
3102         struct object_id oid;
3103         struct object_entry *e;
3104
3105         if (**p == ':') {       /* <mark> */
3106                 e = find_mark(parse_mark_ref_space(p));
3107                 if (!e)
3108                         die("Unknown mark: %s", command_buf.buf);
3109                 oidcpy(&oid, &e->idx.oid);
3110         } else {        /* <sha1> */
3111                 if (parse_oid_hex(*p, &oid, p))
3112                         die("Invalid dataref: %s", command_buf.buf);
3113                 e = find_object(&oid);
3114                 if (*(*p)++ != ' ')
3115                         die("Missing space after tree-ish: %s", command_buf.buf);
3116         }
3117
3118         while (!e || e->type != OBJ_TREE)
3119                 e = dereference(e, &oid);
3120         return e;
3121 }
3122
3123 static void print_ls(int mode, const unsigned char *sha1, const char *path)
3124 {
3125         static struct strbuf line = STRBUF_INIT;
3126
3127         /* See show_tree(). */
3128         const char *type =
3129                 S_ISGITLINK(mode) ? commit_type :
3130                 S_ISDIR(mode) ? tree_type :
3131                 blob_type;
3132
3133         if (!mode) {
3134                 /* missing SP path LF */
3135                 strbuf_reset(&line);
3136                 strbuf_addstr(&line, "missing ");
3137                 quote_c_style(path, &line, NULL, 0);
3138                 strbuf_addch(&line, '\n');
3139         } else {
3140                 /* mode SP type SP object_name TAB path LF */
3141                 strbuf_reset(&line);
3142                 strbuf_addf(&line, "%06o %s %s\t",
3143                                 mode & ~NO_DELTA, type, sha1_to_hex(sha1));
3144                 quote_c_style(path, &line, NULL, 0);
3145                 strbuf_addch(&line, '\n');
3146         }
3147         cat_blob_write(line.buf, line.len);
3148 }
3149
3150 static void parse_ls(const char *p, struct branch *b)
3151 {
3152         struct tree_entry *root = NULL;
3153         struct tree_entry leaf = {NULL};
3154
3155         /* ls SP (<tree-ish> SP)? <path> */
3156         if (*p == '"') {
3157                 if (!b)
3158                         die("Not in a commit: %s", command_buf.buf);
3159                 root = &b->branch_tree;
3160         } else {
3161                 struct object_entry *e = parse_treeish_dataref(&p);
3162                 root = new_tree_entry();
3163                 oidcpy(&root->versions[1].oid, &e->idx.oid);
3164                 if (!is_null_oid(&root->versions[1].oid))
3165                         root->versions[1].mode = S_IFDIR;
3166                 load_tree(root);
3167         }
3168         if (*p == '"') {
3169                 static struct strbuf uq = STRBUF_INIT;
3170                 const char *endp;
3171                 strbuf_reset(&uq);
3172                 if (unquote_c_style(&uq, p, &endp))
3173                         die("Invalid path: %s", command_buf.buf);
3174                 if (*endp)
3175                         die("Garbage after path in: %s", command_buf.buf);
3176                 p = uq.buf;
3177         }
3178         tree_content_get(root, p, &leaf, 1);
3179         /*
3180          * A directory in preparation would have a sha1 of zero
3181          * until it is saved.  Save, for simplicity.
3182          */
3183         if (S_ISDIR(leaf.versions[1].mode))
3184                 store_tree(&leaf);
3185
3186         print_ls(leaf.versions[1].mode, leaf.versions[1].oid.hash, p);
3187         if (leaf.tree)
3188                 release_tree_content_recursive(leaf.tree);
3189         if (!b || root != &b->branch_tree)
3190                 release_tree_entry(root);
3191 }
3192
3193 static void checkpoint(void)
3194 {
3195         checkpoint_requested = 0;
3196         if (object_count) {
3197                 cycle_packfile();
3198         }
3199         dump_branches();
3200         dump_tags();
3201         dump_marks();
3202 }
3203
3204 static void parse_checkpoint(void)
3205 {
3206         checkpoint_requested = 1;
3207         skip_optional_lf();
3208 }
3209
3210 static void parse_progress(void)
3211 {
3212         fwrite(command_buf.buf, 1, command_buf.len, stdout);
3213         fputc('\n', stdout);
3214         fflush(stdout);
3215         skip_optional_lf();
3216 }
3217
3218 static char* make_fast_import_path(const char *path)
3219 {
3220         if (!relative_marks_paths || is_absolute_path(path))
3221                 return xstrdup(path);
3222         return git_pathdup("info/fast-import/%s", path);
3223 }
3224
3225 static void option_import_marks(const char *marks,
3226                                         int from_stream, int ignore_missing)
3227 {
3228         if (import_marks_file) {
3229                 if (from_stream)
3230                         die("Only one import-marks command allowed per stream");
3231
3232                 /* read previous mark file */
3233                 if(!import_marks_file_from_stream)
3234                         read_marks();
3235         }
3236
3237         import_marks_file = make_fast_import_path(marks);
3238         import_marks_file_from_stream = from_stream;
3239         import_marks_file_ignore_missing = ignore_missing;
3240 }
3241
3242 static void option_date_format(const char *fmt)
3243 {
3244         if (!strcmp(fmt, "raw"))
3245                 whenspec = WHENSPEC_RAW;
3246         else if (!strcmp(fmt, "rfc2822"))
3247                 whenspec = WHENSPEC_RFC2822;
3248         else if (!strcmp(fmt, "now"))
3249                 whenspec = WHENSPEC_NOW;
3250         else
3251                 die("unknown --date-format argument %s", fmt);
3252 }
3253
3254 static unsigned long ulong_arg(const char *option, const char *arg)
3255 {
3256         char *endptr;
3257         unsigned long rv = strtoul(arg, &endptr, 0);
3258         if (strchr(arg, '-') || endptr == arg || *endptr)
3259                 die("%s: argument must be a non-negative integer", option);
3260         return rv;
3261 }
3262
3263 static void option_depth(const char *depth)
3264 {
3265         max_depth = ulong_arg("--depth", depth);
3266         if (max_depth > MAX_DEPTH)
3267                 die("--depth cannot exceed %u", MAX_DEPTH);
3268 }
3269
3270 static void option_active_branches(const char *branches)
3271 {
3272         max_active_branches = ulong_arg("--active-branches", branches);
3273 }
3274
3275 static void option_export_marks(const char *marks)
3276 {
3277         export_marks_file = make_fast_import_path(marks);
3278 }
3279
3280 static void option_cat_blob_fd(const char *fd)
3281 {
3282         unsigned long n = ulong_arg("--cat-blob-fd", fd);
3283         if (n > (unsigned long) INT_MAX)
3284                 die("--cat-blob-fd cannot exceed %d", INT_MAX);
3285         cat_blob_fd = (int) n;
3286 }
3287
3288 static void option_export_pack_edges(const char *edges)
3289 {
3290         if (pack_edges)
3291                 fclose(pack_edges);
3292         pack_edges = xfopen(edges, "a");
3293 }
3294
3295 static int parse_one_option(const char *option)
3296 {
3297         if (skip_prefix(option, "max-pack-size=", &option)) {
3298                 unsigned long v;
3299                 if (!git_parse_ulong(option, &v))
3300                         return 0;
3301                 if (v < 8192) {
3302                         warning("max-pack-size is now in bytes, assuming --max-pack-size=%lum", v);
3303                         v *= 1024 * 1024;
3304                 } else if (v < 1024 * 1024) {
3305                         warning("minimum max-pack-size is 1 MiB");
3306                         v = 1024 * 1024;
3307                 }
3308                 max_packsize = v;
3309         } else if (skip_prefix(option, "big-file-threshold=", &option)) {
3310                 unsigned long v;
3311                 if (!git_parse_ulong(option, &v))
3312                         return 0;
3313                 big_file_threshold = v;
3314         } else if (skip_prefix(option, "depth=", &option)) {
3315                 option_depth(option);
3316         } else if (skip_prefix(option, "active-branches=", &option)) {
3317                 option_active_branches(option);
3318         } else if (skip_prefix(option, "export-pack-edges=", &option)) {
3319                 option_export_pack_edges(option);
3320         } else if (!strcmp(option, "quiet")) {
3321                 show_stats = 0;
3322         } else if (!strcmp(option, "stats")) {
3323                 show_stats = 1;
3324         } else if (!strcmp(option, "allow-unsafe-features")) {
3325                 ; /* already handled during early option parsing */
3326         } else {
3327                 return 0;
3328         }
3329
3330         return 1;
3331 }
3332
3333 static void check_unsafe_feature(const char *feature, int from_stream)
3334 {
3335         if (from_stream && !allow_unsafe_features)
3336                 die(_("feature '%s' forbidden in input without --allow-unsafe-features"),
3337                     feature);
3338 }
3339
3340 static int parse_one_feature(const char *feature, int from_stream)
3341 {
3342         const char *arg;
3343
3344         if (skip_prefix(feature, "date-format=", &arg)) {
3345                 option_date_format(arg);
3346         } else if (skip_prefix(feature, "import-marks=", &arg)) {
3347                 check_unsafe_feature("import-marks", from_stream);
3348                 option_import_marks(arg, from_stream, 0);
3349         } else if (skip_prefix(feature, "import-marks-if-exists=", &arg)) {
3350                 check_unsafe_feature("import-marks-if-exists", from_stream);
3351                 option_import_marks(arg, from_stream, 1);
3352         } else if (skip_prefix(feature, "export-marks=", &arg)) {
3353                 check_unsafe_feature(feature, from_stream);
3354                 option_export_marks(arg);
3355         } else if (!strcmp(feature, "get-mark")) {
3356                 ; /* Don't die - this feature is supported */
3357         } else if (!strcmp(feature, "cat-blob")) {
3358                 ; /* Don't die - this feature is supported */
3359         } else if (!strcmp(feature, "relative-marks")) {
3360                 relative_marks_paths = 1;
3361         } else if (!strcmp(feature, "no-relative-marks")) {
3362                 relative_marks_paths = 0;
3363         } else if (!strcmp(feature, "done")) {
3364                 require_explicit_termination = 1;
3365         } else if (!strcmp(feature, "force")) {
3366                 force_update = 1;
3367         } else if (!strcmp(feature, "notes") || !strcmp(feature, "ls")) {
3368                 ; /* do nothing; we have the feature */
3369         } else {
3370                 return 0;
3371         }
3372
3373         return 1;
3374 }
3375
3376 static void parse_feature(const char *feature)
3377 {
3378         if (seen_data_command)
3379                 die("Got feature command '%s' after data command", feature);
3380
3381         if (parse_one_feature(feature, 1))
3382                 return;
3383
3384         die("This version of fast-import does not support feature %s.", feature);
3385 }
3386
3387 static void parse_option(const char *option)
3388 {
3389         if (seen_data_command)
3390                 die("Got option command '%s' after data command", option);
3391
3392         if (parse_one_option(option))
3393                 return;
3394
3395         die("This version of fast-import does not support option: %s", option);
3396 }
3397
3398 static void git_pack_config(void)
3399 {
3400         int indexversion_value;
3401         int limit;
3402         unsigned long packsizelimit_value;
3403
3404         if (!git_config_get_ulong("pack.depth", &max_depth)) {
3405                 if (max_depth > MAX_DEPTH)
3406                         max_depth = MAX_DEPTH;
3407         }
3408         if (!git_config_get_int("pack.indexversion", &indexversion_value)) {
3409                 pack_idx_opts.version = indexversion_value;
3410                 if (pack_idx_opts.version > 2)
3411                         git_die_config("pack.indexversion",
3412                                         "bad pack.indexversion=%"PRIu32, pack_idx_opts.version);
3413         }
3414         if (!git_config_get_ulong("pack.packsizelimit", &packsizelimit_value))
3415                 max_packsize = packsizelimit_value;
3416
3417         if (!git_config_get_int("fastimport.unpacklimit", &limit))
3418                 unpack_limit = limit;
3419         else if (!git_config_get_int("transfer.unpacklimit", &limit))
3420                 unpack_limit = limit;
3421
3422         git_config(git_default_config, NULL);
3423 }
3424
3425 static const char fast_import_usage[] =
3426 "git fast-import [--date-format=<f>] [--max-pack-size=<n>] [--big-file-threshold=<n>] [--depth=<n>] [--active-branches=<n>] [--export-marks=<marks.file>]";
3427
3428 static void parse_argv(void)
3429 {
3430         unsigned int i;
3431
3432         for (i = 1; i < global_argc; i++) {
3433                 const char *a = global_argv[i];
3434
3435                 if (*a != '-' || !strcmp(a, "--"))
3436                         break;
3437
3438                 if (!skip_prefix(a, "--", &a))
3439                         die("unknown option %s", a);
3440
3441                 if (parse_one_option(a))
3442                         continue;
3443
3444                 if (parse_one_feature(a, 0))
3445                         continue;
3446
3447                 if (skip_prefix(a, "cat-blob-fd=", &a)) {
3448                         option_cat_blob_fd(a);
3449                         continue;
3450                 }
3451
3452                 die("unknown option --%s", a);
3453         }
3454         if (i != global_argc)
3455                 usage(fast_import_usage);
3456
3457         seen_data_command = 1;
3458         if (import_marks_file)
3459                 read_marks();
3460 }
3461
3462 int cmd_main(int argc, const char **argv)
3463 {
3464         unsigned int i;
3465
3466         if (argc == 2 && !strcmp(argv[1], "-h"))
3467                 usage(fast_import_usage);
3468
3469         setup_git_directory();
3470         reset_pack_idx_option(&pack_idx_opts);
3471         git_pack_config();
3472
3473         alloc_objects(object_entry_alloc);
3474         strbuf_init(&command_buf, 0);
3475         atom_table = xcalloc(atom_table_sz, sizeof(struct atom_str*));
3476         branch_table = xcalloc(branch_table_sz, sizeof(struct branch*));
3477         avail_tree_table = xcalloc(avail_tree_table_sz, sizeof(struct avail_tree_content*));
3478         marks = pool_calloc(1, sizeof(struct mark_set));
3479
3480         /*
3481          * We don't parse most options until after we've seen the set of
3482          * "feature" lines at the start of the stream (which allows the command
3483          * line to override stream data). But we must do an early parse of any
3484          * command-line options that impact how we interpret the feature lines.
3485          */
3486         for (i = 1; i < argc; i++) {
3487                 const char *arg = argv[i];
3488                 if (*arg != '-' || !strcmp(arg, "--"))
3489                         break;
3490                 if (!strcmp(arg, "--allow-unsafe-features"))
3491                         allow_unsafe_features = 1;
3492         }
3493
3494         global_argc = argc;
3495         global_argv = argv;
3496
3497         rc_free = pool_alloc(cmd_save * sizeof(*rc_free));
3498         for (i = 0; i < (cmd_save - 1); i++)
3499                 rc_free[i].next = &rc_free[i + 1];
3500         rc_free[cmd_save - 1].next = NULL;
3501
3502         prepare_packed_git();
3503         start_packfile();
3504         set_die_routine(die_nicely);
3505         set_checkpoint_signal();
3506         while (read_next_command() != EOF) {
3507                 const char *v;
3508                 if (!strcmp("blob", command_buf.buf))
3509                         parse_new_blob();
3510                 else if (skip_prefix(command_buf.buf, "ls ", &v))
3511                         parse_ls(v, NULL);
3512                 else if (skip_prefix(command_buf.buf, "commit ", &v))
3513                         parse_new_commit(v);
3514                 else if (skip_prefix(command_buf.buf, "tag ", &v))
3515                         parse_new_tag(v);
3516                 else if (skip_prefix(command_buf.buf, "reset ", &v))
3517                         parse_reset_branch(v);
3518                 else if (!strcmp("checkpoint", command_buf.buf))
3519                         parse_checkpoint();
3520                 else if (!strcmp("done", command_buf.buf))
3521                         break;
3522                 else if (starts_with(command_buf.buf, "progress "))
3523                         parse_progress();
3524                 else if (skip_prefix(command_buf.buf, "feature ", &v))
3525                         parse_feature(v);
3526                 else if (skip_prefix(command_buf.buf, "option git ", &v))
3527                         parse_option(v);
3528                 else if (starts_with(command_buf.buf, "option "))
3529                         /* ignore non-git options*/;
3530                 else
3531                         die("Unsupported command: %s", command_buf.buf);
3532
3533                 if (checkpoint_requested)
3534                         checkpoint();
3535         }
3536
3537         /* argv hasn't been parsed yet, do so */
3538         if (!seen_data_command)
3539                 parse_argv();
3540
3541         if (require_explicit_termination && feof(stdin))
3542                 die("stream ends early");
3543
3544         end_packfile();
3545
3546         dump_branches();
3547         dump_tags();
3548         unkeep_all_packs();
3549         dump_marks();
3550
3551         if (pack_edges)
3552                 fclose(pack_edges);
3553
3554         if (show_stats) {
3555                 uintmax_t total_count = 0, duplicate_count = 0;
3556                 for (i = 0; i < ARRAY_SIZE(object_count_by_type); i++)
3557                         total_count += object_count_by_type[i];
3558                 for (i = 0; i < ARRAY_SIZE(duplicate_count_by_type); i++)
3559                         duplicate_count += duplicate_count_by_type[i];
3560
3561                 fprintf(stderr, "%s statistics:\n", argv[0]);
3562                 fprintf(stderr, "---------------------------------------------------------------------\n");
3563                 fprintf(stderr, "Alloc'd objects: %10" PRIuMAX "\n", alloc_count);
3564                 fprintf(stderr, "Total objects:   %10" PRIuMAX " (%10" PRIuMAX " duplicates                  )\n", total_count, duplicate_count);
3565                 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]);
3566                 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]);
3567                 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]);
3568                 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]);
3569                 fprintf(stderr, "Total branches:  %10lu (%10lu loads     )\n", branch_count, branch_load_count);
3570                 fprintf(stderr, "      marks:     %10" PRIuMAX " (%10" PRIuMAX " unique    )\n", (((uintmax_t)1) << marks->shift) * 1024, marks_set_count);
3571                 fprintf(stderr, "      atoms:     %10u\n", atom_cnt);
3572                 fprintf(stderr, "Memory total:    %10" PRIuMAX " KiB\n", (total_allocd + alloc_count*sizeof(struct object_entry))/1024);
3573                 fprintf(stderr, "       pools:    %10lu KiB\n", (unsigned long)(total_allocd/1024));
3574                 fprintf(stderr, "     objects:    %10" PRIuMAX " KiB\n", (alloc_count*sizeof(struct object_entry))/1024);
3575                 fprintf(stderr, "---------------------------------------------------------------------\n");
3576                 pack_report();
3577                 fprintf(stderr, "---------------------------------------------------------------------\n");
3578                 fprintf(stderr, "\n");
3579         }
3580
3581         return failure ? 1 : 0;
3582 }