fast-import.c

   1 /*
   2 Format of STDIN stream:
   3
   4   stream ::= cmd*;
   5
   6   cmd ::= new_blob
   7         | new_commit
   8         | new_tag
   9         | reset_branch
  10         ;
  11
  12   new_blob ::= 'blob' lf
  13         mark?
  14     file_content;
  15   file_content ::= data;
  16
  17   new_commit ::= 'commit' sp ref_str lf
  18     mark?
  19     ('author' sp name '<' email '>' ts tz lf)?
  20     'committer' sp name '<' email '>' ts tz lf
  21     commit_msg
  22     ('from' sp (ref_str | hexsha1 | sha1exp_str | idnum) lf)?
  23     ('merge' sp (ref_str | hexsha1 | sha1exp_str | idnum) lf)*
  24     file_change*
  25     lf;
  26   commit_msg ::= data;
  27
  28   file_change ::= 'M' sp mode sp (hexsha1 | idnum) sp path_str lf
  29                 | 'D' sp path_str lf
  30                 ;
  31   mode ::= '644' | '755';
  32
  33   new_tag ::= 'tag' sp tag_str lf
  34     'from' sp (ref_str | hexsha1 | sha1exp_str | idnum) lf
  35         'tagger' sp name '<' email '>' ts tz lf
  36     tag_msg;
  37   tag_msg ::= data;
  38
  39   reset_branch ::= 'reset' sp ref_str lf
  40     ('from' sp (ref_str | hexsha1 | sha1exp_str | idnum) lf)?
  41     lf;
  42
  43   checkpoint ::= 'checkpoint' lf
  44     lf;
  45
  46      # note: the first idnum in a stream should be 1 and subsequent
  47      # idnums should not have gaps between values as this will cause
  48      # the stream parser to reserve space for the gapped values.  An
  49          # idnum can be updated in the future to a new object by issuing
  50      # a new mark directive with the old idnum.
  51          #
  52   mark ::= 'mark' sp idnum lf;
  53
  54      # note: declen indicates the length of binary_data in bytes.
  55      # declen does not include the lf preceeding or trailing the
  56      # binary data.
  57      #
  58   data ::= 'data' sp declen lf
  59     binary_data
  60         lf;
  61
  62      # note: quoted strings are C-style quoting supporting \c for
  63      # common escapes of 'c' (e..g \n, \t, \\, \") or \nnn where nnn
  64          # is the signed byte value in octal.  Note that the only
  65      # characters which must actually be escaped to protect the
  66      # stream formatting is: \, " and LF.  Otherwise these values
  67          # are UTF8.
  68      #
  69   ref_str     ::= ref     | '"' quoted(ref)     '"' ;
  70   sha1exp_str ::= sha1exp | '"' quoted(sha1exp) '"' ;
  71   tag_str     ::= tag     | '"' quoted(tag)     '"' ;
  72   path_str    ::= path    | '"' quoted(path)    '"' ;
  73
  74   declen ::= # unsigned 32 bit value, ascii base10 notation;
  75   binary_data ::= # file content, not interpreted;
  76
  77   sp ::= # ASCII space character;
  78   lf ::= # ASCII newline (LF) character;
  79
  80      # note: a colon (':') must precede the numerical value assigned to
  81          # an idnum.  This is to distinguish it from a ref or tag name as
  82      # GIT does not permit ':' in ref or tag strings.
  83          #
  84   idnum   ::= ':' declen;
  85   path    ::= # GIT style file path, e.g. "a/b/c";
  86   ref     ::= # GIT ref name, e.g. "refs/heads/MOZ_GECKO_EXPERIMENT";
  87   tag     ::= # GIT tag name, e.g. "FIREFOX_1_5";
  88   sha1exp ::= # Any valid GIT SHA1 expression;
  89   hexsha1 ::= # SHA1 in hexadecimal format;
  90
  91      # note: name and email are UTF8 strings, however name must not
  92          # contain '<' or lf and email must not contain any of the
  93      # following: '<', '>', lf.
  94          #
  95   name  ::= # valid GIT author/committer name;
  96   email ::= # valid GIT author/committer email;
  97   ts    ::= # time since the epoch in seconds, ascii base10 notation;
  98   tz    ::= # GIT style timezone;
  99 */
 100
 101 #include "builtin.h"
 102 #include "cache.h"
 103 #include "object.h"
 104 #include "blob.h"
 105 #include "tree.h"
 106 #include "delta.h"
 107 #include "pack.h"
 108 #include "refs.h"
 109 #include "csum-file.h"
 110 #include "strbuf.h"
 111 #include "quote.h"
 112
 113 struct object_entry
 114 {
 115         struct object_entry *next;
 116         unsigned long offset;
 117         unsigned type : TYPE_BITS;
 118         unsigned pack_id : 16;
 119         unsigned char sha1[20];
 120 };
 121
 122 struct object_entry_pool
 123 {
 124         struct object_entry_pool *next_pool;
 125         struct object_entry *next_free;
 126         struct object_entry *end;
 127         struct object_entry entries[FLEX_ARRAY]; /* more */
 128 };
 129
 130 struct mark_set
 131 {
 132         int shift;
 133         union {
 134                 struct object_entry *marked[1024];
 135                 struct mark_set *sets[1024];
 136         } data;
 137 };
 138
 139 struct last_object
 140 {
 141         void *data;
 142         unsigned long len;
 143         unsigned long offset;
 144         unsigned int depth;
 145         unsigned no_free:1;
 146 };
 147
 148 struct mem_pool
 149 {
 150         struct mem_pool *next_pool;
 151         char *next_free;
 152         char *end;
 153         char space[FLEX_ARRAY]; /* more */
 154 };
 155
 156 struct atom_str
 157 {
 158         struct atom_str *next_atom;
 159         int str_len;
 160         char str_dat[FLEX_ARRAY]; /* more */
 161 };
 162
 163 struct tree_content;
 164 struct tree_entry
 165 {
 166         struct tree_content *tree;
 167         struct atom_str* name;
 168         struct tree_entry_ms
 169         {
 170                 unsigned int mode;
 171                 unsigned char sha1[20];
 172         } versions[2];
 173 };
 174
 175 struct tree_content
 176 {
 177         unsigned int entry_capacity; /* must match avail_tree_content */
 178         unsigned int entry_count;
 179         unsigned int delta_depth;
 180         struct tree_entry *entries[FLEX_ARRAY]; /* more */
 181 };
 182
 183 struct avail_tree_content
 184 {
 185         unsigned int entry_capacity; /* must match tree_content */
 186         struct avail_tree_content *next_avail;
 187 };
 188
 189 struct branch
 190 {
 191         struct branch *table_next_branch;
 192         struct branch *active_next_branch;
 193         const char *name;
 194         unsigned long last_commit;
 195         struct tree_entry branch_tree;
 196         unsigned char sha1[20];
 197 };
 198
 199 struct tag
 200 {
 201         struct tag *next_tag;
 202         const char *name;
 203         unsigned char sha1[20];
 204 };
 205
 206 struct dbuf
 207 {
 208         void *buffer;
 209         size_t capacity;
 210 };
 211
 212 struct hash_list
 213 {
 214         struct hash_list *next;
 215         unsigned char sha1[20];
 216 };
 217
 218 /* Configured limits on output */
 219 static unsigned long max_depth = 10;
 220 static unsigned long max_packsize = -1;
 221 static uintmax_t max_objects = -1;
 222
 223 /* Stats and misc. counters */
 224 static uintmax_t alloc_count;
 225 static uintmax_t object_count;
 226 static uintmax_t marks_set_count;
 227 static uintmax_t object_count_by_type[1 << TYPE_BITS];
 228 static uintmax_t duplicate_count_by_type[1 << TYPE_BITS];
 229 static uintmax_t delta_count_by_type[1 << TYPE_BITS];
 230 static unsigned long branch_count;
 231 static unsigned long branch_load_count;
 232
 233 /* Memory pools */
 234 static size_t mem_pool_alloc = 2*1024*1024 - sizeof(struct mem_pool);
 235 static size_t total_allocd;
 236 static struct mem_pool *mem_pool;
 237
 238 /* Atom management */
 239 static unsigned int atom_table_sz = 4451;
 240 static unsigned int atom_cnt;
 241 static struct atom_str **atom_table;
 242
 243 /* The .pack file being generated */
 244 static unsigned int pack_id;
 245 static struct packed_git *pack_data;
 246 static struct packed_git **all_packs;
 247 static unsigned long pack_size;
 248
 249 /* Table of objects we've written. */
 250 static unsigned int object_entry_alloc = 5000;
 251 static struct object_entry_pool *blocks;
 252 static struct object_entry *object_table[1 << 16];
 253 static struct mark_set *marks;
 254 static const char* mark_file;
 255
 256 /* Our last blob */
 257 static struct last_object last_blob;
 258
 259 /* Tree management */
 260 static unsigned int tree_entry_alloc = 1000;
 261 static void *avail_tree_entry;
 262 static unsigned int avail_tree_table_sz = 100;
 263 static struct avail_tree_content **avail_tree_table;
 264 static struct dbuf old_tree;
 265 static struct dbuf new_tree;
 266
 267 /* Branch data */
 268 static unsigned long max_active_branches = 5;
 269 static unsigned long cur_active_branches;
 270 static unsigned long branch_table_sz = 1039;
 271 static struct branch **branch_table;
 272 static struct branch *active_branches;
 273
 274 /* Tag data */
 275 static struct tag *first_tag;
 276 static struct tag *last_tag;
 277
 278 /* Input stream parsing */
 279 static struct strbuf command_buf;
 280 static uintmax_t next_mark;
 281 static struct dbuf new_data;
 282 static FILE* branch_log;
 283
 284
 285 static void alloc_objects(unsigned int cnt)
 286 {
 287         struct object_entry_pool *b;
 288
 289         b = xmalloc(sizeof(struct object_entry_pool)
 290                 + cnt * sizeof(struct object_entry));
 291         b->next_pool = blocks;
 292         b->next_free = b->entries;
 293         b->end = b->entries + cnt;
 294         blocks = b;
 295         alloc_count += cnt;
 296 }
 297
 298 static struct object_entry* new_object(unsigned char *sha1)
 299 {
 300         struct object_entry *e;
 301
 302         if (blocks->next_free == blocks->end)
 303                 alloc_objects(object_entry_alloc);
 304
 305         e = blocks->next_free++;
 306         hashcpy(e->sha1, sha1);
 307         return e;
 308 }
 309
 310 static struct object_entry* find_object(unsigned char *sha1)
 311 {
 312         unsigned int h = sha1[0] << 8 | sha1[1];
 313         struct object_entry *e;
 314         for (e = object_table[h]; e; e = e->next)
 315                 if (!hashcmp(sha1, e->sha1))
 316                         return e;
 317         return NULL;
 318 }
 319
 320 static struct object_entry* insert_object(unsigned char *sha1)
 321 {
 322         unsigned int h = sha1[0] << 8 | sha1[1];
 323         struct object_entry *e = object_table[h];
 324         struct object_entry *p = NULL;
 325
 326         while (e) {
 327                 if (!hashcmp(sha1, e->sha1))
 328                         return e;
 329                 p = e;
 330                 e = e->next;
 331         }
 332
 333         e = new_object(sha1);
 334         e->next = NULL;
 335         e->offset = 0;
 336         if (p)
 337                 p->next = e;
 338         else
 339                 object_table[h] = e;
 340         return e;
 341 }
 342
 343 static unsigned int hc_str(const char *s, size_t len)
 344 {
 345         unsigned int r = 0;
 346         while (len-- > 0)
 347                 r = r * 31 + *s++;
 348         return r;
 349 }
 350
 351 static void* pool_alloc(size_t len)
 352 {
 353         struct mem_pool *p;
 354         void *r;
 355
 356         for (p = mem_pool; p; p = p->next_pool)
 357                 if ((p->end - p->next_free >= len))
 358                         break;
 359
 360         if (!p) {
 361                 if (len >= (mem_pool_alloc/2)) {
 362                         total_allocd += len;
 363                         return xmalloc(len);
 364                 }
 365                 total_allocd += sizeof(struct mem_pool) + mem_pool_alloc;
 366                 p = xmalloc(sizeof(struct mem_pool) + mem_pool_alloc);
 367                 p->next_pool = mem_pool;
 368                 p->next_free = p->space;
 369                 p->end = p->next_free + mem_pool_alloc;
 370                 mem_pool = p;
 371         }
 372
 373         r = p->next_free;
 374         /* round out to a pointer alignment */
 375         if (len & (sizeof(void*) - 1))
 376                 len += sizeof(void*) - (len & (sizeof(void*) - 1));
 377         p->next_free += len;
 378         return r;
 379 }
 380
 381 static void* pool_calloc(size_t count, size_t size)
 382 {
 383         size_t len = count * size;
 384         void *r = pool_alloc(len);
 385         memset(r, 0, len);
 386         return r;
 387 }
 388
 389 static char* pool_strdup(const char *s)
 390 {
 391         char *r = pool_alloc(strlen(s) + 1);
 392         strcpy(r, s);
 393         return r;
 394 }
 395
 396 static void size_dbuf(struct dbuf *b, size_t maxlen)
 397 {
 398         if (b->buffer) {
 399                 if (b->capacity >= maxlen)
 400                         return;
 401                 free(b->buffer);
 402         }
 403         b->capacity = ((maxlen / 1024) + 1) * 1024;
 404         b->buffer = xmalloc(b->capacity);
 405 }
 406
 407 static void insert_mark(uintmax_t idnum, struct object_entry *oe)
 408 {
 409         struct mark_set *s = marks;
 410         while ((idnum >> s->shift) >= 1024) {
 411                 s = pool_calloc(1, sizeof(struct mark_set));
 412                 s->shift = marks->shift + 10;
 413                 s->data.sets[0] = marks;
 414                 marks = s;
 415         }
 416         while (s->shift) {
 417                 uintmax_t i = idnum >> s->shift;
 418                 idnum -= i << s->shift;
 419                 if (!s->data.sets[i]) {
 420                         s->data.sets[i] = pool_calloc(1, sizeof(struct mark_set));
 421                         s->data.sets[i]->shift = s->shift - 10;
 422                 }
 423                 s = s->data.sets[i];
 424         }
 425         if (!s->data.marked[idnum])
 426                 marks_set_count++;
 427         s->data.marked[idnum] = oe;
 428 }
 429
 430 static struct object_entry* find_mark(uintmax_t idnum)
 431 {
 432         uintmax_t orig_idnum = idnum;
 433         struct mark_set *s = marks;
 434         struct object_entry *oe = NULL;
 435         if ((idnum >> s->shift) < 1024) {
 436                 while (s && s->shift) {
 437                         uintmax_t i = idnum >> s->shift;
 438                         idnum -= i << s->shift;
 439                         s = s->data.sets[i];
 440                 }
 441                 if (s)
 442                         oe = s->data.marked[idnum];
 443         }
 444         if (!oe)
 445                 die("mark :%ju not declared", orig_idnum);
 446         return oe;
 447 }
 448
 449 static struct atom_str* to_atom(const char *s, size_t len)
 450 {
 451         unsigned int hc = hc_str(s, len) % atom_table_sz;
 452         struct atom_str *c;
 453
 454         for (c = atom_table[hc]; c; c = c->next_atom)
 455                 if (c->str_len == len && !strncmp(s, c->str_dat, len))
 456                         return c;
 457
 458         c = pool_alloc(sizeof(struct atom_str) + len + 1);
 459         c->str_len = len;
 460         strncpy(c->str_dat, s, len);
 461         c->str_dat[len] = 0;
 462         c->next_atom = atom_table[hc];
 463         atom_table[hc] = c;
 464         atom_cnt++;
 465         return c;
 466 }
 467
 468 static struct branch* lookup_branch(const char *name)
 469 {
 470         unsigned int hc = hc_str(name, strlen(name)) % branch_table_sz;
 471         struct branch *b;
 472
 473         for (b = branch_table[hc]; b; b = b->table_next_branch)
 474                 if (!strcmp(name, b->name))
 475                         return b;
 476         return NULL;
 477 }
 478
 479 static struct branch* new_branch(const char *name)
 480 {
 481         unsigned int hc = hc_str(name, strlen(name)) % branch_table_sz;
 482         struct branch* b = lookup_branch(name);
 483
 484         if (b)
 485                 die("Invalid attempt to create duplicate branch: %s", name);
 486         if (check_ref_format(name))
 487                 die("Branch name doesn't conform to GIT standards: %s", name);
 488
 489         b = pool_calloc(1, sizeof(struct branch));
 490         b->name = pool_strdup(name);
 491         b->table_next_branch = branch_table[hc];
 492         b->branch_tree.versions[0].mode = S_IFDIR;
 493         b->branch_tree.versions[1].mode = S_IFDIR;
 494         branch_table[hc] = b;
 495         branch_count++;
 496         return b;
 497 }
 498
 499 static unsigned int hc_entries(unsigned int cnt)
 500 {
 501         cnt = cnt & 7 ? (cnt / 8) + 1 : cnt / 8;
 502         return cnt < avail_tree_table_sz ? cnt : avail_tree_table_sz - 1;
 503 }
 504
 505 static struct tree_content* new_tree_content(unsigned int cnt)
 506 {
 507         struct avail_tree_content *f, *l = NULL;
 508         struct tree_content *t;
 509         unsigned int hc = hc_entries(cnt);
 510
 511         for (f = avail_tree_table[hc]; f; l = f, f = f->next_avail)
 512                 if (f->entry_capacity >= cnt)
 513                         break;
 514
 515         if (f) {
 516                 if (l)
 517                         l->next_avail = f->next_avail;
 518                 else
 519                         avail_tree_table[hc] = f->next_avail;
 520         } else {
 521                 cnt = cnt & 7 ? ((cnt / 8) + 1) * 8 : cnt;
 522                 f = pool_alloc(sizeof(*t) + sizeof(t->entries[0]) * cnt);
 523                 f->entry_capacity = cnt;
 524         }
 525
 526         t = (struct tree_content*)f;
 527         t->entry_count = 0;
 528         t->delta_depth = 0;
 529         return t;
 530 }
 531
 532 static void release_tree_entry(struct tree_entry *e);
 533 static void release_tree_content(struct tree_content *t)
 534 {
 535         struct avail_tree_content *f = (struct avail_tree_content*)t;
 536         unsigned int hc = hc_entries(f->entry_capacity);
 537         f->next_avail = avail_tree_table[hc];
 538         avail_tree_table[hc] = f;
 539 }
 540
 541 static void release_tree_content_recursive(struct tree_content *t)
 542 {
 543         unsigned int i;
 544         for (i = 0; i < t->entry_count; i++)
 545                 release_tree_entry(t->entries[i]);
 546         release_tree_content(t);
 547 }
 548
 549 static struct tree_content* grow_tree_content(
 550         struct tree_content *t,
 551         int amt)
 552 {
 553         struct tree_content *r = new_tree_content(t->entry_count + amt);
 554         r->entry_count = t->entry_count;
 555         r->delta_depth = t->delta_depth;
 556         memcpy(r->entries,t->entries,t->entry_count*sizeof(t->entries[0]));
 557         release_tree_content(t);
 558         return r;
 559 }
 560
 561 static struct tree_entry* new_tree_entry()
 562 {
 563         struct tree_entry *e;
 564
 565         if (!avail_tree_entry) {
 566                 unsigned int n = tree_entry_alloc;
 567                 total_allocd += n * sizeof(struct tree_entry);
 568                 avail_tree_entry = e = xmalloc(n * sizeof(struct tree_entry));
 569                 while (n-- > 1) {
 570                         *((void**)e) = e + 1;
 571                         e++;
 572                 }
 573                 *((void**)e) = NULL;
 574         }
 575
 576         e = avail_tree_entry;
 577         avail_tree_entry = *((void**)e);
 578         return e;
 579 }
 580
 581 static void release_tree_entry(struct tree_entry *e)
 582 {
 583         if (e->tree)
 584                 release_tree_content_recursive(e->tree);
 585         *((void**)e) = avail_tree_entry;
 586         avail_tree_entry = e;
 587 }
 588
 589 static void start_packfile()
 590 {
 591         static char tmpfile[PATH_MAX];
 592         struct packed_git *p;
 593         struct pack_header hdr;
 594         int pack_fd;
 595
 596         snprintf(tmpfile, sizeof(tmpfile),
 597                 "%s/pack_XXXXXX", get_object_directory());
 598         pack_fd = mkstemp(tmpfile);
 599         if (pack_fd < 0)
 600                 die("Can't create %s: %s", tmpfile, strerror(errno));
 601         p = xcalloc(1, sizeof(*p) + strlen(tmpfile) + 2);
 602         strcpy(p->pack_name, tmpfile);
 603         p->pack_fd = pack_fd;
 604
 605         hdr.hdr_signature = htonl(PACK_SIGNATURE);
 606         hdr.hdr_version = htonl(2);
 607         hdr.hdr_entries = 0;
 608         write_or_die(p->pack_fd, &hdr, sizeof(hdr));
 609
 610         pack_data = p;
 611         pack_size = sizeof(hdr);
 612         object_count = 0;
 613
 614         all_packs = xrealloc(all_packs, sizeof(*all_packs) * (pack_id + 1));
 615         all_packs[pack_id] = p;
 616 }
 617
 618 static void fixup_header_footer()
 619 {
 620         int pack_fd = pack_data->pack_fd;
 621         SHA_CTX c;
 622         char hdr[8];
 623         unsigned long cnt;
 624         char *buf;
 625
 626         if (lseek(pack_fd, 0, SEEK_SET) != 0)
 627                 die("Failed seeking to start: %s", strerror(errno));
 628
 629         SHA1_Init(&c);
 630         if (read_in_full(pack_fd, hdr, 8) != 8)
 631                 die("Unable to reread header of %s", pack_data->pack_name);
 632         SHA1_Update(&c, hdr, 8);
 633
 634         cnt = htonl(object_count);
 635         SHA1_Update(&c, &cnt, 4);
 636         write_or_die(pack_fd, &cnt, 4);
 637
 638         buf = xmalloc(128 * 1024);
 639         for (;;) {
 640                 size_t n = xread(pack_fd, buf, 128 * 1024);
 641                 if (n <= 0)
 642                         break;
 643                 SHA1_Update(&c, buf, n);
 644         }
 645         free(buf);
 646
 647         SHA1_Final(pack_data->sha1, &c);
 648         write_or_die(pack_fd, pack_data->sha1, sizeof(pack_data->sha1));
 649         close(pack_fd);
 650 }
 651
 652 static int oecmp (const void *a_, const void *b_)
 653 {
 654         struct object_entry *a = *((struct object_entry**)a_);
 655         struct object_entry *b = *((struct object_entry**)b_);
 656         return hashcmp(a->sha1, b->sha1);
 657 }
 658
 659 static char* create_index()
 660 {
 661         static char tmpfile[PATH_MAX];
 662         SHA_CTX ctx;
 663         struct sha1file *f;
 664         struct object_entry **idx, **c, **last, *e;
 665         struct object_entry_pool *o;
 666         unsigned int array[256];
 667         int i, idx_fd;
 668
 669         /* Build the sorted table of object IDs. */
 670         idx = xmalloc(object_count * sizeof(struct object_entry*));
 671         c = idx;
 672         for (o = blocks; o; o = o->next_pool)
 673                 for (e = o->next_free; e-- != o->entries;)
 674                         if (pack_id == e->pack_id)
 675                                 *c++ = e;
 676         last = idx + object_count;
 677         if (c != last)
 678                 die("internal consistency error creating the index");
 679         qsort(idx, object_count, sizeof(struct object_entry*), oecmp);
 680
 681         /* Generate the fan-out array. */
 682         c = idx;
 683         for (i = 0; i < 256; i++) {
 684                 struct object_entry **next = c;;
 685                 while (next < last) {
 686                         if ((*next)->sha1[0] != i)
 687                                 break;
 688                         next++;
 689                 }
 690                 array[i] = htonl(next - idx);
 691                 c = next;
 692         }
 693
 694         snprintf(tmpfile, sizeof(tmpfile),
 695                 "%s/index_XXXXXX", get_object_directory());
 696         idx_fd = mkstemp(tmpfile);
 697         if (idx_fd < 0)
 698                 die("Can't create %s: %s", tmpfile, strerror(errno));
 699         f = sha1fd(idx_fd, tmpfile);
 700         sha1write(f, array, 256 * sizeof(int));
 701         SHA1_Init(&ctx);
 702         for (c = idx; c != last; c++) {
 703                 unsigned int offset = htonl((*c)->offset);
 704                 sha1write(f, &offset, 4);
 705                 sha1write(f, (*c)->sha1, sizeof((*c)->sha1));
 706                 SHA1_Update(&ctx, (*c)->sha1, 20);
 707         }
 708         sha1write(f, pack_data->sha1, sizeof(pack_data->sha1));
 709         sha1close(f, NULL, 1);
 710         free(idx);
 711         SHA1_Final(pack_data->sha1, &ctx);
 712         return tmpfile;
 713 }
 714
 715 static char* keep_pack(char *curr_index_name)
 716 {
 717         static char name[PATH_MAX];
 718         static char *keep_msg = "fast-import";
 719         int keep_fd;
 720
 721         chmod(pack_data->pack_name, 0444);
 722         chmod(curr_index_name, 0444);
 723
 724         snprintf(name, sizeof(name), "%s/pack/pack-%s.keep",
 725                  get_object_directory(), sha1_to_hex(pack_data->sha1));
 726         keep_fd = open(name, O_RDWR|O_CREAT|O_EXCL, 0600);
 727         if (keep_fd < 0)
 728                 die("cannot create keep file");
 729         write(keep_fd, keep_msg, strlen(keep_msg));
 730         close(keep_fd);
 731
 732         snprintf(name, sizeof(name), "%s/pack/pack-%s.pack",
 733                  get_object_directory(), sha1_to_hex(pack_data->sha1));
 734         if (move_temp_to_file(pack_data->pack_name, name))
 735                 die("cannot store pack file");
 736         printf("%s\n", name);
 737
 738         snprintf(name, sizeof(name), "%s/pack/pack-%s.idx",
 739                  get_object_directory(), sha1_to_hex(pack_data->sha1));
 740         if (move_temp_to_file(curr_index_name, name))
 741                 die("cannot store index file");
 742         return name;
 743 }
 744
 745 static void unkeep_all_packs()
 746 {
 747         static char name[PATH_MAX];
 748         int k;
 749
 750         for (k = 0; k < pack_id; k++) {
 751                 struct packed_git *p = all_packs[k];
 752                 snprintf(name, sizeof(name), "%s/pack/pack-%s.keep",
 753                          get_object_directory(), sha1_to_hex(p->sha1));
 754                 unlink(name);
 755         }
 756 }
 757
 758 static void end_packfile()
 759 {
 760         struct packed_git *old_p = pack_data, *new_p;
 761
 762         if (object_count) {
 763                 char *idx_name;
 764
 765                 fixup_header_footer();
 766                 idx_name = keep_pack(create_index());
 767
 768                 /* Register the packfile with core git's machinary. */
 769                 new_p = add_packed_git(idx_name, strlen(idx_name), 1);
 770                 if (!new_p)
 771                         die("core git rejected index %s", idx_name);
 772                 new_p->windows = old_p->windows;
 773                 all_packs[pack_id++] = new_p;
 774                 install_packed_git(new_p);
 775         }
 776         else
 777                 unlink(old_p->pack_name);
 778         free(old_p);
 779
 780         /* We can't carry a delta across packfiles. */
 781         free(last_blob.data);
 782         last_blob.data = NULL;
 783         last_blob.len = 0;
 784         last_blob.offset = 0;
 785         last_blob.depth = 0;
 786 }
 787
 788 static void checkpoint()
 789 {
 790         end_packfile();
 791         start_packfile();
 792 }
 793
 794 static size_t encode_header(
 795         enum object_type type,
 796         size_t size,
 797         unsigned char *hdr)
 798 {
 799         int n = 1;
 800         unsigned char c;
 801
 802         if (type < OBJ_COMMIT || type > OBJ_REF_DELTA)
 803                 die("bad type %d", type);
 804
 805         c = (type << 4) | (size & 15);
 806         size >>= 4;
 807         while (size) {
 808                 *hdr++ = c | 0x80;
 809                 c = size & 0x7f;
 810                 size >>= 7;
 811                 n++;
 812         }
 813         *hdr = c;
 814         return n;
 815 }
 816
 817 static int store_object(
 818         enum object_type type,
 819         void *dat,
 820         size_t datlen,
 821         struct last_object *last,
 822         unsigned char *sha1out,
 823         uintmax_t mark)
 824 {
 825         void *out, *delta;
 826         struct object_entry *e;
 827         unsigned char hdr[96];
 828         unsigned char sha1[20];
 829         unsigned long hdrlen, deltalen;
 830         SHA_CTX c;
 831         z_stream s;
 832
 833         hdrlen = sprintf((char*)hdr,"%s %lu",type_names[type],datlen) + 1;
 834         SHA1_Init(&c);
 835         SHA1_Update(&c, hdr, hdrlen);
 836         SHA1_Update(&c, dat, datlen);
 837         SHA1_Final(sha1, &c);
 838         if (sha1out)
 839                 hashcpy(sha1out, sha1);
 840
 841         e = insert_object(sha1);
 842         if (mark)
 843                 insert_mark(mark, e);
 844         if (e->offset) {
 845                 duplicate_count_by_type[type]++;
 846                 return 1;
 847         }
 848
 849         if (last && last->data && last->depth < max_depth) {
 850                 delta = diff_delta(last->data, last->len,
 851                         dat, datlen,
 852                         &deltalen, 0);
 853                 if (delta && deltalen >= datlen) {
 854                         free(delta);
 855                         delta = NULL;
 856                 }
 857         } else
 858                 delta = NULL;
 859
 860         memset(&s, 0, sizeof(s));
 861         deflateInit(&s, zlib_compression_level);
 862         if (delta) {
 863                 s.next_in = delta;
 864                 s.avail_in = deltalen;
 865         } else {
 866                 s.next_in = dat;
 867                 s.avail_in = datlen;
 868         }
 869         s.avail_out = deflateBound(&s, s.avail_in);
 870         s.next_out = out = xmalloc(s.avail_out);
 871         while (deflate(&s, Z_FINISH) == Z_OK)
 872                 /* nothing */;
 873         deflateEnd(&s);
 874
 875         /* Determine if we should auto-checkpoint. */
 876         if ((object_count + 1) > max_objects
 877                 || (object_count + 1) < object_count
 878                 || (pack_size + 60 + s.total_out) > max_packsize
 879                 || (pack_size + 60 + s.total_out) < pack_size) {
 880
 881                 /* This new object needs to *not* have the current pack_id. */
 882                 e->pack_id = pack_id + 1;
 883                 checkpoint();
 884
 885                 /* We cannot carry a delta into the new pack. */
 886                 if (delta) {
 887                         free(delta);
 888                         delta = NULL;
 889
 890                         memset(&s, 0, sizeof(s));
 891                         deflateInit(&s, zlib_compression_level);
 892                         s.next_in = dat;
 893                         s.avail_in = datlen;
 894                         s.avail_out = deflateBound(&s, s.avail_in);
 895                         s.next_out = out = xrealloc(out, s.avail_out);
 896                         while (deflate(&s, Z_FINISH) == Z_OK)
 897                                 /* nothing */;
 898                         deflateEnd(&s);
 899                 }
 900         }
 901
 902         e->type = type;
 903         e->pack_id = pack_id;
 904         e->offset = pack_size;
 905         object_count++;
 906         object_count_by_type[type]++;
 907
 908         if (delta) {
 909                 unsigned long ofs = e->offset - last->offset;
 910                 unsigned pos = sizeof(hdr) - 1;
 911
 912                 delta_count_by_type[type]++;
 913                 last->depth++;
 914
 915                 hdrlen = encode_header(OBJ_OFS_DELTA, deltalen, hdr);
 916                 write_or_die(pack_data->pack_fd, hdr, hdrlen);
 917                 pack_size += hdrlen;
 918
 919                 hdr[pos] = ofs & 127;
 920                 while (ofs >>= 7)
 921                         hdr[--pos] = 128 | (--ofs & 127);
 922                 write_or_die(pack_data->pack_fd, hdr + pos, sizeof(hdr) - pos);
 923                 pack_size += sizeof(hdr) - pos;
 924         } else {
 925                 if (last)
 926                         last->depth = 0;
 927                 hdrlen = encode_header(type, datlen, hdr);
 928                 write_or_die(pack_data->pack_fd, hdr, hdrlen);
 929                 pack_size += hdrlen;
 930         }
 931
 932         write_or_die(pack_data->pack_fd, out, s.total_out);
 933         pack_size += s.total_out;
 934
 935         free(out);
 936         if (delta)
 937                 free(delta);
 938         if (last) {
 939                 if (last->data && !last->no_free)
 940                         free(last->data);
 941                 last->data = dat;
 942                 last->offset = e->offset;
 943                 last->len = datlen;
 944         }
 945         return 0;
 946 }
 947
 948 static void *gfi_unpack_entry(
 949         struct object_entry *oe,
 950         unsigned long *sizep)
 951 {
 952         static char type[20];
 953         struct packed_git *p = all_packs[oe->pack_id];
 954         if (p == pack_data)
 955                 p->pack_size = pack_size + 20;
 956         return unpack_entry(p, oe->offset, type, sizep);
 957 }
 958
 959 static const char *get_mode(const char *str, unsigned int *modep)
 960 {
 961         unsigned char c;
 962         unsigned int mode = 0;
 963
 964         while ((c = *str++) != ' ') {
 965                 if (c < '0' || c > '7')
 966                         return NULL;
 967                 mode = (mode << 3) + (c - '0');
 968         }
 969         *modep = mode;
 970         return str;
 971 }
 972
 973 static void load_tree(struct tree_entry *root)
 974 {
 975         unsigned char* sha1 = root->versions[1].sha1;
 976         struct object_entry *myoe;
 977         struct tree_content *t;
 978         unsigned long size;
 979         char *buf;
 980         const char *c;
 981
 982         root->tree = t = new_tree_content(8);
 983         if (is_null_sha1(sha1))
 984                 return;
 985
 986         myoe = find_object(sha1);
 987         if (myoe) {
 988                 if (myoe->type != OBJ_TREE)
 989                         die("Not a tree: %s", sha1_to_hex(sha1));
 990                 t->delta_depth = 0;
 991                 buf = gfi_unpack_entry(myoe, &size);
 992         } else {
 993                 char type[20];
 994                 buf = read_sha1_file(sha1, type, &size);
 995                 if (!buf || strcmp(type, tree_type))
 996                         die("Can't load tree %s", sha1_to_hex(sha1));
 997         }
 998
 999         c = buf;
1000         while (c != (buf + size)) {
1001                 struct tree_entry *e = new_tree_entry();
1002
1003                 if (t->entry_count == t->entry_capacity)
1004                         root->tree = t = grow_tree_content(t, 8);
1005                 t->entries[t->entry_count++] = e;
1006
1007                 e->tree = NULL;
1008                 c = get_mode(c, &e->versions[1].mode);
1009                 if (!c)
1010                         die("Corrupt mode in %s", sha1_to_hex(sha1));
1011                 e->versions[0].mode = e->versions[1].mode;
1012                 e->name = to_atom(c, strlen(c));
1013                 c += e->name->str_len + 1;
1014                 hashcpy(e->versions[0].sha1, (unsigned char*)c);
1015                 hashcpy(e->versions[1].sha1, (unsigned char*)c);
1016                 c += 20;
1017         }
1018         free(buf);
1019 }
1020
1021 static int tecmp0 (const void *_a, const void *_b)
1022 {
1023         struct tree_entry *a = *((struct tree_entry**)_a);
1024         struct tree_entry *b = *((struct tree_entry**)_b);
1025         return base_name_compare(
1026                 a->name->str_dat, a->name->str_len, a->versions[0].mode,
1027                 b->name->str_dat, b->name->str_len, b->versions[0].mode);
1028 }
1029
1030 static int tecmp1 (const void *_a, const void *_b)
1031 {
1032         struct tree_entry *a = *((struct tree_entry**)_a);
1033         struct tree_entry *b = *((struct tree_entry**)_b);
1034         return base_name_compare(
1035                 a->name->str_dat, a->name->str_len, a->versions[1].mode,
1036                 b->name->str_dat, b->name->str_len, b->versions[1].mode);
1037 }
1038
1039 static void mktree(struct tree_content *t,
1040         int v,
1041         unsigned long *szp,
1042         struct dbuf *b)
1043 {
1044         size_t maxlen = 0;
1045         unsigned int i;
1046         char *c;
1047
1048         if (!v)
1049                 qsort(t->entries,t->entry_count,sizeof(t->entries[0]),tecmp0);
1050         else
1051                 qsort(t->entries,t->entry_count,sizeof(t->entries[0]),tecmp1);
1052
1053         for (i = 0; i < t->entry_count; i++) {
1054                 if (t->entries[i]->versions[v].mode)
1055                         maxlen += t->entries[i]->name->str_len + 34;
1056         }
1057
1058         size_dbuf(b, maxlen);
1059         c = b->buffer;
1060         for (i = 0; i < t->entry_count; i++) {
1061                 struct tree_entry *e = t->entries[i];
1062                 if (!e->versions[v].mode)
1063                         continue;
1064                 c += sprintf(c, "%o", e->versions[v].mode);
1065                 *c++ = ' ';
1066                 strcpy(c, e->name->str_dat);
1067                 c += e->name->str_len + 1;
1068                 hashcpy((unsigned char*)c, e->versions[v].sha1);
1069                 c += 20;
1070         }
1071         *szp = c - (char*)b->buffer;
1072 }
1073
1074 static void store_tree(struct tree_entry *root)
1075 {
1076         struct tree_content *t = root->tree;
1077         unsigned int i, j, del;
1078         unsigned long new_len;
1079         struct last_object lo;
1080         struct object_entry *le;
1081
1082         if (!is_null_sha1(root->versions[1].sha1))
1083                 return;
1084
1085         for (i = 0; i < t->entry_count; i++) {
1086                 if (t->entries[i]->tree)
1087                         store_tree(t->entries[i]);
1088         }
1089
1090         le = find_object(root->versions[0].sha1);
1091         if (!S_ISDIR(root->versions[0].mode)
1092                 || !le
1093                 || le->pack_id != pack_id) {
1094                 lo.data = NULL;
1095                 lo.depth = 0;
1096         } else {
1097                 mktree(t, 0, &lo.len, &old_tree);
1098                 lo.data = old_tree.buffer;
1099                 lo.offset = le->offset;
1100                 lo.depth = t->delta_depth;
1101                 lo.no_free = 1;
1102         }
1103
1104         mktree(t, 1, &new_len, &new_tree);
1105         store_object(OBJ_TREE, new_tree.buffer, new_len,
1106                 &lo, root->versions[1].sha1, 0);
1107
1108         t->delta_depth = lo.depth;
1109         for (i = 0, j = 0, del = 0; i < t->entry_count; i++) {
1110                 struct tree_entry *e = t->entries[i];
1111                 if (e->versions[1].mode) {
1112                         e->versions[0].mode = e->versions[1].mode;
1113                         hashcpy(e->versions[0].sha1, e->versions[1].sha1);
1114                         t->entries[j++] = e;
1115                 } else {
1116                         release_tree_entry(e);
1117                         del++;
1118                 }
1119         }
1120         t->entry_count -= del;
1121 }
1122
1123 static int tree_content_set(
1124         struct tree_entry *root,
1125         const char *p,
1126         const unsigned char *sha1,
1127         const unsigned int mode)
1128 {
1129         struct tree_content *t = root->tree;
1130         const char *slash1;
1131         unsigned int i, n;
1132         struct tree_entry *e;
1133
1134         slash1 = strchr(p, '/');
1135         if (slash1)
1136                 n = slash1 - p;
1137         else
1138                 n = strlen(p);
1139
1140         for (i = 0; i < t->entry_count; i++) {
1141                 e = t->entries[i];
1142                 if (e->name->str_len == n && !strncmp(p, e->name->str_dat, n)) {
1143                         if (!slash1) {
1144                                 if (e->versions[1].mode == mode
1145                                                 && !hashcmp(e->versions[1].sha1, sha1))
1146                                         return 0;
1147                                 e->versions[1].mode = mode;
1148                                 hashcpy(e->versions[1].sha1, sha1);
1149                                 if (e->tree) {
1150                                         release_tree_content_recursive(e->tree);
1151                                         e->tree = NULL;
1152                                 }
1153                                 hashclr(root->versions[1].sha1);
1154                                 return 1;
1155                         }
1156                         if (!S_ISDIR(e->versions[1].mode)) {
1157                                 e->tree = new_tree_content(8);
1158                                 e->versions[1].mode = S_IFDIR;
1159                         }
1160                         if (!e->tree)
1161                                 load_tree(e);
1162                         if (tree_content_set(e, slash1 + 1, sha1, mode)) {
1163                                 hashclr(root->versions[1].sha1);
1164                                 return 1;
1165                         }
1166                         return 0;
1167                 }
1168         }
1169
1170         if (t->entry_count == t->entry_capacity)
1171                 root->tree = t = grow_tree_content(t, 8);
1172         e = new_tree_entry();
1173         e->name = to_atom(p, n);
1174         e->versions[0].mode = 0;
1175         hashclr(e->versions[0].sha1);
1176         t->entries[t->entry_count++] = e;
1177         if (slash1) {
1178                 e->tree = new_tree_content(8);
1179                 e->versions[1].mode = S_IFDIR;
1180                 tree_content_set(e, slash1 + 1, sha1, mode);
1181         } else {
1182                 e->tree = NULL;
1183                 e->versions[1].mode = mode;
1184                 hashcpy(e->versions[1].sha1, sha1);
1185         }
1186         hashclr(root->versions[1].sha1);
1187         return 1;
1188 }
1189
1190 static int tree_content_remove(struct tree_entry *root, const char *p)
1191 {
1192         struct tree_content *t = root->tree;
1193         const char *slash1;
1194         unsigned int i, n;
1195         struct tree_entry *e;
1196
1197         slash1 = strchr(p, '/');
1198         if (slash1)
1199                 n = slash1 - p;
1200         else
1201                 n = strlen(p);
1202
1203         for (i = 0; i < t->entry_count; i++) {
1204                 e = t->entries[i];
1205                 if (e->name->str_len == n && !strncmp(p, e->name->str_dat, n)) {
1206                         if (!slash1 || !S_ISDIR(e->versions[1].mode))
1207                                 goto del_entry;
1208                         if (!e->tree)
1209                                 load_tree(e);
1210                         if (tree_content_remove(e, slash1 + 1)) {
1211                                 for (n = 0; n < e->tree->entry_count; n++) {
1212                                         if (e->tree->entries[n]->versions[1].mode) {
1213                                                 hashclr(root->versions[1].sha1);
1214                                                 return 1;
1215                                         }
1216                                 }
1217                                 goto del_entry;
1218                         }
1219                         return 0;
1220                 }
1221         }
1222         return 0;
1223
1224 del_entry:
1225         if (e->tree) {
1226                 release_tree_content_recursive(e->tree);
1227                 e->tree = NULL;
1228         }
1229         e->versions[1].mode = 0;
1230         hashclr(e->versions[1].sha1);
1231         hashclr(root->versions[1].sha1);
1232         return 1;
1233 }
1234
1235 static void dump_branches()
1236 {
1237         static const char *msg = "fast-import";
1238         unsigned int i;
1239         struct branch *b;
1240         struct ref_lock *lock;
1241
1242         for (i = 0; i < branch_table_sz; i++) {
1243                 for (b = branch_table[i]; b; b = b->table_next_branch) {
1244                         lock = lock_any_ref_for_update(b->name, NULL);
1245                         if (!lock || write_ref_sha1(lock, b->sha1, msg) < 0)
1246                                 die("Can't write %s", b->name);
1247                 }
1248         }
1249 }
1250
1251 static void dump_tags()
1252 {
1253         static const char *msg = "fast-import";
1254         struct tag *t;
1255         struct ref_lock *lock;
1256         char path[PATH_MAX];
1257
1258         for (t = first_tag; t; t = t->next_tag) {
1259                 sprintf(path, "refs/tags/%s", t->name);
1260                 lock = lock_any_ref_for_update(path, NULL);
1261                 if (!lock || write_ref_sha1(lock, t->sha1, msg) < 0)
1262                         die("Can't write %s", path);
1263         }
1264 }
1265
1266 static void dump_marks_helper(FILE *f,
1267         uintmax_t base,
1268         struct mark_set *m)
1269 {
1270         uintmax_t k;
1271         if (m->shift) {
1272                 for (k = 0; k < 1024; k++) {
1273                         if (m->data.sets[k])
1274                                 dump_marks_helper(f, (base + k) << m->shift,
1275                                         m->data.sets[k]);
1276                 }
1277         } else {
1278                 for (k = 0; k < 1024; k++) {
1279                         if (m->data.marked[k])
1280                                 fprintf(f, ":%ju %s\n", base + k,
1281                                         sha1_to_hex(m->data.marked[k]->sha1));
1282                 }
1283         }
1284 }
1285
1286 static void dump_marks()
1287 {
1288         if (mark_file)
1289         {
1290                 FILE *f = fopen(mark_file, "w");
1291                 dump_marks_helper(f, 0, marks);
1292                 fclose(f);
1293         }
1294 }
1295
1296 static void read_next_command()
1297 {
1298         read_line(&command_buf, stdin, '\n');
1299 }
1300
1301 static void cmd_mark()
1302 {
1303         if (!strncmp("mark :", command_buf.buf, 6)) {
1304                 next_mark = strtoumax(command_buf.buf + 6, NULL, 10);
1305                 read_next_command();
1306         }
1307         else
1308                 next_mark = 0;
1309 }
1310
1311 static void* cmd_data (size_t *size)
1312 {
1313         size_t n = 0;
1314         void *buffer;
1315         size_t length;
1316
1317         if (strncmp("data ", command_buf.buf, 5))
1318                 die("Expected 'data n' command, found: %s", command_buf.buf);
1319
1320         length = strtoul(command_buf.buf + 5, NULL, 10);
1321         buffer = xmalloc(length);
1322
1323         while (n < length) {
1324                 size_t s = fread((char*)buffer + n, 1, length - n, stdin);
1325                 if (!s && feof(stdin))
1326                         die("EOF in data (%lu bytes remaining)", length - n);
1327                 n += s;
1328         }
1329
1330         if (fgetc(stdin) != '\n')
1331                 die("An lf did not trail the binary data as expected.");
1332
1333         *size = length;
1334         return buffer;
1335 }
1336
1337 static void cmd_new_blob()
1338 {
1339         size_t l;
1340         void *d;
1341
1342         read_next_command();
1343         cmd_mark();
1344         d = cmd_data(&l);
1345
1346         if (store_object(OBJ_BLOB, d, l, &last_blob, NULL, next_mark))
1347                 free(d);
1348 }
1349
1350 static void unload_one_branch()
1351 {
1352         while (cur_active_branches
1353                 && cur_active_branches >= max_active_branches) {
1354                 unsigned long min_commit = ULONG_MAX;
1355                 struct branch *e, *l = NULL, *p = NULL;
1356
1357                 for (e = active_branches; e; e = e->active_next_branch) {
1358                         if (e->last_commit < min_commit) {
1359                                 p = l;
1360                                 min_commit = e->last_commit;
1361                         }
1362                         l = e;
1363                 }
1364
1365                 if (p) {
1366                         e = p->active_next_branch;
1367                         p->active_next_branch = e->active_next_branch;
1368                 } else {
1369                         e = active_branches;
1370                         active_branches = e->active_next_branch;
1371                 }
1372                 e->active_next_branch = NULL;
1373                 if (e->branch_tree.tree) {
1374                         release_tree_content_recursive(e->branch_tree.tree);
1375                         e->branch_tree.tree = NULL;
1376                 }
1377                 cur_active_branches--;
1378         }
1379 }
1380
1381 static void load_branch(struct branch *b)
1382 {
1383         load_tree(&b->branch_tree);
1384         b->active_next_branch = active_branches;
1385         active_branches = b;
1386         cur_active_branches++;
1387         branch_load_count++;
1388 }
1389
1390 static void file_change_m(struct branch *b)
1391 {
1392         const char *p = command_buf.buf + 2;
1393         char *p_uq;
1394         const char *endp;
1395         struct object_entry *oe;
1396         unsigned char sha1[20];
1397         unsigned int mode;
1398         char type[20];
1399
1400         p = get_mode(p, &mode);
1401         if (!p)
1402                 die("Corrupt mode: %s", command_buf.buf);
1403         switch (mode) {
1404         case S_IFREG | 0644:
1405         case S_IFREG | 0755:
1406         case S_IFLNK:
1407         case 0644:
1408         case 0755:
1409                 /* ok */
1410                 break;
1411         default:
1412                 die("Corrupt mode: %s", command_buf.buf);
1413         }
1414
1415         if (*p == ':') {
1416                 char *x;
1417                 oe = find_mark(strtoumax(p + 1, &x, 10));
1418                 hashcpy(sha1, oe->sha1);
1419                 p = x;
1420         } else {
1421                 if (get_sha1_hex(p, sha1))
1422                         die("Invalid SHA1: %s", command_buf.buf);
1423                 oe = find_object(sha1);
1424                 p += 40;
1425         }
1426         if (*p++ != ' ')
1427                 die("Missing space after SHA1: %s", command_buf.buf);
1428
1429         p_uq = unquote_c_style(p, &endp);
1430         if (p_uq) {
1431                 if (*endp)
1432                         die("Garbage after path in: %s", command_buf.buf);
1433                 p = p_uq;
1434         }
1435
1436         if (oe) {
1437                 if (oe->type != OBJ_BLOB)
1438                         die("Not a blob (actually a %s): %s",
1439                                 command_buf.buf, type_names[oe->type]);
1440         } else {
1441                 if (sha1_object_info(sha1, type, NULL))
1442                         die("Blob not found: %s", command_buf.buf);
1443                 if (strcmp(blob_type, type))
1444                         die("Not a blob (actually a %s): %s",
1445                                 command_buf.buf, type);
1446         }
1447
1448         tree_content_set(&b->branch_tree, p, sha1, S_IFREG | mode);
1449
1450         if (p_uq)
1451                 free(p_uq);
1452 }
1453
1454 static void file_change_d(struct branch *b)
1455 {
1456         const char *p = command_buf.buf + 2;
1457         char *p_uq;
1458         const char *endp;
1459
1460         p_uq = unquote_c_style(p, &endp);
1461         if (p_uq) {
1462                 if (*endp)
1463                         die("Garbage after path in: %s", command_buf.buf);
1464                 p = p_uq;
1465         }
1466         tree_content_remove(&b->branch_tree, p);
1467         if (p_uq)
1468                 free(p_uq);
1469 }
1470
1471 static void cmd_from(struct branch *b)
1472 {
1473         const char *from, *endp;
1474         char *str_uq;
1475         struct branch *s;
1476
1477         if (strncmp("from ", command_buf.buf, 5))
1478                 return;
1479
1480         if (b->last_commit)
1481                 die("Can't reinitailize branch %s", b->name);
1482
1483         from = strchr(command_buf.buf, ' ') + 1;
1484         str_uq = unquote_c_style(from, &endp);
1485         if (str_uq) {
1486                 if (*endp)
1487                         die("Garbage after string in: %s", command_buf.buf);
1488                 from = str_uq;
1489         }
1490
1491         s = lookup_branch(from);
1492         if (b == s)
1493                 die("Can't create a branch from itself: %s", b->name);
1494         else if (s) {
1495                 unsigned char *t = s->branch_tree.versions[1].sha1;
1496                 hashcpy(b->sha1, s->sha1);
1497                 hashcpy(b->branch_tree.versions[0].sha1, t);
1498                 hashcpy(b->branch_tree.versions[1].sha1, t);
1499         } else if (*from == ':') {
1500                 uintmax_t idnum = strtoumax(from + 1, NULL, 10);
1501                 struct object_entry *oe = find_mark(idnum);
1502                 unsigned long size;
1503                 char *buf;
1504                 if (oe->type != OBJ_COMMIT)
1505                         die("Mark :%ju not a commit", idnum);
1506                 hashcpy(b->sha1, oe->sha1);
1507                 buf = gfi_unpack_entry(oe, &size);
1508                 if (!buf || size < 46)
1509                         die("Not a valid commit: %s", from);
1510                 if (memcmp("tree ", buf, 5)
1511                         || get_sha1_hex(buf + 5, b->branch_tree.versions[1].sha1))
1512                         die("The commit %s is corrupt", sha1_to_hex(b->sha1));
1513                 free(buf);
1514                 hashcpy(b->branch_tree.versions[0].sha1,
1515                         b->branch_tree.versions[1].sha1);
1516         } else if (!get_sha1(from, b->sha1)) {
1517                 if (is_null_sha1(b->sha1)) {
1518                         hashclr(b->branch_tree.versions[0].sha1);
1519                         hashclr(b->branch_tree.versions[1].sha1);
1520                 } else {
1521                         unsigned long size;
1522                         char *buf;
1523
1524                         buf = read_object_with_reference(b->sha1,
1525                                 type_names[OBJ_COMMIT], &size, b->sha1);
1526                         if (!buf || size < 46)
1527                                 die("Not a valid commit: %s", from);
1528                         if (memcmp("tree ", buf, 5)
1529                                 || get_sha1_hex(buf + 5, b->branch_tree.versions[1].sha1))
1530                                 die("The commit %s is corrupt", sha1_to_hex(b->sha1));
1531                         free(buf);
1532                         hashcpy(b->branch_tree.versions[0].sha1,
1533                                 b->branch_tree.versions[1].sha1);
1534                 }
1535         } else
1536                 die("Invalid ref name or SHA1 expression: %s", from);
1537
1538         read_next_command();
1539 }
1540
1541 static struct hash_list* cmd_merge(unsigned int *count)
1542 {
1543         struct hash_list *list = NULL, *n, *e;
1544         const char *from, *endp;
1545         char *str_uq;
1546         struct branch *s;
1547
1548         *count = 0;
1549         while (!strncmp("merge ", command_buf.buf, 6)) {
1550                 from = strchr(command_buf.buf, ' ') + 1;
1551                 str_uq = unquote_c_style(from, &endp);
1552                 if (str_uq) {
1553                         if (*endp)
1554                                 die("Garbage after string in: %s", command_buf.buf);
1555                         from = str_uq;
1556                 }
1557
1558                 n = xmalloc(sizeof(*n));
1559                 s = lookup_branch(from);
1560                 if (s)
1561                         hashcpy(n->sha1, s->sha1);
1562                 else if (*from == ':') {
1563                         uintmax_t idnum = strtoumax(from + 1, NULL, 10);
1564                         struct object_entry *oe = find_mark(idnum);
1565                         if (oe->type != OBJ_COMMIT)
1566                                 die("Mark :%ju not a commit", idnum);
1567                         hashcpy(n->sha1, oe->sha1);
1568                 } else if (get_sha1(from, n->sha1))
1569                         die("Invalid ref name or SHA1 expression: %s", from);
1570
1571                 n->next = NULL;
1572                 if (list)
1573                         e->next = n;
1574                 else
1575                         list = n;
1576                 e = n;
1577                 *count++;
1578                 read_next_command();
1579         }
1580         return list;
1581 }
1582
1583 static void cmd_new_commit()
1584 {
1585         struct branch *b;
1586         void *msg;
1587         size_t msglen;
1588         char *str_uq;
1589         const char *endp;
1590         char *sp;
1591         char *author = NULL;
1592         char *committer = NULL;
1593         struct hash_list *merge_list = NULL;
1594         unsigned int merge_count;
1595
1596         /* Obtain the branch name from the rest of our command */
1597         sp = strchr(command_buf.buf, ' ') + 1;
1598         str_uq = unquote_c_style(sp, &endp);
1599         if (str_uq) {
1600                 if (*endp)
1601                         die("Garbage after ref in: %s", command_buf.buf);
1602                 sp = str_uq;
1603         }
1604         b = lookup_branch(sp);
1605         if (!b)
1606                 b = new_branch(sp);
1607         if (str_uq)
1608                 free(str_uq);
1609
1610         read_next_command();
1611         cmd_mark();
1612         if (!strncmp("author ", command_buf.buf, 7)) {
1613                 author = strdup(command_buf.buf);
1614                 read_next_command();
1615         }
1616         if (!strncmp("committer ", command_buf.buf, 10)) {
1617                 committer = strdup(command_buf.buf);
1618                 read_next_command();
1619         }
1620         if (!committer)
1621                 die("Expected committer but didn't get one");
1622         msg = cmd_data(&msglen);
1623         read_next_command();
1624         cmd_from(b);
1625         merge_list = cmd_merge(&merge_count);
1626
1627         /* ensure the branch is active/loaded */
1628         if (!b->branch_tree.tree || !max_active_branches) {
1629                 unload_one_branch();
1630                 load_branch(b);
1631         }
1632
1633         /* file_change* */
1634         for (;;) {
1635                 if (1 == command_buf.len)
1636                         break;
1637                 else if (!strncmp("M ", command_buf.buf, 2))
1638                         file_change_m(b);
1639                 else if (!strncmp("D ", command_buf.buf, 2))
1640                         file_change_d(b);
1641                 else
1642                         die("Unsupported file_change: %s", command_buf.buf);
1643                 read_next_command();
1644         }
1645
1646         /* build the tree and the commit */
1647         store_tree(&b->branch_tree);
1648         hashcpy(b->branch_tree.versions[0].sha1,
1649                 b->branch_tree.versions[1].sha1);
1650         size_dbuf(&new_data, 97 + msglen
1651                 + merge_count * 49
1652                 + (author
1653                         ? strlen(author) + strlen(committer)
1654                         : 2 * strlen(committer)));
1655         sp = new_data.buffer;
1656         sp += sprintf(sp, "tree %s\n",
1657                 sha1_to_hex(b->branch_tree.versions[1].sha1));
1658         if (!is_null_sha1(b->sha1))
1659                 sp += sprintf(sp, "parent %s\n", sha1_to_hex(b->sha1));
1660         while (merge_list) {
1661                 struct hash_list *next = merge_list->next;
1662                 sp += sprintf(sp, "parent %s\n", sha1_to_hex(merge_list->sha1));
1663                 free(merge_list);
1664                 merge_list = next;
1665         }
1666         if (author)
1667                 sp += sprintf(sp, "%s\n", author);
1668         else
1669                 sp += sprintf(sp, "author %s\n", committer + 10);
1670         sp += sprintf(sp, "%s\n\n", committer);
1671         memcpy(sp, msg, msglen);
1672         sp += msglen;
1673         if (author)
1674                 free(author);
1675         free(committer);
1676         free(msg);
1677
1678         store_object(OBJ_COMMIT,
1679                 new_data.buffer, sp - (char*)new_data.buffer,
1680                 NULL, b->sha1, next_mark);
1681         b->last_commit = object_count_by_type[OBJ_COMMIT];
1682
1683         if (branch_log) {
1684                 int need_dq = quote_c_style(b->name, NULL, NULL, 0);
1685                 fprintf(branch_log, "commit ");
1686                 if (need_dq) {
1687                         fputc('"', branch_log);
1688                         quote_c_style(b->name, NULL, branch_log, 0);
1689                         fputc('"', branch_log);
1690                 } else
1691                         fprintf(branch_log, "%s", b->name);
1692                 fprintf(branch_log," :%ju %s\n",next_mark,sha1_to_hex(b->sha1));
1693         }
1694 }
1695
1696 static void cmd_new_tag()
1697 {
1698         char *str_uq;
1699         const char *endp;
1700         char *sp;
1701         const char *from;
1702         char *tagger;
1703         struct branch *s;
1704         void *msg;
1705         size_t msglen;
1706         struct tag *t;
1707         uintmax_t from_mark = 0;
1708         unsigned char sha1[20];
1709
1710         /* Obtain the new tag name from the rest of our command */
1711         sp = strchr(command_buf.buf, ' ') + 1;
1712         str_uq = unquote_c_style(sp, &endp);
1713         if (str_uq) {
1714                 if (*endp)
1715                         die("Garbage after tag name in: %s", command_buf.buf);
1716                 sp = str_uq;
1717         }
1718         t = pool_alloc(sizeof(struct tag));
1719         t->next_tag = NULL;
1720         t->name = pool_strdup(sp);
1721         if (last_tag)
1722                 last_tag->next_tag = t;
1723         else
1724                 first_tag = t;
1725         last_tag = t;
1726         if (str_uq)
1727                 free(str_uq);
1728         read_next_command();
1729
1730         /* from ... */
1731         if (strncmp("from ", command_buf.buf, 5))
1732                 die("Expected from command, got %s", command_buf.buf);
1733
1734         from = strchr(command_buf.buf, ' ') + 1;
1735         str_uq = unquote_c_style(from, &endp);
1736         if (str_uq) {
1737                 if (*endp)
1738                         die("Garbage after string in: %s", command_buf.buf);
1739                 from = str_uq;
1740         }
1741
1742         s = lookup_branch(from);
1743         if (s) {
1744                 hashcpy(sha1, s->sha1);
1745         } else if (*from == ':') {
1746                 from_mark = strtoumax(from + 1, NULL, 10);
1747                 struct object_entry *oe = find_mark(from_mark);
1748                 if (oe->type != OBJ_COMMIT)
1749                         die("Mark :%ju not a commit", from_mark);
1750                 hashcpy(sha1, oe->sha1);
1751         } else if (!get_sha1(from, sha1)) {
1752                 unsigned long size;
1753                 char *buf;
1754
1755                 buf = read_object_with_reference(sha1,
1756                         type_names[OBJ_COMMIT], &size, sha1);
1757                 if (!buf || size < 46)
1758                         die("Not a valid commit: %s", from);
1759                 free(buf);
1760         } else
1761                 die("Invalid ref name or SHA1 expression: %s", from);
1762
1763         if (str_uq)
1764                 free(str_uq);
1765         read_next_command();
1766
1767         /* tagger ... */
1768         if (strncmp("tagger ", command_buf.buf, 7))
1769                 die("Expected tagger command, got %s", command_buf.buf);
1770         tagger = strdup(command_buf.buf);
1771
1772         /* tag payload/message */
1773         read_next_command();
1774         msg = cmd_data(&msglen);
1775
1776         /* build the tag object */
1777         size_dbuf(&new_data, 67+strlen(t->name)+strlen(tagger)+msglen);
1778         sp = new_data.buffer;
1779         sp += sprintf(sp, "object %s\n", sha1_to_hex(sha1));
1780         sp += sprintf(sp, "type %s\n", type_names[OBJ_COMMIT]);
1781         sp += sprintf(sp, "tag %s\n", t->name);
1782         sp += sprintf(sp, "%s\n\n", tagger);
1783         memcpy(sp, msg, msglen);
1784         sp += msglen;
1785         free(tagger);
1786         free(msg);
1787
1788         store_object(OBJ_TAG, new_data.buffer, sp - (char*)new_data.buffer,
1789                 NULL, t->sha1, 0);
1790
1791         if (branch_log) {
1792                 int need_dq = quote_c_style(t->name, NULL, NULL, 0);
1793                 fprintf(branch_log, "tag ");
1794                 if (need_dq) {
1795                         fputc('"', branch_log);
1796                         quote_c_style(t->name, NULL, branch_log, 0);
1797                         fputc('"', branch_log);
1798                 } else
1799                         fprintf(branch_log, "%s", t->name);
1800                 fprintf(branch_log," :%ju %s\n",from_mark,sha1_to_hex(t->sha1));
1801         }
1802 }
1803
1804 static void cmd_reset_branch()
1805 {
1806         struct branch *b;
1807         char *str_uq;
1808         const char *endp;
1809         char *sp;
1810
1811         /* Obtain the branch name from the rest of our command */
1812         sp = strchr(command_buf.buf, ' ') + 1;
1813         str_uq = unquote_c_style(sp, &endp);
1814         if (str_uq) {
1815                 if (*endp)
1816                         die("Garbage after ref in: %s", command_buf.buf);
1817                 sp = str_uq;
1818         }
1819         b = lookup_branch(sp);
1820         if (b) {
1821                 b->last_commit = 0;
1822                 if (b->branch_tree.tree) {
1823                         release_tree_content_recursive(b->branch_tree.tree);
1824                         b->branch_tree.tree = NULL;
1825                 }
1826         }
1827         else
1828                 b = new_branch(sp);
1829         if (str_uq)
1830                 free(str_uq);
1831         read_next_command();
1832         cmd_from(b);
1833 }
1834
1835 static void cmd_checkpoint()
1836 {
1837         if (object_count)
1838                 checkpoint();
1839         read_next_command();
1840 }
1841
1842 static const char fast_import_usage[] =
1843 "git-fast-import [--objects=n] [--depth=n] [--active-branches=n] [--export-marks=marks.file] [--branch-log=log]";
1844
1845 int main(int argc, const char **argv)
1846 {
1847         int i;
1848         uintmax_t est_obj_cnt = object_entry_alloc;
1849         uintmax_t duplicate_count;
1850
1851         setup_ident();
1852         git_config(git_default_config);
1853
1854         for (i = 1; i < argc; i++) {
1855                 const char *a = argv[i];
1856
1857                 if (*a != '-' || !strcmp(a, "--"))
1858                         break;
1859                 else if (!strncmp(a, "--objects=", 10))
1860                         est_obj_cnt = strtoumax(a + 10, NULL, 0);
1861                 else if (!strncmp(a, "--max-objects-per-pack=", 23))
1862                         max_objects = strtoumax(a + 23, NULL, 0);
1863                 else if (!strncmp(a, "--max-pack-size=", 16))
1864                         max_packsize = strtoumax(a + 16, NULL, 0) * 1024 * 1024;
1865                 else if (!strncmp(a, "--depth=", 8))
1866                         max_depth = strtoul(a + 8, NULL, 0);
1867                 else if (!strncmp(a, "--active-branches=", 18))
1868                         max_active_branches = strtoul(a + 18, NULL, 0);
1869                 else if (!strncmp(a, "--export-marks=", 15))
1870                         mark_file = a + 15;
1871                 else if (!strncmp(a, "--branch-log=", 13)) {
1872                         branch_log = fopen(a + 13, "w");
1873                         if (!branch_log)
1874                                 die("Can't create %s: %s", a + 13, strerror(errno));
1875                 }
1876                 else
1877                         die("unknown option %s", a);
1878         }
1879         if (i != argc)
1880                 usage(fast_import_usage);
1881
1882         alloc_objects(est_obj_cnt);
1883         strbuf_init(&command_buf);
1884
1885         atom_table = xcalloc(atom_table_sz, sizeof(struct atom_str*));
1886         branch_table = xcalloc(branch_table_sz, sizeof(struct branch*));
1887         avail_tree_table = xcalloc(avail_tree_table_sz, sizeof(struct avail_tree_content*));
1888         marks = pool_calloc(1, sizeof(struct mark_set));
1889
1890         start_packfile();
1891         for (;;) {
1892                 read_next_command();
1893                 if (command_buf.eof)
1894                         break;
1895                 else if (!strcmp("blob", command_buf.buf))
1896                         cmd_new_blob();
1897                 else if (!strncmp("commit ", command_buf.buf, 7))
1898                         cmd_new_commit();
1899                 else if (!strncmp("tag ", command_buf.buf, 4))
1900                         cmd_new_tag();
1901                 else if (!strncmp("reset ", command_buf.buf, 6))
1902                         cmd_reset_branch();
1903                 else if (!strcmp("checkpoint", command_buf.buf))
1904                         cmd_checkpoint();
1905                 else
1906                         die("Unsupported command: %s", command_buf.buf);
1907         }
1908         end_packfile();
1909
1910         dump_branches();
1911         dump_tags();
1912         unkeep_all_packs();
1913         dump_marks();
1914         if (branch_log)
1915                 fclose(branch_log);
1916
1917         duplicate_count = 0;
1918         for (i = 0; i < ARRAY_SIZE(duplicate_count_by_type); i++)
1919                 duplicate_count += duplicate_count_by_type[i];
1920
1921         fprintf(stderr, "%s statistics:\n", argv[0]);
1922         fprintf(stderr, "---------------------------------------------------------------------\n");
1923         fprintf(stderr, "Alloc'd objects: %10ju (%10ju overflow  )\n", alloc_count, alloc_count - est_obj_cnt);
1924         fprintf(stderr, "Total objects:   %10ju (%10ju duplicates                  )\n", object_count, duplicate_count);
1925         fprintf(stderr, "      blobs  :   %10ju (%10ju duplicates %10ju deltas)\n", object_count_by_type[OBJ_BLOB], duplicate_count_by_type[OBJ_BLOB], delta_count_by_type[OBJ_BLOB]);
1926         fprintf(stderr, "      trees  :   %10ju (%10ju duplicates %10ju deltas)\n", object_count_by_type[OBJ_TREE], duplicate_count_by_type[OBJ_TREE], delta_count_by_type[OBJ_TREE]);
1927         fprintf(stderr, "      commits:   %10ju (%10ju duplicates %10ju deltas)\n", object_count_by_type[OBJ_COMMIT], duplicate_count_by_type[OBJ_COMMIT], delta_count_by_type[OBJ_COMMIT]);
1928         fprintf(stderr, "      tags   :   %10ju (%10ju duplicates %10ju deltas)\n", object_count_by_type[OBJ_TAG], duplicate_count_by_type[OBJ_TAG], delta_count_by_type[OBJ_TAG]);
1929         fprintf(stderr, "Total branches:  %10lu (%10lu loads     )\n", branch_count, branch_load_count);
1930         fprintf(stderr, "      marks:     %10ju (%10ju unique    )\n", (((uintmax_t)1) << marks->shift) * 1024, marks_set_count);
1931         fprintf(stderr, "      atoms:     %10u\n", atom_cnt);
1932         fprintf(stderr, "Memory total:    %10ju KiB\n", (total_allocd + alloc_count*sizeof(struct object_entry))/1024);
1933         fprintf(stderr, "       pools:    %10lu KiB\n", total_allocd/1024);
1934         fprintf(stderr, "     objects:    %10ju KiB\n", (alloc_count*sizeof(struct object_entry))/1024);
1935         fprintf(stderr, "---------------------------------------------------------------------\n");
1936         pack_report();
1937         fprintf(stderr, "---------------------------------------------------------------------\n");
1938         fprintf(stderr, "\n");
1939
1940         return 0;
1941 }