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 #include "packfile.h"
 171
 172 #define PACK_ID_BITS 16
 173 #define MAX_PACK_ID ((1<<PACK_ID_BITS)-1)
 174 #define DEPTH_BITS 13
 175 #define MAX_DEPTH ((1<<DEPTH_BITS)-1)
 176
 177 /*
 178  * We abuse the setuid bit on directories to mean "do not delta".
 179  */
 180 #define NO_DELTA S_ISUID
 181
 182 struct object_entry {
 183         struct pack_idx_entry idx;
 184         struct object_entry *next;
 185         uint32_t type : TYPE_BITS,
 186                 pack_id : PACK_ID_BITS,
 187                 depth : DEPTH_BITS;
 188 };
 189
 190 struct object_entry_pool {
 191         struct object_entry_pool *next_pool;
 192         struct object_entry *next_free;
 193         struct object_entry *end;
 194         struct object_entry entries[FLEX_ARRAY]; /* more */
 195 };
 196
 197 struct mark_set {
 198         union {
 199                 struct object_entry *marked[1024];
 200                 struct mark_set *sets[1024];
 201         } data;
 202         unsigned int shift;
 203 };
 204
 205 struct last_object {
 206         struct strbuf data;
 207         off_t offset;
 208         unsigned int depth;
 209         unsigned no_swap : 1;
 210 };
 211
 212 struct mem_pool {
 213         struct mem_pool *next_pool;
 214         char *next_free;
 215         char *end;
 216         uintmax_t space[FLEX_ARRAY]; /* more */
 217 };
 218
 219 struct atom_str {
 220         struct atom_str *next_atom;
 221         unsigned short str_len;
 222         char str_dat[FLEX_ARRAY]; /* more */
 223 };
 224
 225 struct tree_content;
 226 struct tree_entry {
 227         struct tree_content *tree;
 228         struct atom_str *name;
 229         struct tree_entry_ms {
 230                 uint16_t mode;
 231                 struct object_id oid;
 232         } versions[2];
 233 };
 234
 235 struct tree_content {
 236         unsigned int entry_capacity; /* must match avail_tree_content */
 237         unsigned int entry_count;
 238         unsigned int delta_depth;
 239         struct tree_entry *entries[FLEX_ARRAY]; /* more */
 240 };
 241
 242 struct avail_tree_content {
 243         unsigned int entry_capacity; /* must match tree_content */
 244         struct avail_tree_content *next_avail;
 245 };
 246
 247 struct branch {
 248         struct branch *table_next_branch;
 249         struct branch *active_next_branch;
 250         const char *name;
 251         struct tree_entry branch_tree;
 252         uintmax_t last_commit;
 253         uintmax_t num_notes;
 254         unsigned active : 1;
 255         unsigned delete : 1;
 256         unsigned pack_id : PACK_ID_BITS;
 257         struct object_id oid;
 258 };
 259
 260 struct tag {
 261         struct tag *next_tag;
 262         const char *name;
 263         unsigned int pack_id;
 264         struct object_id oid;
 265 };
 266
 267 struct hash_list {
 268         struct hash_list *next;
 269         struct object_id oid;
 270 };
 271
 272 typedef enum {
 273         WHENSPEC_RAW = 1,
 274         WHENSPEC_RFC2822,
 275         WHENSPEC_NOW
 276 } whenspec_type;
 277
 278 struct recent_command {
 279         struct recent_command *prev;
 280         struct recent_command *next;
 281         char *buf;
 282 };
 283
 284 /* Configured limits on output */
 285 static unsigned long max_depth = 50;
 286 static off_t max_packsize;
 287 static int unpack_limit = 100;
 288 static int force_update;
 289
 290 /* Stats and misc. counters */
 291 static uintmax_t alloc_count;
 292 static uintmax_t marks_set_count;
 293 static uintmax_t object_count_by_type[1 << TYPE_BITS];
 294 static uintmax_t duplicate_count_by_type[1 << TYPE_BITS];
 295 static uintmax_t delta_count_by_type[1 << TYPE_BITS];
 296 static uintmax_t delta_count_attempts_by_type[1 << TYPE_BITS];
 297 static unsigned long object_count;
 298 static unsigned long branch_count;
 299 static unsigned long branch_load_count;
 300 static int failure;
 301 static FILE *pack_edges;
 302 static unsigned int show_stats = 1;
 303 static int global_argc;
 304 static const char **global_argv;
 305
 306 /* Memory pools */
 307 static size_t mem_pool_alloc = 2*1024*1024 - sizeof(struct mem_pool);
 308 static size_t total_allocd;
 309 static struct mem_pool *mem_pool;
 310
 311 /* Atom management */
 312 static unsigned int atom_table_sz = 4451;
 313 static unsigned int atom_cnt;
 314 static struct atom_str **atom_table;
 315
 316 /* The .pack file being generated */
 317 static struct pack_idx_option pack_idx_opts;
 318 static unsigned int pack_id;
 319 static struct sha1file *pack_file;
 320 static struct packed_git *pack_data;
 321 static struct packed_git **all_packs;
 322 static off_t pack_size;
 323
 324 /* Table of objects we've written. */
 325 static unsigned int object_entry_alloc = 5000;
 326 static struct object_entry_pool *blocks;
 327 static struct object_entry *object_table[1 << 16];
 328 static struct mark_set *marks;
 329 static const char *export_marks_file;
 330 static const char *import_marks_file;
 331 static int import_marks_file_from_stream;
 332 static int import_marks_file_ignore_missing;
 333 static int import_marks_file_done;
 334 static int relative_marks_paths;
 335
 336 /* Our last blob */
 337 static struct last_object last_blob = { STRBUF_INIT, 0, 0, 0 };
 338
 339 /* Tree management */
 340 static unsigned int tree_entry_alloc = 1000;
 341 static void *avail_tree_entry;
 342 static unsigned int avail_tree_table_sz = 100;
 343 static struct avail_tree_content **avail_tree_table;
 344 static struct strbuf old_tree = STRBUF_INIT;
 345 static struct strbuf new_tree = STRBUF_INIT;
 346
 347 /* Branch data */
 348 static unsigned long max_active_branches = 5;
 349 static unsigned long cur_active_branches;
 350 static unsigned long branch_table_sz = 1039;
 351 static struct branch **branch_table;
 352 static struct branch *active_branches;
 353
 354 /* Tag data */
 355 static struct tag *first_tag;
 356 static struct tag *last_tag;
 357
 358 /* Input stream parsing */
 359 static whenspec_type whenspec = WHENSPEC_RAW;
 360 static struct strbuf command_buf = STRBUF_INIT;
 361 static int unread_command_buf;
 362 static struct recent_command cmd_hist = {&cmd_hist, &cmd_hist, NULL};
 363 static struct recent_command *cmd_tail = &cmd_hist;
 364 static struct recent_command *rc_free;
 365 static unsigned int cmd_save = 100;
 366 static uintmax_t next_mark;
 367 static struct strbuf new_data = STRBUF_INIT;
 368 static int seen_data_command;
 369 static int require_explicit_termination;
 370 static int allow_unsafe_features;
 371
 372 /* Signal handling */
 373 static volatile sig_atomic_t checkpoint_requested;
 374
 375 /* Where to write output of cat-blob commands */
 376 static int cat_blob_fd = STDOUT_FILENO;
 377
 378 static void parse_argv(void);
 379 static void parse_get_mark(const char *p);
 380 static void parse_cat_blob(const char *p);
 381 static void parse_ls(const char *p, struct branch *b);
 382
 383 static void write_branch_report(FILE *rpt, struct branch *b)
 384 {
 385         fprintf(rpt, "%s:\n", b->name);
 386
 387         fprintf(rpt, "  status      :");
 388         if (b->active)
 389                 fputs(" active", rpt);
 390         if (b->branch_tree.tree)
 391                 fputs(" loaded", rpt);
 392         if (is_null_oid(&b->branch_tree.versions[1].oid))
 393                 fputs(" dirty", rpt);
 394         fputc('\n', rpt);
 395
 396         fprintf(rpt, "  tip commit  : %s\n", oid_to_hex(&b->oid));
 397         fprintf(rpt, "  old tree    : %s\n",
 398                 oid_to_hex(&b->branch_tree.versions[0].oid));
 399         fprintf(rpt, "  cur tree    : %s\n",
 400                 oid_to_hex(&b->branch_tree.versions[1].oid));
 401         fprintf(rpt, "  commit clock: %" PRIuMAX "\n", b->last_commit);
 402
 403         fputs("  last pack   : ", rpt);
 404         if (b->pack_id < MAX_PACK_ID)
 405                 fprintf(rpt, "%u", b->pack_id);
 406         fputc('\n', rpt);
 407
 408         fputc('\n', rpt);
 409 }
 410
 411 static void dump_marks_helper(FILE *, uintmax_t, struct mark_set *);
 412
 413 static void write_crash_report(const char *err)
 414 {
 415         char *loc = git_pathdup("fast_import_crash_%"PRIuMAX, (uintmax_t) getpid());
 416         FILE *rpt = fopen(loc, "w");
 417         struct branch *b;
 418         unsigned long lu;
 419         struct recent_command *rc;
 420
 421         if (!rpt) {
 422                 error_errno("can't write crash report %s", loc);
 423                 free(loc);
 424                 return;
 425         }
 426
 427         fprintf(stderr, "fast-import: dumping crash report to %s\n", loc);
 428
 429         fprintf(rpt, "fast-import crash report:\n");
 430         fprintf(rpt, "    fast-import process: %"PRIuMAX"\n", (uintmax_t) getpid());
 431         fprintf(rpt, "    parent process     : %"PRIuMAX"\n", (uintmax_t) getppid());
 432         fprintf(rpt, "    at %s\n", show_date(time(NULL), 0, DATE_MODE(ISO8601)));
 433         fputc('\n', rpt);
 434
 435         fputs("fatal: ", rpt);
 436         fputs(err, rpt);
 437         fputc('\n', rpt);
 438
 439         fputc('\n', rpt);
 440         fputs("Most Recent Commands Before Crash\n", rpt);
 441         fputs("---------------------------------\n", rpt);
 442         for (rc = cmd_hist.next; rc != &cmd_hist; rc = rc->next) {
 443                 if (rc->next == &cmd_hist)
 444                         fputs("* ", rpt);
 445                 else
 446                         fputs("  ", rpt);
 447                 fputs(rc->buf, rpt);
 448                 fputc('\n', rpt);
 449         }
 450
 451         fputc('\n', rpt);
 452         fputs("Active Branch LRU\n", rpt);
 453         fputs("-----------------\n", rpt);
 454         fprintf(rpt, "    active_branches = %lu cur, %lu max\n",
 455                 cur_active_branches,
 456                 max_active_branches);
 457         fputc('\n', rpt);
 458         fputs("  pos  clock name\n", rpt);
 459         fputs("  ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~\n", rpt);
 460         for (b = active_branches, lu = 0; b; b = b->active_next_branch)
 461                 fprintf(rpt, "  %2lu) %6" PRIuMAX" %s\n",
 462                         ++lu, b->last_commit, b->name);
 463
 464         fputc('\n', rpt);
 465         fputs("Inactive Branches\n", rpt);
 466         fputs("-----------------\n", rpt);
 467         for (lu = 0; lu < branch_table_sz; lu++) {
 468                 for (b = branch_table[lu]; b; b = b->table_next_branch)
 469                         write_branch_report(rpt, b);
 470         }
 471
 472         if (first_tag) {
 473                 struct tag *tg;
 474                 fputc('\n', rpt);
 475                 fputs("Annotated Tags\n", rpt);
 476                 fputs("--------------\n", rpt);
 477                 for (tg = first_tag; tg; tg = tg->next_tag) {
 478                         fputs(oid_to_hex(&tg->oid), rpt);
 479                         fputc(' ', rpt);
 480                         fputs(tg->name, rpt);
 481                         fputc('\n', rpt);
 482                 }
 483         }
 484
 485         fputc('\n', rpt);
 486         fputs("Marks\n", rpt);
 487         fputs("-----\n", rpt);
 488         if (export_marks_file)
 489                 fprintf(rpt, "  exported to %s\n", export_marks_file);
 490         else
 491                 dump_marks_helper(rpt, 0, marks);
 492
 493         fputc('\n', rpt);
 494         fputs("-------------------\n", rpt);
 495         fputs("END OF CRASH REPORT\n", rpt);
 496         fclose(rpt);
 497         free(loc);
 498 }
 499
 500 static void end_packfile(void);
 501 static void unkeep_all_packs(void);
 502 static void dump_marks(void);
 503
 504 static NORETURN void die_nicely(const char *err, va_list params)
 505 {
 506         static int zombie;
 507         char message[2 * PATH_MAX];
 508
 509         vsnprintf(message, sizeof(message), err, params);
 510         fputs("fatal: ", stderr);
 511         fputs(message, stderr);
 512         fputc('\n', stderr);
 513
 514         if (!zombie) {
 515                 zombie = 1;
 516                 write_crash_report(message);
 517                 end_packfile();
 518                 unkeep_all_packs();
 519                 dump_marks();
 520         }
 521         exit(128);
 522 }
 523
 524 #ifndef SIGUSR1 /* Windows, for example */
 525
 526 static void set_checkpoint_signal(void)
 527 {
 528 }
 529
 530 #else
 531
 532 static void checkpoint_signal(int signo)
 533 {
 534         checkpoint_requested = 1;
 535 }
 536
 537 static void set_checkpoint_signal(void)
 538 {
 539         struct sigaction sa;
 540
 541         memset(&sa, 0, sizeof(sa));
 542         sa.sa_handler = checkpoint_signal;
 543         sigemptyset(&sa.sa_mask);
 544         sa.sa_flags = SA_RESTART;
 545         sigaction(SIGUSR1, &sa, NULL);
 546 }
 547
 548 #endif
 549
 550 static void alloc_objects(unsigned int cnt)
 551 {
 552         struct object_entry_pool *b;
 553
 554         b = xmalloc(sizeof(struct object_entry_pool)
 555                 + cnt * sizeof(struct object_entry));
 556         b->next_pool = blocks;
 557         b->next_free = b->entries;
 558         b->end = b->entries + cnt;
 559         blocks = b;
 560         alloc_count += cnt;
 561 }
 562
 563 static struct object_entry *new_object(struct object_id *oid)
 564 {
 565         struct object_entry *e;
 566
 567         if (blocks->next_free == blocks->end)
 568                 alloc_objects(object_entry_alloc);
 569
 570         e = blocks->next_free++;
 571         oidcpy(&e->idx.oid, oid);
 572         return e;
 573 }
 574
 575 static struct object_entry *find_object(struct object_id *oid)
 576 {
 577         unsigned int h = oid->hash[0] << 8 | oid->hash[1];
 578         struct object_entry *e;
 579         for (e = object_table[h]; e; e = e->next)
 580                 if (!oidcmp(oid, &e->idx.oid))
 581                         return e;
 582         return NULL;
 583 }
 584
 585 static struct object_entry *insert_object(struct object_id *oid)
 586 {
 587         unsigned int h = oid->hash[0] << 8 | oid->hash[1];
 588         struct object_entry *e = object_table[h];
 589
 590         while (e) {
 591                 if (!oidcmp(oid, &e->idx.oid))
 592                         return e;
 593                 e = e->next;
 594         }
 595
 596         e = new_object(oid);
 597         e->next = object_table[h];
 598         e->idx.offset = 0;
 599         object_table[h] = e;
 600         return e;
 601 }
 602
 603 static void invalidate_pack_id(unsigned int id)
 604 {
 605         unsigned int h;
 606         unsigned long lu;
 607         struct tag *t;
 608
 609         for (h = 0; h < ARRAY_SIZE(object_table); h++) {
 610                 struct object_entry *e;
 611
 612                 for (e = object_table[h]; e; e = e->next)
 613                         if (e->pack_id == id)
 614                                 e->pack_id = MAX_PACK_ID;
 615         }
 616
 617         for (lu = 0; lu < branch_table_sz; lu++) {
 618                 struct branch *b;
 619
 620                 for (b = branch_table[lu]; b; b = b->table_next_branch)
 621                         if (b->pack_id == id)
 622                                 b->pack_id = MAX_PACK_ID;
 623         }
 624
 625         for (t = first_tag; t; t = t->next_tag)
 626                 if (t->pack_id == id)
 627                         t->pack_id = MAX_PACK_ID;
 628 }
 629
 630 static unsigned int hc_str(const char *s, size_t len)
 631 {
 632         unsigned int r = 0;
 633         while (len-- > 0)
 634                 r = r * 31 + *s++;
 635         return r;
 636 }
 637
 638 static void *pool_alloc(size_t len)
 639 {
 640         struct mem_pool *p;
 641         void *r;
 642
 643         /* round up to a 'uintmax_t' alignment */
 644         if (len & (sizeof(uintmax_t) - 1))
 645                 len += sizeof(uintmax_t) - (len & (sizeof(uintmax_t) - 1));
 646
 647         for (p = mem_pool; p; p = p->next_pool)
 648                 if ((p->end - p->next_free >= len))
 649                         break;
 650
 651         if (!p) {
 652                 if (len >= (mem_pool_alloc/2)) {
 653                         total_allocd += len;
 654                         return xmalloc(len);
 655                 }
 656                 total_allocd += sizeof(struct mem_pool) + mem_pool_alloc;
 657                 p = xmalloc(st_add(sizeof(struct mem_pool), mem_pool_alloc));
 658                 p->next_pool = mem_pool;
 659                 p->next_free = (char *) p->space;
 660                 p->end = p->next_free + mem_pool_alloc;
 661                 mem_pool = p;
 662         }
 663
 664         r = p->next_free;
 665         p->next_free += len;
 666         return r;
 667 }
 668
 669 static void *pool_calloc(size_t count, size_t size)
 670 {
 671         size_t len = count * size;
 672         void *r = pool_alloc(len);
 673         memset(r, 0, len);
 674         return r;
 675 }
 676
 677 static char *pool_strdup(const char *s)
 678 {
 679         size_t len = strlen(s) + 1;
 680         char *r = pool_alloc(len);
 681         memcpy(r, s, len);
 682         return r;
 683 }
 684
 685 static void insert_mark(uintmax_t idnum, struct object_entry *oe)
 686 {
 687         struct mark_set *s = marks;
 688         while ((idnum >> s->shift) >= 1024) {
 689                 s = pool_calloc(1, sizeof(struct mark_set));
 690                 s->shift = marks->shift + 10;
 691                 s->data.sets[0] = marks;
 692                 marks = s;
 693         }
 694         while (s->shift) {
 695                 uintmax_t i = idnum >> s->shift;
 696                 idnum -= i << s->shift;
 697                 if (!s->data.sets[i]) {
 698                         s->data.sets[i] = pool_calloc(1, sizeof(struct mark_set));
 699                         s->data.sets[i]->shift = s->shift - 10;
 700                 }
 701                 s = s->data.sets[i];
 702         }
 703         if (!s->data.marked[idnum])
 704                 marks_set_count++;
 705         s->data.marked[idnum] = oe;
 706 }
 707
 708 static struct object_entry *find_mark(uintmax_t idnum)
 709 {
 710         uintmax_t orig_idnum = idnum;
 711         struct mark_set *s = marks;
 712         struct object_entry *oe = NULL;
 713         if ((idnum >> s->shift) < 1024) {
 714                 while (s && s->shift) {
 715                         uintmax_t i = idnum >> s->shift;
 716                         idnum -= i << s->shift;
 717                         s = s->data.sets[i];
 718                 }
 719                 if (s)
 720                         oe = s->data.marked[idnum];
 721         }
 722         if (!oe)
 723                 die("mark :%" PRIuMAX " not declared", orig_idnum);
 724         return oe;
 725 }
 726
 727 static struct atom_str *to_atom(const char *s, unsigned short len)
 728 {
 729         unsigned int hc = hc_str(s, len) % atom_table_sz;
 730         struct atom_str *c;
 731
 732         for (c = atom_table[hc]; c; c = c->next_atom)
 733                 if (c->str_len == len && !strncmp(s, c->str_dat, len))
 734                         return c;
 735
 736         c = pool_alloc(sizeof(struct atom_str) + len + 1);
 737         c->str_len = len;
 738         memcpy(c->str_dat, s, len);
 739         c->str_dat[len] = 0;
 740         c->next_atom = atom_table[hc];
 741         atom_table[hc] = c;
 742         atom_cnt++;
 743         return c;
 744 }
 745
 746 static struct branch *lookup_branch(const char *name)
 747 {
 748         unsigned int hc = hc_str(name, strlen(name)) % branch_table_sz;
 749         struct branch *b;
 750
 751         for (b = branch_table[hc]; b; b = b->table_next_branch)
 752                 if (!strcmp(name, b->name))
 753                         return b;
 754         return NULL;
 755 }
 756
 757 static struct branch *new_branch(const char *name)
 758 {
 759         unsigned int hc = hc_str(name, strlen(name)) % branch_table_sz;
 760         struct branch *b = lookup_branch(name);
 761
 762         if (b)
 763                 die("Invalid attempt to create duplicate branch: %s", name);
 764         if (check_refname_format(name, REFNAME_ALLOW_ONELEVEL))
 765                 die("Branch name doesn't conform to GIT standards: %s", name);
 766
 767         b = pool_calloc(1, sizeof(struct branch));
 768         b->name = pool_strdup(name);
 769         b->table_next_branch = branch_table[hc];
 770         b->branch_tree.versions[0].mode = S_IFDIR;
 771         b->branch_tree.versions[1].mode = S_IFDIR;
 772         b->num_notes = 0;
 773         b->active = 0;
 774         b->pack_id = MAX_PACK_ID;
 775         branch_table[hc] = b;
 776         branch_count++;
 777         return b;
 778 }
 779
 780 static unsigned int hc_entries(unsigned int cnt)
 781 {
 782         cnt = cnt & 7 ? (cnt / 8) + 1 : cnt / 8;
 783         return cnt < avail_tree_table_sz ? cnt : avail_tree_table_sz - 1;
 784 }
 785
 786 static struct tree_content *new_tree_content(unsigned int cnt)
 787 {
 788         struct avail_tree_content *f, *l = NULL;
 789         struct tree_content *t;
 790         unsigned int hc = hc_entries(cnt);
 791
 792         for (f = avail_tree_table[hc]; f; l = f, f = f->next_avail)
 793                 if (f->entry_capacity >= cnt)
 794                         break;
 795
 796         if (f) {
 797                 if (l)
 798                         l->next_avail = f->next_avail;
 799                 else
 800                         avail_tree_table[hc] = f->next_avail;
 801         } else {
 802                 cnt = cnt & 7 ? ((cnt / 8) + 1) * 8 : cnt;
 803                 f = pool_alloc(sizeof(*t) + sizeof(t->entries[0]) * cnt);
 804                 f->entry_capacity = cnt;
 805         }
 806
 807         t = (struct tree_content*)f;
 808         t->entry_count = 0;
 809         t->delta_depth = 0;
 810         return t;
 811 }
 812
 813 static void release_tree_entry(struct tree_entry *e);
 814 static void release_tree_content(struct tree_content *t)
 815 {
 816         struct avail_tree_content *f = (struct avail_tree_content*)t;
 817         unsigned int hc = hc_entries(f->entry_capacity);
 818         f->next_avail = avail_tree_table[hc];
 819         avail_tree_table[hc] = f;
 820 }
 821
 822 static void release_tree_content_recursive(struct tree_content *t)
 823 {
 824         unsigned int i;
 825         for (i = 0; i < t->entry_count; i++)
 826                 release_tree_entry(t->entries[i]);
 827         release_tree_content(t);
 828 }
 829
 830 static struct tree_content *grow_tree_content(
 831         struct tree_content *t,
 832         int amt)
 833 {
 834         struct tree_content *r = new_tree_content(t->entry_count + amt);
 835         r->entry_count = t->entry_count;
 836         r->delta_depth = t->delta_depth;
 837         memcpy(r->entries,t->entries,t->entry_count*sizeof(t->entries[0]));
 838         release_tree_content(t);
 839         return r;
 840 }
 841
 842 static struct tree_entry *new_tree_entry(void)
 843 {
 844         struct tree_entry *e;
 845
 846         if (!avail_tree_entry) {
 847                 unsigned int n = tree_entry_alloc;
 848                 total_allocd += n * sizeof(struct tree_entry);
 849                 ALLOC_ARRAY(e, n);
 850                 avail_tree_entry = e;
 851                 while (n-- > 1) {
 852                         *((void**)e) = e + 1;
 853                         e++;
 854                 }
 855                 *((void**)e) = NULL;
 856         }
 857
 858         e = avail_tree_entry;
 859         avail_tree_entry = *((void**)e);
 860         return e;
 861 }
 862
 863 static void release_tree_entry(struct tree_entry *e)
 864 {
 865         if (e->tree)
 866                 release_tree_content_recursive(e->tree);
 867         *((void**)e) = avail_tree_entry;
 868         avail_tree_entry = e;
 869 }
 870
 871 static struct tree_content *dup_tree_content(struct tree_content *s)
 872 {
 873         struct tree_content *d;
 874         struct tree_entry *a, *b;
 875         unsigned int i;
 876
 877         if (!s)
 878                 return NULL;
 879         d = new_tree_content(s->entry_count);
 880         for (i = 0; i < s->entry_count; i++) {
 881                 a = s->entries[i];
 882                 b = new_tree_entry();
 883                 memcpy(b, a, sizeof(*a));
 884                 if (a->tree && is_null_oid(&b->versions[1].oid))
 885                         b->tree = dup_tree_content(a->tree);
 886                 else
 887                         b->tree = NULL;
 888                 d->entries[i] = b;
 889         }
 890         d->entry_count = s->entry_count;
 891         d->delta_depth = s->delta_depth;
 892
 893         return d;
 894 }
 895
 896 static void start_packfile(void)
 897 {
 898         struct strbuf tmp_file = STRBUF_INIT;
 899         struct packed_git *p;
 900         struct pack_header hdr;
 901         int pack_fd;
 902
 903         pack_fd = odb_mkstemp(&tmp_file, "pack/tmp_pack_XXXXXX");
 904         FLEX_ALLOC_STR(p, pack_name, tmp_file.buf);
 905         strbuf_release(&tmp_file);
 906
 907         p->pack_fd = pack_fd;
 908         p->do_not_close = 1;
 909         pack_file = sha1fd(pack_fd, p->pack_name);
 910
 911         hdr.hdr_signature = htonl(PACK_SIGNATURE);
 912         hdr.hdr_version = htonl(2);
 913         hdr.hdr_entries = 0;
 914         sha1write(pack_file, &hdr, sizeof(hdr));
 915
 916         pack_data = p;
 917         pack_size = sizeof(hdr);
 918         object_count = 0;
 919
 920         REALLOC_ARRAY(all_packs, pack_id + 1);
 921         all_packs[pack_id] = p;
 922 }
 923
 924 static const char *create_index(void)
 925 {
 926         const char *tmpfile;
 927         struct pack_idx_entry **idx, **c, **last;
 928         struct object_entry *e;
 929         struct object_entry_pool *o;
 930
 931         /* Build the table of object IDs. */
 932         ALLOC_ARRAY(idx, object_count);
 933         c = idx;
 934         for (o = blocks; o; o = o->next_pool)
 935                 for (e = o->next_free; e-- != o->entries;)
 936                         if (pack_id == e->pack_id)
 937                                 *c++ = &e->idx;
 938         last = idx + object_count;
 939         if (c != last)
 940                 die("internal consistency error creating the index");
 941
 942         tmpfile = write_idx_file(NULL, idx, object_count, &pack_idx_opts, pack_data->sha1);
 943         free(idx);
 944         return tmpfile;
 945 }
 946
 947 static char *keep_pack(const char *curr_index_name)
 948 {
 949         static const char *keep_msg = "fast-import";
 950         struct strbuf name = STRBUF_INIT;
 951         int keep_fd;
 952
 953         odb_pack_name(&name, pack_data->sha1, "keep");
 954         keep_fd = odb_pack_keep(name.buf);
 955         if (keep_fd < 0)
 956                 die_errno("cannot create keep file");
 957         write_or_die(keep_fd, keep_msg, strlen(keep_msg));
 958         if (close(keep_fd))
 959                 die_errno("failed to write keep file");
 960
 961         odb_pack_name(&name, pack_data->sha1, "pack");
 962         if (finalize_object_file(pack_data->pack_name, name.buf))
 963                 die("cannot store pack file");
 964
 965         odb_pack_name(&name, pack_data->sha1, "idx");
 966         if (finalize_object_file(curr_index_name, name.buf))
 967                 die("cannot store index file");
 968         free((void *)curr_index_name);
 969         return strbuf_detach(&name, NULL);
 970 }
 971
 972 static void unkeep_all_packs(void)
 973 {
 974         struct strbuf name = STRBUF_INIT;
 975         int k;
 976
 977         for (k = 0; k < pack_id; k++) {
 978                 struct packed_git *p = all_packs[k];
 979                 odb_pack_name(&name, p->sha1, "keep");
 980                 unlink_or_warn(name.buf);
 981         }
 982         strbuf_release(&name);
 983 }
 984
 985 static int loosen_small_pack(const struct packed_git *p)
 986 {
 987         struct child_process unpack = CHILD_PROCESS_INIT;
 988
 989         if (lseek(p->pack_fd, 0, SEEK_SET) < 0)
 990                 die_errno("Failed seeking to start of '%s'", p->pack_name);
 991
 992         unpack.in = p->pack_fd;
 993         unpack.git_cmd = 1;
 994         unpack.stdout_to_stderr = 1;
 995         argv_array_push(&unpack.args, "unpack-objects");
 996         if (!show_stats)
 997                 argv_array_push(&unpack.args, "-q");
 998
 999         return run_command(&unpack);
1000 }
1001
1002 static void end_packfile(void)
1003 {
1004         static int running;
1005
1006         if (running || !pack_data)
1007                 return;
1008
1009         running = 1;
1010         clear_delta_base_cache();
1011         if (object_count) {
1012                 struct packed_git *new_p;
1013                 struct object_id cur_pack_oid;
1014                 char *idx_name;
1015                 int i;
1016                 struct branch *b;
1017                 struct tag *t;
1018
1019                 close_pack_windows(pack_data);
1020                 sha1close(pack_file, cur_pack_oid.hash, 0);
1021                 fixup_pack_header_footer(pack_data->pack_fd, pack_data->sha1,
1022                                     pack_data->pack_name, object_count,
1023                                     cur_pack_oid.hash, pack_size);
1024
1025                 if (object_count <= unpack_limit) {
1026                         if (!loosen_small_pack(pack_data)) {
1027                                 invalidate_pack_id(pack_id);
1028                                 goto discard_pack;
1029                         }
1030                 }
1031
1032                 close(pack_data->pack_fd);
1033                 idx_name = keep_pack(create_index());
1034
1035                 /* Register the packfile with core git's machinery. */
1036                 new_p = add_packed_git(idx_name, strlen(idx_name), 1);
1037                 if (!new_p)
1038                         die("core git rejected index %s", idx_name);
1039                 all_packs[pack_id] = new_p;
1040                 install_packed_git(new_p);
1041                 free(idx_name);
1042
1043                 /* Print the boundary */
1044                 if (pack_edges) {
1045                         fprintf(pack_edges, "%s:", new_p->pack_name);
1046                         for (i = 0; i < branch_table_sz; i++) {
1047                                 for (b = branch_table[i]; b; b = b->table_next_branch) {
1048                                         if (b->pack_id == pack_id)
1049                                                 fprintf(pack_edges, " %s",
1050                                                         oid_to_hex(&b->oid));
1051                                 }
1052                         }
1053                         for (t = first_tag; t; t = t->next_tag) {
1054                                 if (t->pack_id == pack_id)
1055                                         fprintf(pack_edges, " %s",
1056                                                 oid_to_hex(&t->oid));
1057                         }
1058                         fputc('\n', pack_edges);
1059                         fflush(pack_edges);
1060                 }
1061
1062                 pack_id++;
1063         }
1064         else {
1065 discard_pack:
1066                 close(pack_data->pack_fd);
1067                 unlink_or_warn(pack_data->pack_name);
1068         }
1069         FREE_AND_NULL(pack_data);
1070         running = 0;
1071
1072         /* We can't carry a delta across packfiles. */
1073         strbuf_release(&last_blob.data);
1074         last_blob.offset = 0;
1075         last_blob.depth = 0;
1076 }
1077
1078 static void cycle_packfile(void)
1079 {
1080         end_packfile();
1081         start_packfile();
1082 }
1083
1084 static int store_object(
1085         enum object_type type,
1086         struct strbuf *dat,
1087         struct last_object *last,
1088         struct object_id *oidout,
1089         uintmax_t mark)
1090 {
1091         void *out, *delta;
1092         struct object_entry *e;
1093         unsigned char hdr[96];
1094         struct object_id oid;
1095         unsigned long hdrlen, deltalen;
1096         git_SHA_CTX c;
1097         git_zstream s;
1098
1099         hdrlen = xsnprintf((char *)hdr, sizeof(hdr), "%s %lu",
1100                            typename(type), (unsigned long)dat->len) + 1;
1101         git_SHA1_Init(&c);
1102         git_SHA1_Update(&c, hdr, hdrlen);
1103         git_SHA1_Update(&c, dat->buf, dat->len);
1104         git_SHA1_Final(oid.hash, &c);
1105         if (oidout)
1106                 oidcpy(oidout, &oid);
1107
1108         e = insert_object(&oid);
1109         if (mark)
1110                 insert_mark(mark, e);
1111         if (e->idx.offset) {
1112                 duplicate_count_by_type[type]++;
1113                 return 1;
1114         } else if (find_sha1_pack(oid.hash, packed_git)) {
1115                 e->type = type;
1116                 e->pack_id = MAX_PACK_ID;
1117                 e->idx.offset = 1; /* just not zero! */
1118                 duplicate_count_by_type[type]++;
1119                 return 1;
1120         }
1121
1122         if (last && last->data.buf && last->depth < max_depth && dat->len > 20) {
1123                 delta_count_attempts_by_type[type]++;
1124                 delta = diff_delta(last->data.buf, last->data.len,
1125                         dat->buf, dat->len,
1126                         &deltalen, dat->len - 20);
1127         } else
1128                 delta = NULL;
1129
1130         git_deflate_init(&s, pack_compression_level);
1131         if (delta) {
1132                 s.next_in = delta;
1133                 s.avail_in = deltalen;
1134         } else {
1135                 s.next_in = (void *)dat->buf;
1136                 s.avail_in = dat->len;
1137         }
1138         s.avail_out = git_deflate_bound(&s, s.avail_in);
1139         s.next_out = out = xmalloc(s.avail_out);
1140         while (git_deflate(&s, Z_FINISH) == Z_OK)
1141                 ; /* nothing */
1142         git_deflate_end(&s);
1143
1144         /* Determine if we should auto-checkpoint. */
1145         if ((max_packsize && (pack_size + 60 + s.total_out) > max_packsize)
1146                 || (pack_size + 60 + s.total_out) < pack_size) {
1147
1148                 /* This new object needs to *not* have the current pack_id. */
1149                 e->pack_id = pack_id + 1;
1150                 cycle_packfile();
1151
1152                 /* We cannot carry a delta into the new pack. */
1153                 if (delta) {
1154                         FREE_AND_NULL(delta);
1155
1156                         git_deflate_init(&s, pack_compression_level);
1157                         s.next_in = (void *)dat->buf;
1158                         s.avail_in = dat->len;
1159                         s.avail_out = git_deflate_bound(&s, s.avail_in);
1160                         s.next_out = out = xrealloc(out, s.avail_out);
1161                         while (git_deflate(&s, Z_FINISH) == Z_OK)
1162                                 ; /* nothing */
1163                         git_deflate_end(&s);
1164                 }
1165         }
1166
1167         e->type = type;
1168         e->pack_id = pack_id;
1169         e->idx.offset = pack_size;
1170         object_count++;
1171         object_count_by_type[type]++;
1172
1173         crc32_begin(pack_file);
1174
1175         if (delta) {
1176                 off_t ofs = e->idx.offset - last->offset;
1177                 unsigned pos = sizeof(hdr) - 1;
1178
1179                 delta_count_by_type[type]++;
1180                 e->depth = last->depth + 1;
1181
1182                 hdrlen = encode_in_pack_object_header(hdr, sizeof(hdr),
1183                                                       OBJ_OFS_DELTA, deltalen);
1184                 sha1write(pack_file, hdr, hdrlen);
1185                 pack_size += hdrlen;
1186
1187                 hdr[pos] = ofs & 127;
1188                 while (ofs >>= 7)
1189                         hdr[--pos] = 128 | (--ofs & 127);
1190                 sha1write(pack_file, hdr + pos, sizeof(hdr) - pos);
1191                 pack_size += sizeof(hdr) - pos;
1192         } else {
1193                 e->depth = 0;
1194                 hdrlen = encode_in_pack_object_header(hdr, sizeof(hdr),
1195                                                       type, dat->len);
1196                 sha1write(pack_file, hdr, hdrlen);
1197                 pack_size += hdrlen;
1198         }
1199
1200         sha1write(pack_file, out, s.total_out);
1201         pack_size += s.total_out;
1202
1203         e->idx.crc32 = crc32_end(pack_file);
1204
1205         free(out);
1206         free(delta);
1207         if (last) {
1208                 if (last->no_swap) {
1209                         last->data = *dat;
1210                 } else {
1211                         strbuf_swap(&last->data, dat);
1212                 }
1213                 last->offset = e->idx.offset;
1214                 last->depth = e->depth;
1215         }
1216         return 0;
1217 }
1218
1219 static void truncate_pack(struct sha1file_checkpoint *checkpoint)
1220 {
1221         if (sha1file_truncate(pack_file, checkpoint))
1222                 die_errno("cannot truncate pack to skip duplicate");
1223         pack_size = checkpoint->offset;
1224 }
1225
1226 static void stream_blob(uintmax_t len, struct object_id *oidout, uintmax_t mark)
1227 {
1228         size_t in_sz = 64 * 1024, out_sz = 64 * 1024;
1229         unsigned char *in_buf = xmalloc(in_sz);
1230         unsigned char *out_buf = xmalloc(out_sz);
1231         struct object_entry *e;
1232         struct object_id oid;
1233         unsigned long hdrlen;
1234         off_t offset;
1235         git_SHA_CTX c;
1236         git_zstream s;
1237         struct sha1file_checkpoint checkpoint;
1238         int status = Z_OK;
1239
1240         /* Determine if we should auto-checkpoint. */
1241         if ((max_packsize && (pack_size + 60 + len) > max_packsize)
1242                 || (pack_size + 60 + len) < pack_size)
1243                 cycle_packfile();
1244
1245         sha1file_checkpoint(pack_file, &checkpoint);
1246         offset = checkpoint.offset;
1247
1248         hdrlen = xsnprintf((char *)out_buf, out_sz, "blob %" PRIuMAX, len) + 1;
1249
1250         git_SHA1_Init(&c);
1251         git_SHA1_Update(&c, out_buf, hdrlen);
1252
1253         crc32_begin(pack_file);
1254
1255         git_deflate_init(&s, pack_compression_level);
1256
1257         hdrlen = encode_in_pack_object_header(out_buf, out_sz, OBJ_BLOB, len);
1258
1259         s.next_out = out_buf + hdrlen;
1260         s.avail_out = out_sz - hdrlen;
1261
1262         while (status != Z_STREAM_END) {
1263                 if (0 < len && !s.avail_in) {
1264                         size_t cnt = in_sz < len ? in_sz : (size_t)len;
1265                         size_t n = fread(in_buf, 1, cnt, stdin);
1266                         if (!n && feof(stdin))
1267                                 die("EOF in data (%" PRIuMAX " bytes remaining)", len);
1268
1269                         git_SHA1_Update(&c, in_buf, n);
1270                         s.next_in = in_buf;
1271                         s.avail_in = n;
1272                         len -= n;
1273                 }
1274
1275                 status = git_deflate(&s, len ? 0 : Z_FINISH);
1276
1277                 if (!s.avail_out || status == Z_STREAM_END) {
1278                         size_t n = s.next_out - out_buf;
1279                         sha1write(pack_file, out_buf, n);
1280                         pack_size += n;
1281                         s.next_out = out_buf;
1282                         s.avail_out = out_sz;
1283                 }
1284
1285                 switch (status) {
1286                 case Z_OK:
1287                 case Z_BUF_ERROR:
1288                 case Z_STREAM_END:
1289                         continue;
1290                 default:
1291                         die("unexpected deflate failure: %d", status);
1292                 }
1293         }
1294         git_deflate_end(&s);
1295         git_SHA1_Final(oid.hash, &c);
1296
1297         if (oidout)
1298                 oidcpy(oidout, &oid);
1299
1300         e = insert_object(&oid);
1301
1302         if (mark)
1303                 insert_mark(mark, e);
1304
1305         if (e->idx.offset) {
1306                 duplicate_count_by_type[OBJ_BLOB]++;
1307                 truncate_pack(&checkpoint);
1308
1309         } else if (find_sha1_pack(oid.hash, packed_git)) {
1310                 e->type = OBJ_BLOB;
1311                 e->pack_id = MAX_PACK_ID;
1312                 e->idx.offset = 1; /* just not zero! */
1313                 duplicate_count_by_type[OBJ_BLOB]++;
1314                 truncate_pack(&checkpoint);
1315
1316         } else {
1317                 e->depth = 0;
1318                 e->type = OBJ_BLOB;
1319                 e->pack_id = pack_id;
1320                 e->idx.offset = offset;
1321                 e->idx.crc32 = crc32_end(pack_file);
1322                 object_count++;
1323                 object_count_by_type[OBJ_BLOB]++;
1324         }
1325
1326         free(in_buf);
1327         free(out_buf);
1328 }
1329
1330 /* All calls must be guarded by find_object() or find_mark() to
1331  * ensure the 'struct object_entry' passed was written by this
1332  * process instance.  We unpack the entry by the offset, avoiding
1333  * the need for the corresponding .idx file.  This unpacking rule
1334  * works because we only use OBJ_REF_DELTA within the packfiles
1335  * created by fast-import.
1336  *
1337  * oe must not be NULL.  Such an oe usually comes from giving
1338  * an unknown SHA-1 to find_object() or an undefined mark to
1339  * find_mark().  Callers must test for this condition and use
1340  * the standard read_sha1_file() when it happens.
1341  *
1342  * oe->pack_id must not be MAX_PACK_ID.  Such an oe is usually from
1343  * find_mark(), where the mark was reloaded from an existing marks
1344  * file and is referencing an object that this fast-import process
1345  * instance did not write out to a packfile.  Callers must test for
1346  * this condition and use read_sha1_file() instead.
1347  */
1348 static void *gfi_unpack_entry(
1349         struct object_entry *oe,
1350         unsigned long *sizep)
1351 {
1352         enum object_type type;
1353         struct packed_git *p = all_packs[oe->pack_id];
1354         if (p == pack_data && p->pack_size < (pack_size + 20)) {
1355                 /* The object is stored in the packfile we are writing to
1356                  * and we have modified it since the last time we scanned
1357                  * back to read a previously written object.  If an old
1358                  * window covered [p->pack_size, p->pack_size + 20) its
1359                  * data is stale and is not valid.  Closing all windows
1360                  * and updating the packfile length ensures we can read
1361                  * the newly written data.
1362                  */
1363                 close_pack_windows(p);
1364                 sha1flush(pack_file);
1365
1366                 /* We have to offer 20 bytes additional on the end of
1367                  * the packfile as the core unpacker code assumes the
1368                  * footer is present at the file end and must promise
1369                  * at least 20 bytes within any window it maps.  But
1370                  * we don't actually create the footer here.
1371                  */
1372                 p->pack_size = pack_size + 20;
1373         }
1374         return unpack_entry(p, oe->idx.offset, &type, sizep);
1375 }
1376
1377 static const char *get_mode(const char *str, uint16_t *modep)
1378 {
1379         unsigned char c;
1380         uint16_t mode = 0;
1381
1382         while ((c = *str++) != ' ') {
1383                 if (c < '0' || c > '7')
1384                         return NULL;
1385                 mode = (mode << 3) + (c - '0');
1386         }
1387         *modep = mode;
1388         return str;
1389 }
1390
1391 static void load_tree(struct tree_entry *root)
1392 {
1393         struct object_id *oid = &root->versions[1].oid;
1394         struct object_entry *myoe;
1395         struct tree_content *t;
1396         unsigned long size;
1397         char *buf;
1398         const char *c;
1399
1400         root->tree = t = new_tree_content(8);
1401         if (is_null_oid(oid))
1402                 return;
1403
1404         myoe = find_object(oid);
1405         if (myoe && myoe->pack_id != MAX_PACK_ID) {
1406                 if (myoe->type != OBJ_TREE)
1407                         die("Not a tree: %s", oid_to_hex(oid));
1408                 t->delta_depth = myoe->depth;
1409                 buf = gfi_unpack_entry(myoe, &size);
1410                 if (!buf)
1411                         die("Can't load tree %s", oid_to_hex(oid));
1412         } else {
1413                 enum object_type type;
1414                 buf = read_sha1_file(oid->hash, &type, &size);
1415                 if (!buf || type != OBJ_TREE)
1416                         die("Can't load tree %s", oid_to_hex(oid));
1417         }
1418
1419         c = buf;
1420         while (c != (buf + size)) {
1421                 struct tree_entry *e = new_tree_entry();
1422
1423                 if (t->entry_count == t->entry_capacity)
1424                         root->tree = t = grow_tree_content(t, t->entry_count);
1425                 t->entries[t->entry_count++] = e;
1426
1427                 e->tree = NULL;
1428                 c = get_mode(c, &e->versions[1].mode);
1429                 if (!c)
1430                         die("Corrupt mode in %s", oid_to_hex(oid));
1431                 e->versions[0].mode = e->versions[1].mode;
1432                 e->name = to_atom(c, strlen(c));
1433                 c += e->name->str_len + 1;
1434                 hashcpy(e->versions[0].oid.hash, (unsigned char *)c);
1435                 hashcpy(e->versions[1].oid.hash, (unsigned char *)c);
1436                 c += GIT_SHA1_RAWSZ;
1437         }
1438         free(buf);
1439 }
1440
1441 static int tecmp0 (const void *_a, const void *_b)
1442 {
1443         struct tree_entry *a = *((struct tree_entry**)_a);
1444         struct tree_entry *b = *((struct tree_entry**)_b);
1445         return base_name_compare(
1446                 a->name->str_dat, a->name->str_len, a->versions[0].mode,
1447                 b->name->str_dat, b->name->str_len, b->versions[0].mode);
1448 }
1449
1450 static int tecmp1 (const void *_a, const void *_b)
1451 {
1452         struct tree_entry *a = *((struct tree_entry**)_a);
1453         struct tree_entry *b = *((struct tree_entry**)_b);
1454         return base_name_compare(
1455                 a->name->str_dat, a->name->str_len, a->versions[1].mode,
1456                 b->name->str_dat, b->name->str_len, b->versions[1].mode);
1457 }
1458
1459 static void mktree(struct tree_content *t, int v, struct strbuf *b)
1460 {
1461         size_t maxlen = 0;
1462         unsigned int i;
1463
1464         if (!v)
1465                 QSORT(t->entries, t->entry_count, tecmp0);
1466         else
1467                 QSORT(t->entries, t->entry_count, tecmp1);
1468
1469         for (i = 0; i < t->entry_count; i++) {
1470                 if (t->entries[i]->versions[v].mode)
1471                         maxlen += t->entries[i]->name->str_len + 34;
1472         }
1473
1474         strbuf_reset(b);
1475         strbuf_grow(b, maxlen);
1476         for (i = 0; i < t->entry_count; i++) {
1477                 struct tree_entry *e = t->entries[i];
1478                 if (!e->versions[v].mode)
1479                         continue;
1480                 strbuf_addf(b, "%o %s%c",
1481                         (unsigned int)(e->versions[v].mode & ~NO_DELTA),
1482                         e->name->str_dat, '\0');
1483                 strbuf_add(b, e->versions[v].oid.hash, GIT_SHA1_RAWSZ);
1484         }
1485 }
1486
1487 static void store_tree(struct tree_entry *root)
1488 {
1489         struct tree_content *t;
1490         unsigned int i, j, del;
1491         struct last_object lo = { STRBUF_INIT, 0, 0, /* no_swap */ 1 };
1492         struct object_entry *le = NULL;
1493
1494         if (!is_null_oid(&root->versions[1].oid))
1495                 return;
1496
1497         if (!root->tree)
1498                 load_tree(root);
1499         t = root->tree;
1500
1501         for (i = 0; i < t->entry_count; i++) {
1502                 if (t->entries[i]->tree)
1503                         store_tree(t->entries[i]);
1504         }
1505
1506         if (!(root->versions[0].mode & NO_DELTA))
1507                 le = find_object(&root->versions[0].oid);
1508         if (S_ISDIR(root->versions[0].mode) && le && le->pack_id == pack_id) {
1509                 mktree(t, 0, &old_tree);
1510                 lo.data = old_tree;
1511                 lo.offset = le->idx.offset;
1512                 lo.depth = t->delta_depth;
1513         }
1514
1515         mktree(t, 1, &new_tree);
1516         store_object(OBJ_TREE, &new_tree, &lo, &root->versions[1].oid, 0);
1517
1518         t->delta_depth = lo.depth;
1519         for (i = 0, j = 0, del = 0; i < t->entry_count; i++) {
1520                 struct tree_entry *e = t->entries[i];
1521                 if (e->versions[1].mode) {
1522                         e->versions[0].mode = e->versions[1].mode;
1523                         oidcpy(&e->versions[0].oid, &e->versions[1].oid);
1524                         t->entries[j++] = e;
1525                 } else {
1526                         release_tree_entry(e);
1527                         del++;
1528                 }
1529         }
1530         t->entry_count -= del;
1531 }
1532
1533 static void tree_content_replace(
1534         struct tree_entry *root,
1535         const struct object_id *oid,
1536         const uint16_t mode,
1537         struct tree_content *newtree)
1538 {
1539         if (!S_ISDIR(mode))
1540                 die("Root cannot be a non-directory");
1541         oidclr(&root->versions[0].oid);
1542         oidcpy(&root->versions[1].oid, oid);
1543         if (root->tree)
1544                 release_tree_content_recursive(root->tree);
1545         root->tree = newtree;
1546 }
1547
1548 static int tree_content_set(
1549         struct tree_entry *root,
1550         const char *p,
1551         const struct object_id *oid,
1552         const uint16_t mode,
1553         struct tree_content *subtree)
1554 {
1555         struct tree_content *t;
1556         const char *slash1;
1557         unsigned int i, n;
1558         struct tree_entry *e;
1559
1560         slash1 = strchrnul(p, '/');
1561         n = slash1 - p;
1562         if (!n)
1563                 die("Empty path component found in input");
1564         if (!*slash1 && !S_ISDIR(mode) && subtree)
1565                 die("Non-directories cannot have subtrees");
1566
1567         if (!root->tree)
1568                 load_tree(root);
1569         t = root->tree;
1570         for (i = 0; i < t->entry_count; i++) {
1571                 e = t->entries[i];
1572                 if (e->name->str_len == n && !fspathncmp(p, e->name->str_dat, n)) {
1573                         if (!*slash1) {
1574                                 if (!S_ISDIR(mode)
1575                                                 && e->versions[1].mode == mode
1576                                                 && !oidcmp(&e->versions[1].oid, oid))
1577                                         return 0;
1578                                 e->versions[1].mode = mode;
1579                                 oidcpy(&e->versions[1].oid, oid);
1580                                 if (e->tree)
1581                                         release_tree_content_recursive(e->tree);
1582                                 e->tree = subtree;
1583
1584                                 /*
1585                                  * We need to leave e->versions[0].sha1 alone
1586                                  * to avoid modifying the preimage tree used
1587                                  * when writing out the parent directory.
1588                                  * But after replacing the subdir with a
1589                                  * completely different one, it's not a good
1590                                  * delta base any more, and besides, we've
1591                                  * thrown away the tree entries needed to
1592                                  * make a delta against it.
1593                                  *
1594                                  * So let's just explicitly disable deltas
1595                                  * for the subtree.
1596                                  */
1597                                 if (S_ISDIR(e->versions[0].mode))
1598                                         e->versions[0].mode |= NO_DELTA;
1599
1600                                 oidclr(&root->versions[1].oid);
1601                                 return 1;
1602                         }
1603                         if (!S_ISDIR(e->versions[1].mode)) {
1604                                 e->tree = new_tree_content(8);
1605                                 e->versions[1].mode = S_IFDIR;
1606                         }
1607                         if (!e->tree)
1608                                 load_tree(e);
1609                         if (tree_content_set(e, slash1 + 1, oid, mode, subtree)) {
1610                                 oidclr(&root->versions[1].oid);
1611                                 return 1;
1612                         }
1613                         return 0;
1614                 }
1615         }
1616
1617         if (t->entry_count == t->entry_capacity)
1618                 root->tree = t = grow_tree_content(t, t->entry_count);
1619         e = new_tree_entry();
1620         e->name = to_atom(p, n);
1621         e->versions[0].mode = 0;
1622         oidclr(&e->versions[0].oid);
1623         t->entries[t->entry_count++] = e;
1624         if (*slash1) {
1625                 e->tree = new_tree_content(8);
1626                 e->versions[1].mode = S_IFDIR;
1627                 tree_content_set(e, slash1 + 1, oid, mode, subtree);
1628         } else {
1629                 e->tree = subtree;
1630                 e->versions[1].mode = mode;
1631                 oidcpy(&e->versions[1].oid, oid);
1632         }
1633         oidclr(&root->versions[1].oid);
1634         return 1;
1635 }
1636
1637 static int tree_content_remove(
1638         struct tree_entry *root,
1639         const char *p,
1640         struct tree_entry *backup_leaf,
1641         int allow_root)
1642 {
1643         struct tree_content *t;
1644         const char *slash1;
1645         unsigned int i, n;
1646         struct tree_entry *e;
1647
1648         slash1 = strchrnul(p, '/');
1649         n = slash1 - p;
1650
1651         if (!root->tree)
1652                 load_tree(root);
1653
1654         if (!*p && allow_root) {
1655                 e = root;
1656                 goto del_entry;
1657         }
1658
1659         t = root->tree;
1660         for (i = 0; i < t->entry_count; i++) {
1661                 e = t->entries[i];
1662                 if (e->name->str_len == n && !fspathncmp(p, e->name->str_dat, n)) {
1663                         if (*slash1 && !S_ISDIR(e->versions[1].mode))
1664                                 /*
1665                                  * If p names a file in some subdirectory, and a
1666                                  * file or symlink matching the name of the
1667                                  * parent directory of p exists, then p cannot
1668                                  * exist and need not be deleted.
1669                                  */
1670                                 return 1;
1671                         if (!*slash1 || !S_ISDIR(e->versions[1].mode))
1672                                 goto del_entry;
1673                         if (!e->tree)
1674                                 load_tree(e);
1675                         if (tree_content_remove(e, slash1 + 1, backup_leaf, 0)) {
1676                                 for (n = 0; n < e->tree->entry_count; n++) {
1677                                         if (e->tree->entries[n]->versions[1].mode) {
1678                                                 oidclr(&root->versions[1].oid);
1679                                                 return 1;
1680                                         }
1681                                 }
1682                                 backup_leaf = NULL;
1683                                 goto del_entry;
1684                         }
1685                         return 0;
1686                 }
1687         }
1688         return 0;
1689
1690 del_entry:
1691         if (backup_leaf)
1692                 memcpy(backup_leaf, e, sizeof(*backup_leaf));
1693         else if (e->tree)
1694                 release_tree_content_recursive(e->tree);
1695         e->tree = NULL;
1696         e->versions[1].mode = 0;
1697         oidclr(&e->versions[1].oid);
1698         oidclr(&root->versions[1].oid);
1699         return 1;
1700 }
1701
1702 static int tree_content_get(
1703         struct tree_entry *root,
1704         const char *p,
1705         struct tree_entry *leaf,
1706         int allow_root)
1707 {
1708         struct tree_content *t;
1709         const char *slash1;
1710         unsigned int i, n;
1711         struct tree_entry *e;
1712
1713         slash1 = strchrnul(p, '/');
1714         n = slash1 - p;
1715         if (!n && !allow_root)
1716                 die("Empty path component found in input");
1717
1718         if (!root->tree)
1719                 load_tree(root);
1720
1721         if (!n) {
1722                 e = root;
1723                 goto found_entry;
1724         }
1725
1726         t = root->tree;
1727         for (i = 0; i < t->entry_count; i++) {
1728                 e = t->entries[i];
1729                 if (e->name->str_len == n && !fspathncmp(p, e->name->str_dat, n)) {
1730                         if (!*slash1)
1731                                 goto found_entry;
1732                         if (!S_ISDIR(e->versions[1].mode))
1733                                 return 0;
1734                         if (!e->tree)
1735                                 load_tree(e);
1736                         return tree_content_get(e, slash1 + 1, leaf, 0);
1737                 }
1738         }
1739         return 0;
1740
1741 found_entry:
1742         memcpy(leaf, e, sizeof(*leaf));
1743         if (e->tree && is_null_oid(&e->versions[1].oid))
1744                 leaf->tree = dup_tree_content(e->tree);
1745         else
1746                 leaf->tree = NULL;
1747         return 1;
1748 }
1749
1750 static int update_branch(struct branch *b)
1751 {
1752         static const char *msg = "fast-import";
1753         struct ref_transaction *transaction;
1754         struct object_id old_oid;
1755         struct strbuf err = STRBUF_INIT;
1756
1757         if (is_null_oid(&b->oid)) {
1758                 if (b->delete)
1759                         delete_ref(NULL, b->name, NULL, 0);
1760                 return 0;
1761         }
1762         if (read_ref(b->name, &old_oid))
1763                 oidclr(&old_oid);
1764         if (!force_update && !is_null_oid(&old_oid)) {
1765                 struct commit *old_cmit, *new_cmit;
1766
1767                 old_cmit = lookup_commit_reference_gently(&old_oid, 0);
1768                 new_cmit = lookup_commit_reference_gently(&b->oid, 0);
1769                 if (!old_cmit || !new_cmit)
1770                         return error("Branch %s is missing commits.", b->name);
1771
1772                 if (!in_merge_bases(old_cmit, new_cmit)) {
1773                         warning("Not updating %s"
1774                                 " (new tip %s does not contain %s)",
1775                                 b->name, oid_to_hex(&b->oid),
1776                                 oid_to_hex(&old_oid));
1777                         return -1;
1778                 }
1779         }
1780         transaction = ref_transaction_begin(&err);
1781         if (!transaction ||
1782             ref_transaction_update(transaction, b->name, &b->oid, &old_oid,
1783                                    0, msg, &err) ||
1784             ref_transaction_commit(transaction, &err)) {
1785                 ref_transaction_free(transaction);
1786                 error("%s", err.buf);
1787                 strbuf_release(&err);
1788                 return -1;
1789         }
1790         ref_transaction_free(transaction);
1791         strbuf_release(&err);
1792         return 0;
1793 }
1794
1795 static void dump_branches(void)
1796 {
1797         unsigned int i;
1798         struct branch *b;
1799
1800         for (i = 0; i < branch_table_sz; i++) {
1801                 for (b = branch_table[i]; b; b = b->table_next_branch)
1802                         failure |= update_branch(b);
1803         }
1804 }
1805
1806 static void dump_tags(void)
1807 {
1808         static const char *msg = "fast-import";
1809         struct tag *t;
1810         struct strbuf ref_name = STRBUF_INIT;
1811         struct strbuf err = STRBUF_INIT;
1812         struct ref_transaction *transaction;
1813
1814         transaction = ref_transaction_begin(&err);
1815         if (!transaction) {
1816                 failure |= error("%s", err.buf);
1817                 goto cleanup;
1818         }
1819         for (t = first_tag; t; t = t->next_tag) {
1820                 strbuf_reset(&ref_name);
1821                 strbuf_addf(&ref_name, "refs/tags/%s", t->name);
1822
1823                 if (ref_transaction_update(transaction, ref_name.buf,
1824                                            &t->oid, NULL, 0, msg, &err)) {
1825                         failure |= error("%s", err.buf);
1826                         goto cleanup;
1827                 }
1828         }
1829         if (ref_transaction_commit(transaction, &err))
1830                 failure |= error("%s", err.buf);
1831
1832  cleanup:
1833         ref_transaction_free(transaction);
1834         strbuf_release(&ref_name);
1835         strbuf_release(&err);
1836 }
1837
1838 static void dump_marks_helper(FILE *f,
1839         uintmax_t base,
1840         struct mark_set *m)
1841 {
1842         uintmax_t k;
1843         if (m->shift) {
1844                 for (k = 0; k < 1024; k++) {
1845                         if (m->data.sets[k])
1846                                 dump_marks_helper(f, base + (k << m->shift),
1847                                         m->data.sets[k]);
1848                 }
1849         } else {
1850                 for (k = 0; k < 1024; k++) {
1851                         if (m->data.marked[k])
1852                                 fprintf(f, ":%" PRIuMAX " %s\n", base + k,
1853                                         oid_to_hex(&m->data.marked[k]->idx.oid));
1854                 }
1855         }
1856 }
1857
1858 static void dump_marks(void)
1859 {
1860         static struct lock_file mark_lock;
1861         FILE *f;
1862
1863         if (!export_marks_file || (import_marks_file && !import_marks_file_done))
1864                 return;
1865
1866         if (safe_create_leading_directories_const(export_marks_file)) {
1867                 failure |= error_errno("unable to create leading directories of %s",
1868                                        export_marks_file);
1869                 return;
1870         }
1871
1872         if (hold_lock_file_for_update(&mark_lock, export_marks_file, 0) < 0) {
1873                 failure |= error_errno("Unable to write marks file %s",
1874                                        export_marks_file);
1875                 return;
1876         }
1877
1878         f = fdopen_lock_file(&mark_lock, "w");
1879         if (!f) {
1880                 int saved_errno = errno;
1881                 rollback_lock_file(&mark_lock);
1882                 failure |= error("Unable to write marks file %s: %s",
1883                         export_marks_file, strerror(saved_errno));
1884                 return;
1885         }
1886
1887         dump_marks_helper(f, 0, marks);
1888         if (commit_lock_file(&mark_lock)) {
1889                 failure |= error_errno("Unable to write file %s",
1890                                        export_marks_file);
1891                 return;
1892         }
1893 }
1894
1895 static void read_marks(void)
1896 {
1897         char line[512];
1898         FILE *f = fopen(import_marks_file, "r");
1899         if (f)
1900                 ;
1901         else if (import_marks_file_ignore_missing && errno == ENOENT)
1902                 goto done; /* Marks file does not exist */
1903         else
1904                 die_errno("cannot read '%s'", import_marks_file);
1905         while (fgets(line, sizeof(line), f)) {
1906                 uintmax_t mark;
1907                 char *end;
1908                 struct object_id oid;
1909                 struct object_entry *e;
1910
1911                 end = strchr(line, '\n');
1912                 if (line[0] != ':' || !end)
1913                         die("corrupt mark line: %s", line);
1914                 *end = 0;
1915                 mark = strtoumax(line + 1, &end, 10);
1916                 if (!mark || end == line + 1
1917                         || *end != ' ' || get_oid_hex(end + 1, &oid))
1918                         die("corrupt mark line: %s", line);
1919                 e = find_object(&oid);
1920                 if (!e) {
1921                         enum object_type type = sha1_object_info(oid.hash, NULL);
1922                         if (type < 0)
1923                                 die("object not found: %s", oid_to_hex(&oid));
1924                         e = insert_object(&oid);
1925                         e->type = type;
1926                         e->pack_id = MAX_PACK_ID;
1927                         e->idx.offset = 1; /* just not zero! */
1928                 }
1929                 insert_mark(mark, e);
1930         }
1931         fclose(f);
1932 done:
1933         import_marks_file_done = 1;
1934 }
1935
1936
1937 static int read_next_command(void)
1938 {
1939         static int stdin_eof = 0;
1940
1941         if (stdin_eof) {
1942                 unread_command_buf = 0;
1943                 return EOF;
1944         }
1945
1946         for (;;) {
1947                 const char *p;
1948
1949                 if (unread_command_buf) {
1950                         unread_command_buf = 0;
1951                 } else {
1952                         struct recent_command *rc;
1953
1954                         strbuf_detach(&command_buf, NULL);
1955                         stdin_eof = strbuf_getline_lf(&command_buf, stdin);
1956                         if (stdin_eof)
1957                                 return EOF;
1958
1959                         if (!seen_data_command
1960                                 && !starts_with(command_buf.buf, "feature ")
1961                                 && !starts_with(command_buf.buf, "option ")) {
1962                                 parse_argv();
1963                         }
1964
1965                         rc = rc_free;
1966                         if (rc)
1967                                 rc_free = rc->next;
1968                         else {
1969                                 rc = cmd_hist.next;
1970                                 cmd_hist.next = rc->next;
1971                                 cmd_hist.next->prev = &cmd_hist;
1972                                 free(rc->buf);
1973                         }
1974
1975                         rc->buf = command_buf.buf;
1976                         rc->prev = cmd_tail;
1977                         rc->next = cmd_hist.prev;
1978                         rc->prev->next = rc;
1979                         cmd_tail = rc;
1980                 }
1981                 if (skip_prefix(command_buf.buf, "get-mark ", &p)) {
1982                         parse_get_mark(p);
1983                         continue;
1984                 }
1985                 if (skip_prefix(command_buf.buf, "cat-blob ", &p)) {
1986                         parse_cat_blob(p);
1987                         continue;
1988                 }
1989                 if (command_buf.buf[0] == '#')
1990                         continue;
1991                 return 0;
1992         }
1993 }
1994
1995 static void skip_optional_lf(void)
1996 {
1997         int term_char = fgetc(stdin);
1998         if (term_char != '\n' && term_char != EOF)
1999                 ungetc(term_char, stdin);
2000 }
2001
2002 static void parse_mark(void)
2003 {
2004         const char *v;
2005         if (skip_prefix(command_buf.buf, "mark :", &v)) {
2006                 next_mark = strtoumax(v, NULL, 10);
2007                 read_next_command();
2008         }
2009         else
2010                 next_mark = 0;
2011 }
2012
2013 static int parse_data(struct strbuf *sb, uintmax_t limit, uintmax_t *len_res)
2014 {
2015         const char *data;
2016         strbuf_reset(sb);
2017
2018         if (!skip_prefix(command_buf.buf, "data ", &data))
2019                 die("Expected 'data n' command, found: %s", command_buf.buf);
2020
2021         if (skip_prefix(data, "<<", &data)) {
2022                 char *term = xstrdup(data);
2023                 size_t term_len = command_buf.len - (data - command_buf.buf);
2024
2025                 strbuf_detach(&command_buf, NULL);
2026                 for (;;) {
2027                         if (strbuf_getline_lf(&command_buf, stdin) == EOF)
2028                                 die("EOF in data (terminator '%s' not found)", term);
2029                         if (term_len == command_buf.len
2030                                 && !strcmp(term, command_buf.buf))
2031                                 break;
2032                         strbuf_addbuf(sb, &command_buf);
2033                         strbuf_addch(sb, '\n');
2034                 }
2035                 free(term);
2036         }
2037         else {
2038                 uintmax_t len = strtoumax(data, NULL, 10);
2039                 size_t n = 0, length = (size_t)len;
2040
2041                 if (limit && limit < len) {
2042                         *len_res = len;
2043                         return 0;
2044                 }
2045                 if (length < len)
2046                         die("data is too large to use in this context");
2047
2048                 while (n < length) {
2049                         size_t s = strbuf_fread(sb, length - n, stdin);
2050                         if (!s && feof(stdin))
2051                                 die("EOF in data (%lu bytes remaining)",
2052                                         (unsigned long)(length - n));
2053                         n += s;
2054                 }
2055         }
2056
2057         skip_optional_lf();
2058         return 1;
2059 }
2060
2061 static int validate_raw_date(const char *src, struct strbuf *result)
2062 {
2063         const char *orig_src = src;
2064         char *endp;
2065         unsigned long num;
2066
2067         errno = 0;
2068
2069         num = strtoul(src, &endp, 10);
2070         /* NEEDSWORK: perhaps check for reasonable values? */
2071         if (errno || endp == src || *endp != ' ')
2072                 return -1;
2073
2074         src = endp + 1;
2075         if (*src != '-' && *src != '+')
2076                 return -1;
2077
2078         num = strtoul(src + 1, &endp, 10);
2079         if (errno || endp == src + 1 || *endp || 1400 < num)
2080                 return -1;
2081
2082         strbuf_addstr(result, orig_src);
2083         return 0;
2084 }
2085
2086 static char *parse_ident(const char *buf)
2087 {
2088         const char *ltgt;
2089         size_t name_len;
2090         struct strbuf ident = STRBUF_INIT;
2091
2092         /* ensure there is a space delimiter even if there is no name */
2093         if (*buf == '<')
2094                 --buf;
2095
2096         ltgt = buf + strcspn(buf, "<>");
2097         if (*ltgt != '<')
2098                 die("Missing < in ident string: %s", buf);
2099         if (ltgt != buf && ltgt[-1] != ' ')
2100                 die("Missing space before < in ident string: %s", buf);
2101         ltgt = ltgt + 1 + strcspn(ltgt + 1, "<>");
2102         if (*ltgt != '>')
2103                 die("Missing > in ident string: %s", buf);
2104         ltgt++;
2105         if (*ltgt != ' ')
2106                 die("Missing space after > in ident string: %s", buf);
2107         ltgt++;
2108         name_len = ltgt - buf;
2109         strbuf_add(&ident, buf, name_len);
2110
2111         switch (whenspec) {
2112         case WHENSPEC_RAW:
2113                 if (validate_raw_date(ltgt, &ident) < 0)
2114                         die("Invalid raw date \"%s\" in ident: %s", ltgt, buf);
2115                 break;
2116         case WHENSPEC_RFC2822:
2117                 if (parse_date(ltgt, &ident) < 0)
2118                         die("Invalid rfc2822 date \"%s\" in ident: %s", ltgt, buf);
2119                 break;
2120         case WHENSPEC_NOW:
2121                 if (strcmp("now", ltgt))
2122                         die("Date in ident must be 'now': %s", buf);
2123                 datestamp(&ident);
2124                 break;
2125         }
2126
2127         return strbuf_detach(&ident, NULL);
2128 }
2129
2130 static void parse_and_store_blob(
2131         struct last_object *last,
2132         struct object_id *oidout,
2133         uintmax_t mark)
2134 {
2135         static struct strbuf buf = STRBUF_INIT;
2136         uintmax_t len;
2137
2138         if (parse_data(&buf, big_file_threshold, &len))
2139                 store_object(OBJ_BLOB, &buf, last, oidout, mark);
2140         else {
2141                 if (last) {
2142                         strbuf_release(&last->data);
2143                         last->offset = 0;
2144                         last->depth = 0;
2145                 }
2146                 stream_blob(len, oidout, mark);
2147                 skip_optional_lf();
2148         }
2149 }
2150
2151 static void parse_new_blob(void)
2152 {
2153         read_next_command();
2154         parse_mark();
2155         parse_and_store_blob(&last_blob, NULL, next_mark);
2156 }
2157
2158 static void unload_one_branch(void)
2159 {
2160         while (cur_active_branches
2161                 && cur_active_branches >= max_active_branches) {
2162                 uintmax_t min_commit = ULONG_MAX;
2163                 struct branch *e, *l = NULL, *p = NULL;
2164
2165                 for (e = active_branches; e; e = e->active_next_branch) {
2166                         if (e->last_commit < min_commit) {
2167                                 p = l;
2168                                 min_commit = e->last_commit;
2169                         }
2170                         l = e;
2171                 }
2172
2173                 if (p) {
2174                         e = p->active_next_branch;
2175                         p->active_next_branch = e->active_next_branch;
2176                 } else {
2177                         e = active_branches;
2178                         active_branches = e->active_next_branch;
2179                 }
2180                 e->active = 0;
2181                 e->active_next_branch = NULL;
2182                 if (e->branch_tree.tree) {
2183                         release_tree_content_recursive(e->branch_tree.tree);
2184                         e->branch_tree.tree = NULL;
2185                 }
2186                 cur_active_branches--;
2187         }
2188 }
2189
2190 static void load_branch(struct branch *b)
2191 {
2192         load_tree(&b->branch_tree);
2193         if (!b->active) {
2194                 b->active = 1;
2195                 b->active_next_branch = active_branches;
2196                 active_branches = b;
2197                 cur_active_branches++;
2198                 branch_load_count++;
2199         }
2200 }
2201
2202 static unsigned char convert_num_notes_to_fanout(uintmax_t num_notes)
2203 {
2204         unsigned char fanout = 0;
2205         while ((num_notes >>= 8))
2206                 fanout++;
2207         return fanout;
2208 }
2209
2210 static void construct_path_with_fanout(const char *hex_sha1,
2211                 unsigned char fanout, char *path)
2212 {
2213         unsigned int i = 0, j = 0;
2214         if (fanout >= 20)
2215                 die("Too large fanout (%u)", fanout);
2216         while (fanout) {
2217                 path[i++] = hex_sha1[j++];
2218                 path[i++] = hex_sha1[j++];
2219                 path[i++] = '/';
2220                 fanout--;
2221         }
2222         memcpy(path + i, hex_sha1 + j, GIT_SHA1_HEXSZ - j);
2223         path[i + GIT_SHA1_HEXSZ - j] = '\0';
2224 }
2225
2226 static uintmax_t do_change_note_fanout(
2227                 struct tree_entry *orig_root, struct tree_entry *root,
2228                 char *hex_oid, unsigned int hex_oid_len,
2229                 char *fullpath, unsigned int fullpath_len,
2230                 unsigned char fanout)
2231 {
2232         struct tree_content *t;
2233         struct tree_entry *e, leaf;
2234         unsigned int i, tmp_hex_oid_len, tmp_fullpath_len;
2235         uintmax_t num_notes = 0;
2236         struct object_id oid;
2237         char realpath[60];
2238
2239         if (!root->tree)
2240                 load_tree(root);
2241         t = root->tree;
2242
2243         for (i = 0; t && i < t->entry_count; i++) {
2244                 e = t->entries[i];
2245                 tmp_hex_oid_len = hex_oid_len + e->name->str_len;
2246                 tmp_fullpath_len = fullpath_len;
2247
2248                 /*
2249                  * We're interested in EITHER existing note entries (entries
2250                  * with exactly 40 hex chars in path, not including directory
2251                  * separators), OR directory entries that may contain note
2252                  * entries (with < 40 hex chars in path).
2253                  * Also, each path component in a note entry must be a multiple
2254                  * of 2 chars.
2255                  */
2256                 if (!e->versions[1].mode ||
2257                     tmp_hex_oid_len > GIT_SHA1_HEXSZ ||
2258                     e->name->str_len % 2)
2259                         continue;
2260
2261                 /* This _may_ be a note entry, or a subdir containing notes */
2262                 memcpy(hex_oid + hex_oid_len, e->name->str_dat,
2263                        e->name->str_len);
2264                 if (tmp_fullpath_len)
2265                         fullpath[tmp_fullpath_len++] = '/';
2266                 memcpy(fullpath + tmp_fullpath_len, e->name->str_dat,
2267                        e->name->str_len);
2268                 tmp_fullpath_len += e->name->str_len;
2269                 fullpath[tmp_fullpath_len] = '\0';
2270
2271                 if (tmp_hex_oid_len == GIT_SHA1_HEXSZ && !get_oid_hex(hex_oid, &oid)) {
2272                         /* This is a note entry */
2273                         if (fanout == 0xff) {
2274                                 /* Counting mode, no rename */
2275                                 num_notes++;
2276                                 continue;
2277                         }
2278                         construct_path_with_fanout(hex_oid, fanout, realpath);
2279                         if (!strcmp(fullpath, realpath)) {
2280                                 /* Note entry is in correct location */
2281                                 num_notes++;
2282                                 continue;
2283                         }
2284
2285                         /* Rename fullpath to realpath */
2286                         if (!tree_content_remove(orig_root, fullpath, &leaf, 0))
2287                                 die("Failed to remove path %s", fullpath);
2288                         tree_content_set(orig_root, realpath,
2289                                 &leaf.versions[1].oid,
2290                                 leaf.versions[1].mode,
2291                                 leaf.tree);
2292                 } else if (S_ISDIR(e->versions[1].mode)) {
2293                         /* This is a subdir that may contain note entries */
2294                         num_notes += do_change_note_fanout(orig_root, e,
2295                                 hex_oid, tmp_hex_oid_len,
2296                                 fullpath, tmp_fullpath_len, fanout);
2297                 }
2298
2299                 /* The above may have reallocated the current tree_content */
2300                 t = root->tree;
2301         }
2302         return num_notes;
2303 }
2304
2305 static uintmax_t change_note_fanout(struct tree_entry *root,
2306                 unsigned char fanout)
2307 {
2308         /*
2309          * The size of path is due to one slash between every two hex digits,
2310          * plus the terminating NUL.  Note that there is no slash at the end, so
2311          * the number of slashes is one less than half the number of hex
2312          * characters.
2313          */
2314         char hex_oid[GIT_MAX_HEXSZ], path[GIT_MAX_HEXSZ + (GIT_MAX_HEXSZ / 2) - 1 + 1];
2315         return do_change_note_fanout(root, root, hex_oid, 0, path, 0, fanout);
2316 }
2317
2318 /*
2319  * Given a pointer into a string, parse a mark reference:
2320  *
2321  *   idnum ::= ':' bigint;
2322  *
2323  * Return the first character after the value in *endptr.
2324  *
2325  * Complain if the following character is not what is expected,
2326  * either a space or end of the string.
2327  */
2328 static uintmax_t parse_mark_ref(const char *p, char **endptr)
2329 {
2330         uintmax_t mark;
2331
2332         assert(*p == ':');
2333         p++;
2334         mark = strtoumax(p, endptr, 10);
2335         if (*endptr == p)
2336                 die("No value after ':' in mark: %s", command_buf.buf);
2337         return mark;
2338 }
2339
2340 /*
2341  * Parse the mark reference, and complain if this is not the end of
2342  * the string.
2343  */
2344 static uintmax_t parse_mark_ref_eol(const char *p)
2345 {
2346         char *end;
2347         uintmax_t mark;
2348
2349         mark = parse_mark_ref(p, &end);
2350         if (*end != '\0')
2351                 die("Garbage after mark: %s", command_buf.buf);
2352         return mark;
2353 }
2354
2355 /*
2356  * Parse the mark reference, demanding a trailing space.  Return a
2357  * pointer to the space.
2358  */
2359 static uintmax_t parse_mark_ref_space(const char **p)
2360 {
2361         uintmax_t mark;
2362         char *end;
2363
2364         mark = parse_mark_ref(*p, &end);
2365         if (*end++ != ' ')
2366                 die("Missing space after mark: %s", command_buf.buf);
2367         *p = end;
2368         return mark;
2369 }
2370
2371 static void file_change_m(const char *p, struct branch *b)
2372 {
2373         static struct strbuf uq = STRBUF_INIT;
2374         const char *endp;
2375         struct object_entry *oe;
2376         struct object_id oid;
2377         uint16_t mode, inline_data = 0;
2378
2379         p = get_mode(p, &mode);
2380         if (!p)
2381                 die("Corrupt mode: %s", command_buf.buf);
2382         switch (mode) {
2383         case 0644:
2384         case 0755:
2385                 mode |= S_IFREG;
2386         case S_IFREG | 0644:
2387         case S_IFREG | 0755:
2388         case S_IFLNK:
2389         case S_IFDIR:
2390         case S_IFGITLINK:
2391                 /* ok */
2392                 break;
2393         default:
2394                 die("Corrupt mode: %s", command_buf.buf);
2395         }
2396
2397         if (*p == ':') {
2398                 oe = find_mark(parse_mark_ref_space(&p));
2399                 oidcpy(&oid, &oe->idx.oid);
2400         } else if (skip_prefix(p, "inline ", &p)) {
2401                 inline_data = 1;
2402                 oe = NULL; /* not used with inline_data, but makes gcc happy */
2403         } else {
2404                 if (parse_oid_hex(p, &oid, &p))
2405                         die("Invalid dataref: %s", command_buf.buf);
2406                 oe = find_object(&oid);
2407                 if (*p++ != ' ')
2408                         die("Missing space after SHA1: %s", command_buf.buf);
2409         }
2410
2411         strbuf_reset(&uq);
2412         if (!unquote_c_style(&uq, p, &endp)) {
2413                 if (*endp)
2414                         die("Garbage after path in: %s", command_buf.buf);
2415                 p = uq.buf;
2416         }
2417
2418         /* Git does not track empty, non-toplevel directories. */
2419         if (S_ISDIR(mode) && is_empty_tree_oid(&oid) && *p) {
2420                 tree_content_remove(&b->branch_tree, p, NULL, 0);
2421                 return;
2422         }
2423
2424         if (S_ISGITLINK(mode)) {
2425                 if (inline_data)
2426                         die("Git links cannot be specified 'inline': %s",
2427                                 command_buf.buf);
2428                 else if (oe) {
2429                         if (oe->type != OBJ_COMMIT)
2430                                 die("Not a commit (actually a %s): %s",
2431                                         typename(oe->type), command_buf.buf);
2432                 }
2433                 /*
2434                  * Accept the sha1 without checking; it expected to be in
2435                  * another repository.
2436                  */
2437         } else if (inline_data) {
2438                 if (S_ISDIR(mode))
2439                         die("Directories cannot be specified 'inline': %s",
2440                                 command_buf.buf);
2441                 if (p != uq.buf) {
2442                         strbuf_addstr(&uq, p);
2443                         p = uq.buf;
2444                 }
2445                 read_next_command();
2446                 parse_and_store_blob(&last_blob, &oid, 0);
2447         } else {
2448                 enum object_type expected = S_ISDIR(mode) ?
2449                                                 OBJ_TREE: OBJ_BLOB;
2450                 enum object_type type = oe ? oe->type :
2451                                         sha1_object_info(oid.hash, NULL);
2452                 if (type < 0)
2453                         die("%s not found: %s",
2454                                         S_ISDIR(mode) ?  "Tree" : "Blob",
2455                                         command_buf.buf);
2456                 if (type != expected)
2457                         die("Not a %s (actually a %s): %s",
2458                                 typename(expected), typename(type),
2459                                 command_buf.buf);
2460         }
2461
2462         if (!*p) {
2463                 tree_content_replace(&b->branch_tree, &oid, mode, NULL);
2464                 return;
2465         }
2466         tree_content_set(&b->branch_tree, p, &oid, mode, NULL);
2467 }
2468
2469 static void file_change_d(const char *p, struct branch *b)
2470 {
2471         static struct strbuf uq = STRBUF_INIT;
2472         const char *endp;
2473
2474         strbuf_reset(&uq);
2475         if (!unquote_c_style(&uq, p, &endp)) {
2476                 if (*endp)
2477                         die("Garbage after path in: %s", command_buf.buf);
2478                 p = uq.buf;
2479         }
2480         tree_content_remove(&b->branch_tree, p, NULL, 1);
2481 }
2482
2483 static void file_change_cr(const char *s, struct branch *b, int rename)
2484 {
2485         const char *d;
2486         static struct strbuf s_uq = STRBUF_INIT;
2487         static struct strbuf d_uq = STRBUF_INIT;
2488         const char *endp;
2489         struct tree_entry leaf;
2490
2491         strbuf_reset(&s_uq);
2492         if (!unquote_c_style(&s_uq, s, &endp)) {
2493                 if (*endp != ' ')
2494                         die("Missing space after source: %s", command_buf.buf);
2495         } else {
2496                 endp = strchr(s, ' ');
2497                 if (!endp)
2498                         die("Missing space after source: %s", command_buf.buf);
2499                 strbuf_add(&s_uq, s, endp - s);
2500         }
2501         s = s_uq.buf;
2502
2503         endp++;
2504         if (!*endp)
2505                 die("Missing dest: %s", command_buf.buf);
2506
2507         d = endp;
2508         strbuf_reset(&d_uq);
2509         if (!unquote_c_style(&d_uq, d, &endp)) {
2510                 if (*endp)
2511                         die("Garbage after dest in: %s", command_buf.buf);
2512                 d = d_uq.buf;
2513         }
2514
2515         memset(&leaf, 0, sizeof(leaf));
2516         if (rename)
2517                 tree_content_remove(&b->branch_tree, s, &leaf, 1);
2518         else
2519                 tree_content_get(&b->branch_tree, s, &leaf, 1);
2520         if (!leaf.versions[1].mode)
2521                 die("Path %s not in branch", s);
2522         if (!*d) {      /* C "path/to/subdir" "" */
2523                 tree_content_replace(&b->branch_tree,
2524                         &leaf.versions[1].oid,
2525                         leaf.versions[1].mode,
2526                         leaf.tree);
2527                 return;
2528         }
2529         tree_content_set(&b->branch_tree, d,
2530                 &leaf.versions[1].oid,
2531                 leaf.versions[1].mode,
2532                 leaf.tree);
2533 }
2534
2535 static void note_change_n(const char *p, struct branch *b, unsigned char *old_fanout)
2536 {
2537         static struct strbuf uq = STRBUF_INIT;
2538         struct object_entry *oe;
2539         struct branch *s;
2540         struct object_id oid, commit_oid;
2541         char path[60];
2542         uint16_t inline_data = 0;
2543         unsigned char new_fanout;
2544
2545         /*
2546          * When loading a branch, we don't traverse its tree to count the real
2547          * number of notes (too expensive to do this for all non-note refs).
2548          * This means that recently loaded notes refs might incorrectly have
2549          * b->num_notes == 0, and consequently, old_fanout might be wrong.
2550          *
2551          * Fix this by traversing the tree and counting the number of notes
2552          * when b->num_notes == 0. If the notes tree is truly empty, the
2553          * calculation should not take long.
2554          */
2555         if (b->num_notes == 0 && *old_fanout == 0) {
2556                 /* Invoke change_note_fanout() in "counting mode". */
2557                 b->num_notes = change_note_fanout(&b->branch_tree, 0xff);
2558                 *old_fanout = convert_num_notes_to_fanout(b->num_notes);
2559         }
2560
2561         /* Now parse the notemodify command. */
2562         /* <dataref> or 'inline' */
2563         if (*p == ':') {
2564                 oe = find_mark(parse_mark_ref_space(&p));
2565                 oidcpy(&oid, &oe->idx.oid);
2566         } else if (skip_prefix(p, "inline ", &p)) {
2567                 inline_data = 1;
2568                 oe = NULL; /* not used with inline_data, but makes gcc happy */
2569         } else {
2570                 if (parse_oid_hex(p, &oid, &p))
2571                         die("Invalid dataref: %s", command_buf.buf);
2572                 oe = find_object(&oid);
2573                 if (*p++ != ' ')
2574                         die("Missing space after SHA1: %s", command_buf.buf);
2575         }
2576
2577         /* <commit-ish> */
2578         s = lookup_branch(p);
2579         if (s) {
2580                 if (is_null_oid(&s->oid))
2581                         die("Can't add a note on empty branch.");
2582                 oidcpy(&commit_oid, &s->oid);
2583         } else if (*p == ':') {
2584                 uintmax_t commit_mark = parse_mark_ref_eol(p);
2585                 struct object_entry *commit_oe = find_mark(commit_mark);
2586                 if (commit_oe->type != OBJ_COMMIT)
2587                         die("Mark :%" PRIuMAX " not a commit", commit_mark);
2588                 oidcpy(&commit_oid, &commit_oe->idx.oid);
2589         } else if (!get_oid(p, &commit_oid)) {
2590                 unsigned long size;
2591                 char *buf = read_object_with_reference(commit_oid.hash,
2592                         commit_type, &size, commit_oid.hash);
2593                 if (!buf || size < 46)
2594                         die("Not a valid commit: %s", p);
2595                 free(buf);
2596         } else
2597                 die("Invalid ref name or SHA1 expression: %s", p);
2598
2599         if (inline_data) {
2600                 if (p != uq.buf) {
2601                         strbuf_addstr(&uq, p);
2602                         p = uq.buf;
2603                 }
2604                 read_next_command();
2605                 parse_and_store_blob(&last_blob, &oid, 0);
2606         } else if (oe) {
2607                 if (oe->type != OBJ_BLOB)
2608                         die("Not a blob (actually a %s): %s",
2609                                 typename(oe->type), command_buf.buf);
2610         } else if (!is_null_oid(&oid)) {
2611                 enum object_type type = sha1_object_info(oid.hash, NULL);
2612                 if (type < 0)
2613                         die("Blob not found: %s", command_buf.buf);
2614                 if (type != OBJ_BLOB)
2615                         die("Not a blob (actually a %s): %s",
2616                             typename(type), command_buf.buf);
2617         }
2618
2619         construct_path_with_fanout(oid_to_hex(&commit_oid), *old_fanout, path);
2620         if (tree_content_remove(&b->branch_tree, path, NULL, 0))
2621                 b->num_notes--;
2622
2623         if (is_null_oid(&oid))
2624                 return; /* nothing to insert */
2625
2626         b->num_notes++;
2627         new_fanout = convert_num_notes_to_fanout(b->num_notes);
2628         construct_path_with_fanout(oid_to_hex(&commit_oid), new_fanout, path);
2629         tree_content_set(&b->branch_tree, path, &oid, S_IFREG | 0644, NULL);
2630 }
2631
2632 static void file_change_deleteall(struct branch *b)
2633 {
2634         release_tree_content_recursive(b->branch_tree.tree);
2635         oidclr(&b->branch_tree.versions[0].oid);
2636         oidclr(&b->branch_tree.versions[1].oid);
2637         load_tree(&b->branch_tree);
2638         b->num_notes = 0;
2639 }
2640
2641 static void parse_from_commit(struct branch *b, char *buf, unsigned long size)
2642 {
2643         if (!buf || size < GIT_SHA1_HEXSZ + 6)
2644                 die("Not a valid commit: %s", oid_to_hex(&b->oid));
2645         if (memcmp("tree ", buf, 5)
2646                 || get_oid_hex(buf + 5, &b->branch_tree.versions[1].oid))
2647                 die("The commit %s is corrupt", oid_to_hex(&b->oid));
2648         oidcpy(&b->branch_tree.versions[0].oid,
2649                &b->branch_tree.versions[1].oid);
2650 }
2651
2652 static void parse_from_existing(struct branch *b)
2653 {
2654         if (is_null_oid(&b->oid)) {
2655                 oidclr(&b->branch_tree.versions[0].oid);
2656                 oidclr(&b->branch_tree.versions[1].oid);
2657         } else {
2658                 unsigned long size;
2659                 char *buf;
2660
2661                 buf = read_object_with_reference(b->oid.hash,
2662                                                  commit_type, &size,
2663                                                  b->oid.hash);
2664                 parse_from_commit(b, buf, size);
2665                 free(buf);
2666         }
2667 }
2668
2669 static int parse_from(struct branch *b)
2670 {
2671         const char *from;
2672         struct branch *s;
2673         struct object_id oid;
2674
2675         if (!skip_prefix(command_buf.buf, "from ", &from))
2676                 return 0;
2677
2678         oidcpy(&oid, &b->branch_tree.versions[1].oid);
2679
2680         s = lookup_branch(from);
2681         if (b == s)
2682                 die("Can't create a branch from itself: %s", b->name);
2683         else if (s) {
2684                 struct object_id *t = &s->branch_tree.versions[1].oid;
2685                 oidcpy(&b->oid, &s->oid);
2686                 oidcpy(&b->branch_tree.versions[0].oid, t);
2687                 oidcpy(&b->branch_tree.versions[1].oid, t);
2688         } else if (*from == ':') {
2689                 uintmax_t idnum = parse_mark_ref_eol(from);
2690                 struct object_entry *oe = find_mark(idnum);
2691                 if (oe->type != OBJ_COMMIT)
2692                         die("Mark :%" PRIuMAX " not a commit", idnum);
2693                 if (oidcmp(&b->oid, &oe->idx.oid)) {
2694                         oidcpy(&b->oid, &oe->idx.oid);
2695                         if (oe->pack_id != MAX_PACK_ID) {
2696                                 unsigned long size;
2697                                 char *buf = gfi_unpack_entry(oe, &size);
2698                                 parse_from_commit(b, buf, size);
2699                                 free(buf);
2700                         } else
2701                                 parse_from_existing(b);
2702                 }
2703         } else if (!get_oid(from, &b->oid)) {
2704                 parse_from_existing(b);
2705                 if (is_null_oid(&b->oid))
2706                         b->delete = 1;
2707         }
2708         else
2709                 die("Invalid ref name or SHA1 expression: %s", from);
2710
2711         if (b->branch_tree.tree && oidcmp(&oid, &b->branch_tree.versions[1].oid)) {
2712                 release_tree_content_recursive(b->branch_tree.tree);
2713                 b->branch_tree.tree = NULL;
2714         }
2715
2716         read_next_command();
2717         return 1;
2718 }
2719
2720 static struct hash_list *parse_merge(unsigned int *count)
2721 {
2722         struct hash_list *list = NULL, **tail = &list, *n;
2723         const char *from;
2724         struct branch *s;
2725
2726         *count = 0;
2727         while (skip_prefix(command_buf.buf, "merge ", &from)) {
2728                 n = xmalloc(sizeof(*n));
2729                 s = lookup_branch(from);
2730                 if (s)
2731                         oidcpy(&n->oid, &s->oid);
2732                 else if (*from == ':') {
2733                         uintmax_t idnum = parse_mark_ref_eol(from);
2734                         struct object_entry *oe = find_mark(idnum);
2735                         if (oe->type != OBJ_COMMIT)
2736                                 die("Mark :%" PRIuMAX " not a commit", idnum);
2737                         oidcpy(&n->oid, &oe->idx.oid);
2738                 } else if (!get_oid(from, &n->oid)) {
2739                         unsigned long size;
2740                         char *buf = read_object_with_reference(n->oid.hash,
2741                                 commit_type, &size, n->oid.hash);
2742                         if (!buf || size < 46)
2743                                 die("Not a valid commit: %s", from);
2744                         free(buf);
2745                 } else
2746                         die("Invalid ref name or SHA1 expression: %s", from);
2747
2748                 n->next = NULL;
2749                 *tail = n;
2750                 tail = &n->next;
2751
2752                 (*count)++;
2753                 read_next_command();
2754         }
2755         return list;
2756 }
2757
2758 static void parse_new_commit(const char *arg)
2759 {
2760         static struct strbuf msg = STRBUF_INIT;
2761         struct branch *b;
2762         char *author = NULL;
2763         char *committer = NULL;
2764         struct hash_list *merge_list = NULL;
2765         unsigned int merge_count;
2766         unsigned char prev_fanout, new_fanout;
2767         const char *v;
2768
2769         b = lookup_branch(arg);
2770         if (!b)
2771                 b = new_branch(arg);
2772
2773         read_next_command();
2774         parse_mark();
2775         if (skip_prefix(command_buf.buf, "author ", &v)) {
2776                 author = parse_ident(v);
2777                 read_next_command();
2778         }
2779         if (skip_prefix(command_buf.buf, "committer ", &v)) {
2780                 committer = parse_ident(v);
2781                 read_next_command();
2782         }
2783         if (!committer)
2784                 die("Expected committer but didn't get one");
2785         parse_data(&msg, 0, NULL);
2786         read_next_command();
2787         parse_from(b);
2788         merge_list = parse_merge(&merge_count);
2789
2790         /* ensure the branch is active/loaded */
2791         if (!b->branch_tree.tree || !max_active_branches) {
2792                 unload_one_branch();
2793                 load_branch(b);
2794         }
2795
2796         prev_fanout = convert_num_notes_to_fanout(b->num_notes);
2797
2798         /* file_change* */
2799         while (command_buf.len > 0) {
2800                 if (skip_prefix(command_buf.buf, "M ", &v))
2801                         file_change_m(v, b);
2802                 else if (skip_prefix(command_buf.buf, "D ", &v))
2803                         file_change_d(v, b);
2804                 else if (skip_prefix(command_buf.buf, "R ", &v))
2805                         file_change_cr(v, b, 1);
2806                 else if (skip_prefix(command_buf.buf, "C ", &v))
2807                         file_change_cr(v, b, 0);
2808                 else if (skip_prefix(command_buf.buf, "N ", &v))
2809                         note_change_n(v, b, &prev_fanout);
2810                 else if (!strcmp("deleteall", command_buf.buf))
2811                         file_change_deleteall(b);
2812                 else if (skip_prefix(command_buf.buf, "ls ", &v))
2813                         parse_ls(v, b);
2814                 else {
2815                         unread_command_buf = 1;
2816                         break;
2817                 }
2818                 if (read_next_command() == EOF)
2819                         break;
2820         }
2821
2822         new_fanout = convert_num_notes_to_fanout(b->num_notes);
2823         if (new_fanout != prev_fanout)
2824                 b->num_notes = change_note_fanout(&b->branch_tree, new_fanout);
2825
2826         /* build the tree and the commit */
2827         store_tree(&b->branch_tree);
2828         oidcpy(&b->branch_tree.versions[0].oid,
2829                &b->branch_tree.versions[1].oid);
2830
2831         strbuf_reset(&new_data);
2832         strbuf_addf(&new_data, "tree %s\n",
2833                 oid_to_hex(&b->branch_tree.versions[1].oid));
2834         if (!is_null_oid(&b->oid))
2835                 strbuf_addf(&new_data, "parent %s\n",
2836                             oid_to_hex(&b->oid));
2837         while (merge_list) {
2838                 struct hash_list *next = merge_list->next;
2839                 strbuf_addf(&new_data, "parent %s\n",
2840                             oid_to_hex(&merge_list->oid));
2841                 free(merge_list);
2842                 merge_list = next;
2843         }
2844         strbuf_addf(&new_data,
2845                 "author %s\n"
2846                 "committer %s\n"
2847                 "\n",
2848                 author ? author : committer, committer);
2849         strbuf_addbuf(&new_data, &msg);
2850         free(author);
2851         free(committer);
2852
2853         if (!store_object(OBJ_COMMIT, &new_data, NULL, &b->oid, next_mark))
2854                 b->pack_id = pack_id;
2855         b->last_commit = object_count_by_type[OBJ_COMMIT];
2856 }
2857
2858 static void parse_new_tag(const char *arg)
2859 {
2860         static struct strbuf msg = STRBUF_INIT;
2861         const char *from;
2862         char *tagger;
2863         struct branch *s;
2864         struct tag *t;
2865         uintmax_t from_mark = 0;
2866         struct object_id oid;
2867         enum object_type type;
2868         const char *v;
2869
2870         t = pool_alloc(sizeof(struct tag));
2871         memset(t, 0, sizeof(struct tag));
2872         t->name = pool_strdup(arg);
2873         if (last_tag)
2874                 last_tag->next_tag = t;
2875         else
2876                 first_tag = t;
2877         last_tag = t;
2878         read_next_command();
2879
2880         /* from ... */
2881         if (!skip_prefix(command_buf.buf, "from ", &from))
2882                 die("Expected from command, got %s", command_buf.buf);
2883         s = lookup_branch(from);
2884         if (s) {
2885                 if (is_null_oid(&s->oid))
2886                         die("Can't tag an empty branch.");
2887                 oidcpy(&oid, &s->oid);
2888                 type = OBJ_COMMIT;
2889         } else if (*from == ':') {
2890                 struct object_entry *oe;
2891                 from_mark = parse_mark_ref_eol(from);
2892                 oe = find_mark(from_mark);
2893                 type = oe->type;
2894                 oidcpy(&oid, &oe->idx.oid);
2895         } else if (!get_oid(from, &oid)) {
2896                 struct object_entry *oe = find_object(&oid);
2897                 if (!oe) {
2898                         type = sha1_object_info(oid.hash, NULL);
2899                         if (type < 0)
2900                                 die("Not a valid object: %s", from);
2901                 } else
2902                         type = oe->type;
2903         } else
2904                 die("Invalid ref name or SHA1 expression: %s", from);
2905         read_next_command();
2906
2907         /* tagger ... */
2908         if (skip_prefix(command_buf.buf, "tagger ", &v)) {
2909                 tagger = parse_ident(v);
2910                 read_next_command();
2911         } else
2912                 tagger = NULL;
2913
2914         /* tag payload/message */
2915         parse_data(&msg, 0, NULL);
2916
2917         /* build the tag object */
2918         strbuf_reset(&new_data);
2919
2920         strbuf_addf(&new_data,
2921                     "object %s\n"
2922                     "type %s\n"
2923                     "tag %s\n",
2924                     oid_to_hex(&oid), typename(type), t->name);
2925         if (tagger)
2926                 strbuf_addf(&new_data,
2927                             "tagger %s\n", tagger);
2928         strbuf_addch(&new_data, '\n');
2929         strbuf_addbuf(&new_data, &msg);
2930         free(tagger);
2931
2932         if (store_object(OBJ_TAG, &new_data, NULL, &t->oid, 0))
2933                 t->pack_id = MAX_PACK_ID;
2934         else
2935                 t->pack_id = pack_id;
2936 }
2937
2938 static void parse_reset_branch(const char *arg)
2939 {
2940         struct branch *b;
2941
2942         b = lookup_branch(arg);
2943         if (b) {
2944                 oidclr(&b->oid);
2945                 oidclr(&b->branch_tree.versions[0].oid);
2946                 oidclr(&b->branch_tree.versions[1].oid);
2947                 if (b->branch_tree.tree) {
2948                         release_tree_content_recursive(b->branch_tree.tree);
2949                         b->branch_tree.tree = NULL;
2950                 }
2951         }
2952         else
2953                 b = new_branch(arg);
2954         read_next_command();
2955         parse_from(b);
2956         if (command_buf.len > 0)
2957                 unread_command_buf = 1;
2958 }
2959
2960 static void cat_blob_write(const char *buf, unsigned long size)
2961 {
2962         if (write_in_full(cat_blob_fd, buf, size) < 0)
2963                 die_errno("Write to frontend failed");
2964 }
2965
2966 static void cat_blob(struct object_entry *oe, struct object_id *oid)
2967 {
2968         struct strbuf line = STRBUF_INIT;
2969         unsigned long size;
2970         enum object_type type = 0;
2971         char *buf;
2972
2973         if (!oe || oe->pack_id == MAX_PACK_ID) {
2974                 buf = read_sha1_file(oid->hash, &type, &size);
2975         } else {
2976                 type = oe->type;
2977                 buf = gfi_unpack_entry(oe, &size);
2978         }
2979
2980         /*
2981          * Output based on batch_one_object() from cat-file.c.
2982          */
2983         if (type <= 0) {
2984                 strbuf_reset(&line);
2985                 strbuf_addf(&line, "%s missing\n", oid_to_hex(oid));
2986                 cat_blob_write(line.buf, line.len);
2987                 strbuf_release(&line);
2988                 free(buf);
2989                 return;
2990         }
2991         if (!buf)
2992                 die("Can't read object %s", oid_to_hex(oid));
2993         if (type != OBJ_BLOB)
2994                 die("Object %s is a %s but a blob was expected.",
2995                     oid_to_hex(oid), typename(type));
2996         strbuf_reset(&line);
2997         strbuf_addf(&line, "%s %s %lu\n", oid_to_hex(oid),
2998                                                 typename(type), size);
2999         cat_blob_write(line.buf, line.len);
3000         strbuf_release(&line);
3001         cat_blob_write(buf, size);
3002         cat_blob_write("\n", 1);
3003         if (oe && oe->pack_id == pack_id) {
3004                 last_blob.offset = oe->idx.offset;
3005                 strbuf_attach(&last_blob.data, buf, size, size);
3006                 last_blob.depth = oe->depth;
3007         } else
3008                 free(buf);
3009 }
3010
3011 static void parse_get_mark(const char *p)
3012 {
3013         struct object_entry *oe = oe;
3014         char output[GIT_MAX_HEXSZ + 2];
3015
3016         /* get-mark SP <object> LF */
3017         if (*p != ':')
3018                 die("Not a mark: %s", p);
3019
3020         oe = find_mark(parse_mark_ref_eol(p));
3021         if (!oe)
3022                 die("Unknown mark: %s", command_buf.buf);
3023
3024         xsnprintf(output, sizeof(output), "%s\n", oid_to_hex(&oe->idx.oid));
3025         cat_blob_write(output, GIT_SHA1_HEXSZ + 1);
3026 }
3027
3028 static void parse_cat_blob(const char *p)
3029 {
3030         struct object_entry *oe = oe;
3031         struct object_id oid;
3032
3033         /* cat-blob SP <object> LF */
3034         if (*p == ':') {
3035                 oe = find_mark(parse_mark_ref_eol(p));
3036                 if (!oe)
3037                         die("Unknown mark: %s", command_buf.buf);
3038                 oidcpy(&oid, &oe->idx.oid);
3039         } else {
3040                 if (parse_oid_hex(p, &oid, &p))
3041                         die("Invalid dataref: %s", command_buf.buf);
3042                 if (*p)
3043                         die("Garbage after SHA1: %s", command_buf.buf);
3044                 oe = find_object(&oid);
3045         }
3046
3047         cat_blob(oe, &oid);
3048 }
3049
3050 static struct object_entry *dereference(struct object_entry *oe,
3051                                         struct object_id *oid)
3052 {
3053         unsigned long size;
3054         char *buf = NULL;
3055         if (!oe) {
3056                 enum object_type type = sha1_object_info(oid->hash, NULL);
3057                 if (type < 0)
3058                         die("object not found: %s", oid_to_hex(oid));
3059                 /* cache it! */
3060                 oe = insert_object(oid);
3061                 oe->type = type;
3062                 oe->pack_id = MAX_PACK_ID;
3063                 oe->idx.offset = 1;
3064         }
3065         switch (oe->type) {
3066         case OBJ_TREE:  /* easy case. */
3067                 return oe;
3068         case OBJ_COMMIT:
3069         case OBJ_TAG:
3070                 break;
3071         default:
3072                 die("Not a tree-ish: %s", command_buf.buf);
3073         }
3074
3075         if (oe->pack_id != MAX_PACK_ID) {       /* in a pack being written */
3076                 buf = gfi_unpack_entry(oe, &size);
3077         } else {
3078                 enum object_type unused;
3079                 buf = read_sha1_file(oid->hash, &unused, &size);
3080         }
3081         if (!buf)
3082                 die("Can't load object %s", oid_to_hex(oid));
3083
3084         /* Peel one layer. */
3085         switch (oe->type) {
3086         case OBJ_TAG:
3087                 if (size < GIT_SHA1_HEXSZ + strlen("object ") ||
3088                     get_oid_hex(buf + strlen("object "), oid))
3089                         die("Invalid SHA1 in tag: %s", command_buf.buf);
3090                 break;
3091         case OBJ_COMMIT:
3092                 if (size < GIT_SHA1_HEXSZ + strlen("tree ") ||
3093                     get_oid_hex(buf + strlen("tree "), oid))
3094                         die("Invalid SHA1 in commit: %s", command_buf.buf);
3095         }
3096
3097         free(buf);
3098         return find_object(oid);
3099 }
3100
3101 static struct object_entry *parse_treeish_dataref(const char **p)
3102 {
3103         struct object_id oid;
3104         struct object_entry *e;
3105
3106         if (**p == ':') {       /* <mark> */
3107                 e = find_mark(parse_mark_ref_space(p));
3108                 if (!e)
3109                         die("Unknown mark: %s", command_buf.buf);
3110                 oidcpy(&oid, &e->idx.oid);
3111         } else {        /* <sha1> */
3112                 if (parse_oid_hex(*p, &oid, p))
3113                         die("Invalid dataref: %s", command_buf.buf);
3114                 e = find_object(&oid);
3115                 if (*(*p)++ != ' ')
3116                         die("Missing space after tree-ish: %s", command_buf.buf);
3117         }
3118
3119         while (!e || e->type != OBJ_TREE)
3120                 e = dereference(e, &oid);
3121         return e;
3122 }
3123
3124 static void print_ls(int mode, const unsigned char *sha1, const char *path)
3125 {
3126         static struct strbuf line = STRBUF_INIT;
3127
3128         /* See show_tree(). */
3129         const char *type =
3130                 S_ISGITLINK(mode) ? commit_type :
3131                 S_ISDIR(mode) ? tree_type :
3132                 blob_type;
3133
3134         if (!mode) {
3135                 /* missing SP path LF */
3136                 strbuf_reset(&line);
3137                 strbuf_addstr(&line, "missing ");
3138                 quote_c_style(path, &line, NULL, 0);
3139                 strbuf_addch(&line, '\n');
3140         } else {
3141                 /* mode SP type SP object_name TAB path LF */
3142                 strbuf_reset(&line);
3143                 strbuf_addf(&line, "%06o %s %s\t",
3144                                 mode & ~NO_DELTA, type, sha1_to_hex(sha1));
3145                 quote_c_style(path, &line, NULL, 0);
3146                 strbuf_addch(&line, '\n');
3147         }
3148         cat_blob_write(line.buf, line.len);
3149 }
3150
3151 static void parse_ls(const char *p, struct branch *b)
3152 {
3153         struct tree_entry *root = NULL;
3154         struct tree_entry leaf = {NULL};
3155
3156         /* ls SP (<tree-ish> SP)? <path> */
3157         if (*p == '"') {
3158                 if (!b)
3159                         die("Not in a commit: %s", command_buf.buf);
3160                 root = &b->branch_tree;
3161         } else {
3162                 struct object_entry *e = parse_treeish_dataref(&p);
3163                 root = new_tree_entry();
3164                 oidcpy(&root->versions[1].oid, &e->idx.oid);
3165                 if (!is_null_oid(&root->versions[1].oid))
3166                         root->versions[1].mode = S_IFDIR;
3167                 load_tree(root);
3168         }
3169         if (*p == '"') {
3170                 static struct strbuf uq = STRBUF_INIT;
3171                 const char *endp;
3172                 strbuf_reset(&uq);
3173                 if (unquote_c_style(&uq, p, &endp))
3174                         die("Invalid path: %s", command_buf.buf);
3175                 if (*endp)
3176                         die("Garbage after path in: %s", command_buf.buf);
3177                 p = uq.buf;
3178         }
3179         tree_content_get(root, p, &leaf, 1);
3180         /*
3181          * A directory in preparation would have a sha1 of zero
3182          * until it is saved.  Save, for simplicity.
3183          */
3184         if (S_ISDIR(leaf.versions[1].mode))
3185                 store_tree(&leaf);
3186
3187         print_ls(leaf.versions[1].mode, leaf.versions[1].oid.hash, p);
3188         if (leaf.tree)
3189                 release_tree_content_recursive(leaf.tree);
3190         if (!b || root != &b->branch_tree)
3191                 release_tree_entry(root);
3192 }
3193
3194 static void checkpoint(void)
3195 {
3196         checkpoint_requested = 0;
3197         if (object_count) {
3198                 cycle_packfile();
3199         }
3200         dump_branches();
3201         dump_tags();
3202         dump_marks();
3203 }
3204
3205 static void parse_checkpoint(void)
3206 {
3207         checkpoint_requested = 1;
3208         skip_optional_lf();
3209 }
3210
3211 static void parse_progress(void)
3212 {
3213         fwrite(command_buf.buf, 1, command_buf.len, stdout);
3214         fputc('\n', stdout);
3215         fflush(stdout);
3216         skip_optional_lf();
3217 }
3218
3219 static char* make_fast_import_path(const char *path)
3220 {
3221         if (!relative_marks_paths || is_absolute_path(path))
3222                 return xstrdup(path);
3223         return git_pathdup("info/fast-import/%s", path);
3224 }
3225
3226 static void option_import_marks(const char *marks,
3227                                         int from_stream, int ignore_missing)
3228 {
3229         if (import_marks_file) {
3230                 if (from_stream)
3231                         die("Only one import-marks command allowed per stream");
3232
3233                 /* read previous mark file */
3234                 if(!import_marks_file_from_stream)
3235                         read_marks();
3236         }
3237
3238         import_marks_file = make_fast_import_path(marks);
3239         import_marks_file_from_stream = from_stream;
3240         import_marks_file_ignore_missing = ignore_missing;
3241 }
3242
3243 static void option_date_format(const char *fmt)
3244 {
3245         if (!strcmp(fmt, "raw"))
3246                 whenspec = WHENSPEC_RAW;
3247         else if (!strcmp(fmt, "rfc2822"))
3248                 whenspec = WHENSPEC_RFC2822;
3249         else if (!strcmp(fmt, "now"))
3250                 whenspec = WHENSPEC_NOW;
3251         else
3252                 die("unknown --date-format argument %s", fmt);
3253 }
3254
3255 static unsigned long ulong_arg(const char *option, const char *arg)
3256 {
3257         char *endptr;
3258         unsigned long rv = strtoul(arg, &endptr, 0);
3259         if (strchr(arg, '-') || endptr == arg || *endptr)
3260                 die("%s: argument must be a non-negative integer", option);
3261         return rv;
3262 }
3263
3264 static void option_depth(const char *depth)
3265 {
3266         max_depth = ulong_arg("--depth", depth);
3267         if (max_depth > MAX_DEPTH)
3268                 die("--depth cannot exceed %u", MAX_DEPTH);
3269 }
3270
3271 static void option_active_branches(const char *branches)
3272 {
3273         max_active_branches = ulong_arg("--active-branches", branches);
3274 }
3275
3276 static void option_export_marks(const char *marks)
3277 {
3278         export_marks_file = make_fast_import_path(marks);
3279 }
3280
3281 static void option_cat_blob_fd(const char *fd)
3282 {
3283         unsigned long n = ulong_arg("--cat-blob-fd", fd);
3284         if (n > (unsigned long) INT_MAX)
3285                 die("--cat-blob-fd cannot exceed %d", INT_MAX);
3286         cat_blob_fd = (int) n;
3287 }
3288
3289 static void option_export_pack_edges(const char *edges)
3290 {
3291         if (pack_edges)
3292                 fclose(pack_edges);
3293         pack_edges = xfopen(edges, "a");
3294 }
3295
3296 static int parse_one_option(const char *option)
3297 {
3298         if (skip_prefix(option, "max-pack-size=", &option)) {
3299                 unsigned long v;
3300                 if (!git_parse_ulong(option, &v))
3301                         return 0;
3302                 if (v < 8192) {
3303                         warning("max-pack-size is now in bytes, assuming --max-pack-size=%lum", v);
3304                         v *= 1024 * 1024;
3305                 } else if (v < 1024 * 1024) {
3306                         warning("minimum max-pack-size is 1 MiB");
3307                         v = 1024 * 1024;
3308                 }
3309                 max_packsize = v;
3310         } else if (skip_prefix(option, "big-file-threshold=", &option)) {
3311                 unsigned long v;
3312                 if (!git_parse_ulong(option, &v))
3313                         return 0;
3314                 big_file_threshold = v;
3315         } else if (skip_prefix(option, "depth=", &option)) {
3316                 option_depth(option);
3317         } else if (skip_prefix(option, "active-branches=", &option)) {
3318                 option_active_branches(option);
3319         } else if (skip_prefix(option, "export-pack-edges=", &option)) {
3320                 option_export_pack_edges(option);
3321         } else if (!strcmp(option, "quiet")) {
3322                 show_stats = 0;
3323         } else if (!strcmp(option, "stats")) {
3324                 show_stats = 1;
3325         } else if (!strcmp(option, "allow-unsafe-features")) {
3326                 ; /* already handled during early option parsing */
3327         } else {
3328                 return 0;
3329         }
3330
3331         return 1;
3332 }
3333
3334 static void check_unsafe_feature(const char *feature, int from_stream)
3335 {
3336         if (from_stream && !allow_unsafe_features)
3337                 die(_("feature '%s' forbidden in input without --allow-unsafe-features"),
3338                     feature);
3339 }
3340
3341 static int parse_one_feature(const char *feature, int from_stream)
3342 {
3343         const char *arg;
3344
3345         if (skip_prefix(feature, "date-format=", &arg)) {
3346                 option_date_format(arg);
3347         } else if (skip_prefix(feature, "import-marks=", &arg)) {
3348                 check_unsafe_feature("import-marks", from_stream);
3349                 option_import_marks(arg, from_stream, 0);
3350         } else if (skip_prefix(feature, "import-marks-if-exists=", &arg)) {
3351                 check_unsafe_feature("import-marks-if-exists", from_stream);
3352                 option_import_marks(arg, from_stream, 1);
3353         } else if (skip_prefix(feature, "export-marks=", &arg)) {
3354                 check_unsafe_feature(feature, from_stream);
3355                 option_export_marks(arg);
3356         } else if (!strcmp(feature, "get-mark")) {
3357                 ; /* Don't die - this feature is supported */
3358         } else if (!strcmp(feature, "cat-blob")) {
3359                 ; /* Don't die - this feature is supported */
3360         } else if (!strcmp(feature, "relative-marks")) {
3361                 relative_marks_paths = 1;
3362         } else if (!strcmp(feature, "no-relative-marks")) {
3363                 relative_marks_paths = 0;
3364         } else if (!strcmp(feature, "done")) {
3365                 require_explicit_termination = 1;
3366         } else if (!strcmp(feature, "force")) {
3367                 force_update = 1;
3368         } else if (!strcmp(feature, "notes") || !strcmp(feature, "ls")) {
3369                 ; /* do nothing; we have the feature */
3370         } else {
3371                 return 0;
3372         }
3373
3374         return 1;
3375 }
3376
3377 static void parse_feature(const char *feature)
3378 {
3379         if (seen_data_command)
3380                 die("Got feature command '%s' after data command", feature);
3381
3382         if (parse_one_feature(feature, 1))
3383                 return;
3384
3385         die("This version of fast-import does not support feature %s.", feature);
3386 }
3387
3388 static void parse_option(const char *option)
3389 {
3390         if (seen_data_command)
3391                 die("Got option command '%s' after data command", option);
3392
3393         if (parse_one_option(option))
3394                 return;
3395
3396         die("This version of fast-import does not support option: %s", option);
3397 }
3398
3399 static void git_pack_config(void)
3400 {
3401         int indexversion_value;
3402         int limit;
3403         unsigned long packsizelimit_value;
3404
3405         if (!git_config_get_ulong("pack.depth", &max_depth)) {
3406                 if (max_depth > MAX_DEPTH)
3407                         max_depth = MAX_DEPTH;
3408         }
3409         if (!git_config_get_int("pack.indexversion", &indexversion_value)) {
3410                 pack_idx_opts.version = indexversion_value;
3411                 if (pack_idx_opts.version > 2)
3412                         git_die_config("pack.indexversion",
3413                                         "bad pack.indexversion=%"PRIu32, pack_idx_opts.version);
3414         }
3415         if (!git_config_get_ulong("pack.packsizelimit", &packsizelimit_value))
3416                 max_packsize = packsizelimit_value;
3417
3418         if (!git_config_get_int("fastimport.unpacklimit", &limit))
3419                 unpack_limit = limit;
3420         else if (!git_config_get_int("transfer.unpacklimit", &limit))
3421                 unpack_limit = limit;
3422
3423         git_config(git_default_config, NULL);
3424 }
3425
3426 static const char fast_import_usage[] =
3427 "git fast-import [--date-format=<f>] [--max-pack-size=<n>] [--big-file-threshold=<n>] [--depth=<n>] [--active-branches=<n>] [--export-marks=<marks.file>]";
3428
3429 static void parse_argv(void)
3430 {
3431         unsigned int i;
3432
3433         for (i = 1; i < global_argc; i++) {
3434                 const char *a = global_argv[i];
3435
3436                 if (*a != '-' || !strcmp(a, "--"))
3437                         break;
3438
3439                 if (!skip_prefix(a, "--", &a))
3440                         die("unknown option %s", a);
3441
3442                 if (parse_one_option(a))
3443                         continue;
3444
3445                 if (parse_one_feature(a, 0))
3446                         continue;
3447
3448                 if (skip_prefix(a, "cat-blob-fd=", &a)) {
3449                         option_cat_blob_fd(a);
3450                         continue;
3451                 }
3452
3453                 die("unknown option --%s", a);
3454         }
3455         if (i != global_argc)
3456                 usage(fast_import_usage);
3457
3458         seen_data_command = 1;
3459         if (import_marks_file)
3460                 read_marks();
3461 }
3462
3463 int cmd_main(int argc, const char **argv)
3464 {
3465         unsigned int i;
3466
3467         if (argc == 2 && !strcmp(argv[1], "-h"))
3468                 usage(fast_import_usage);
3469
3470         setup_git_directory();
3471         reset_pack_idx_option(&pack_idx_opts);
3472         git_pack_config();
3473
3474         alloc_objects(object_entry_alloc);
3475         strbuf_init(&command_buf, 0);
3476         atom_table = xcalloc(atom_table_sz, sizeof(struct atom_str*));
3477         branch_table = xcalloc(branch_table_sz, sizeof(struct branch*));
3478         avail_tree_table = xcalloc(avail_tree_table_sz, sizeof(struct avail_tree_content*));
3479         marks = pool_calloc(1, sizeof(struct mark_set));
3480
3481         /*
3482          * We don't parse most options until after we've seen the set of
3483          * "feature" lines at the start of the stream (which allows the command
3484          * line to override stream data). But we must do an early parse of any
3485          * command-line options that impact how we interpret the feature lines.
3486          */
3487         for (i = 1; i < argc; i++) {
3488                 const char *arg = argv[i];
3489                 if (*arg != '-' || !strcmp(arg, "--"))
3490                         break;
3491                 if (!strcmp(arg, "--allow-unsafe-features"))
3492                         allow_unsafe_features = 1;
3493         }
3494
3495         global_argc = argc;
3496         global_argv = argv;
3497
3498         rc_free = pool_alloc(cmd_save * sizeof(*rc_free));
3499         for (i = 0; i < (cmd_save - 1); i++)
3500                 rc_free[i].next = &rc_free[i + 1];
3501         rc_free[cmd_save - 1].next = NULL;
3502
3503         prepare_packed_git();
3504         start_packfile();
3505         set_die_routine(die_nicely);
3506         set_checkpoint_signal();
3507         while (read_next_command() != EOF) {
3508                 const char *v;
3509                 if (!strcmp("blob", command_buf.buf))
3510                         parse_new_blob();
3511                 else if (skip_prefix(command_buf.buf, "ls ", &v))
3512                         parse_ls(v, NULL);
3513                 else if (skip_prefix(command_buf.buf, "commit ", &v))
3514                         parse_new_commit(v);
3515                 else if (skip_prefix(command_buf.buf, "tag ", &v))
3516                         parse_new_tag(v);
3517                 else if (skip_prefix(command_buf.buf, "reset ", &v))
3518                         parse_reset_branch(v);
3519                 else if (!strcmp("checkpoint", command_buf.buf))
3520                         parse_checkpoint();
3521                 else if (!strcmp("done", command_buf.buf))
3522                         break;
3523                 else if (starts_with(command_buf.buf, "progress "))
3524                         parse_progress();
3525                 else if (skip_prefix(command_buf.buf, "feature ", &v))
3526                         parse_feature(v);
3527                 else if (skip_prefix(command_buf.buf, "option git ", &v))
3528                         parse_option(v);
3529                 else if (starts_with(command_buf.buf, "option "))
3530                         /* ignore non-git options*/;
3531                 else
3532                         die("Unsupported command: %s", command_buf.buf);
3533
3534                 if (checkpoint_requested)
3535                         checkpoint();
3536         }
3537
3538         /* argv hasn't been parsed yet, do so */
3539         if (!seen_data_command)
3540                 parse_argv();
3541
3542         if (require_explicit_termination && feof(stdin))
3543                 die("stream ends early");
3544
3545         end_packfile();
3546
3547         dump_branches();
3548         dump_tags();
3549         unkeep_all_packs();
3550         dump_marks();
3551
3552         if (pack_edges)
3553                 fclose(pack_edges);
3554
3555         if (show_stats) {
3556                 uintmax_t total_count = 0, duplicate_count = 0;
3557                 for (i = 0; i < ARRAY_SIZE(object_count_by_type); i++)
3558                         total_count += object_count_by_type[i];
3559                 for (i = 0; i < ARRAY_SIZE(duplicate_count_by_type); i++)
3560                         duplicate_count += duplicate_count_by_type[i];
3561
3562                 fprintf(stderr, "%s statistics:\n", argv[0]);
3563                 fprintf(stderr, "---------------------------------------------------------------------\n");
3564                 fprintf(stderr, "Alloc'd objects: %10" PRIuMAX "\n", alloc_count);
3565                 fprintf(stderr, "Total objects:   %10" PRIuMAX " (%10" PRIuMAX " duplicates                  )\n", total_count, duplicate_count);
3566                 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]);
3567                 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]);
3568                 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]);
3569                 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]);
3570                 fprintf(stderr, "Total branches:  %10lu (%10lu loads     )\n", branch_count, branch_load_count);
3571                 fprintf(stderr, "      marks:     %10" PRIuMAX " (%10" PRIuMAX " unique    )\n", (((uintmax_t)1) << marks->shift) * 1024, marks_set_count);
3572                 fprintf(stderr, "      atoms:     %10u\n", atom_cnt);
3573                 fprintf(stderr, "Memory total:    %10" PRIuMAX " KiB\n", (total_allocd + alloc_count*sizeof(struct object_entry))/1024);
3574                 fprintf(stderr, "       pools:    %10lu KiB\n", (unsigned long)(total_allocd/1024));
3575                 fprintf(stderr, "     objects:    %10" PRIuMAX " KiB\n", (alloc_count*sizeof(struct object_entry))/1024);
3576                 fprintf(stderr, "---------------------------------------------------------------------\n");
3577                 pack_report();
3578                 fprintf(stderr, "---------------------------------------------------------------------\n");
3579                 fprintf(stderr, "\n");
3580         }
3581
3582         return failure ? 1 : 0;
3583 }