git.oblomov.eu Git - git/blob - builtin/fast-import.c

   1 #include "builtin.h"
   2 #include "cache.h"
   3 #include "repository.h"
   4 #include "config.h"
   5 #include "lockfile.h"
   6 #include "object.h"
   7 #include "blob.h"
   8 #include "tree.h"
   9 #include "commit.h"
  10 #include "delta.h"
  11 #include "pack.h"
  12 #include "refs.h"
  13 #include "csum-file.h"
  14 #include "quote.h"
  15 #include "dir.h"
  16 #include "run-command.h"
  17 #include "packfile.h"
  18 #include "object-store.h"
  19 #include "mem-pool.h"
  20 #include "commit-reach.h"
  21 #include "khash.h"
  22
  23 #define PACK_ID_BITS 16
  24 #define MAX_PACK_ID ((1<<PACK_ID_BITS)-1)
  25 #define DEPTH_BITS 13
  26 #define MAX_DEPTH ((1<<DEPTH_BITS)-1)
  27
  28 /*
  29  * We abuse the setuid bit on directories to mean "do not delta".
  30  */
  31 #define NO_DELTA S_ISUID
  32
  33 /*
  34  * The amount of additional space required in order to write an object into the
  35  * current pack. This is the hash lengths at the end of the pack, plus the
  36  * length of one object ID.
  37  */
  38 #define PACK_SIZE_THRESHOLD (the_hash_algo->rawsz * 3)
  39
  40 struct object_entry {
  41         struct pack_idx_entry idx;
  42         struct hashmap_entry ent;
  43         uint32_t type : TYPE_BITS,
  44                 pack_id : PACK_ID_BITS,
  45                 depth : DEPTH_BITS;
  46 };
  47
  48 static int object_entry_hashcmp(const void *map_data,
  49                                 const struct hashmap_entry *eptr,
  50                                 const struct hashmap_entry *entry_or_key,
  51                                 const void *keydata)
  52 {
  53         const struct object_id *oid = keydata;
  54         const struct object_entry *e1, *e2;
  55
  56         e1 = container_of(eptr, const struct object_entry, ent);
  57         if (oid)
  58                 return oidcmp(&e1->idx.oid, oid);
  59
  60         e2 = container_of(entry_or_key, const struct object_entry, ent);
  61         return oidcmp(&e1->idx.oid, &e2->idx.oid);
  62 }
  63
  64 struct object_entry_pool {
  65         struct object_entry_pool *next_pool;
  66         struct object_entry *next_free;
  67         struct object_entry *end;
  68         struct object_entry entries[FLEX_ARRAY]; /* more */
  69 };
  70
  71 struct mark_set {
  72         union {
  73                 struct object_id *oids[1024];
  74                 struct object_entry *marked[1024];
  75                 struct mark_set *sets[1024];
  76         } data;
  77         unsigned int shift;
  78 };
  79
  80 struct last_object {
  81         struct strbuf data;
  82         off_t offset;
  83         unsigned int depth;
  84         unsigned no_swap : 1;
  85 };
  86
  87 struct atom_str {
  88         struct atom_str *next_atom;
  89         unsigned short str_len;
  90         char str_dat[FLEX_ARRAY]; /* more */
  91 };
  92
  93 struct tree_content;
  94 struct tree_entry {
  95         struct tree_content *tree;
  96         struct atom_str *name;
  97         struct tree_entry_ms {
  98                 uint16_t mode;
  99                 struct object_id oid;
 100         } versions[2];
 101 };
 102
 103 struct tree_content {
 104         unsigned int entry_capacity; /* must match avail_tree_content */
 105         unsigned int entry_count;
 106         unsigned int delta_depth;
 107         struct tree_entry *entries[FLEX_ARRAY]; /* more */
 108 };
 109
 110 struct avail_tree_content {
 111         unsigned int entry_capacity; /* must match tree_content */
 112         struct avail_tree_content *next_avail;
 113 };
 114
 115 struct branch {
 116         struct branch *table_next_branch;
 117         struct branch *active_next_branch;
 118         const char *name;
 119         struct tree_entry branch_tree;
 120         uintmax_t last_commit;
 121         uintmax_t num_notes;
 122         unsigned active : 1;
 123         unsigned delete : 1;
 124         unsigned pack_id : PACK_ID_BITS;
 125         struct object_id oid;
 126 };
 127
 128 struct tag {
 129         struct tag *next_tag;
 130         const char *name;
 131         unsigned int pack_id;
 132         struct object_id oid;
 133 };
 134
 135 struct hash_list {
 136         struct hash_list *next;
 137         struct object_id oid;
 138 };
 139
 140 typedef enum {
 141         WHENSPEC_RAW = 1,
 142         WHENSPEC_RAW_PERMISSIVE,
 143         WHENSPEC_RFC2822,
 144         WHENSPEC_NOW
 145 } whenspec_type;
 146
 147 struct recent_command {
 148         struct recent_command *prev;
 149         struct recent_command *next;
 150         char *buf;
 151 };
 152
 153 typedef void (*mark_set_inserter_t)(struct mark_set *s, struct object_id *oid, uintmax_t mark);
 154 typedef void (*each_mark_fn_t)(uintmax_t mark, void *obj, void *cbp);
 155
 156 /* Configured limits on output */
 157 static unsigned long max_depth = 50;
 158 static off_t max_packsize;
 159 static int unpack_limit = 100;
 160 static int force_update;
 161
 162 /* Stats and misc. counters */
 163 static uintmax_t alloc_count;
 164 static uintmax_t marks_set_count;
 165 static uintmax_t object_count_by_type[1 << TYPE_BITS];
 166 static uintmax_t duplicate_count_by_type[1 << TYPE_BITS];
 167 static uintmax_t delta_count_by_type[1 << TYPE_BITS];
 168 static uintmax_t delta_count_attempts_by_type[1 << TYPE_BITS];
 169 static unsigned long object_count;
 170 static unsigned long branch_count;
 171 static unsigned long branch_load_count;
 172 static int failure;
 173 static FILE *pack_edges;
 174 static unsigned int show_stats = 1;
 175 static int global_argc;
 176 static const char **global_argv;
 177
 178 /* Memory pools */
 179 static struct mem_pool fi_mem_pool =  {NULL, 2*1024*1024 -
 180                                        sizeof(struct mp_block), 0 };
 181
 182 /* Atom management */
 183 static unsigned int atom_table_sz = 4451;
 184 static unsigned int atom_cnt;
 185 static struct atom_str **atom_table;
 186
 187 /* The .pack file being generated */
 188 static struct pack_idx_option pack_idx_opts;
 189 static unsigned int pack_id;
 190 static struct hashfile *pack_file;
 191 static struct packed_git *pack_data;
 192 static struct packed_git **all_packs;
 193 static off_t pack_size;
 194
 195 /* Table of objects we've written. */
 196 static unsigned int object_entry_alloc = 5000;
 197 static struct object_entry_pool *blocks;
 198 static struct hashmap object_table;
 199 static struct mark_set *marks;
 200 static const char *export_marks_file;
 201 static const char *import_marks_file;
 202 static int import_marks_file_from_stream;
 203 static int import_marks_file_ignore_missing;
 204 static int import_marks_file_done;
 205 static int relative_marks_paths;
 206
 207 /* Our last blob */
 208 static struct last_object last_blob = { STRBUF_INIT, 0, 0, 0 };
 209
 210 /* Tree management */
 211 static unsigned int tree_entry_alloc = 1000;
 212 static void *avail_tree_entry;
 213 static unsigned int avail_tree_table_sz = 100;
 214 static struct avail_tree_content **avail_tree_table;
 215 static size_t tree_entry_allocd;
 216 static struct strbuf old_tree = STRBUF_INIT;
 217 static struct strbuf new_tree = STRBUF_INIT;
 218
 219 /* Branch data */
 220 static unsigned long max_active_branches = 5;
 221 static unsigned long cur_active_branches;
 222 static unsigned long branch_table_sz = 1039;
 223 static struct branch **branch_table;
 224 static struct branch *active_branches;
 225
 226 /* Tag data */
 227 static struct tag *first_tag;
 228 static struct tag *last_tag;
 229
 230 /* Input stream parsing */
 231 static whenspec_type whenspec = WHENSPEC_RAW;
 232 static struct strbuf command_buf = STRBUF_INIT;
 233 static int unread_command_buf;
 234 static struct recent_command cmd_hist = {&cmd_hist, &cmd_hist, NULL};
 235 static struct recent_command *cmd_tail = &cmd_hist;
 236 static struct recent_command *rc_free;
 237 static unsigned int cmd_save = 100;
 238 static uintmax_t next_mark;
 239 static struct strbuf new_data = STRBUF_INIT;
 240 static int seen_data_command;
 241 static int require_explicit_termination;
 242 static int allow_unsafe_features;
 243
 244 /* Signal handling */
 245 static volatile sig_atomic_t checkpoint_requested;
 246
 247 /* Submodule marks */
 248 static struct string_list sub_marks_from = STRING_LIST_INIT_DUP;
 249 static struct string_list sub_marks_to = STRING_LIST_INIT_DUP;
 250 static kh_oid_map_t *sub_oid_map;
 251
 252 /* Where to write output of cat-blob commands */
 253 static int cat_blob_fd = STDOUT_FILENO;
 254
 255 static void parse_argv(void);
 256 static void parse_get_mark(const char *p);
 257 static void parse_cat_blob(const char *p);
 258 static void parse_ls(const char *p, struct branch *b);
 259
 260 static void for_each_mark(struct mark_set *m, uintmax_t base, each_mark_fn_t callback, void *p)
 261 {
 262         uintmax_t k;
 263         if (m->shift) {
 264                 for (k = 0; k < 1024; k++) {
 265                         if (m->data.sets[k])
 266                                 for_each_mark(m->data.sets[k], base + (k << m->shift), callback, p);
 267                 }
 268         } else {
 269                 for (k = 0; k < 1024; k++) {
 270                         if (m->data.marked[k])
 271                                 callback(base + k, m->data.marked[k], p);
 272                 }
 273         }
 274 }
 275
 276 static void dump_marks_fn(uintmax_t mark, void *object, void *cbp) {
 277         struct object_entry *e = object;
 278         FILE *f = cbp;
 279
 280         fprintf(f, ":%" PRIuMAX " %s\n", mark, oid_to_hex(&e->idx.oid));
 281 }
 282
 283 static void write_branch_report(FILE *rpt, struct branch *b)
 284 {
 285         fprintf(rpt, "%s:\n", b->name);
 286
 287         fprintf(rpt, "  status      :");
 288         if (b->active)
 289                 fputs(" active", rpt);
 290         if (b->branch_tree.tree)
 291                 fputs(" loaded", rpt);
 292         if (is_null_oid(&b->branch_tree.versions[1].oid))
 293                 fputs(" dirty", rpt);
 294         fputc('\n', rpt);
 295
 296         fprintf(rpt, "  tip commit  : %s\n", oid_to_hex(&b->oid));
 297         fprintf(rpt, "  old tree    : %s\n",
 298                 oid_to_hex(&b->branch_tree.versions[0].oid));
 299         fprintf(rpt, "  cur tree    : %s\n",
 300                 oid_to_hex(&b->branch_tree.versions[1].oid));
 301         fprintf(rpt, "  commit clock: %" PRIuMAX "\n", b->last_commit);
 302
 303         fputs("  last pack   : ", rpt);
 304         if (b->pack_id < MAX_PACK_ID)
 305                 fprintf(rpt, "%u", b->pack_id);
 306         fputc('\n', rpt);
 307
 308         fputc('\n', rpt);
 309 }
 310
 311 static void write_crash_report(const char *err)
 312 {
 313         char *loc = git_pathdup("fast_import_crash_%"PRIuMAX, (uintmax_t) getpid());
 314         FILE *rpt = fopen(loc, "w");
 315         struct branch *b;
 316         unsigned long lu;
 317         struct recent_command *rc;
 318
 319         if (!rpt) {
 320                 error_errno("can't write crash report %s", loc);
 321                 free(loc);
 322                 return;
 323         }
 324
 325         fprintf(stderr, "fast-import: dumping crash report to %s\n", loc);
 326
 327         fprintf(rpt, "fast-import crash report:\n");
 328         fprintf(rpt, "    fast-import process: %"PRIuMAX"\n", (uintmax_t) getpid());
 329         fprintf(rpt, "    parent process     : %"PRIuMAX"\n", (uintmax_t) getppid());
 330         fprintf(rpt, "    at %s\n", show_date(time(NULL), 0, DATE_MODE(ISO8601)));
 331         fputc('\n', rpt);
 332
 333         fputs("fatal: ", rpt);
 334         fputs(err, rpt);
 335         fputc('\n', rpt);
 336
 337         fputc('\n', rpt);
 338         fputs("Most Recent Commands Before Crash\n", rpt);
 339         fputs("---------------------------------\n", rpt);
 340         for (rc = cmd_hist.next; rc != &cmd_hist; rc = rc->next) {
 341                 if (rc->next == &cmd_hist)
 342                         fputs("* ", rpt);
 343                 else
 344                         fputs("  ", rpt);
 345                 fputs(rc->buf, rpt);
 346                 fputc('\n', rpt);
 347         }
 348
 349         fputc('\n', rpt);
 350         fputs("Active Branch LRU\n", rpt);
 351         fputs("-----------------\n", rpt);
 352         fprintf(rpt, "    active_branches = %lu cur, %lu max\n",
 353                 cur_active_branches,
 354                 max_active_branches);
 355         fputc('\n', rpt);
 356         fputs("  pos  clock name\n", rpt);
 357         fputs("  ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~\n", rpt);
 358         for (b = active_branches, lu = 0; b; b = b->active_next_branch)
 359                 fprintf(rpt, "  %2lu) %6" PRIuMAX" %s\n",
 360                         ++lu, b->last_commit, b->name);
 361
 362         fputc('\n', rpt);
 363         fputs("Inactive Branches\n", rpt);
 364         fputs("-----------------\n", rpt);
 365         for (lu = 0; lu < branch_table_sz; lu++) {
 366                 for (b = branch_table[lu]; b; b = b->table_next_branch)
 367                         write_branch_report(rpt, b);
 368         }
 369
 370         if (first_tag) {
 371                 struct tag *tg;
 372                 fputc('\n', rpt);
 373                 fputs("Annotated Tags\n", rpt);
 374                 fputs("--------------\n", rpt);
 375                 for (tg = first_tag; tg; tg = tg->next_tag) {
 376                         fputs(oid_to_hex(&tg->oid), rpt);
 377                         fputc(' ', rpt);
 378                         fputs(tg->name, rpt);
 379                         fputc('\n', rpt);
 380                 }
 381         }
 382
 383         fputc('\n', rpt);
 384         fputs("Marks\n", rpt);
 385         fputs("-----\n", rpt);
 386         if (export_marks_file)
 387                 fprintf(rpt, "  exported to %s\n", export_marks_file);
 388         else
 389                 for_each_mark(marks, 0, dump_marks_fn, rpt);
 390
 391         fputc('\n', rpt);
 392         fputs("-------------------\n", rpt);
 393         fputs("END OF CRASH REPORT\n", rpt);
 394         fclose(rpt);
 395         free(loc);
 396 }
 397
 398 static void end_packfile(void);
 399 static void unkeep_all_packs(void);
 400 static void dump_marks(void);
 401
 402 static NORETURN void die_nicely(const char *err, va_list params)
 403 {
 404         static int zombie;
 405         char message[2 * PATH_MAX];
 406
 407         vsnprintf(message, sizeof(message), err, params);
 408         fputs("fatal: ", stderr);
 409         fputs(message, stderr);
 410         fputc('\n', stderr);
 411
 412         if (!zombie) {
 413                 zombie = 1;
 414                 write_crash_report(message);
 415                 end_packfile();
 416                 unkeep_all_packs();
 417                 dump_marks();
 418         }
 419         exit(128);
 420 }
 421
 422 #ifndef SIGUSR1 /* Windows, for example */
 423
 424 static void set_checkpoint_signal(void)
 425 {
 426 }
 427
 428 #else
 429
 430 static void checkpoint_signal(int signo)
 431 {
 432         checkpoint_requested = 1;
 433 }
 434
 435 static void set_checkpoint_signal(void)
 436 {
 437         struct sigaction sa;
 438
 439         memset(&sa, 0, sizeof(sa));
 440         sa.sa_handler = checkpoint_signal;
 441         sigemptyset(&sa.sa_mask);
 442         sa.sa_flags = SA_RESTART;
 443         sigaction(SIGUSR1, &sa, NULL);
 444 }
 445
 446 #endif
 447
 448 static void alloc_objects(unsigned int cnt)
 449 {
 450         struct object_entry_pool *b;
 451
 452         b = xmalloc(sizeof(struct object_entry_pool)
 453                 + cnt * sizeof(struct object_entry));
 454         b->next_pool = blocks;
 455         b->next_free = b->entries;
 456         b->end = b->entries + cnt;
 457         blocks = b;
 458         alloc_count += cnt;
 459 }
 460
 461 static struct object_entry *new_object(struct object_id *oid)
 462 {
 463         struct object_entry *e;
 464
 465         if (blocks->next_free == blocks->end)
 466                 alloc_objects(object_entry_alloc);
 467
 468         e = blocks->next_free++;
 469         oidcpy(&e->idx.oid, oid);
 470         return e;
 471 }
 472
 473 static struct object_entry *find_object(struct object_id *oid)
 474 {
 475         return hashmap_get_entry_from_hash(&object_table, oidhash(oid), oid,
 476                                            struct object_entry, ent);
 477 }
 478
 479 static struct object_entry *insert_object(struct object_id *oid)
 480 {
 481         struct object_entry *e;
 482         unsigned int hash = oidhash(oid);
 483
 484         e = hashmap_get_entry_from_hash(&object_table, hash, oid,
 485                                         struct object_entry, ent);
 486         if (!e) {
 487                 e = new_object(oid);
 488                 e->idx.offset = 0;
 489                 hashmap_entry_init(&e->ent, hash);
 490                 hashmap_add(&object_table, &e->ent);
 491         }
 492
 493         return e;
 494 }
 495
 496 static void invalidate_pack_id(unsigned int id)
 497 {
 498         unsigned long lu;
 499         struct tag *t;
 500         struct hashmap_iter iter;
 501         struct object_entry *e;
 502
 503         hashmap_for_each_entry(&object_table, &iter, e, ent) {
 504                 if (e->pack_id == id)
 505                         e->pack_id = MAX_PACK_ID;
 506         }
 507
 508         for (lu = 0; lu < branch_table_sz; lu++) {
 509                 struct branch *b;
 510
 511                 for (b = branch_table[lu]; b; b = b->table_next_branch)
 512                         if (b->pack_id == id)
 513                                 b->pack_id = MAX_PACK_ID;
 514         }
 515
 516         for (t = first_tag; t; t = t->next_tag)
 517                 if (t->pack_id == id)
 518                         t->pack_id = MAX_PACK_ID;
 519 }
 520
 521 static unsigned int hc_str(const char *s, size_t len)
 522 {
 523         unsigned int r = 0;
 524         while (len-- > 0)
 525                 r = r * 31 + *s++;
 526         return r;
 527 }
 528
 529 static void insert_mark(struct mark_set *s, uintmax_t idnum, struct object_entry *oe)
 530 {
 531         while ((idnum >> s->shift) >= 1024) {
 532                 s = mem_pool_calloc(&fi_mem_pool, 1, sizeof(struct mark_set));
 533                 s->shift = marks->shift + 10;
 534                 s->data.sets[0] = marks;
 535                 marks = s;
 536         }
 537         while (s->shift) {
 538                 uintmax_t i = idnum >> s->shift;
 539                 idnum -= i << s->shift;
 540                 if (!s->data.sets[i]) {
 541                         s->data.sets[i] = mem_pool_calloc(&fi_mem_pool, 1, sizeof(struct mark_set));
 542                         s->data.sets[i]->shift = s->shift - 10;
 543                 }
 544                 s = s->data.sets[i];
 545         }
 546         if (!s->data.marked[idnum])
 547                 marks_set_count++;
 548         s->data.marked[idnum] = oe;
 549 }
 550
 551 static void *find_mark(struct mark_set *s, uintmax_t idnum)
 552 {
 553         uintmax_t orig_idnum = idnum;
 554         struct object_entry *oe = NULL;
 555         if ((idnum >> s->shift) < 1024) {
 556                 while (s && s->shift) {
 557                         uintmax_t i = idnum >> s->shift;
 558                         idnum -= i << s->shift;
 559                         s = s->data.sets[i];
 560                 }
 561                 if (s)
 562                         oe = s->data.marked[idnum];
 563         }
 564         if (!oe)
 565                 die("mark :%" PRIuMAX " not declared", orig_idnum);
 566         return oe;
 567 }
 568
 569 static struct atom_str *to_atom(const char *s, unsigned short len)
 570 {
 571         unsigned int hc = hc_str(s, len) % atom_table_sz;
 572         struct atom_str *c;
 573
 574         for (c = atom_table[hc]; c; c = c->next_atom)
 575                 if (c->str_len == len && !strncmp(s, c->str_dat, len))
 576                         return c;
 577
 578         c = mem_pool_alloc(&fi_mem_pool, sizeof(struct atom_str) + len + 1);
 579         c->str_len = len;
 580         memcpy(c->str_dat, s, len);
 581         c->str_dat[len] = 0;
 582         c->next_atom = atom_table[hc];
 583         atom_table[hc] = c;
 584         atom_cnt++;
 585         return c;
 586 }
 587
 588 static struct branch *lookup_branch(const char *name)
 589 {
 590         unsigned int hc = hc_str(name, strlen(name)) % branch_table_sz;
 591         struct branch *b;
 592
 593         for (b = branch_table[hc]; b; b = b->table_next_branch)
 594                 if (!strcmp(name, b->name))
 595                         return b;
 596         return NULL;
 597 }
 598
 599 static struct branch *new_branch(const char *name)
 600 {
 601         unsigned int hc = hc_str(name, strlen(name)) % branch_table_sz;
 602         struct branch *b = lookup_branch(name);
 603
 604         if (b)
 605                 die("Invalid attempt to create duplicate branch: %s", name);
 606         if (check_refname_format(name, REFNAME_ALLOW_ONELEVEL))
 607                 die("Branch name doesn't conform to GIT standards: %s", name);
 608
 609         b = mem_pool_calloc(&fi_mem_pool, 1, sizeof(struct branch));
 610         b->name = mem_pool_strdup(&fi_mem_pool, name);
 611         b->table_next_branch = branch_table[hc];
 612         b->branch_tree.versions[0].mode = S_IFDIR;
 613         b->branch_tree.versions[1].mode = S_IFDIR;
 614         b->num_notes = 0;
 615         b->active = 0;
 616         b->pack_id = MAX_PACK_ID;
 617         branch_table[hc] = b;
 618         branch_count++;
 619         return b;
 620 }
 621
 622 static unsigned int hc_entries(unsigned int cnt)
 623 {
 624         cnt = cnt & 7 ? (cnt / 8) + 1 : cnt / 8;
 625         return cnt < avail_tree_table_sz ? cnt : avail_tree_table_sz - 1;
 626 }
 627
 628 static struct tree_content *new_tree_content(unsigned int cnt)
 629 {
 630         struct avail_tree_content *f, *l = NULL;
 631         struct tree_content *t;
 632         unsigned int hc = hc_entries(cnt);
 633
 634         for (f = avail_tree_table[hc]; f; l = f, f = f->next_avail)
 635                 if (f->entry_capacity >= cnt)
 636                         break;
 637
 638         if (f) {
 639                 if (l)
 640                         l->next_avail = f->next_avail;
 641                 else
 642                         avail_tree_table[hc] = f->next_avail;
 643         } else {
 644                 cnt = cnt & 7 ? ((cnt / 8) + 1) * 8 : cnt;
 645                 f = mem_pool_alloc(&fi_mem_pool, sizeof(*t) + sizeof(t->entries[0]) * cnt);
 646                 f->entry_capacity = cnt;
 647         }
 648
 649         t = (struct tree_content*)f;
 650         t->entry_count = 0;
 651         t->delta_depth = 0;
 652         return t;
 653 }
 654
 655 static void release_tree_entry(struct tree_entry *e);
 656 static void release_tree_content(struct tree_content *t)
 657 {
 658         struct avail_tree_content *f = (struct avail_tree_content*)t;
 659         unsigned int hc = hc_entries(f->entry_capacity);
 660         f->next_avail = avail_tree_table[hc];
 661         avail_tree_table[hc] = f;
 662 }
 663
 664 static void release_tree_content_recursive(struct tree_content *t)
 665 {
 666         unsigned int i;
 667         for (i = 0; i < t->entry_count; i++)
 668                 release_tree_entry(t->entries[i]);
 669         release_tree_content(t);
 670 }
 671
 672 static struct tree_content *grow_tree_content(
 673         struct tree_content *t,
 674         int amt)
 675 {
 676         struct tree_content *r = new_tree_content(t->entry_count + amt);
 677         r->entry_count = t->entry_count;
 678         r->delta_depth = t->delta_depth;
 679         COPY_ARRAY(r->entries, t->entries, t->entry_count);
 680         release_tree_content(t);
 681         return r;
 682 }
 683
 684 static struct tree_entry *new_tree_entry(void)
 685 {
 686         struct tree_entry *e;
 687
 688         if (!avail_tree_entry) {
 689                 unsigned int n = tree_entry_alloc;
 690                 tree_entry_allocd += n * sizeof(struct tree_entry);
 691                 ALLOC_ARRAY(e, n);
 692                 avail_tree_entry = e;
 693                 while (n-- > 1) {
 694                         *((void**)e) = e + 1;
 695                         e++;
 696                 }
 697                 *((void**)e) = NULL;
 698         }
 699
 700         e = avail_tree_entry;
 701         avail_tree_entry = *((void**)e);
 702         return e;
 703 }
 704
 705 static void release_tree_entry(struct tree_entry *e)
 706 {
 707         if (e->tree)
 708                 release_tree_content_recursive(e->tree);
 709         *((void**)e) = avail_tree_entry;
 710         avail_tree_entry = e;
 711 }
 712
 713 static struct tree_content *dup_tree_content(struct tree_content *s)
 714 {
 715         struct tree_content *d;
 716         struct tree_entry *a, *b;
 717         unsigned int i;
 718
 719         if (!s)
 720                 return NULL;
 721         d = new_tree_content(s->entry_count);
 722         for (i = 0; i < s->entry_count; i++) {
 723                 a = s->entries[i];
 724                 b = new_tree_entry();
 725                 memcpy(b, a, sizeof(*a));
 726                 if (a->tree && is_null_oid(&b->versions[1].oid))
 727                         b->tree = dup_tree_content(a->tree);
 728                 else
 729                         b->tree = NULL;
 730                 d->entries[i] = b;
 731         }
 732         d->entry_count = s->entry_count;
 733         d->delta_depth = s->delta_depth;
 734
 735         return d;
 736 }
 737
 738 static void start_packfile(void)
 739 {
 740         struct strbuf tmp_file = STRBUF_INIT;
 741         struct packed_git *p;
 742         struct pack_header hdr;
 743         int pack_fd;
 744
 745         pack_fd = odb_mkstemp(&tmp_file, "pack/tmp_pack_XXXXXX");
 746         FLEX_ALLOC_STR(p, pack_name, tmp_file.buf);
 747         strbuf_release(&tmp_file);
 748
 749         p->pack_fd = pack_fd;
 750         p->do_not_close = 1;
 751         pack_file = hashfd(pack_fd, p->pack_name);
 752
 753         hdr.hdr_signature = htonl(PACK_SIGNATURE);
 754         hdr.hdr_version = htonl(2);
 755         hdr.hdr_entries = 0;
 756         hashwrite(pack_file, &hdr, sizeof(hdr));
 757
 758         pack_data = p;
 759         pack_size = sizeof(hdr);
 760         object_count = 0;
 761
 762         REALLOC_ARRAY(all_packs, pack_id + 1);
 763         all_packs[pack_id] = p;
 764 }
 765
 766 static const char *create_index(void)
 767 {
 768         const char *tmpfile;
 769         struct pack_idx_entry **idx, **c, **last;
 770         struct object_entry *e;
 771         struct object_entry_pool *o;
 772
 773         /* Build the table of object IDs. */
 774         ALLOC_ARRAY(idx, object_count);
 775         c = idx;
 776         for (o = blocks; o; o = o->next_pool)
 777                 for (e = o->next_free; e-- != o->entries;)
 778                         if (pack_id == e->pack_id)
 779                                 *c++ = &e->idx;
 780         last = idx + object_count;
 781         if (c != last)
 782                 die("internal consistency error creating the index");
 783
 784         tmpfile = write_idx_file(NULL, idx, object_count, &pack_idx_opts,
 785                                  pack_data->hash);
 786         free(idx);
 787         return tmpfile;
 788 }
 789
 790 static char *keep_pack(const char *curr_index_name)
 791 {
 792         static const char *keep_msg = "fast-import";
 793         struct strbuf name = STRBUF_INIT;
 794         int keep_fd;
 795
 796         odb_pack_name(&name, pack_data->hash, "keep");
 797         keep_fd = odb_pack_keep(name.buf);
 798         if (keep_fd < 0)
 799                 die_errno("cannot create keep file");
 800         write_or_die(keep_fd, keep_msg, strlen(keep_msg));
 801         if (close(keep_fd))
 802                 die_errno("failed to write keep file");
 803
 804         odb_pack_name(&name, pack_data->hash, "pack");
 805         if (finalize_object_file(pack_data->pack_name, name.buf))
 806                 die("cannot store pack file");
 807
 808         odb_pack_name(&name, pack_data->hash, "idx");
 809         if (finalize_object_file(curr_index_name, name.buf))
 810                 die("cannot store index file");
 811         free((void *)curr_index_name);
 812         return strbuf_detach(&name, NULL);
 813 }
 814
 815 static void unkeep_all_packs(void)
 816 {
 817         struct strbuf name = STRBUF_INIT;
 818         int k;
 819
 820         for (k = 0; k < pack_id; k++) {
 821                 struct packed_git *p = all_packs[k];
 822                 odb_pack_name(&name, p->hash, "keep");
 823                 unlink_or_warn(name.buf);
 824         }
 825         strbuf_release(&name);
 826 }
 827
 828 static int loosen_small_pack(const struct packed_git *p)
 829 {
 830         struct child_process unpack = CHILD_PROCESS_INIT;
 831
 832         if (lseek(p->pack_fd, 0, SEEK_SET) < 0)
 833                 die_errno("Failed seeking to start of '%s'", p->pack_name);
 834
 835         unpack.in = p->pack_fd;
 836         unpack.git_cmd = 1;
 837         unpack.stdout_to_stderr = 1;
 838         strvec_push(&unpack.args, "unpack-objects");
 839         if (!show_stats)
 840                 strvec_push(&unpack.args, "-q");
 841
 842         return run_command(&unpack);
 843 }
 844
 845 static void end_packfile(void)
 846 {
 847         static int running;
 848
 849         if (running || !pack_data)
 850                 return;
 851
 852         running = 1;
 853         clear_delta_base_cache();
 854         if (object_count) {
 855                 struct packed_git *new_p;
 856                 struct object_id cur_pack_oid;
 857                 char *idx_name;
 858                 int i;
 859                 struct branch *b;
 860                 struct tag *t;
 861
 862                 close_pack_windows(pack_data);
 863                 finalize_hashfile(pack_file, cur_pack_oid.hash, 0);
 864                 fixup_pack_header_footer(pack_data->pack_fd, pack_data->hash,
 865                                          pack_data->pack_name, object_count,
 866                                          cur_pack_oid.hash, pack_size);
 867
 868                 if (object_count <= unpack_limit) {
 869                         if (!loosen_small_pack(pack_data)) {
 870                                 invalidate_pack_id(pack_id);
 871                                 goto discard_pack;
 872                         }
 873                 }
 874
 875                 close(pack_data->pack_fd);
 876                 idx_name = keep_pack(create_index());
 877
 878                 /* Register the packfile with core git's machinery. */
 879                 new_p = add_packed_git(idx_name, strlen(idx_name), 1);
 880                 if (!new_p)
 881                         die("core git rejected index %s", idx_name);
 882                 all_packs[pack_id] = new_p;
 883                 install_packed_git(the_repository, new_p);
 884                 free(idx_name);
 885
 886                 /* Print the boundary */
 887                 if (pack_edges) {
 888                         fprintf(pack_edges, "%s:", new_p->pack_name);
 889                         for (i = 0; i < branch_table_sz; i++) {
 890                                 for (b = branch_table[i]; b; b = b->table_next_branch) {
 891                                         if (b->pack_id == pack_id)
 892                                                 fprintf(pack_edges, " %s",
 893                                                         oid_to_hex(&b->oid));
 894                                 }
 895                         }
 896                         for (t = first_tag; t; t = t->next_tag) {
 897                                 if (t->pack_id == pack_id)
 898                                         fprintf(pack_edges, " %s",
 899                                                 oid_to_hex(&t->oid));
 900                         }
 901                         fputc('\n', pack_edges);
 902                         fflush(pack_edges);
 903                 }
 904
 905                 pack_id++;
 906         }
 907         else {
 908 discard_pack:
 909                 close(pack_data->pack_fd);
 910                 unlink_or_warn(pack_data->pack_name);
 911         }
 912         FREE_AND_NULL(pack_data);
 913         running = 0;
 914
 915         /* We can't carry a delta across packfiles. */
 916         strbuf_release(&last_blob.data);
 917         last_blob.offset = 0;
 918         last_blob.depth = 0;
 919 }
 920
 921 static void cycle_packfile(void)
 922 {
 923         end_packfile();
 924         start_packfile();
 925 }
 926
 927 static int store_object(
 928         enum object_type type,
 929         struct strbuf *dat,
 930         struct last_object *last,
 931         struct object_id *oidout,
 932         uintmax_t mark)
 933 {
 934         void *out, *delta;
 935         struct object_entry *e;
 936         unsigned char hdr[96];
 937         struct object_id oid;
 938         unsigned long hdrlen, deltalen;
 939         git_hash_ctx c;
 940         git_zstream s;
 941
 942         hdrlen = xsnprintf((char *)hdr, sizeof(hdr), "%s %lu",
 943                            type_name(type), (unsigned long)dat->len) + 1;
 944         the_hash_algo->init_fn(&c);
 945         the_hash_algo->update_fn(&c, hdr, hdrlen);
 946         the_hash_algo->update_fn(&c, dat->buf, dat->len);
 947         the_hash_algo->final_fn(oid.hash, &c);
 948         if (oidout)
 949                 oidcpy(oidout, &oid);
 950
 951         e = insert_object(&oid);
 952         if (mark)
 953                 insert_mark(marks, mark, e);
 954         if (e->idx.offset) {
 955                 duplicate_count_by_type[type]++;
 956                 return 1;
 957         } else if (find_sha1_pack(oid.hash,
 958                                   get_all_packs(the_repository))) {
 959                 e->type = type;
 960                 e->pack_id = MAX_PACK_ID;
 961                 e->idx.offset = 1; /* just not zero! */
 962                 duplicate_count_by_type[type]++;
 963                 return 1;
 964         }
 965
 966         if (last && last->data.len && last->data.buf && last->depth < max_depth
 967                 && dat->len > the_hash_algo->rawsz) {
 968
 969                 delta_count_attempts_by_type[type]++;
 970                 delta = diff_delta(last->data.buf, last->data.len,
 971                         dat->buf, dat->len,
 972                         &deltalen, dat->len - the_hash_algo->rawsz);
 973         } else
 974                 delta = NULL;
 975
 976         git_deflate_init(&s, pack_compression_level);
 977         if (delta) {
 978                 s.next_in = delta;
 979                 s.avail_in = deltalen;
 980         } else {
 981                 s.next_in = (void *)dat->buf;
 982                 s.avail_in = dat->len;
 983         }
 984         s.avail_out = git_deflate_bound(&s, s.avail_in);
 985         s.next_out = out = xmalloc(s.avail_out);
 986         while (git_deflate(&s, Z_FINISH) == Z_OK)
 987                 ; /* nothing */
 988         git_deflate_end(&s);
 989
 990         /* Determine if we should auto-checkpoint. */
 991         if ((max_packsize
 992                 && (pack_size + PACK_SIZE_THRESHOLD + s.total_out) > max_packsize)
 993                 || (pack_size + PACK_SIZE_THRESHOLD + s.total_out) < pack_size) {
 994
 995                 /* This new object needs to *not* have the current pack_id. */
 996                 e->pack_id = pack_id + 1;
 997                 cycle_packfile();
 998
 999                 /* We cannot carry a delta into the new pack. */
1000                 if (delta) {
1001                         FREE_AND_NULL(delta);
1002
1003                         git_deflate_init(&s, pack_compression_level);
1004                         s.next_in = (void *)dat->buf;
1005                         s.avail_in = dat->len;
1006                         s.avail_out = git_deflate_bound(&s, s.avail_in);
1007                         s.next_out = out = xrealloc(out, s.avail_out);
1008                         while (git_deflate(&s, Z_FINISH) == Z_OK)
1009                                 ; /* nothing */
1010                         git_deflate_end(&s);
1011                 }
1012         }
1013
1014         e->type = type;
1015         e->pack_id = pack_id;
1016         e->idx.offset = pack_size;
1017         object_count++;
1018         object_count_by_type[type]++;
1019
1020         crc32_begin(pack_file);
1021
1022         if (delta) {
1023                 off_t ofs = e->idx.offset - last->offset;
1024                 unsigned pos = sizeof(hdr) - 1;
1025
1026                 delta_count_by_type[type]++;
1027                 e->depth = last->depth + 1;
1028
1029                 hdrlen = encode_in_pack_object_header(hdr, sizeof(hdr),
1030                                                       OBJ_OFS_DELTA, deltalen);
1031                 hashwrite(pack_file, hdr, hdrlen);
1032                 pack_size += hdrlen;
1033
1034                 hdr[pos] = ofs & 127;
1035                 while (ofs >>= 7)
1036                         hdr[--pos] = 128 | (--ofs & 127);
1037                 hashwrite(pack_file, hdr + pos, sizeof(hdr) - pos);
1038                 pack_size += sizeof(hdr) - pos;
1039         } else {
1040                 e->depth = 0;
1041                 hdrlen = encode_in_pack_object_header(hdr, sizeof(hdr),
1042                                                       type, dat->len);
1043                 hashwrite(pack_file, hdr, hdrlen);
1044                 pack_size += hdrlen;
1045         }
1046
1047         hashwrite(pack_file, out, s.total_out);
1048         pack_size += s.total_out;
1049
1050         e->idx.crc32 = crc32_end(pack_file);
1051
1052         free(out);
1053         free(delta);
1054         if (last) {
1055                 if (last->no_swap) {
1056                         last->data = *dat;
1057                 } else {
1058                         strbuf_swap(&last->data, dat);
1059                 }
1060                 last->offset = e->idx.offset;
1061                 last->depth = e->depth;
1062         }
1063         return 0;
1064 }
1065
1066 static void truncate_pack(struct hashfile_checkpoint *checkpoint)
1067 {
1068         if (hashfile_truncate(pack_file, checkpoint))
1069                 die_errno("cannot truncate pack to skip duplicate");
1070         pack_size = checkpoint->offset;
1071 }
1072
1073 static void stream_blob(uintmax_t len, struct object_id *oidout, uintmax_t mark)
1074 {
1075         size_t in_sz = 64 * 1024, out_sz = 64 * 1024;
1076         unsigned char *in_buf = xmalloc(in_sz);
1077         unsigned char *out_buf = xmalloc(out_sz);
1078         struct object_entry *e;
1079         struct object_id oid;
1080         unsigned long hdrlen;
1081         off_t offset;
1082         git_hash_ctx c;
1083         git_zstream s;
1084         struct hashfile_checkpoint checkpoint;
1085         int status = Z_OK;
1086
1087         /* Determine if we should auto-checkpoint. */
1088         if ((max_packsize
1089                 && (pack_size + PACK_SIZE_THRESHOLD + len) > max_packsize)
1090                 || (pack_size + PACK_SIZE_THRESHOLD + len) < pack_size)
1091                 cycle_packfile();
1092
1093         hashfile_checkpoint(pack_file, &checkpoint);
1094         offset = checkpoint.offset;
1095
1096         hdrlen = xsnprintf((char *)out_buf, out_sz, "blob %" PRIuMAX, len) + 1;
1097
1098         the_hash_algo->init_fn(&c);
1099         the_hash_algo->update_fn(&c, out_buf, hdrlen);
1100
1101         crc32_begin(pack_file);
1102
1103         git_deflate_init(&s, pack_compression_level);
1104
1105         hdrlen = encode_in_pack_object_header(out_buf, out_sz, OBJ_BLOB, len);
1106
1107         s.next_out = out_buf + hdrlen;
1108         s.avail_out = out_sz - hdrlen;
1109
1110         while (status != Z_STREAM_END) {
1111                 if (0 < len && !s.avail_in) {
1112                         size_t cnt = in_sz < len ? in_sz : (size_t)len;
1113                         size_t n = fread(in_buf, 1, cnt, stdin);
1114                         if (!n && feof(stdin))
1115                                 die("EOF in data (%" PRIuMAX " bytes remaining)", len);
1116
1117                         the_hash_algo->update_fn(&c, in_buf, n);
1118                         s.next_in = in_buf;
1119                         s.avail_in = n;
1120                         len -= n;
1121                 }
1122
1123                 status = git_deflate(&s, len ? 0 : Z_FINISH);
1124
1125                 if (!s.avail_out || status == Z_STREAM_END) {
1126                         size_t n = s.next_out - out_buf;
1127                         hashwrite(pack_file, out_buf, n);
1128                         pack_size += n;
1129                         s.next_out = out_buf;
1130                         s.avail_out = out_sz;
1131                 }
1132
1133                 switch (status) {
1134                 case Z_OK:
1135                 case Z_BUF_ERROR:
1136                 case Z_STREAM_END:
1137                         continue;
1138                 default:
1139                         die("unexpected deflate failure: %d", status);
1140                 }
1141         }
1142         git_deflate_end(&s);
1143         the_hash_algo->final_fn(oid.hash, &c);
1144
1145         if (oidout)
1146                 oidcpy(oidout, &oid);
1147
1148         e = insert_object(&oid);
1149
1150         if (mark)
1151                 insert_mark(marks, mark, e);
1152
1153         if (e->idx.offset) {
1154                 duplicate_count_by_type[OBJ_BLOB]++;
1155                 truncate_pack(&checkpoint);
1156
1157         } else if (find_sha1_pack(oid.hash,
1158                                   get_all_packs(the_repository))) {
1159                 e->type = OBJ_BLOB;
1160                 e->pack_id = MAX_PACK_ID;
1161                 e->idx.offset = 1; /* just not zero! */
1162                 duplicate_count_by_type[OBJ_BLOB]++;
1163                 truncate_pack(&checkpoint);
1164
1165         } else {
1166                 e->depth = 0;
1167                 e->type = OBJ_BLOB;
1168                 e->pack_id = pack_id;
1169                 e->idx.offset = offset;
1170                 e->idx.crc32 = crc32_end(pack_file);
1171                 object_count++;
1172                 object_count_by_type[OBJ_BLOB]++;
1173         }
1174
1175         free(in_buf);
1176         free(out_buf);
1177 }
1178
1179 /* All calls must be guarded by find_object() or find_mark() to
1180  * ensure the 'struct object_entry' passed was written by this
1181  * process instance.  We unpack the entry by the offset, avoiding
1182  * the need for the corresponding .idx file.  This unpacking rule
1183  * works because we only use OBJ_REF_DELTA within the packfiles
1184  * created by fast-import.
1185  *
1186  * oe must not be NULL.  Such an oe usually comes from giving
1187  * an unknown SHA-1 to find_object() or an undefined mark to
1188  * find_mark().  Callers must test for this condition and use
1189  * the standard read_sha1_file() when it happens.
1190  *
1191  * oe->pack_id must not be MAX_PACK_ID.  Such an oe is usually from
1192  * find_mark(), where the mark was reloaded from an existing marks
1193  * file and is referencing an object that this fast-import process
1194  * instance did not write out to a packfile.  Callers must test for
1195  * this condition and use read_sha1_file() instead.
1196  */
1197 static void *gfi_unpack_entry(
1198         struct object_entry *oe,
1199         unsigned long *sizep)
1200 {
1201         enum object_type type;
1202         struct packed_git *p = all_packs[oe->pack_id];
1203         if (p == pack_data && p->pack_size < (pack_size + the_hash_algo->rawsz)) {
1204                 /* The object is stored in the packfile we are writing to
1205                  * and we have modified it since the last time we scanned
1206                  * back to read a previously written object.  If an old
1207                  * window covered [p->pack_size, p->pack_size + rawsz) its
1208                  * data is stale and is not valid.  Closing all windows
1209                  * and updating the packfile length ensures we can read
1210                  * the newly written data.
1211                  */
1212                 close_pack_windows(p);
1213                 hashflush(pack_file);
1214
1215                 /* We have to offer rawsz bytes additional on the end of
1216                  * the packfile as the core unpacker code assumes the
1217                  * footer is present at the file end and must promise
1218                  * at least rawsz bytes within any window it maps.  But
1219                  * we don't actually create the footer here.
1220                  */
1221                 p->pack_size = pack_size + the_hash_algo->rawsz;
1222         }
1223         return unpack_entry(the_repository, p, oe->idx.offset, &type, sizep);
1224 }
1225
1226 static const char *get_mode(const char *str, uint16_t *modep)
1227 {
1228         unsigned char c;
1229         uint16_t mode = 0;
1230
1231         while ((c = *str++) != ' ') {
1232                 if (c < '0' || c > '7')
1233                         return NULL;
1234                 mode = (mode << 3) + (c - '0');
1235         }
1236         *modep = mode;
1237         return str;
1238 }
1239
1240 static void load_tree(struct tree_entry *root)
1241 {
1242         struct object_id *oid = &root->versions[1].oid;
1243         struct object_entry *myoe;
1244         struct tree_content *t;
1245         unsigned long size;
1246         char *buf;
1247         const char *c;
1248
1249         root->tree = t = new_tree_content(8);
1250         if (is_null_oid(oid))
1251                 return;
1252
1253         myoe = find_object(oid);
1254         if (myoe && myoe->pack_id != MAX_PACK_ID) {
1255                 if (myoe->type != OBJ_TREE)
1256                         die("Not a tree: %s", oid_to_hex(oid));
1257                 t->delta_depth = myoe->depth;
1258                 buf = gfi_unpack_entry(myoe, &size);
1259                 if (!buf)
1260                         die("Can't load tree %s", oid_to_hex(oid));
1261         } else {
1262                 enum object_type type;
1263                 buf = read_object_file(oid, &type, &size);
1264                 if (!buf || type != OBJ_TREE)
1265                         die("Can't load tree %s", oid_to_hex(oid));
1266         }
1267
1268         c = buf;
1269         while (c != (buf + size)) {
1270                 struct tree_entry *e = new_tree_entry();
1271
1272                 if (t->entry_count == t->entry_capacity)
1273                         root->tree = t = grow_tree_content(t, t->entry_count);
1274                 t->entries[t->entry_count++] = e;
1275
1276                 e->tree = NULL;
1277                 c = get_mode(c, &e->versions[1].mode);
1278                 if (!c)
1279                         die("Corrupt mode in %s", oid_to_hex(oid));
1280                 e->versions[0].mode = e->versions[1].mode;
1281                 e->name = to_atom(c, strlen(c));
1282                 c += e->name->str_len + 1;
1283                 hashcpy(e->versions[0].oid.hash, (unsigned char *)c);
1284                 hashcpy(e->versions[1].oid.hash, (unsigned char *)c);
1285                 c += the_hash_algo->rawsz;
1286         }
1287         free(buf);
1288 }
1289
1290 static int tecmp0 (const void *_a, const void *_b)
1291 {
1292         struct tree_entry *a = *((struct tree_entry**)_a);
1293         struct tree_entry *b = *((struct tree_entry**)_b);
1294         return base_name_compare(
1295                 a->name->str_dat, a->name->str_len, a->versions[0].mode,
1296                 b->name->str_dat, b->name->str_len, b->versions[0].mode);
1297 }
1298
1299 static int tecmp1 (const void *_a, const void *_b)
1300 {
1301         struct tree_entry *a = *((struct tree_entry**)_a);
1302         struct tree_entry *b = *((struct tree_entry**)_b);
1303         return base_name_compare(
1304                 a->name->str_dat, a->name->str_len, a->versions[1].mode,
1305                 b->name->str_dat, b->name->str_len, b->versions[1].mode);
1306 }
1307
1308 static void mktree(struct tree_content *t, int v, struct strbuf *b)
1309 {
1310         size_t maxlen = 0;
1311         unsigned int i;
1312
1313         if (!v)
1314                 QSORT(t->entries, t->entry_count, tecmp0);
1315         else
1316                 QSORT(t->entries, t->entry_count, tecmp1);
1317
1318         for (i = 0; i < t->entry_count; i++) {
1319                 if (t->entries[i]->versions[v].mode)
1320                         maxlen += t->entries[i]->name->str_len + 34;
1321         }
1322
1323         strbuf_reset(b);
1324         strbuf_grow(b, maxlen);
1325         for (i = 0; i < t->entry_count; i++) {
1326                 struct tree_entry *e = t->entries[i];
1327                 if (!e->versions[v].mode)
1328                         continue;
1329                 strbuf_addf(b, "%o %s%c",
1330                         (unsigned int)(e->versions[v].mode & ~NO_DELTA),
1331                         e->name->str_dat, '\0');
1332                 strbuf_add(b, e->versions[v].oid.hash, the_hash_algo->rawsz);
1333         }
1334 }
1335
1336 static void store_tree(struct tree_entry *root)
1337 {
1338         struct tree_content *t;
1339         unsigned int i, j, del;
1340         struct last_object lo = { STRBUF_INIT, 0, 0, /* no_swap */ 1 };
1341         struct object_entry *le = NULL;
1342
1343         if (!is_null_oid(&root->versions[1].oid))
1344                 return;
1345
1346         if (!root->tree)
1347                 load_tree(root);
1348         t = root->tree;
1349
1350         for (i = 0; i < t->entry_count; i++) {
1351                 if (t->entries[i]->tree)
1352                         store_tree(t->entries[i]);
1353         }
1354
1355         if (!(root->versions[0].mode & NO_DELTA))
1356                 le = find_object(&root->versions[0].oid);
1357         if (S_ISDIR(root->versions[0].mode) && le && le->pack_id == pack_id) {
1358                 mktree(t, 0, &old_tree);
1359                 lo.data = old_tree;
1360                 lo.offset = le->idx.offset;
1361                 lo.depth = t->delta_depth;
1362         }
1363
1364         mktree(t, 1, &new_tree);
1365         store_object(OBJ_TREE, &new_tree, &lo, &root->versions[1].oid, 0);
1366
1367         t->delta_depth = lo.depth;
1368         for (i = 0, j = 0, del = 0; i < t->entry_count; i++) {
1369                 struct tree_entry *e = t->entries[i];
1370                 if (e->versions[1].mode) {
1371                         e->versions[0].mode = e->versions[1].mode;
1372                         oidcpy(&e->versions[0].oid, &e->versions[1].oid);
1373                         t->entries[j++] = e;
1374                 } else {
1375                         release_tree_entry(e);
1376                         del++;
1377                 }
1378         }
1379         t->entry_count -= del;
1380 }
1381
1382 static void tree_content_replace(
1383         struct tree_entry *root,
1384         const struct object_id *oid,
1385         const uint16_t mode,
1386         struct tree_content *newtree)
1387 {
1388         if (!S_ISDIR(mode))
1389                 die("Root cannot be a non-directory");
1390         oidclr(&root->versions[0].oid);
1391         oidcpy(&root->versions[1].oid, oid);
1392         if (root->tree)
1393                 release_tree_content_recursive(root->tree);
1394         root->tree = newtree;
1395 }
1396
1397 static int tree_content_set(
1398         struct tree_entry *root,
1399         const char *p,
1400         const struct object_id *oid,
1401         const uint16_t mode,
1402         struct tree_content *subtree)
1403 {
1404         struct tree_content *t;
1405         const char *slash1;
1406         unsigned int i, n;
1407         struct tree_entry *e;
1408
1409         slash1 = strchrnul(p, '/');
1410         n = slash1 - p;
1411         if (!n)
1412                 die("Empty path component found in input");
1413         if (!*slash1 && !S_ISDIR(mode) && subtree)
1414                 die("Non-directories cannot have subtrees");
1415
1416         if (!root->tree)
1417                 load_tree(root);
1418         t = root->tree;
1419         for (i = 0; i < t->entry_count; i++) {
1420                 e = t->entries[i];
1421                 if (e->name->str_len == n && !fspathncmp(p, e->name->str_dat, n)) {
1422                         if (!*slash1) {
1423                                 if (!S_ISDIR(mode)
1424                                                 && e->versions[1].mode == mode
1425                                                 && oideq(&e->versions[1].oid, oid))
1426                                         return 0;
1427                                 e->versions[1].mode = mode;
1428                                 oidcpy(&e->versions[1].oid, oid);
1429                                 if (e->tree)
1430                                         release_tree_content_recursive(e->tree);
1431                                 e->tree = subtree;
1432
1433                                 /*
1434                                  * We need to leave e->versions[0].sha1 alone
1435                                  * to avoid modifying the preimage tree used
1436                                  * when writing out the parent directory.
1437                                  * But after replacing the subdir with a
1438                                  * completely different one, it's not a good
1439                                  * delta base any more, and besides, we've
1440                                  * thrown away the tree entries needed to
1441                                  * make a delta against it.
1442                                  *
1443                                  * So let's just explicitly disable deltas
1444                                  * for the subtree.
1445                                  */
1446                                 if (S_ISDIR(e->versions[0].mode))
1447                                         e->versions[0].mode |= NO_DELTA;
1448
1449                                 oidclr(&root->versions[1].oid);
1450                                 return 1;
1451                         }
1452                         if (!S_ISDIR(e->versions[1].mode)) {
1453                                 e->tree = new_tree_content(8);
1454                                 e->versions[1].mode = S_IFDIR;
1455                         }
1456                         if (!e->tree)
1457                                 load_tree(e);
1458                         if (tree_content_set(e, slash1 + 1, oid, mode, subtree)) {
1459                                 oidclr(&root->versions[1].oid);
1460                                 return 1;
1461                         }
1462                         return 0;
1463                 }
1464         }
1465
1466         if (t->entry_count == t->entry_capacity)
1467                 root->tree = t = grow_tree_content(t, t->entry_count);
1468         e = new_tree_entry();
1469         e->name = to_atom(p, n);
1470         e->versions[0].mode = 0;
1471         oidclr(&e->versions[0].oid);
1472         t->entries[t->entry_count++] = e;
1473         if (*slash1) {
1474                 e->tree = new_tree_content(8);
1475                 e->versions[1].mode = S_IFDIR;
1476                 tree_content_set(e, slash1 + 1, oid, mode, subtree);
1477         } else {
1478                 e->tree = subtree;
1479                 e->versions[1].mode = mode;
1480                 oidcpy(&e->versions[1].oid, oid);
1481         }
1482         oidclr(&root->versions[1].oid);
1483         return 1;
1484 }
1485
1486 static int tree_content_remove(
1487         struct tree_entry *root,
1488         const char *p,
1489         struct tree_entry *backup_leaf,
1490         int allow_root)
1491 {
1492         struct tree_content *t;
1493         const char *slash1;
1494         unsigned int i, n;
1495         struct tree_entry *e;
1496
1497         slash1 = strchrnul(p, '/');
1498         n = slash1 - p;
1499
1500         if (!root->tree)
1501                 load_tree(root);
1502
1503         if (!*p && allow_root) {
1504                 e = root;
1505                 goto del_entry;
1506         }
1507
1508         t = root->tree;
1509         for (i = 0; i < t->entry_count; i++) {
1510                 e = t->entries[i];
1511                 if (e->name->str_len == n && !fspathncmp(p, e->name->str_dat, n)) {
1512                         if (*slash1 && !S_ISDIR(e->versions[1].mode))
1513                                 /*
1514                                  * If p names a file in some subdirectory, and a
1515                                  * file or symlink matching the name of the
1516                                  * parent directory of p exists, then p cannot
1517                                  * exist and need not be deleted.
1518                                  */
1519                                 return 1;
1520                         if (!*slash1 || !S_ISDIR(e->versions[1].mode))
1521                                 goto del_entry;
1522                         if (!e->tree)
1523                                 load_tree(e);
1524                         if (tree_content_remove(e, slash1 + 1, backup_leaf, 0)) {
1525                                 for (n = 0; n < e->tree->entry_count; n++) {
1526                                         if (e->tree->entries[n]->versions[1].mode) {
1527                                                 oidclr(&root->versions[1].oid);
1528                                                 return 1;
1529                                         }
1530                                 }
1531                                 backup_leaf = NULL;
1532                                 goto del_entry;
1533                         }
1534                         return 0;
1535                 }
1536         }
1537         return 0;
1538
1539 del_entry:
1540         if (backup_leaf)
1541                 memcpy(backup_leaf, e, sizeof(*backup_leaf));
1542         else if (e->tree)
1543                 release_tree_content_recursive(e->tree);
1544         e->tree = NULL;
1545         e->versions[1].mode = 0;
1546         oidclr(&e->versions[1].oid);
1547         oidclr(&root->versions[1].oid);
1548         return 1;
1549 }
1550
1551 static int tree_content_get(
1552         struct tree_entry *root,
1553         const char *p,
1554         struct tree_entry *leaf,
1555         int allow_root)
1556 {
1557         struct tree_content *t;
1558         const char *slash1;
1559         unsigned int i, n;
1560         struct tree_entry *e;
1561
1562         slash1 = strchrnul(p, '/');
1563         n = slash1 - p;
1564         if (!n && !allow_root)
1565                 die("Empty path component found in input");
1566
1567         if (!root->tree)
1568                 load_tree(root);
1569
1570         if (!n) {
1571                 e = root;
1572                 goto found_entry;
1573         }
1574
1575         t = root->tree;
1576         for (i = 0; i < t->entry_count; i++) {
1577                 e = t->entries[i];
1578                 if (e->name->str_len == n && !fspathncmp(p, e->name->str_dat, n)) {
1579                         if (!*slash1)
1580                                 goto found_entry;
1581                         if (!S_ISDIR(e->versions[1].mode))
1582                                 return 0;
1583                         if (!e->tree)
1584                                 load_tree(e);
1585                         return tree_content_get(e, slash1 + 1, leaf, 0);
1586                 }
1587         }
1588         return 0;
1589
1590 found_entry:
1591         memcpy(leaf, e, sizeof(*leaf));
1592         if (e->tree && is_null_oid(&e->versions[1].oid))
1593                 leaf->tree = dup_tree_content(e->tree);
1594         else
1595                 leaf->tree = NULL;
1596         return 1;
1597 }
1598
1599 static int update_branch(struct branch *b)
1600 {
1601         static const char *msg = "fast-import";
1602         struct ref_transaction *transaction;
1603         struct object_id old_oid;
1604         struct strbuf err = STRBUF_INIT;
1605
1606         if (is_null_oid(&b->oid)) {
1607                 if (b->delete)
1608                         delete_ref(NULL, b->name, NULL, 0);
1609                 return 0;
1610         }
1611         if (read_ref(b->name, &old_oid))
1612                 oidclr(&old_oid);
1613         if (!force_update && !is_null_oid(&old_oid)) {
1614                 struct commit *old_cmit, *new_cmit;
1615
1616                 old_cmit = lookup_commit_reference_gently(the_repository,
1617                                                           &old_oid, 0);
1618                 new_cmit = lookup_commit_reference_gently(the_repository,
1619                                                           &b->oid, 0);
1620                 if (!old_cmit || !new_cmit)
1621                         return error("Branch %s is missing commits.", b->name);
1622
1623                 if (!in_merge_bases(old_cmit, new_cmit)) {
1624                         warning("Not updating %s"
1625                                 " (new tip %s does not contain %s)",
1626                                 b->name, oid_to_hex(&b->oid),
1627                                 oid_to_hex(&old_oid));
1628                         return -1;
1629                 }
1630         }
1631         transaction = ref_transaction_begin(&err);
1632         if (!transaction ||
1633             ref_transaction_update(transaction, b->name, &b->oid, &old_oid,
1634                                    0, msg, &err) ||
1635             ref_transaction_commit(transaction, &err)) {
1636                 ref_transaction_free(transaction);
1637                 error("%s", err.buf);
1638                 strbuf_release(&err);
1639                 return -1;
1640         }
1641         ref_transaction_free(transaction);
1642         strbuf_release(&err);
1643         return 0;
1644 }
1645
1646 static void dump_branches(void)
1647 {
1648         unsigned int i;
1649         struct branch *b;
1650
1651         for (i = 0; i < branch_table_sz; i++) {
1652                 for (b = branch_table[i]; b; b = b->table_next_branch)
1653                         failure |= update_branch(b);
1654         }
1655 }
1656
1657 static void dump_tags(void)
1658 {
1659         static const char *msg = "fast-import";
1660         struct tag *t;
1661         struct strbuf ref_name = STRBUF_INIT;
1662         struct strbuf err = STRBUF_INIT;
1663         struct ref_transaction *transaction;
1664
1665         transaction = ref_transaction_begin(&err);
1666         if (!transaction) {
1667                 failure |= error("%s", err.buf);
1668                 goto cleanup;
1669         }
1670         for (t = first_tag; t; t = t->next_tag) {
1671                 strbuf_reset(&ref_name);
1672                 strbuf_addf(&ref_name, "refs/tags/%s", t->name);
1673
1674                 if (ref_transaction_update(transaction, ref_name.buf,
1675                                            &t->oid, NULL, 0, msg, &err)) {
1676                         failure |= error("%s", err.buf);
1677                         goto cleanup;
1678                 }
1679         }
1680         if (ref_transaction_commit(transaction, &err))
1681                 failure |= error("%s", err.buf);
1682
1683  cleanup:
1684         ref_transaction_free(transaction);
1685         strbuf_release(&ref_name);
1686         strbuf_release(&err);
1687 }
1688
1689 static void dump_marks(void)
1690 {
1691         struct lock_file mark_lock = LOCK_INIT;
1692         FILE *f;
1693
1694         if (!export_marks_file || (import_marks_file && !import_marks_file_done))
1695                 return;
1696
1697         if (safe_create_leading_directories_const(export_marks_file)) {
1698                 failure |= error_errno("unable to create leading directories of %s",
1699                                        export_marks_file);
1700                 return;
1701         }
1702
1703         if (hold_lock_file_for_update(&mark_lock, export_marks_file, 0) < 0) {
1704                 failure |= error_errno("Unable to write marks file %s",
1705                                        export_marks_file);
1706                 return;
1707         }
1708
1709         f = fdopen_lock_file(&mark_lock, "w");
1710         if (!f) {
1711                 int saved_errno = errno;
1712                 rollback_lock_file(&mark_lock);
1713                 failure |= error("Unable to write marks file %s: %s",
1714                         export_marks_file, strerror(saved_errno));
1715                 return;
1716         }
1717
1718         for_each_mark(marks, 0, dump_marks_fn, f);
1719         if (commit_lock_file(&mark_lock)) {
1720                 failure |= error_errno("Unable to write file %s",
1721                                        export_marks_file);
1722                 return;
1723         }
1724 }
1725
1726 static void insert_object_entry(struct mark_set *s, struct object_id *oid, uintmax_t mark)
1727 {
1728         struct object_entry *e;
1729         e = find_object(oid);
1730         if (!e) {
1731                 enum object_type type = oid_object_info(the_repository,
1732                                                         oid, NULL);
1733                 if (type < 0)
1734                         die("object not found: %s", oid_to_hex(oid));
1735                 e = insert_object(oid);
1736                 e->type = type;
1737                 e->pack_id = MAX_PACK_ID;
1738                 e->idx.offset = 1; /* just not zero! */
1739         }
1740         insert_mark(s, mark, e);
1741 }
1742
1743 static void insert_oid_entry(struct mark_set *s, struct object_id *oid, uintmax_t mark)
1744 {
1745         insert_mark(s, mark, xmemdupz(oid, sizeof(*oid)));
1746 }
1747
1748 static void read_mark_file(struct mark_set *s, FILE *f, mark_set_inserter_t inserter)
1749 {
1750         char line[512];
1751         while (fgets(line, sizeof(line), f)) {
1752                 uintmax_t mark;
1753                 char *end;
1754                 struct object_id oid;
1755
1756                 /* Ensure SHA-1 objects are padded with zeros. */
1757                 memset(oid.hash, 0, sizeof(oid.hash));
1758
1759                 end = strchr(line, '\n');
1760                 if (line[0] != ':' || !end)
1761                         die("corrupt mark line: %s", line);
1762                 *end = 0;
1763                 mark = strtoumax(line + 1, &end, 10);
1764                 if (!mark || end == line + 1
1765                         || *end != ' '
1766                         || get_oid_hex_any(end + 1, &oid) == GIT_HASH_UNKNOWN)
1767                         die("corrupt mark line: %s", line);
1768                 inserter(s, &oid, mark);
1769         }
1770 }
1771
1772 static void read_marks(void)
1773 {
1774         FILE *f = fopen(import_marks_file, "r");
1775         if (f)
1776                 ;
1777         else if (import_marks_file_ignore_missing && errno == ENOENT)
1778                 goto done; /* Marks file does not exist */
1779         else
1780                 die_errno("cannot read '%s'", import_marks_file);
1781         read_mark_file(marks, f, insert_object_entry);
1782         fclose(f);
1783 done:
1784         import_marks_file_done = 1;
1785 }
1786
1787
1788 static int read_next_command(void)
1789 {
1790         static int stdin_eof = 0;
1791
1792         if (stdin_eof) {
1793                 unread_command_buf = 0;
1794                 return EOF;
1795         }
1796
1797         for (;;) {
1798                 if (unread_command_buf) {
1799                         unread_command_buf = 0;
1800                 } else {
1801                         struct recent_command *rc;
1802
1803                         stdin_eof = strbuf_getline_lf(&command_buf, stdin);
1804                         if (stdin_eof)
1805                                 return EOF;
1806
1807                         if (!seen_data_command
1808                                 && !starts_with(command_buf.buf, "feature ")
1809                                 && !starts_with(command_buf.buf, "option ")) {
1810                                 parse_argv();
1811                         }
1812
1813                         rc = rc_free;
1814                         if (rc)
1815                                 rc_free = rc->next;
1816                         else {
1817                                 rc = cmd_hist.next;
1818                                 cmd_hist.next = rc->next;
1819                                 cmd_hist.next->prev = &cmd_hist;
1820                                 free(rc->buf);
1821                         }
1822
1823                         rc->buf = xstrdup(command_buf.buf);
1824                         rc->prev = cmd_tail;
1825                         rc->next = cmd_hist.prev;
1826                         rc->prev->next = rc;
1827                         cmd_tail = rc;
1828                 }
1829                 if (command_buf.buf[0] == '#')
1830                         continue;
1831                 return 0;
1832         }
1833 }
1834
1835 static void skip_optional_lf(void)
1836 {
1837         int term_char = fgetc(stdin);
1838         if (term_char != '\n' && term_char != EOF)
1839                 ungetc(term_char, stdin);
1840 }
1841
1842 static void parse_mark(void)
1843 {
1844         const char *v;
1845         if (skip_prefix(command_buf.buf, "mark :", &v)) {
1846                 next_mark = strtoumax(v, NULL, 10);
1847                 read_next_command();
1848         }
1849         else
1850                 next_mark = 0;
1851 }
1852
1853 static void parse_original_identifier(void)
1854 {
1855         const char *v;
1856         if (skip_prefix(command_buf.buf, "original-oid ", &v))
1857                 read_next_command();
1858 }
1859
1860 static int parse_data(struct strbuf *sb, uintmax_t limit, uintmax_t *len_res)
1861 {
1862         const char *data;
1863         strbuf_reset(sb);
1864
1865         if (!skip_prefix(command_buf.buf, "data ", &data))
1866                 die("Expected 'data n' command, found: %s", command_buf.buf);
1867
1868         if (skip_prefix(data, "<<", &data)) {
1869                 char *term = xstrdup(data);
1870                 size_t term_len = command_buf.len - (data - command_buf.buf);
1871
1872                 for (;;) {
1873                         if (strbuf_getline_lf(&command_buf, stdin) == EOF)
1874                                 die("EOF in data (terminator '%s' not found)", term);
1875                         if (term_len == command_buf.len
1876                                 && !strcmp(term, command_buf.buf))
1877                                 break;
1878                         strbuf_addbuf(sb, &command_buf);
1879                         strbuf_addch(sb, '\n');
1880                 }
1881                 free(term);
1882         }
1883         else {
1884                 uintmax_t len = strtoumax(data, NULL, 10);
1885                 size_t n = 0, length = (size_t)len;
1886
1887                 if (limit && limit < len) {
1888                         *len_res = len;
1889                         return 0;
1890                 }
1891                 if (length < len)
1892                         die("data is too large to use in this context");
1893
1894                 while (n < length) {
1895                         size_t s = strbuf_fread(sb, length - n, stdin);
1896                         if (!s && feof(stdin))
1897                                 die("EOF in data (%lu bytes remaining)",
1898                                         (unsigned long)(length - n));
1899                         n += s;
1900                 }
1901         }
1902
1903         skip_optional_lf();
1904         return 1;
1905 }
1906
1907 static int validate_raw_date(const char *src, struct strbuf *result, int strict)
1908 {
1909         const char *orig_src = src;
1910         char *endp;
1911         unsigned long num;
1912
1913         errno = 0;
1914
1915         num = strtoul(src, &endp, 10);
1916         /*
1917          * NEEDSWORK: perhaps check for reasonable values? For example, we
1918          *            could error on values representing times more than a
1919          *            day in the future.
1920          */
1921         if (errno || endp == src || *endp != ' ')
1922                 return -1;
1923
1924         src = endp + 1;
1925         if (*src != '-' && *src != '+')
1926                 return -1;
1927
1928         num = strtoul(src + 1, &endp, 10);
1929         /*
1930          * NEEDSWORK: check for brokenness other than num > 1400, such as
1931          *            (num % 100) >= 60, or ((num % 100) % 15) != 0 ?
1932          */
1933         if (errno || endp == src + 1 || *endp || /* did not parse */
1934             (strict && (1400 < num))             /* parsed a broken timezone */
1935            )
1936                 return -1;
1937
1938         strbuf_addstr(result, orig_src);
1939         return 0;
1940 }
1941
1942 static char *parse_ident(const char *buf)
1943 {
1944         const char *ltgt;
1945         size_t name_len;
1946         struct strbuf ident = STRBUF_INIT;
1947
1948         /* ensure there is a space delimiter even if there is no name */
1949         if (*buf == '<')
1950                 --buf;
1951
1952         ltgt = buf + strcspn(buf, "<>");
1953         if (*ltgt != '<')
1954                 die("Missing < in ident string: %s", buf);
1955         if (ltgt != buf && ltgt[-1] != ' ')
1956                 die("Missing space before < in ident string: %s", buf);
1957         ltgt = ltgt + 1 + strcspn(ltgt + 1, "<>");
1958         if (*ltgt != '>')
1959                 die("Missing > in ident string: %s", buf);
1960         ltgt++;
1961         if (*ltgt != ' ')
1962                 die("Missing space after > in ident string: %s", buf);
1963         ltgt++;
1964         name_len = ltgt - buf;
1965         strbuf_add(&ident, buf, name_len);
1966
1967         switch (whenspec) {
1968         case WHENSPEC_RAW:
1969                 if (validate_raw_date(ltgt, &ident, 1) < 0)
1970                         die("Invalid raw date \"%s\" in ident: %s", ltgt, buf);
1971                 break;
1972         case WHENSPEC_RAW_PERMISSIVE:
1973                 if (validate_raw_date(ltgt, &ident, 0) < 0)
1974                         die("Invalid raw date \"%s\" in ident: %s", ltgt, buf);
1975                 break;
1976         case WHENSPEC_RFC2822:
1977                 if (parse_date(ltgt, &ident) < 0)
1978                         die("Invalid rfc2822 date \"%s\" in ident: %s", ltgt, buf);
1979                 break;
1980         case WHENSPEC_NOW:
1981                 if (strcmp("now", ltgt))
1982                         die("Date in ident must be 'now': %s", buf);
1983                 datestamp(&ident);
1984                 break;
1985         }
1986
1987         return strbuf_detach(&ident, NULL);
1988 }
1989
1990 static void parse_and_store_blob(
1991         struct last_object *last,
1992         struct object_id *oidout,
1993         uintmax_t mark)
1994 {
1995         static struct strbuf buf = STRBUF_INIT;
1996         uintmax_t len;
1997
1998         if (parse_data(&buf, big_file_threshold, &len))
1999                 store_object(OBJ_BLOB, &buf, last, oidout, mark);
2000         else {
2001                 if (last) {
2002                         strbuf_release(&last->data);
2003                         last->offset = 0;
2004                         last->depth = 0;
2005                 }
2006                 stream_blob(len, oidout, mark);
2007                 skip_optional_lf();
2008         }
2009 }
2010
2011 static void parse_new_blob(void)
2012 {
2013         read_next_command();
2014         parse_mark();
2015         parse_original_identifier();
2016         parse_and_store_blob(&last_blob, NULL, next_mark);
2017 }
2018
2019 static void unload_one_branch(void)
2020 {
2021         while (cur_active_branches
2022                 && cur_active_branches >= max_active_branches) {
2023                 uintmax_t min_commit = ULONG_MAX;
2024                 struct branch *e, *l = NULL, *p = NULL;
2025
2026                 for (e = active_branches; e; e = e->active_next_branch) {
2027                         if (e->last_commit < min_commit) {
2028                                 p = l;
2029                                 min_commit = e->last_commit;
2030                         }
2031                         l = e;
2032                 }
2033
2034                 if (p) {
2035                         e = p->active_next_branch;
2036                         p->active_next_branch = e->active_next_branch;
2037                 } else {
2038                         e = active_branches;
2039                         active_branches = e->active_next_branch;
2040                 }
2041                 e->active = 0;
2042                 e->active_next_branch = NULL;
2043                 if (e->branch_tree.tree) {
2044                         release_tree_content_recursive(e->branch_tree.tree);
2045                         e->branch_tree.tree = NULL;
2046                 }
2047                 cur_active_branches--;
2048         }
2049 }
2050
2051 static void load_branch(struct branch *b)
2052 {
2053         load_tree(&b->branch_tree);
2054         if (!b->active) {
2055                 b->active = 1;
2056                 b->active_next_branch = active_branches;
2057                 active_branches = b;
2058                 cur_active_branches++;
2059                 branch_load_count++;
2060         }
2061 }
2062
2063 static unsigned char convert_num_notes_to_fanout(uintmax_t num_notes)
2064 {
2065         unsigned char fanout = 0;
2066         while ((num_notes >>= 8))
2067                 fanout++;
2068         return fanout;
2069 }
2070
2071 static void construct_path_with_fanout(const char *hex_sha1,
2072                 unsigned char fanout, char *path)
2073 {
2074         unsigned int i = 0, j = 0;
2075         if (fanout >= the_hash_algo->rawsz)
2076                 die("Too large fanout (%u)", fanout);
2077         while (fanout) {
2078                 path[i++] = hex_sha1[j++];
2079                 path[i++] = hex_sha1[j++];
2080                 path[i++] = '/';
2081                 fanout--;
2082         }
2083         memcpy(path + i, hex_sha1 + j, the_hash_algo->hexsz - j);
2084         path[i + the_hash_algo->hexsz - j] = '\0';
2085 }
2086
2087 static uintmax_t do_change_note_fanout(
2088                 struct tree_entry *orig_root, struct tree_entry *root,
2089                 char *hex_oid, unsigned int hex_oid_len,
2090                 char *fullpath, unsigned int fullpath_len,
2091                 unsigned char fanout)
2092 {
2093         struct tree_content *t;
2094         struct tree_entry *e, leaf;
2095         unsigned int i, tmp_hex_oid_len, tmp_fullpath_len;
2096         uintmax_t num_notes = 0;
2097         struct object_id oid;
2098         /* hex oid + '/' between each pair of hex digits + NUL */
2099         char realpath[GIT_MAX_HEXSZ + ((GIT_MAX_HEXSZ / 2) - 1) + 1];
2100         const unsigned hexsz = the_hash_algo->hexsz;
2101
2102         if (!root->tree)
2103                 load_tree(root);
2104         t = root->tree;
2105
2106         for (i = 0; t && i < t->entry_count; i++) {
2107                 e = t->entries[i];
2108                 tmp_hex_oid_len = hex_oid_len + e->name->str_len;
2109                 tmp_fullpath_len = fullpath_len;
2110
2111                 /*
2112                  * We're interested in EITHER existing note entries (entries
2113                  * with exactly 40 hex chars in path, not including directory
2114                  * separators), OR directory entries that may contain note
2115                  * entries (with < 40 hex chars in path).
2116                  * Also, each path component in a note entry must be a multiple
2117                  * of 2 chars.
2118                  */
2119                 if (!e->versions[1].mode ||
2120                     tmp_hex_oid_len > hexsz ||
2121                     e->name->str_len % 2)
2122                         continue;
2123
2124                 /* This _may_ be a note entry, or a subdir containing notes */
2125                 memcpy(hex_oid + hex_oid_len, e->name->str_dat,
2126                        e->name->str_len);
2127                 if (tmp_fullpath_len)
2128                         fullpath[tmp_fullpath_len++] = '/';
2129                 memcpy(fullpath + tmp_fullpath_len, e->name->str_dat,
2130                        e->name->str_len);
2131                 tmp_fullpath_len += e->name->str_len;
2132                 fullpath[tmp_fullpath_len] = '\0';
2133
2134                 if (tmp_hex_oid_len == hexsz && !get_oid_hex(hex_oid, &oid)) {
2135                         /* This is a note entry */
2136                         if (fanout == 0xff) {
2137                                 /* Counting mode, no rename */
2138                                 num_notes++;
2139                                 continue;
2140                         }
2141                         construct_path_with_fanout(hex_oid, fanout, realpath);
2142                         if (!strcmp(fullpath, realpath)) {
2143                                 /* Note entry is in correct location */
2144                                 num_notes++;
2145                                 continue;
2146                         }
2147
2148                         /* Rename fullpath to realpath */
2149                         if (!tree_content_remove(orig_root, fullpath, &leaf, 0))
2150                                 die("Failed to remove path %s", fullpath);
2151                         tree_content_set(orig_root, realpath,
2152                                 &leaf.versions[1].oid,
2153                                 leaf.versions[1].mode,
2154                                 leaf.tree);
2155                 } else if (S_ISDIR(e->versions[1].mode)) {
2156                         /* This is a subdir that may contain note entries */
2157                         num_notes += do_change_note_fanout(orig_root, e,
2158                                 hex_oid, tmp_hex_oid_len,
2159                                 fullpath, tmp_fullpath_len, fanout);
2160                 }
2161
2162                 /* The above may have reallocated the current tree_content */
2163                 t = root->tree;
2164         }
2165         return num_notes;
2166 }
2167
2168 static uintmax_t change_note_fanout(struct tree_entry *root,
2169                 unsigned char fanout)
2170 {
2171         /*
2172          * The size of path is due to one slash between every two hex digits,
2173          * plus the terminating NUL.  Note that there is no slash at the end, so
2174          * the number of slashes is one less than half the number of hex
2175          * characters.
2176          */
2177         char hex_oid[GIT_MAX_HEXSZ], path[GIT_MAX_HEXSZ + (GIT_MAX_HEXSZ / 2) - 1 + 1];
2178         return do_change_note_fanout(root, root, hex_oid, 0, path, 0, fanout);
2179 }
2180
2181 static int parse_mapped_oid_hex(const char *hex, struct object_id *oid, const char **end)
2182 {
2183         int algo;
2184         khiter_t it;
2185
2186         /* Make SHA-1 object IDs have all-zero padding. */
2187         memset(oid->hash, 0, sizeof(oid->hash));
2188
2189         algo = parse_oid_hex_any(hex, oid, end);
2190         if (algo == GIT_HASH_UNKNOWN)
2191                 return -1;
2192
2193         it = kh_get_oid_map(sub_oid_map, *oid);
2194         /* No such object? */
2195         if (it == kh_end(sub_oid_map)) {
2196                 /* If we're using the same algorithm, pass it through. */
2197                 if (hash_algos[algo].format_id == the_hash_algo->format_id)
2198                         return 0;
2199                 return -1;
2200         }
2201         oidcpy(oid, kh_value(sub_oid_map, it));
2202         return 0;
2203 }
2204
2205 /*
2206  * Given a pointer into a string, parse a mark reference:
2207  *
2208  *   idnum ::= ':' bigint;
2209  *
2210  * Return the first character after the value in *endptr.
2211  *
2212  * Complain if the following character is not what is expected,
2213  * either a space or end of the string.
2214  */
2215 static uintmax_t parse_mark_ref(const char *p, char **endptr)
2216 {
2217         uintmax_t mark;
2218
2219         assert(*p == ':');
2220         p++;
2221         mark = strtoumax(p, endptr, 10);
2222         if (*endptr == p)
2223                 die("No value after ':' in mark: %s", command_buf.buf);
2224         return mark;
2225 }
2226
2227 /*
2228  * Parse the mark reference, and complain if this is not the end of
2229  * the string.
2230  */
2231 static uintmax_t parse_mark_ref_eol(const char *p)
2232 {
2233         char *end;
2234         uintmax_t mark;
2235
2236         mark = parse_mark_ref(p, &end);
2237         if (*end != '\0')
2238                 die("Garbage after mark: %s", command_buf.buf);
2239         return mark;
2240 }
2241
2242 /*
2243  * Parse the mark reference, demanding a trailing space.  Return a
2244  * pointer to the space.
2245  */
2246 static uintmax_t parse_mark_ref_space(const char **p)
2247 {
2248         uintmax_t mark;
2249         char *end;
2250
2251         mark = parse_mark_ref(*p, &end);
2252         if (*end++ != ' ')
2253                 die("Missing space after mark: %s", command_buf.buf);
2254         *p = end;
2255         return mark;
2256 }
2257
2258 static void file_change_m(const char *p, struct branch *b)
2259 {
2260         static struct strbuf uq = STRBUF_INIT;
2261         const char *endp;
2262         struct object_entry *oe;
2263         struct object_id oid;
2264         uint16_t mode, inline_data = 0;
2265
2266         p = get_mode(p, &mode);
2267         if (!p)
2268                 die("Corrupt mode: %s", command_buf.buf);
2269         switch (mode) {
2270         case 0644:
2271         case 0755:
2272                 mode |= S_IFREG;
2273         case S_IFREG | 0644:
2274         case S_IFREG | 0755:
2275         case S_IFLNK:
2276         case S_IFDIR:
2277         case S_IFGITLINK:
2278                 /* ok */
2279                 break;
2280         default:
2281                 die("Corrupt mode: %s", command_buf.buf);
2282         }
2283
2284         if (*p == ':') {
2285                 oe = find_mark(marks, parse_mark_ref_space(&p));
2286                 oidcpy(&oid, &oe->idx.oid);
2287         } else if (skip_prefix(p, "inline ", &p)) {
2288                 inline_data = 1;
2289                 oe = NULL; /* not used with inline_data, but makes gcc happy */
2290         } else {
2291                 if (parse_mapped_oid_hex(p, &oid, &p))
2292                         die("Invalid dataref: %s", command_buf.buf);
2293                 oe = find_object(&oid);
2294                 if (*p++ != ' ')
2295                         die("Missing space after SHA1: %s", command_buf.buf);
2296         }
2297
2298         strbuf_reset(&uq);
2299         if (!unquote_c_style(&uq, p, &endp)) {
2300                 if (*endp)
2301                         die("Garbage after path in: %s", command_buf.buf);
2302                 p = uq.buf;
2303         }
2304
2305         /* Git does not track empty, non-toplevel directories. */
2306         if (S_ISDIR(mode) && is_empty_tree_oid(&oid) && *p) {
2307                 tree_content_remove(&b->branch_tree, p, NULL, 0);
2308                 return;
2309         }
2310
2311         if (S_ISGITLINK(mode)) {
2312                 if (inline_data)
2313                         die("Git links cannot be specified 'inline': %s",
2314                                 command_buf.buf);
2315                 else if (oe) {
2316                         if (oe->type != OBJ_COMMIT)
2317                                 die("Not a commit (actually a %s): %s",
2318                                         type_name(oe->type), command_buf.buf);
2319                 }
2320                 /*
2321                  * Accept the sha1 without checking; it expected to be in
2322                  * another repository.
2323                  */
2324         } else if (inline_data) {
2325                 if (S_ISDIR(mode))
2326                         die("Directories cannot be specified 'inline': %s",
2327                                 command_buf.buf);
2328                 if (p != uq.buf) {
2329                         strbuf_addstr(&uq, p);
2330                         p = uq.buf;
2331                 }
2332                 while (read_next_command() != EOF) {
2333                         const char *v;
2334                         if (skip_prefix(command_buf.buf, "cat-blob ", &v))
2335                                 parse_cat_blob(v);
2336                         else {
2337                                 parse_and_store_blob(&last_blob, &oid, 0);
2338                                 break;
2339                         }
2340                 }
2341         } else {
2342                 enum object_type expected = S_ISDIR(mode) ?
2343                                                 OBJ_TREE: OBJ_BLOB;
2344                 enum object_type type = oe ? oe->type :
2345                                         oid_object_info(the_repository, &oid,
2346                                                         NULL);
2347                 if (type < 0)
2348                         die("%s not found: %s",
2349                                         S_ISDIR(mode) ?  "Tree" : "Blob",
2350                                         command_buf.buf);
2351                 if (type != expected)
2352                         die("Not a %s (actually a %s): %s",
2353                                 type_name(expected), type_name(type),
2354                                 command_buf.buf);
2355         }
2356
2357         if (!*p) {
2358                 tree_content_replace(&b->branch_tree, &oid, mode, NULL);
2359                 return;
2360         }
2361         tree_content_set(&b->branch_tree, p, &oid, mode, NULL);
2362 }
2363
2364 static void file_change_d(const char *p, struct branch *b)
2365 {
2366         static struct strbuf uq = STRBUF_INIT;
2367         const char *endp;
2368
2369         strbuf_reset(&uq);
2370         if (!unquote_c_style(&uq, p, &endp)) {
2371                 if (*endp)
2372                         die("Garbage after path in: %s", command_buf.buf);
2373                 p = uq.buf;
2374         }
2375         tree_content_remove(&b->branch_tree, p, NULL, 1);
2376 }
2377
2378 static void file_change_cr(const char *s, struct branch *b, int rename)
2379 {
2380         const char *d;
2381         static struct strbuf s_uq = STRBUF_INIT;
2382         static struct strbuf d_uq = STRBUF_INIT;
2383         const char *endp;
2384         struct tree_entry leaf;
2385
2386         strbuf_reset(&s_uq);
2387         if (!unquote_c_style(&s_uq, s, &endp)) {
2388                 if (*endp != ' ')
2389                         die("Missing space after source: %s", command_buf.buf);
2390         } else {
2391                 endp = strchr(s, ' ');
2392                 if (!endp)
2393                         die("Missing space after source: %s", command_buf.buf);
2394                 strbuf_add(&s_uq, s, endp - s);
2395         }
2396         s = s_uq.buf;
2397
2398         endp++;
2399         if (!*endp)
2400                 die("Missing dest: %s", command_buf.buf);
2401
2402         d = endp;
2403         strbuf_reset(&d_uq);
2404         if (!unquote_c_style(&d_uq, d, &endp)) {
2405                 if (*endp)
2406                         die("Garbage after dest in: %s", command_buf.buf);
2407                 d = d_uq.buf;
2408         }
2409
2410         memset(&leaf, 0, sizeof(leaf));
2411         if (rename)
2412                 tree_content_remove(&b->branch_tree, s, &leaf, 1);
2413         else
2414                 tree_content_get(&b->branch_tree, s, &leaf, 1);
2415         if (!leaf.versions[1].mode)
2416                 die("Path %s not in branch", s);
2417         if (!*d) {      /* C "path/to/subdir" "" */
2418                 tree_content_replace(&b->branch_tree,
2419                         &leaf.versions[1].oid,
2420                         leaf.versions[1].mode,
2421                         leaf.tree);
2422                 return;
2423         }
2424         tree_content_set(&b->branch_tree, d,
2425                 &leaf.versions[1].oid,
2426                 leaf.versions[1].mode,
2427                 leaf.tree);
2428 }
2429
2430 static void note_change_n(const char *p, struct branch *b, unsigned char *old_fanout)
2431 {
2432         static struct strbuf uq = STRBUF_INIT;
2433         struct object_entry *oe;
2434         struct branch *s;
2435         struct object_id oid, commit_oid;
2436         char path[GIT_MAX_RAWSZ * 3];
2437         uint16_t inline_data = 0;
2438         unsigned char new_fanout;
2439
2440         /*
2441          * When loading a branch, we don't traverse its tree to count the real
2442          * number of notes (too expensive to do this for all non-note refs).
2443          * This means that recently loaded notes refs might incorrectly have
2444          * b->num_notes == 0, and consequently, old_fanout might be wrong.
2445          *
2446          * Fix this by traversing the tree and counting the number of notes
2447          * when b->num_notes == 0. If the notes tree is truly empty, the
2448          * calculation should not take long.
2449          */
2450         if (b->num_notes == 0 && *old_fanout == 0) {
2451                 /* Invoke change_note_fanout() in "counting mode". */
2452                 b->num_notes = change_note_fanout(&b->branch_tree, 0xff);
2453                 *old_fanout = convert_num_notes_to_fanout(b->num_notes);
2454         }
2455
2456         /* Now parse the notemodify command. */
2457         /* <dataref> or 'inline' */
2458         if (*p == ':') {
2459                 oe = find_mark(marks, parse_mark_ref_space(&p));
2460                 oidcpy(&oid, &oe->idx.oid);
2461         } else if (skip_prefix(p, "inline ", &p)) {
2462                 inline_data = 1;
2463                 oe = NULL; /* not used with inline_data, but makes gcc happy */
2464         } else {
2465                 if (parse_mapped_oid_hex(p, &oid, &p))
2466                         die("Invalid dataref: %s", command_buf.buf);
2467                 oe = find_object(&oid);
2468                 if (*p++ != ' ')
2469                         die("Missing space after SHA1: %s", command_buf.buf);
2470         }
2471
2472         /* <commit-ish> */
2473         s = lookup_branch(p);
2474         if (s) {
2475                 if (is_null_oid(&s->oid))
2476                         die("Can't add a note on empty branch.");
2477                 oidcpy(&commit_oid, &s->oid);
2478         } else if (*p == ':') {
2479                 uintmax_t commit_mark = parse_mark_ref_eol(p);
2480                 struct object_entry *commit_oe = find_mark(marks, commit_mark);
2481                 if (commit_oe->type != OBJ_COMMIT)
2482                         die("Mark :%" PRIuMAX " not a commit", commit_mark);
2483                 oidcpy(&commit_oid, &commit_oe->idx.oid);
2484         } else if (!get_oid(p, &commit_oid)) {
2485                 unsigned long size;
2486                 char *buf = read_object_with_reference(the_repository,
2487                                                        &commit_oid,
2488                                                        commit_type, &size,
2489                                                        &commit_oid);
2490                 if (!buf || size < the_hash_algo->hexsz + 6)
2491                         die("Not a valid commit: %s", p);
2492                 free(buf);
2493         } else
2494                 die("Invalid ref name or SHA1 expression: %s", p);
2495
2496         if (inline_data) {
2497                 if (p != uq.buf) {
2498                         strbuf_addstr(&uq, p);
2499                         p = uq.buf;
2500                 }
2501                 read_next_command();
2502                 parse_and_store_blob(&last_blob, &oid, 0);
2503         } else if (oe) {
2504                 if (oe->type != OBJ_BLOB)
2505                         die("Not a blob (actually a %s): %s",
2506                                 type_name(oe->type), command_buf.buf);
2507         } else if (!is_null_oid(&oid)) {
2508                 enum object_type type = oid_object_info(the_repository, &oid,
2509                                                         NULL);
2510                 if (type < 0)
2511                         die("Blob not found: %s", command_buf.buf);
2512                 if (type != OBJ_BLOB)
2513                         die("Not a blob (actually a %s): %s",
2514                             type_name(type), command_buf.buf);
2515         }
2516
2517         construct_path_with_fanout(oid_to_hex(&commit_oid), *old_fanout, path);
2518         if (tree_content_remove(&b->branch_tree, path, NULL, 0))
2519                 b->num_notes--;
2520
2521         if (is_null_oid(&oid))
2522                 return; /* nothing to insert */
2523
2524         b->num_notes++;
2525         new_fanout = convert_num_notes_to_fanout(b->num_notes);
2526         construct_path_with_fanout(oid_to_hex(&commit_oid), new_fanout, path);
2527         tree_content_set(&b->branch_tree, path, &oid, S_IFREG | 0644, NULL);
2528 }
2529
2530 static void file_change_deleteall(struct branch *b)
2531 {
2532         release_tree_content_recursive(b->branch_tree.tree);
2533         oidclr(&b->branch_tree.versions[0].oid);
2534         oidclr(&b->branch_tree.versions[1].oid);
2535         load_tree(&b->branch_tree);
2536         b->num_notes = 0;
2537 }
2538
2539 static void parse_from_commit(struct branch *b, char *buf, unsigned long size)
2540 {
2541         if (!buf || size < the_hash_algo->hexsz + 6)
2542                 die("Not a valid commit: %s", oid_to_hex(&b->oid));
2543         if (memcmp("tree ", buf, 5)
2544                 || get_oid_hex(buf + 5, &b->branch_tree.versions[1].oid))
2545                 die("The commit %s is corrupt", oid_to_hex(&b->oid));
2546         oidcpy(&b->branch_tree.versions[0].oid,
2547                &b->branch_tree.versions[1].oid);
2548 }
2549
2550 static void parse_from_existing(struct branch *b)
2551 {
2552         if (is_null_oid(&b->oid)) {
2553                 oidclr(&b->branch_tree.versions[0].oid);
2554                 oidclr(&b->branch_tree.versions[1].oid);
2555         } else {
2556                 unsigned long size;
2557                 char *buf;
2558
2559                 buf = read_object_with_reference(the_repository,
2560                                                  &b->oid, commit_type, &size,
2561                                                  &b->oid);
2562                 parse_from_commit(b, buf, size);
2563                 free(buf);
2564         }
2565 }
2566
2567 static int parse_objectish(struct branch *b, const char *objectish)
2568 {
2569         struct branch *s;
2570         struct object_id oid;
2571
2572         oidcpy(&oid, &b->branch_tree.versions[1].oid);
2573
2574         s = lookup_branch(objectish);
2575         if (b == s)
2576                 die("Can't create a branch from itself: %s", b->name);
2577         else if (s) {
2578                 struct object_id *t = &s->branch_tree.versions[1].oid;
2579                 oidcpy(&b->oid, &s->oid);
2580                 oidcpy(&b->branch_tree.versions[0].oid, t);
2581                 oidcpy(&b->branch_tree.versions[1].oid, t);
2582         } else if (*objectish == ':') {
2583                 uintmax_t idnum = parse_mark_ref_eol(objectish);
2584                 struct object_entry *oe = find_mark(marks, idnum);
2585                 if (oe->type != OBJ_COMMIT)
2586                         die("Mark :%" PRIuMAX " not a commit", idnum);
2587                 if (!oideq(&b->oid, &oe->idx.oid)) {
2588                         oidcpy(&b->oid, &oe->idx.oid);
2589                         if (oe->pack_id != MAX_PACK_ID) {
2590                                 unsigned long size;
2591                                 char *buf = gfi_unpack_entry(oe, &size);
2592                                 parse_from_commit(b, buf, size);
2593                                 free(buf);
2594                         } else
2595                                 parse_from_existing(b);
2596                 }
2597         } else if (!get_oid(objectish, &b->oid)) {
2598                 parse_from_existing(b);
2599                 if (is_null_oid(&b->oid))
2600                         b->delete = 1;
2601         }
2602         else
2603                 die("Invalid ref name or SHA1 expression: %s", objectish);
2604
2605         if (b->branch_tree.tree && !oideq(&oid, &b->branch_tree.versions[1].oid)) {
2606                 release_tree_content_recursive(b->branch_tree.tree);
2607                 b->branch_tree.tree = NULL;
2608         }
2609
2610         read_next_command();
2611         return 1;
2612 }
2613
2614 static int parse_from(struct branch *b)
2615 {
2616         const char *from;
2617
2618         if (!skip_prefix(command_buf.buf, "from ", &from))
2619                 return 0;
2620
2621         return parse_objectish(b, from);
2622 }
2623
2624 static int parse_objectish_with_prefix(struct branch *b, const char *prefix)
2625 {
2626         const char *base;
2627
2628         if (!skip_prefix(command_buf.buf, prefix, &base))
2629                 return 0;
2630
2631         return parse_objectish(b, base);
2632 }
2633
2634 static struct hash_list *parse_merge(unsigned int *count)
2635 {
2636         struct hash_list *list = NULL, **tail = &list, *n;
2637         const char *from;
2638         struct branch *s;
2639
2640         *count = 0;
2641         while (skip_prefix(command_buf.buf, "merge ", &from)) {
2642                 n = xmalloc(sizeof(*n));
2643                 s = lookup_branch(from);
2644                 if (s)
2645                         oidcpy(&n->oid, &s->oid);
2646                 else if (*from == ':') {
2647                         uintmax_t idnum = parse_mark_ref_eol(from);
2648                         struct object_entry *oe = find_mark(marks, idnum);
2649                         if (oe->type != OBJ_COMMIT)
2650                                 die("Mark :%" PRIuMAX " not a commit", idnum);
2651                         oidcpy(&n->oid, &oe->idx.oid);
2652                 } else if (!get_oid(from, &n->oid)) {
2653                         unsigned long size;
2654                         char *buf = read_object_with_reference(the_repository,
2655                                                                &n->oid,
2656                                                                commit_type,
2657                                                                &size, &n->oid);
2658                         if (!buf || size < the_hash_algo->hexsz + 6)
2659                                 die("Not a valid commit: %s", from);
2660                         free(buf);
2661                 } else
2662                         die("Invalid ref name or SHA1 expression: %s", from);
2663
2664                 n->next = NULL;
2665                 *tail = n;
2666                 tail = &n->next;
2667
2668                 (*count)++;
2669                 read_next_command();
2670         }
2671         return list;
2672 }
2673
2674 static void parse_new_commit(const char *arg)
2675 {
2676         static struct strbuf msg = STRBUF_INIT;
2677         struct branch *b;
2678         char *author = NULL;
2679         char *committer = NULL;
2680         char *encoding = NULL;
2681         struct hash_list *merge_list = NULL;
2682         unsigned int merge_count;
2683         unsigned char prev_fanout, new_fanout;
2684         const char *v;
2685
2686         b = lookup_branch(arg);
2687         if (!b)
2688                 b = new_branch(arg);
2689
2690         read_next_command();
2691         parse_mark();
2692         parse_original_identifier();
2693         if (skip_prefix(command_buf.buf, "author ", &v)) {
2694                 author = parse_ident(v);
2695                 read_next_command();
2696         }
2697         if (skip_prefix(command_buf.buf, "committer ", &v)) {
2698                 committer = parse_ident(v);
2699                 read_next_command();
2700         }
2701         if (!committer)
2702                 die("Expected committer but didn't get one");
2703         if (skip_prefix(command_buf.buf, "encoding ", &v)) {
2704                 encoding = xstrdup(v);
2705                 read_next_command();
2706         }
2707         parse_data(&msg, 0, NULL);
2708         read_next_command();
2709         parse_from(b);
2710         merge_list = parse_merge(&merge_count);
2711
2712         /* ensure the branch is active/loaded */
2713         if (!b->branch_tree.tree || !max_active_branches) {
2714                 unload_one_branch();
2715                 load_branch(b);
2716         }
2717
2718         prev_fanout = convert_num_notes_to_fanout(b->num_notes);
2719
2720         /* file_change* */
2721         while (command_buf.len > 0) {
2722                 if (skip_prefix(command_buf.buf, "M ", &v))
2723                         file_change_m(v, b);
2724                 else if (skip_prefix(command_buf.buf, "D ", &v))
2725                         file_change_d(v, b);
2726                 else if (skip_prefix(command_buf.buf, "R ", &v))
2727                         file_change_cr(v, b, 1);
2728                 else if (skip_prefix(command_buf.buf, "C ", &v))
2729                         file_change_cr(v, b, 0);
2730                 else if (skip_prefix(command_buf.buf, "N ", &v))
2731                         note_change_n(v, b, &prev_fanout);
2732                 else if (!strcmp("deleteall", command_buf.buf))
2733                         file_change_deleteall(b);
2734                 else if (skip_prefix(command_buf.buf, "ls ", &v))
2735                         parse_ls(v, b);
2736                 else if (skip_prefix(command_buf.buf, "cat-blob ", &v))
2737                         parse_cat_blob(v);
2738                 else {
2739                         unread_command_buf = 1;
2740                         break;
2741                 }
2742                 if (read_next_command() == EOF)
2743                         break;
2744         }
2745
2746         new_fanout = convert_num_notes_to_fanout(b->num_notes);
2747         if (new_fanout != prev_fanout)
2748                 b->num_notes = change_note_fanout(&b->branch_tree, new_fanout);
2749
2750         /* build the tree and the commit */
2751         store_tree(&b->branch_tree);
2752         oidcpy(&b->branch_tree.versions[0].oid,
2753                &b->branch_tree.versions[1].oid);
2754
2755         strbuf_reset(&new_data);
2756         strbuf_addf(&new_data, "tree %s\n",
2757                 oid_to_hex(&b->branch_tree.versions[1].oid));
2758         if (!is_null_oid(&b->oid))
2759                 strbuf_addf(&new_data, "parent %s\n",
2760                             oid_to_hex(&b->oid));
2761         while (merge_list) {
2762                 struct hash_list *next = merge_list->next;
2763                 strbuf_addf(&new_data, "parent %s\n",
2764                             oid_to_hex(&merge_list->oid));
2765                 free(merge_list);
2766                 merge_list = next;
2767         }
2768         strbuf_addf(&new_data,
2769                 "author %s\n"
2770                 "committer %s\n",
2771                 author ? author : committer, committer);
2772         if (encoding)
2773                 strbuf_addf(&new_data,
2774                         "encoding %s\n",
2775                         encoding);
2776         strbuf_addch(&new_data, '\n');
2777         strbuf_addbuf(&new_data, &msg);
2778         free(author);
2779         free(committer);
2780         free(encoding);
2781
2782         if (!store_object(OBJ_COMMIT, &new_data, NULL, &b->oid, next_mark))
2783                 b->pack_id = pack_id;
2784         b->last_commit = object_count_by_type[OBJ_COMMIT];
2785 }
2786
2787 static void parse_new_tag(const char *arg)
2788 {
2789         static struct strbuf msg = STRBUF_INIT;
2790         const char *from;
2791         char *tagger;
2792         struct branch *s;
2793         struct tag *t;
2794         uintmax_t from_mark = 0;
2795         struct object_id oid;
2796         enum object_type type;
2797         const char *v;
2798
2799         t = mem_pool_alloc(&fi_mem_pool, sizeof(struct tag));
2800         memset(t, 0, sizeof(struct tag));
2801         t->name = mem_pool_strdup(&fi_mem_pool, arg);
2802         if (last_tag)
2803                 last_tag->next_tag = t;
2804         else
2805                 first_tag = t;
2806         last_tag = t;
2807         read_next_command();
2808         parse_mark();
2809
2810         /* from ... */
2811         if (!skip_prefix(command_buf.buf, "from ", &from))
2812                 die("Expected from command, got %s", command_buf.buf);
2813         s = lookup_branch(from);
2814         if (s) {
2815                 if (is_null_oid(&s->oid))
2816                         die("Can't tag an empty branch.");
2817                 oidcpy(&oid, &s->oid);
2818                 type = OBJ_COMMIT;
2819         } else if (*from == ':') {
2820                 struct object_entry *oe;
2821                 from_mark = parse_mark_ref_eol(from);
2822                 oe = find_mark(marks, from_mark);
2823                 type = oe->type;
2824                 oidcpy(&oid, &oe->idx.oid);
2825         } else if (!get_oid(from, &oid)) {
2826                 struct object_entry *oe = find_object(&oid);
2827                 if (!oe) {
2828                         type = oid_object_info(the_repository, &oid, NULL);
2829                         if (type < 0)
2830                                 die("Not a valid object: %s", from);
2831                 } else
2832                         type = oe->type;
2833         } else
2834                 die("Invalid ref name or SHA1 expression: %s", from);
2835         read_next_command();
2836
2837         /* original-oid ... */
2838         parse_original_identifier();
2839
2840         /* tagger ... */
2841         if (skip_prefix(command_buf.buf, "tagger ", &v)) {
2842                 tagger = parse_ident(v);
2843                 read_next_command();
2844         } else
2845                 tagger = NULL;
2846
2847         /* tag payload/message */
2848         parse_data(&msg, 0, NULL);
2849
2850         /* build the tag object */
2851         strbuf_reset(&new_data);
2852
2853         strbuf_addf(&new_data,
2854                     "object %s\n"
2855                     "type %s\n"
2856                     "tag %s\n",
2857                     oid_to_hex(&oid), type_name(type), t->name);
2858         if (tagger)
2859                 strbuf_addf(&new_data,
2860                             "tagger %s\n", tagger);
2861         strbuf_addch(&new_data, '\n');
2862         strbuf_addbuf(&new_data, &msg);
2863         free(tagger);
2864
2865         if (store_object(OBJ_TAG, &new_data, NULL, &t->oid, next_mark))
2866                 t->pack_id = MAX_PACK_ID;
2867         else
2868                 t->pack_id = pack_id;
2869 }
2870
2871 static void parse_reset_branch(const char *arg)
2872 {
2873         struct branch *b;
2874         const char *tag_name;
2875
2876         b = lookup_branch(arg);
2877         if (b) {
2878                 oidclr(&b->oid);
2879                 oidclr(&b->branch_tree.versions[0].oid);
2880                 oidclr(&b->branch_tree.versions[1].oid);
2881                 if (b->branch_tree.tree) {
2882                         release_tree_content_recursive(b->branch_tree.tree);
2883                         b->branch_tree.tree = NULL;
2884                 }
2885         }
2886         else
2887                 b = new_branch(arg);
2888         read_next_command();
2889         parse_from(b);
2890         if (b->delete && skip_prefix(b->name, "refs/tags/", &tag_name)) {
2891                 /*
2892                  * Elsewhere, we call dump_branches() before dump_tags(),
2893                  * and dump_branches() will handle ref deletions first, so
2894                  * in order to make sure the deletion actually takes effect,
2895                  * we need to remove the tag from our list of tags to update.
2896                  *
2897                  * NEEDSWORK: replace list of tags with hashmap for faster
2898                  * deletion?
2899                  */
2900                 struct tag *t, *prev = NULL;
2901                 for (t = first_tag; t; t = t->next_tag) {
2902                         if (!strcmp(t->name, tag_name))
2903                                 break;
2904                         prev = t;
2905                 }
2906                 if (t) {
2907                         if (prev)
2908                                 prev->next_tag = t->next_tag;
2909                         else
2910                                 first_tag = t->next_tag;
2911                         if (!t->next_tag)
2912                                 last_tag = prev;
2913                         /* There is no mem_pool_free(t) function to call. */
2914                 }
2915         }
2916         if (command_buf.len > 0)
2917                 unread_command_buf = 1;
2918 }
2919
2920 static void cat_blob_write(const char *buf, unsigned long size)
2921 {
2922         if (write_in_full(cat_blob_fd, buf, size) < 0)
2923                 die_errno("Write to frontend failed");
2924 }
2925
2926 static void cat_blob(struct object_entry *oe, struct object_id *oid)
2927 {
2928         struct strbuf line = STRBUF_INIT;
2929         unsigned long size;
2930         enum object_type type = 0;
2931         char *buf;
2932
2933         if (!oe || oe->pack_id == MAX_PACK_ID) {
2934                 buf = read_object_file(oid, &type, &size);
2935         } else {
2936                 type = oe->type;
2937                 buf = gfi_unpack_entry(oe, &size);
2938         }
2939
2940         /*
2941          * Output based on batch_one_object() from cat-file.c.
2942          */
2943         if (type <= 0) {
2944                 strbuf_reset(&line);
2945                 strbuf_addf(&line, "%s missing\n", oid_to_hex(oid));
2946                 cat_blob_write(line.buf, line.len);
2947                 strbuf_release(&line);
2948                 free(buf);
2949                 return;
2950         }
2951         if (!buf)
2952                 die("Can't read object %s", oid_to_hex(oid));
2953         if (type != OBJ_BLOB)
2954                 die("Object %s is a %s but a blob was expected.",
2955                     oid_to_hex(oid), type_name(type));
2956         strbuf_reset(&line);
2957         strbuf_addf(&line, "%s %s %"PRIuMAX"\n", oid_to_hex(oid),
2958                     type_name(type), (uintmax_t)size);
2959         cat_blob_write(line.buf, line.len);
2960         strbuf_release(&line);
2961         cat_blob_write(buf, size);
2962         cat_blob_write("\n", 1);
2963         if (oe && oe->pack_id == pack_id) {
2964                 last_blob.offset = oe->idx.offset;
2965                 strbuf_attach(&last_blob.data, buf, size, size);
2966                 last_blob.depth = oe->depth;
2967         } else
2968                 free(buf);
2969 }
2970
2971 static void parse_get_mark(const char *p)
2972 {
2973         struct object_entry *oe;
2974         char output[GIT_MAX_HEXSZ + 2];
2975
2976         /* get-mark SP <object> LF */
2977         if (*p != ':')
2978                 die("Not a mark: %s", p);
2979
2980         oe = find_mark(marks, parse_mark_ref_eol(p));
2981         if (!oe)
2982                 die("Unknown mark: %s", command_buf.buf);
2983
2984         xsnprintf(output, sizeof(output), "%s\n", oid_to_hex(&oe->idx.oid));
2985         cat_blob_write(output, the_hash_algo->hexsz + 1);
2986 }
2987
2988 static void parse_cat_blob(const char *p)
2989 {
2990         struct object_entry *oe;
2991         struct object_id oid;
2992
2993         /* cat-blob SP <object> LF */
2994         if (*p == ':') {
2995                 oe = find_mark(marks, parse_mark_ref_eol(p));
2996                 if (!oe)
2997                         die("Unknown mark: %s", command_buf.buf);
2998                 oidcpy(&oid, &oe->idx.oid);
2999         } else {
3000                 if (parse_mapped_oid_hex(p, &oid, &p))
3001                         die("Invalid dataref: %s", command_buf.buf);
3002                 if (*p)
3003                         die("Garbage after SHA1: %s", command_buf.buf);
3004                 oe = find_object(&oid);
3005         }
3006
3007         cat_blob(oe, &oid);
3008 }
3009
3010 static struct object_entry *dereference(struct object_entry *oe,
3011                                         struct object_id *oid)
3012 {
3013         unsigned long size;
3014         char *buf = NULL;
3015         const unsigned hexsz = the_hash_algo->hexsz;
3016
3017         if (!oe) {
3018                 enum object_type type = oid_object_info(the_repository, oid,
3019                                                         NULL);
3020                 if (type < 0)
3021                         die("object not found: %s", oid_to_hex(oid));
3022                 /* cache it! */
3023                 oe = insert_object(oid);
3024                 oe->type = type;
3025                 oe->pack_id = MAX_PACK_ID;
3026                 oe->idx.offset = 1;
3027         }
3028         switch (oe->type) {
3029         case OBJ_TREE:  /* easy case. */
3030                 return oe;
3031         case OBJ_COMMIT:
3032         case OBJ_TAG:
3033                 break;
3034         default:
3035                 die("Not a tree-ish: %s", command_buf.buf);
3036         }
3037
3038         if (oe->pack_id != MAX_PACK_ID) {       /* in a pack being written */
3039                 buf = gfi_unpack_entry(oe, &size);
3040         } else {
3041                 enum object_type unused;
3042                 buf = read_object_file(oid, &unused, &size);
3043         }
3044         if (!buf)
3045                 die("Can't load object %s", oid_to_hex(oid));
3046
3047         /* Peel one layer. */
3048         switch (oe->type) {
3049         case OBJ_TAG:
3050                 if (size < hexsz + strlen("object ") ||
3051                     get_oid_hex(buf + strlen("object "), oid))
3052                         die("Invalid SHA1 in tag: %s", command_buf.buf);
3053                 break;
3054         case OBJ_COMMIT:
3055                 if (size < hexsz + strlen("tree ") ||
3056                     get_oid_hex(buf + strlen("tree "), oid))
3057                         die("Invalid SHA1 in commit: %s", command_buf.buf);
3058         }
3059
3060         free(buf);
3061         return find_object(oid);
3062 }
3063
3064 static void insert_mapped_mark(uintmax_t mark, void *object, void *cbp)
3065 {
3066         struct object_id *fromoid = object;
3067         struct object_id *tooid = find_mark(cbp, mark);
3068         int ret;
3069         khiter_t it;
3070
3071         it = kh_put_oid_map(sub_oid_map, *fromoid, &ret);
3072         /* We've already seen this object. */
3073         if (ret == 0)
3074                 return;
3075         kh_value(sub_oid_map, it) = tooid;
3076 }
3077
3078 static void build_mark_map_one(struct mark_set *from, struct mark_set *to)
3079 {
3080         for_each_mark(from, 0, insert_mapped_mark, to);
3081 }
3082
3083 static void build_mark_map(struct string_list *from, struct string_list *to)
3084 {
3085         struct string_list_item *fromp, *top;
3086
3087         sub_oid_map = kh_init_oid_map();
3088
3089         for_each_string_list_item(fromp, from) {
3090                 top = string_list_lookup(to, fromp->string);
3091                 if (!fromp->util) {
3092                         die(_("Missing from marks for submodule '%s'"), fromp->string);
3093                 } else if (!top || !top->util) {
3094                         die(_("Missing to marks for submodule '%s'"), fromp->string);
3095                 }
3096                 build_mark_map_one(fromp->util, top->util);
3097         }
3098 }
3099
3100 static struct object_entry *parse_treeish_dataref(const char **p)
3101 {
3102         struct object_id oid;
3103         struct object_entry *e;
3104
3105         if (**p == ':') {       /* <mark> */
3106                 e = find_mark(marks, parse_mark_ref_space(p));
3107                 if (!e)
3108                         die("Unknown mark: %s", command_buf.buf);
3109                 oidcpy(&oid, &e->idx.oid);
3110         } else {        /* <sha1> */
3111                 if (parse_mapped_oid_hex(*p, &oid, p))
3112                         die("Invalid dataref: %s", command_buf.buf);
3113                 e = find_object(&oid);
3114                 if (*(*p)++ != ' ')
3115                         die("Missing space after tree-ish: %s", command_buf.buf);
3116         }
3117
3118         while (!e || e->type != OBJ_TREE)
3119                 e = dereference(e, &oid);
3120         return e;
3121 }
3122
3123 static void print_ls(int mode, const unsigned char *hash, const char *path)
3124 {
3125         static struct strbuf line = STRBUF_INIT;
3126
3127         /* See show_tree(). */
3128         const char *type =
3129                 S_ISGITLINK(mode) ? commit_type :
3130                 S_ISDIR(mode) ? tree_type :
3131                 blob_type;
3132
3133         if (!mode) {
3134                 /* missing SP path LF */
3135                 strbuf_reset(&line);
3136                 strbuf_addstr(&line, "missing ");
3137                 quote_c_style(path, &line, NULL, 0);
3138                 strbuf_addch(&line, '\n');
3139         } else {
3140                 /* mode SP type SP object_name TAB path LF */
3141                 strbuf_reset(&line);
3142                 strbuf_addf(&line, "%06o %s %s\t",
3143                                 mode & ~NO_DELTA, type, hash_to_hex(hash));
3144                 quote_c_style(path, &line, NULL, 0);
3145                 strbuf_addch(&line, '\n');
3146         }
3147         cat_blob_write(line.buf, line.len);
3148 }
3149
3150 static void parse_ls(const char *p, struct branch *b)
3151 {
3152         struct tree_entry *root = NULL;
3153         struct tree_entry leaf = {NULL};
3154
3155         /* ls SP (<tree-ish> SP)? <path> */
3156         if (*p == '"') {
3157                 if (!b)
3158                         die("Not in a commit: %s", command_buf.buf);
3159                 root = &b->branch_tree;
3160         } else {
3161                 struct object_entry *e = parse_treeish_dataref(&p);
3162                 root = new_tree_entry();
3163                 oidcpy(&root->versions[1].oid, &e->idx.oid);
3164                 if (!is_null_oid(&root->versions[1].oid))
3165                         root->versions[1].mode = S_IFDIR;
3166                 load_tree(root);
3167         }
3168         if (*p == '"') {
3169                 static struct strbuf uq = STRBUF_INIT;
3170                 const char *endp;
3171                 strbuf_reset(&uq);
3172                 if (unquote_c_style(&uq, p, &endp))
3173                         die("Invalid path: %s", command_buf.buf);
3174                 if (*endp)
3175                         die("Garbage after path in: %s", command_buf.buf);
3176                 p = uq.buf;
3177         }
3178         tree_content_get(root, p, &leaf, 1);
3179         /*
3180          * A directory in preparation would have a sha1 of zero
3181          * until it is saved.  Save, for simplicity.
3182          */
3183         if (S_ISDIR(leaf.versions[1].mode))
3184                 store_tree(&leaf);
3185
3186         print_ls(leaf.versions[1].mode, leaf.versions[1].oid.hash, p);
3187         if (leaf.tree)
3188                 release_tree_content_recursive(leaf.tree);
3189         if (!b || root != &b->branch_tree)
3190                 release_tree_entry(root);
3191 }
3192
3193 static void checkpoint(void)
3194 {
3195         checkpoint_requested = 0;
3196         if (object_count) {
3197                 cycle_packfile();
3198         }
3199         dump_branches();
3200         dump_tags();
3201         dump_marks();
3202 }
3203
3204 static void parse_checkpoint(void)
3205 {
3206         checkpoint_requested = 1;
3207         skip_optional_lf();
3208 }
3209
3210 static void parse_progress(void)
3211 {
3212         fwrite(command_buf.buf, 1, command_buf.len, stdout);
3213         fputc('\n', stdout);
3214         fflush(stdout);
3215         skip_optional_lf();
3216 }
3217
3218 static void parse_alias(void)
3219 {
3220         struct object_entry *e;
3221         struct branch b;
3222
3223         skip_optional_lf();
3224         read_next_command();
3225
3226         /* mark ... */
3227         parse_mark();
3228         if (!next_mark)
3229                 die(_("Expected 'mark' command, got %s"), command_buf.buf);
3230
3231         /* to ... */
3232         memset(&b, 0, sizeof(b));
3233         if (!parse_objectish_with_prefix(&b, "to "))
3234                 die(_("Expected 'to' command, got %s"), command_buf.buf);
3235         e = find_object(&b.oid);
3236         assert(e);
3237         insert_mark(marks, next_mark, e);
3238 }
3239
3240 static char* make_fast_import_path(const char *path)
3241 {
3242         if (!relative_marks_paths || is_absolute_path(path))
3243                 return xstrdup(path);
3244         return git_pathdup("info/fast-import/%s", path);
3245 }
3246
3247 static void option_import_marks(const char *marks,
3248                                         int from_stream, int ignore_missing)
3249 {
3250         if (import_marks_file) {
3251                 if (from_stream)
3252                         die("Only one import-marks command allowed per stream");
3253
3254                 /* read previous mark file */
3255                 if(!import_marks_file_from_stream)
3256                         read_marks();
3257         }
3258
3259         import_marks_file = make_fast_import_path(marks);
3260         import_marks_file_from_stream = from_stream;
3261         import_marks_file_ignore_missing = ignore_missing;
3262 }
3263
3264 static void option_date_format(const char *fmt)
3265 {
3266         if (!strcmp(fmt, "raw"))
3267                 whenspec = WHENSPEC_RAW;
3268         else if (!strcmp(fmt, "raw-permissive"))
3269                 whenspec = WHENSPEC_RAW_PERMISSIVE;
3270         else if (!strcmp(fmt, "rfc2822"))
3271                 whenspec = WHENSPEC_RFC2822;
3272         else if (!strcmp(fmt, "now"))
3273                 whenspec = WHENSPEC_NOW;
3274         else
3275                 die("unknown --date-format argument %s", fmt);
3276 }
3277
3278 static unsigned long ulong_arg(const char *option, const char *arg)
3279 {
3280         char *endptr;
3281         unsigned long rv = strtoul(arg, &endptr, 0);
3282         if (strchr(arg, '-') || endptr == arg || *endptr)
3283                 die("%s: argument must be a non-negative integer", option);
3284         return rv;
3285 }
3286
3287 static void option_depth(const char *depth)
3288 {
3289         max_depth = ulong_arg("--depth", depth);
3290         if (max_depth > MAX_DEPTH)
3291                 die("--depth cannot exceed %u", MAX_DEPTH);
3292 }
3293
3294 static void option_active_branches(const char *branches)
3295 {
3296         max_active_branches = ulong_arg("--active-branches", branches);
3297 }
3298
3299 static void option_export_marks(const char *marks)
3300 {
3301         export_marks_file = make_fast_import_path(marks);
3302 }
3303
3304 static void option_cat_blob_fd(const char *fd)
3305 {
3306         unsigned long n = ulong_arg("--cat-blob-fd", fd);
3307         if (n > (unsigned long) INT_MAX)
3308                 die("--cat-blob-fd cannot exceed %d", INT_MAX);
3309         cat_blob_fd = (int) n;
3310 }
3311
3312 static void option_export_pack_edges(const char *edges)
3313 {
3314         if (pack_edges)
3315                 fclose(pack_edges);
3316         pack_edges = xfopen(edges, "a");
3317 }
3318
3319 static void option_rewrite_submodules(const char *arg, struct string_list *list)
3320 {
3321         struct mark_set *ms;
3322         FILE *fp;
3323         char *s = xstrdup(arg);
3324         char *f = strchr(s, ':');
3325         if (!f)
3326                 die(_("Expected format name:filename for submodule rewrite option"));
3327         *f = '\0';
3328         f++;
3329         ms = xcalloc(1, sizeof(*ms));
3330         string_list_insert(list, s)->util = ms;
3331
3332         fp = fopen(f, "r");
3333         if (!fp)
3334                 die_errno("cannot read '%s'", f);
3335         read_mark_file(ms, fp, insert_oid_entry);
3336         fclose(fp);
3337 }
3338
3339 static int parse_one_option(const char *option)
3340 {
3341         if (skip_prefix(option, "max-pack-size=", &option)) {
3342                 unsigned long v;
3343                 if (!git_parse_ulong(option, &v))
3344                         return 0;
3345                 if (v < 8192) {
3346                         warning("max-pack-size is now in bytes, assuming --max-pack-size=%lum", v);
3347                         v *= 1024 * 1024;
3348                 } else if (v < 1024 * 1024) {
3349                         warning("minimum max-pack-size is 1 MiB");
3350                         v = 1024 * 1024;
3351                 }
3352                 max_packsize = v;
3353         } else if (skip_prefix(option, "big-file-threshold=", &option)) {
3354                 unsigned long v;
3355                 if (!git_parse_ulong(option, &v))
3356                         return 0;
3357                 big_file_threshold = v;
3358         } else if (skip_prefix(option, "depth=", &option)) {
3359                 option_depth(option);
3360         } else if (skip_prefix(option, "active-branches=", &option)) {
3361                 option_active_branches(option);
3362         } else if (skip_prefix(option, "export-pack-edges=", &option)) {
3363                 option_export_pack_edges(option);
3364         } else if (!strcmp(option, "quiet")) {
3365                 show_stats = 0;
3366         } else if (!strcmp(option, "stats")) {
3367                 show_stats = 1;
3368         } else if (!strcmp(option, "allow-unsafe-features")) {
3369                 ; /* already handled during early option parsing */
3370         } else {
3371                 return 0;
3372         }
3373
3374         return 1;
3375 }
3376
3377 static void check_unsafe_feature(const char *feature, int from_stream)
3378 {
3379         if (from_stream && !allow_unsafe_features)
3380                 die(_("feature '%s' forbidden in input without --allow-unsafe-features"),
3381                     feature);
3382 }
3383
3384 static int parse_one_feature(const char *feature, int from_stream)
3385 {
3386         const char *arg;
3387
3388         if (skip_prefix(feature, "date-format=", &arg)) {
3389                 option_date_format(arg);
3390         } else if (skip_prefix(feature, "import-marks=", &arg)) {
3391                 check_unsafe_feature("import-marks", from_stream);
3392                 option_import_marks(arg, from_stream, 0);
3393         } else if (skip_prefix(feature, "import-marks-if-exists=", &arg)) {
3394                 check_unsafe_feature("import-marks-if-exists", from_stream);
3395                 option_import_marks(arg, from_stream, 1);
3396         } else if (skip_prefix(feature, "export-marks=", &arg)) {
3397                 check_unsafe_feature(feature, from_stream);
3398                 option_export_marks(arg);
3399         } else if (!strcmp(feature, "alias")) {
3400                 ; /* Don't die - this feature is supported */
3401         } else if (skip_prefix(feature, "rewrite-submodules-to=", &arg)) {
3402                 option_rewrite_submodules(arg, &sub_marks_to);
3403         } else if (skip_prefix(feature, "rewrite-submodules-from=", &arg)) {
3404                 option_rewrite_submodules(arg, &sub_marks_from);
3405         } else if (skip_prefix(feature, "rewrite-submodules-from=", &arg)) {
3406         } else if (!strcmp(feature, "get-mark")) {
3407                 ; /* Don't die - this feature is supported */
3408         } else if (!strcmp(feature, "cat-blob")) {
3409                 ; /* Don't die - this feature is supported */
3410         } else if (!strcmp(feature, "relative-marks")) {
3411                 relative_marks_paths = 1;
3412         } else if (!strcmp(feature, "no-relative-marks")) {
3413                 relative_marks_paths = 0;
3414         } else if (!strcmp(feature, "done")) {
3415                 require_explicit_termination = 1;
3416         } else if (!strcmp(feature, "force")) {
3417                 force_update = 1;
3418         } else if (!strcmp(feature, "notes") || !strcmp(feature, "ls")) {
3419                 ; /* do nothing; we have the feature */
3420         } else {
3421                 return 0;
3422         }
3423
3424         return 1;
3425 }
3426
3427 static void parse_feature(const char *feature)
3428 {
3429         if (seen_data_command)
3430                 die("Got feature command '%s' after data command", feature);
3431
3432         if (parse_one_feature(feature, 1))
3433                 return;
3434
3435         die("This version of fast-import does not support feature %s.", feature);
3436 }
3437
3438 static void parse_option(const char *option)
3439 {
3440         if (seen_data_command)
3441                 die("Got option command '%s' after data command", option);
3442
3443         if (parse_one_option(option))
3444                 return;
3445
3446         die("This version of fast-import does not support option: %s", option);
3447 }
3448
3449 static void git_pack_config(void)
3450 {
3451         int indexversion_value;
3452         int limit;
3453         unsigned long packsizelimit_value;
3454
3455         if (!git_config_get_ulong("pack.depth", &max_depth)) {
3456                 if (max_depth > MAX_DEPTH)
3457                         max_depth = MAX_DEPTH;
3458         }
3459         if (!git_config_get_int("pack.indexversion", &indexversion_value)) {
3460                 pack_idx_opts.version = indexversion_value;
3461                 if (pack_idx_opts.version > 2)
3462                         git_die_config("pack.indexversion",
3463                                         "bad pack.indexversion=%"PRIu32, pack_idx_opts.version);
3464         }
3465         if (!git_config_get_ulong("pack.packsizelimit", &packsizelimit_value))
3466                 max_packsize = packsizelimit_value;
3467
3468         if (!git_config_get_int("fastimport.unpacklimit", &limit))
3469                 unpack_limit = limit;
3470         else if (!git_config_get_int("transfer.unpacklimit", &limit))
3471                 unpack_limit = limit;
3472
3473         git_config(git_default_config, NULL);
3474 }
3475
3476 static const char fast_import_usage[] =
3477 "git fast-import [--date-format=<f>] [--max-pack-size=<n>] [--big-file-threshold=<n>] [--depth=<n>] [--active-branches=<n>] [--export-marks=<marks.file>]";
3478
3479 static void parse_argv(void)
3480 {
3481         unsigned int i;
3482
3483         for (i = 1; i < global_argc; i++) {
3484                 const char *a = global_argv[i];
3485
3486                 if (*a != '-' || !strcmp(a, "--"))
3487                         break;
3488
3489                 if (!skip_prefix(a, "--", &a))
3490                         die("unknown option %s", a);
3491
3492                 if (parse_one_option(a))
3493                         continue;
3494
3495                 if (parse_one_feature(a, 0))
3496                         continue;
3497
3498                 if (skip_prefix(a, "cat-blob-fd=", &a)) {
3499                         option_cat_blob_fd(a);
3500                         continue;
3501                 }
3502
3503                 die("unknown option --%s", a);
3504         }
3505         if (i != global_argc)
3506                 usage(fast_import_usage);
3507
3508         seen_data_command = 1;
3509         if (import_marks_file)
3510                 read_marks();
3511         build_mark_map(&sub_marks_from, &sub_marks_to);
3512 }
3513
3514 int cmd_fast_import(int argc, const char **argv, const char *prefix)
3515 {
3516         unsigned int i;
3517
3518         if (argc == 2 && !strcmp(argv[1], "-h"))
3519                 usage(fast_import_usage);
3520
3521         reset_pack_idx_option(&pack_idx_opts);
3522         git_pack_config();
3523
3524         alloc_objects(object_entry_alloc);
3525         strbuf_init(&command_buf, 0);
3526         atom_table = xcalloc(atom_table_sz, sizeof(struct atom_str*));
3527         branch_table = xcalloc(branch_table_sz, sizeof(struct branch*));
3528         avail_tree_table = xcalloc(avail_tree_table_sz, sizeof(struct avail_tree_content*));
3529         marks = mem_pool_calloc(&fi_mem_pool, 1, sizeof(struct mark_set));
3530
3531         hashmap_init(&object_table, object_entry_hashcmp, NULL, 0);
3532
3533         /*
3534          * We don't parse most options until after we've seen the set of
3535          * "feature" lines at the start of the stream (which allows the command
3536          * line to override stream data). But we must do an early parse of any
3537          * command-line options that impact how we interpret the feature lines.
3538          */
3539         for (i = 1; i < argc; i++) {
3540                 const char *arg = argv[i];
3541                 if (*arg != '-' || !strcmp(arg, "--"))
3542                         break;
3543                 if (!strcmp(arg, "--allow-unsafe-features"))
3544                         allow_unsafe_features = 1;
3545         }
3546
3547         global_argc = argc;
3548         global_argv = argv;
3549
3550         rc_free = mem_pool_alloc(&fi_mem_pool, cmd_save * sizeof(*rc_free));
3551         for (i = 0; i < (cmd_save - 1); i++)
3552                 rc_free[i].next = &rc_free[i + 1];
3553         rc_free[cmd_save - 1].next = NULL;
3554
3555         start_packfile();
3556         set_die_routine(die_nicely);
3557         set_checkpoint_signal();
3558         while (read_next_command() != EOF) {
3559                 const char *v;
3560                 if (!strcmp("blob", command_buf.buf))
3561                         parse_new_blob();
3562                 else if (skip_prefix(command_buf.buf, "commit ", &v))
3563                         parse_new_commit(v);
3564                 else if (skip_prefix(command_buf.buf, "tag ", &v))
3565                         parse_new_tag(v);
3566                 else if (skip_prefix(command_buf.buf, "reset ", &v))
3567                         parse_reset_branch(v);
3568                 else if (skip_prefix(command_buf.buf, "ls ", &v))
3569                         parse_ls(v, NULL);
3570                 else if (skip_prefix(command_buf.buf, "cat-blob ", &v))
3571                         parse_cat_blob(v);
3572                 else if (skip_prefix(command_buf.buf, "get-mark ", &v))
3573                         parse_get_mark(v);
3574                 else if (!strcmp("checkpoint", command_buf.buf))
3575                         parse_checkpoint();
3576                 else if (!strcmp("done", command_buf.buf))
3577                         break;
3578                 else if (!strcmp("alias", command_buf.buf))
3579                         parse_alias();
3580                 else if (starts_with(command_buf.buf, "progress "))
3581                         parse_progress();
3582                 else if (skip_prefix(command_buf.buf, "feature ", &v))
3583                         parse_feature(v);
3584                 else if (skip_prefix(command_buf.buf, "option git ", &v))
3585                         parse_option(v);
3586                 else if (starts_with(command_buf.buf, "option "))
3587                         /* ignore non-git options*/;
3588                 else
3589                         die("Unsupported command: %s", command_buf.buf);
3590
3591                 if (checkpoint_requested)
3592                         checkpoint();
3593         }
3594
3595         /* argv hasn't been parsed yet, do so */
3596         if (!seen_data_command)
3597                 parse_argv();
3598
3599         if (require_explicit_termination && feof(stdin))
3600                 die("stream ends early");
3601
3602         end_packfile();
3603
3604         dump_branches();
3605         dump_tags();
3606         unkeep_all_packs();
3607         dump_marks();
3608
3609         if (pack_edges)
3610                 fclose(pack_edges);
3611
3612         if (show_stats) {
3613                 uintmax_t total_count = 0, duplicate_count = 0;
3614                 for (i = 0; i < ARRAY_SIZE(object_count_by_type); i++)
3615                         total_count += object_count_by_type[i];
3616                 for (i = 0; i < ARRAY_SIZE(duplicate_count_by_type); i++)
3617                         duplicate_count += duplicate_count_by_type[i];
3618
3619                 fprintf(stderr, "%s statistics:\n", argv[0]);
3620                 fprintf(stderr, "---------------------------------------------------------------------\n");
3621                 fprintf(stderr, "Alloc'd objects: %10" PRIuMAX "\n", alloc_count);
3622                 fprintf(stderr, "Total objects:   %10" PRIuMAX " (%10" PRIuMAX " duplicates                  )\n", total_count, duplicate_count);
3623                 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]);
3624                 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]);
3625                 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]);
3626                 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]);
3627                 fprintf(stderr, "Total branches:  %10lu (%10lu loads     )\n", branch_count, branch_load_count);
3628                 fprintf(stderr, "      marks:     %10" PRIuMAX " (%10" PRIuMAX " unique    )\n", (((uintmax_t)1) << marks->shift) * 1024, marks_set_count);
3629                 fprintf(stderr, "      atoms:     %10u\n", atom_cnt);
3630                 fprintf(stderr, "Memory total:    %10" PRIuMAX " KiB\n", (tree_entry_allocd + fi_mem_pool.pool_alloc + alloc_count*sizeof(struct object_entry))/1024);
3631                 fprintf(stderr, "       pools:    %10lu KiB\n", (unsigned long)((tree_entry_allocd + fi_mem_pool.pool_alloc) /1024));
3632                 fprintf(stderr, "     objects:    %10" PRIuMAX " KiB\n", (alloc_count*sizeof(struct object_entry))/1024);
3633                 fprintf(stderr, "---------------------------------------------------------------------\n");
3634                 pack_report();
3635                 fprintf(stderr, "---------------------------------------------------------------------\n");
3636                 fprintf(stderr, "\n");
3637         }
3638
3639         return failure ? 1 : 0;
3640 }