commit.c

   1 #include "cache.h"
   2 #include "tag.h"
   3 #include "commit.h"
   4 #include "commit-graph.h"
   5 #include "pkt-line.h"
   6 #include "utf8.h"
   7 #include "diff.h"
   8 #include "revision.h"
   9 #include "notes.h"
  10 #include "gpg-interface.h"
  11 #include "mergesort.h"
  12 #include "commit-slab.h"
  13 #include "prio-queue.h"
  14 #include "sha1-lookup.h"
  15 #include "wt-status.h"
  16 #include "advice.h"
  17
  18 static struct commit_extra_header *read_commit_extra_header_lines(const char *buf, size_t len, const char **);
  19
  20 int save_commit_buffer = 1;
  21
  22 const char *commit_type = "commit";
  23
  24 struct commit *lookup_commit_reference_gently(const struct object_id *oid,
  25                                               int quiet)
  26 {
  27         struct object *obj = deref_tag(parse_object(oid), NULL, 0);
  28
  29         if (!obj)
  30                 return NULL;
  31         return object_as_type(obj, OBJ_COMMIT, quiet);
  32 }
  33
  34 struct commit *lookup_commit_reference(const struct object_id *oid)
  35 {
  36         return lookup_commit_reference_gently(oid, 0);
  37 }
  38
  39 struct commit *lookup_commit_or_die(const struct object_id *oid, const char *ref_name)
  40 {
  41         struct commit *c = lookup_commit_reference(oid);
  42         if (!c)
  43                 die(_("could not parse %s"), ref_name);
  44         if (oidcmp(oid, &c->object.oid)) {
  45                 warning(_("%s %s is not a commit!"),
  46                         ref_name, oid_to_hex(oid));
  47         }
  48         return c;
  49 }
  50
  51 struct commit *lookup_commit(const struct object_id *oid)
  52 {
  53         struct object *obj = lookup_object(oid->hash);
  54         if (!obj)
  55                 return create_object(oid->hash, alloc_commit_node());
  56         return object_as_type(obj, OBJ_COMMIT, 0);
  57 }
  58
  59 struct commit *lookup_commit_reference_by_name(const char *name)
  60 {
  61         struct object_id oid;
  62         struct commit *commit;
  63
  64         if (get_oid_committish(name, &oid))
  65                 return NULL;
  66         commit = lookup_commit_reference(&oid);
  67         if (parse_commit(commit))
  68                 return NULL;
  69         return commit;
  70 }
  71
  72 static timestamp_t parse_commit_date(const char *buf, const char *tail)
  73 {
  74         const char *dateptr;
  75
  76         if (buf + 6 >= tail)
  77                 return 0;
  78         if (memcmp(buf, "author", 6))
  79                 return 0;
  80         while (buf < tail && *buf++ != '\n')
  81                 /* nada */;
  82         if (buf + 9 >= tail)
  83                 return 0;
  84         if (memcmp(buf, "committer", 9))
  85                 return 0;
  86         while (buf < tail && *buf++ != '>')
  87                 /* nada */;
  88         if (buf >= tail)
  89                 return 0;
  90         dateptr = buf;
  91         while (buf < tail && *buf++ != '\n')
  92                 /* nada */;
  93         if (buf >= tail)
  94                 return 0;
  95         /* dateptr < buf && buf[-1] == '\n', so parsing will stop at buf-1 */
  96         return parse_timestamp(dateptr, NULL, 10);
  97 }
  98
  99 static struct commit_graft **commit_graft;
 100 static int commit_graft_alloc, commit_graft_nr;
 101
 102 static const unsigned char *commit_graft_sha1_access(size_t index, void *table)
 103 {
 104         struct commit_graft **commit_graft_table = table;
 105         return commit_graft_table[index]->oid.hash;
 106 }
 107
 108 static int commit_graft_pos(const unsigned char *sha1)
 109 {
 110         return sha1_pos(sha1, commit_graft, commit_graft_nr,
 111                         commit_graft_sha1_access);
 112 }
 113
 114 int register_commit_graft(struct commit_graft *graft, int ignore_dups)
 115 {
 116         int pos = commit_graft_pos(graft->oid.hash);
 117
 118         if (0 <= pos) {
 119                 if (ignore_dups)
 120                         free(graft);
 121                 else {
 122                         free(commit_graft[pos]);
 123                         commit_graft[pos] = graft;
 124                 }
 125                 return 1;
 126         }
 127         pos = -pos - 1;
 128         ALLOC_GROW(commit_graft, commit_graft_nr + 1, commit_graft_alloc);
 129         commit_graft_nr++;
 130         if (pos < commit_graft_nr)
 131                 MOVE_ARRAY(commit_graft + pos + 1, commit_graft + pos,
 132                            commit_graft_nr - pos - 1);
 133         commit_graft[pos] = graft;
 134         return 0;
 135 }
 136
 137 struct commit_graft *read_graft_line(struct strbuf *line)
 138 {
 139         /* The format is just "Commit Parent1 Parent2 ...\n" */
 140         int i, phase;
 141         const char *tail = NULL;
 142         struct commit_graft *graft = NULL;
 143         struct object_id dummy_oid, *oid;
 144
 145         strbuf_rtrim(line);
 146         if (!line->len || line->buf[0] == '#')
 147                 return NULL;
 148         /*
 149          * phase 0 verifies line, counts hashes in line and allocates graft
 150          * phase 1 fills graft
 151          */
 152         for (phase = 0; phase < 2; phase++) {
 153                 oid = graft ? &graft->oid : &dummy_oid;
 154                 if (parse_oid_hex(line->buf, oid, &tail))
 155                         goto bad_graft_data;
 156                 for (i = 0; *tail != '\0'; i++) {
 157                         oid = graft ? &graft->parent[i] : &dummy_oid;
 158                         if (!isspace(*tail++) || parse_oid_hex(tail, oid, &tail))
 159                                 goto bad_graft_data;
 160                 }
 161                 if (!graft) {
 162                         graft = xmalloc(st_add(sizeof(*graft),
 163                                                st_mult(sizeof(struct object_id), i)));
 164                         graft->nr_parent = i;
 165                 }
 166         }
 167         return graft;
 168
 169 bad_graft_data:
 170         error("bad graft data: %s", line->buf);
 171         assert(!graft);
 172         return NULL;
 173 }
 174
 175 static int read_graft_file(const char *graft_file)
 176 {
 177         FILE *fp = fopen_or_warn(graft_file, "r");
 178         struct strbuf buf = STRBUF_INIT;
 179         if (!fp)
 180                 return -1;
 181         if (advice_graft_file_deprecated)
 182                 advise(_("Support for <GIT_DIR>/info/grafts is deprecated\n"
 183                          "and will be removed in a future Git version.\n"
 184                          "\n"
 185                          "Please use \"git replace --convert-graft-file\"\n"
 186                          "to convert the grafts into replace refs.\n"
 187                          "\n"
 188                          "Turn this message off by running\n"
 189                          "\"git config advice.graftFileDeprecated false\""));
 190         while (!strbuf_getwholeline(&buf, fp, '\n')) {
 191                 /* The format is just "Commit Parent1 Parent2 ...\n" */
 192                 struct commit_graft *graft = read_graft_line(&buf);
 193                 if (!graft)
 194                         continue;
 195                 if (register_commit_graft(graft, 1))
 196                         error("duplicate graft data: %s", buf.buf);
 197         }
 198         fclose(fp);
 199         strbuf_release(&buf);
 200         return 0;
 201 }
 202
 203 static void prepare_commit_graft(void)
 204 {
 205         static int commit_graft_prepared;
 206         char *graft_file;
 207
 208         if (commit_graft_prepared)
 209                 return;
 210         graft_file = get_graft_file();
 211         read_graft_file(graft_file);
 212         /* make sure shallows are read */
 213         is_repository_shallow();
 214         commit_graft_prepared = 1;
 215 }
 216
 217 struct commit_graft *lookup_commit_graft(const struct object_id *oid)
 218 {
 219         int pos;
 220         prepare_commit_graft();
 221         pos = commit_graft_pos(oid->hash);
 222         if (pos < 0)
 223                 return NULL;
 224         return commit_graft[pos];
 225 }
 226
 227 int for_each_commit_graft(each_commit_graft_fn fn, void *cb_data)
 228 {
 229         int i, ret;
 230         for (i = ret = 0; i < commit_graft_nr && !ret; i++)
 231                 ret = fn(commit_graft[i], cb_data);
 232         return ret;
 233 }
 234
 235 int unregister_shallow(const struct object_id *oid)
 236 {
 237         int pos = commit_graft_pos(oid->hash);
 238         if (pos < 0)
 239                 return -1;
 240         if (pos + 1 < commit_graft_nr)
 241                 MOVE_ARRAY(commit_graft + pos, commit_graft + pos + 1,
 242                            commit_graft_nr - pos - 1);
 243         commit_graft_nr--;
 244         return 0;
 245 }
 246
 247 struct commit_buffer {
 248         void *buffer;
 249         unsigned long size;
 250 };
 251 define_commit_slab(buffer_slab, struct commit_buffer);
 252 static struct buffer_slab buffer_slab = COMMIT_SLAB_INIT(1, buffer_slab);
 253
 254 void set_commit_buffer(struct commit *commit, void *buffer, unsigned long size)
 255 {
 256         struct commit_buffer *v = buffer_slab_at(&buffer_slab, commit);
 257         v->buffer = buffer;
 258         v->size = size;
 259 }
 260
 261 const void *get_cached_commit_buffer(const struct commit *commit, unsigned long *sizep)
 262 {
 263         struct commit_buffer *v = buffer_slab_peek(&buffer_slab, commit);
 264         if (!v) {
 265                 if (sizep)
 266                         *sizep = 0;
 267                 return NULL;
 268         }
 269         if (sizep)
 270                 *sizep = v->size;
 271         return v->buffer;
 272 }
 273
 274 const void *get_commit_buffer(const struct commit *commit, unsigned long *sizep)
 275 {
 276         const void *ret = get_cached_commit_buffer(commit, sizep);
 277         if (!ret) {
 278                 enum object_type type;
 279                 unsigned long size;
 280                 ret = read_object_file(&commit->object.oid, &type, &size);
 281                 if (!ret)
 282                         die("cannot read commit object %s",
 283                             oid_to_hex(&commit->object.oid));
 284                 if (type != OBJ_COMMIT)
 285                         die("expected commit for %s, got %s",
 286                             oid_to_hex(&commit->object.oid), type_name(type));
 287                 if (sizep)
 288                         *sizep = size;
 289         }
 290         return ret;
 291 }
 292
 293 void unuse_commit_buffer(const struct commit *commit, const void *buffer)
 294 {
 295         struct commit_buffer *v = buffer_slab_peek(&buffer_slab, commit);
 296         if (!(v && v->buffer == buffer))
 297                 free((void *)buffer);
 298 }
 299
 300 void free_commit_buffer(struct commit *commit)
 301 {
 302         struct commit_buffer *v = buffer_slab_peek(&buffer_slab, commit);
 303         if (v) {
 304                 FREE_AND_NULL(v->buffer);
 305                 v->size = 0;
 306         }
 307 }
 308
 309 struct tree *get_commit_tree(const struct commit *commit)
 310 {
 311         if (commit->maybe_tree || !commit->object.parsed)
 312                 return commit->maybe_tree;
 313
 314         if (commit->graph_pos == COMMIT_NOT_FROM_GRAPH)
 315                 BUG("commit has NULL tree, but was not loaded from commit-graph");
 316
 317         return get_commit_tree_in_graph(commit);
 318 }
 319
 320 struct object_id *get_commit_tree_oid(const struct commit *commit)
 321 {
 322         return &get_commit_tree(commit)->object.oid;
 323 }
 324
 325 const void *detach_commit_buffer(struct commit *commit, unsigned long *sizep)
 326 {
 327         struct commit_buffer *v = buffer_slab_peek(&buffer_slab, commit);
 328         void *ret;
 329
 330         if (!v) {
 331                 if (sizep)
 332                         *sizep = 0;
 333                 return NULL;
 334         }
 335         ret = v->buffer;
 336         if (sizep)
 337                 *sizep = v->size;
 338
 339         v->buffer = NULL;
 340         v->size = 0;
 341         return ret;
 342 }
 343
 344 int parse_commit_buffer(struct commit *item, const void *buffer, unsigned long size)
 345 {
 346         const char *tail = buffer;
 347         const char *bufptr = buffer;
 348         struct object_id parent;
 349         struct commit_list **pptr;
 350         struct commit_graft *graft;
 351         const int tree_entry_len = GIT_SHA1_HEXSZ + 5;
 352         const int parent_entry_len = GIT_SHA1_HEXSZ + 7;
 353
 354         if (item->object.parsed)
 355                 return 0;
 356         item->object.parsed = 1;
 357         tail += size;
 358         if (tail <= bufptr + tree_entry_len + 1 || memcmp(bufptr, "tree ", 5) ||
 359                         bufptr[tree_entry_len] != '\n')
 360                 return error("bogus commit object %s", oid_to_hex(&item->object.oid));
 361         if (get_oid_hex(bufptr + 5, &parent) < 0)
 362                 return error("bad tree pointer in commit %s",
 363                              oid_to_hex(&item->object.oid));
 364         item->maybe_tree = lookup_tree(&parent);
 365         bufptr += tree_entry_len + 1; /* "tree " + "hex sha1" + "\n" */
 366         pptr = &item->parents;
 367
 368         graft = lookup_commit_graft(&item->object.oid);
 369         while (bufptr + parent_entry_len < tail && !memcmp(bufptr, "parent ", 7)) {
 370                 struct commit *new_parent;
 371
 372                 if (tail <= bufptr + parent_entry_len + 1 ||
 373                     get_oid_hex(bufptr + 7, &parent) ||
 374                     bufptr[parent_entry_len] != '\n')
 375                         return error("bad parents in commit %s", oid_to_hex(&item->object.oid));
 376                 bufptr += parent_entry_len + 1;
 377                 /*
 378                  * The clone is shallow if nr_parent < 0, and we must
 379                  * not traverse its real parents even when we unhide them.
 380                  */
 381                 if (graft && (graft->nr_parent < 0 || grafts_replace_parents))
 382                         continue;
 383                 new_parent = lookup_commit(&parent);
 384                 if (new_parent)
 385                         pptr = &commit_list_insert(new_parent, pptr)->next;
 386         }
 387         if (graft) {
 388                 int i;
 389                 struct commit *new_parent;
 390                 for (i = 0; i < graft->nr_parent; i++) {
 391                         new_parent = lookup_commit(&graft->parent[i]);
 392                         if (!new_parent)
 393                                 continue;
 394                         pptr = &commit_list_insert(new_parent, pptr)->next;
 395                 }
 396         }
 397         item->date = parse_commit_date(bufptr, tail);
 398
 399         return 0;
 400 }
 401
 402 int parse_commit_gently(struct commit *item, int quiet_on_missing)
 403 {
 404         enum object_type type;
 405         void *buffer;
 406         unsigned long size;
 407         int ret;
 408
 409         if (!item)
 410                 return -1;
 411         if (item->object.parsed)
 412                 return 0;
 413         if (parse_commit_in_graph(item))
 414                 return 0;
 415         buffer = read_object_file(&item->object.oid, &type, &size);
 416         if (!buffer)
 417                 return quiet_on_missing ? -1 :
 418                         error("Could not read %s",
 419                              oid_to_hex(&item->object.oid));
 420         if (type != OBJ_COMMIT) {
 421                 free(buffer);
 422                 return error("Object %s not a commit",
 423                              oid_to_hex(&item->object.oid));
 424         }
 425         ret = parse_commit_buffer(item, buffer, size);
 426         if (save_commit_buffer && !ret) {
 427                 set_commit_buffer(item, buffer, size);
 428                 return 0;
 429         }
 430         free(buffer);
 431         return ret;
 432 }
 433
 434 void parse_commit_or_die(struct commit *item)
 435 {
 436         if (parse_commit(item))
 437                 die("unable to parse commit %s",
 438                     item ? oid_to_hex(&item->object.oid) : "(null)");
 439 }
 440
 441 int find_commit_subject(const char *commit_buffer, const char **subject)
 442 {
 443         const char *eol;
 444         const char *p = commit_buffer;
 445
 446         while (*p && (*p != '\n' || p[1] != '\n'))
 447                 p++;
 448         if (*p) {
 449                 p = skip_blank_lines(p + 2);
 450                 eol = strchrnul(p, '\n');
 451         } else
 452                 eol = p;
 453
 454         *subject = p;
 455
 456         return eol - p;
 457 }
 458
 459 struct commit_list *commit_list_insert(struct commit *item, struct commit_list **list_p)
 460 {
 461         struct commit_list *new_list = xmalloc(sizeof(struct commit_list));
 462         new_list->item = item;
 463         new_list->next = *list_p;
 464         *list_p = new_list;
 465         return new_list;
 466 }
 467
 468 unsigned commit_list_count(const struct commit_list *l)
 469 {
 470         unsigned c = 0;
 471         for (; l; l = l->next )
 472                 c++;
 473         return c;
 474 }
 475
 476 struct commit_list *copy_commit_list(struct commit_list *list)
 477 {
 478         struct commit_list *head = NULL;
 479         struct commit_list **pp = &head;
 480         while (list) {
 481                 pp = commit_list_append(list->item, pp);
 482                 list = list->next;
 483         }
 484         return head;
 485 }
 486
 487 void free_commit_list(struct commit_list *list)
 488 {
 489         while (list)
 490                 pop_commit(&list);
 491 }
 492
 493 struct commit_list * commit_list_insert_by_date(struct commit *item, struct commit_list **list)
 494 {
 495         struct commit_list **pp = list;
 496         struct commit_list *p;
 497         while ((p = *pp) != NULL) {
 498                 if (p->item->date < item->date) {
 499                         break;
 500                 }
 501                 pp = &p->next;
 502         }
 503         return commit_list_insert(item, pp);
 504 }
 505
 506 static int commit_list_compare_by_date(const void *a, const void *b)
 507 {
 508         timestamp_t a_date = ((const struct commit_list *)a)->item->date;
 509         timestamp_t b_date = ((const struct commit_list *)b)->item->date;
 510         if (a_date < b_date)
 511                 return 1;
 512         if (a_date > b_date)
 513                 return -1;
 514         return 0;
 515 }
 516
 517 static void *commit_list_get_next(const void *a)
 518 {
 519         return ((const struct commit_list *)a)->next;
 520 }
 521
 522 static void commit_list_set_next(void *a, void *next)
 523 {
 524         ((struct commit_list *)a)->next = next;
 525 }
 526
 527 void commit_list_sort_by_date(struct commit_list **list)
 528 {
 529         *list = llist_mergesort(*list, commit_list_get_next, commit_list_set_next,
 530                                 commit_list_compare_by_date);
 531 }
 532
 533 struct commit *pop_most_recent_commit(struct commit_list **list,
 534                                       unsigned int mark)
 535 {
 536         struct commit *ret = pop_commit(list);
 537         struct commit_list *parents = ret->parents;
 538
 539         while (parents) {
 540                 struct commit *commit = parents->item;
 541                 if (!parse_commit(commit) && !(commit->object.flags & mark)) {
 542                         commit->object.flags |= mark;
 543                         commit_list_insert_by_date(commit, list);
 544                 }
 545                 parents = parents->next;
 546         }
 547         return ret;
 548 }
 549
 550 static void clear_commit_marks_1(struct commit_list **plist,
 551                                  struct commit *commit, unsigned int mark)
 552 {
 553         while (commit) {
 554                 struct commit_list *parents;
 555
 556                 if (!(mark & commit->object.flags))
 557                         return;
 558
 559                 commit->object.flags &= ~mark;
 560
 561                 parents = commit->parents;
 562                 if (!parents)
 563                         return;
 564
 565                 while ((parents = parents->next))
 566                         commit_list_insert(parents->item, plist);
 567
 568                 commit = commit->parents->item;
 569         }
 570 }
 571
 572 void clear_commit_marks_many(int nr, struct commit **commit, unsigned int mark)
 573 {
 574         struct commit_list *list = NULL;
 575
 576         while (nr--) {
 577                 clear_commit_marks_1(&list, *commit, mark);
 578                 commit++;
 579         }
 580         while (list)
 581                 clear_commit_marks_1(&list, pop_commit(&list), mark);
 582 }
 583
 584 void clear_commit_marks(struct commit *commit, unsigned int mark)
 585 {
 586         clear_commit_marks_many(1, &commit, mark);
 587 }
 588
 589 struct commit *pop_commit(struct commit_list **stack)
 590 {
 591         struct commit_list *top = *stack;
 592         struct commit *item = top ? top->item : NULL;
 593
 594         if (top) {
 595                 *stack = top->next;
 596                 free(top);
 597         }
 598         return item;
 599 }
 600
 601 /*
 602  * Topological sort support
 603  */
 604
 605 /* count number of children that have not been emitted */
 606 define_commit_slab(indegree_slab, int);
 607
 608 /* record author-date for each commit object */
 609 define_commit_slab(author_date_slab, unsigned long);
 610
 611 static void record_author_date(struct author_date_slab *author_date,
 612                                struct commit *commit)
 613 {
 614         const char *buffer = get_commit_buffer(commit, NULL);
 615         struct ident_split ident;
 616         const char *ident_line;
 617         size_t ident_len;
 618         char *date_end;
 619         timestamp_t date;
 620
 621         ident_line = find_commit_header(buffer, "author", &ident_len);
 622         if (!ident_line)
 623                 goto fail_exit; /* no author line */
 624         if (split_ident_line(&ident, ident_line, ident_len) ||
 625             !ident.date_begin || !ident.date_end)
 626                 goto fail_exit; /* malformed "author" line */
 627
 628         date = parse_timestamp(ident.date_begin, &date_end, 10);
 629         if (date_end != ident.date_end)
 630                 goto fail_exit; /* malformed date */
 631         *(author_date_slab_at(author_date, commit)) = date;
 632
 633 fail_exit:
 634         unuse_commit_buffer(commit, buffer);
 635 }
 636
 637 static int compare_commits_by_author_date(const void *a_, const void *b_,
 638                                           void *cb_data)
 639 {
 640         const struct commit *a = a_, *b = b_;
 641         struct author_date_slab *author_date = cb_data;
 642         timestamp_t a_date = *(author_date_slab_at(author_date, a));
 643         timestamp_t b_date = *(author_date_slab_at(author_date, b));
 644
 645         /* newer commits with larger date first */
 646         if (a_date < b_date)
 647                 return 1;
 648         else if (a_date > b_date)
 649                 return -1;
 650         return 0;
 651 }
 652
 653 int compare_commits_by_commit_date(const void *a_, const void *b_, void *unused)
 654 {
 655         const struct commit *a = a_, *b = b_;
 656         /* newer commits with larger date first */
 657         if (a->date < b->date)
 658                 return 1;
 659         else if (a->date > b->date)
 660                 return -1;
 661         return 0;
 662 }
 663
 664 /*
 665  * Performs an in-place topological sort on the list supplied.
 666  */
 667 void sort_in_topological_order(struct commit_list **list, enum rev_sort_order sort_order)
 668 {
 669         struct commit_list *next, *orig = *list;
 670         struct commit_list **pptr;
 671         struct indegree_slab indegree;
 672         struct prio_queue queue;
 673         struct commit *commit;
 674         struct author_date_slab author_date;
 675
 676         if (!orig)
 677                 return;
 678         *list = NULL;
 679
 680         init_indegree_slab(&indegree);
 681         memset(&queue, '\0', sizeof(queue));
 682
 683         switch (sort_order) {
 684         default: /* REV_SORT_IN_GRAPH_ORDER */
 685                 queue.compare = NULL;
 686                 break;
 687         case REV_SORT_BY_COMMIT_DATE:
 688                 queue.compare = compare_commits_by_commit_date;
 689                 break;
 690         case REV_SORT_BY_AUTHOR_DATE:
 691                 init_author_date_slab(&author_date);
 692                 queue.compare = compare_commits_by_author_date;
 693                 queue.cb_data = &author_date;
 694                 break;
 695         }
 696
 697         /* Mark them and clear the indegree */
 698         for (next = orig; next; next = next->next) {
 699                 struct commit *commit = next->item;
 700                 *(indegree_slab_at(&indegree, commit)) = 1;
 701                 /* also record the author dates, if needed */
 702                 if (sort_order == REV_SORT_BY_AUTHOR_DATE)
 703                         record_author_date(&author_date, commit);
 704         }
 705
 706         /* update the indegree */
 707         for (next = orig; next; next = next->next) {
 708                 struct commit_list *parents = next->item->parents;
 709                 while (parents) {
 710                         struct commit *parent = parents->item;
 711                         int *pi = indegree_slab_at(&indegree, parent);
 712
 713                         if (*pi)
 714                                 (*pi)++;
 715                         parents = parents->next;
 716                 }
 717         }
 718
 719         /*
 720          * find the tips
 721          *
 722          * tips are nodes not reachable from any other node in the list
 723          *
 724          * the tips serve as a starting set for the work queue.
 725          */
 726         for (next = orig; next; next = next->next) {
 727                 struct commit *commit = next->item;
 728
 729                 if (*(indegree_slab_at(&indegree, commit)) == 1)
 730                         prio_queue_put(&queue, commit);
 731         }
 732
 733         /*
 734          * This is unfortunate; the initial tips need to be shown
 735          * in the order given from the revision traversal machinery.
 736          */
 737         if (sort_order == REV_SORT_IN_GRAPH_ORDER)
 738                 prio_queue_reverse(&queue);
 739
 740         /* We no longer need the commit list */
 741         free_commit_list(orig);
 742
 743         pptr = list;
 744         *list = NULL;
 745         while ((commit = prio_queue_get(&queue)) != NULL) {
 746                 struct commit_list *parents;
 747
 748                 for (parents = commit->parents; parents ; parents = parents->next) {
 749                         struct commit *parent = parents->item;
 750                         int *pi = indegree_slab_at(&indegree, parent);
 751
 752                         if (!*pi)
 753                                 continue;
 754
 755                         /*
 756                          * parents are only enqueued for emission
 757                          * when all their children have been emitted thereby
 758                          * guaranteeing topological order.
 759                          */
 760                         if (--(*pi) == 1)
 761                                 prio_queue_put(&queue, parent);
 762                 }
 763                 /*
 764                  * all children of commit have already been
 765                  * emitted. we can emit it now.
 766                  */
 767                 *(indegree_slab_at(&indegree, commit)) = 0;
 768
 769                 pptr = &commit_list_insert(commit, pptr)->next;
 770         }
 771
 772         clear_indegree_slab(&indegree);
 773         clear_prio_queue(&queue);
 774         if (sort_order == REV_SORT_BY_AUTHOR_DATE)
 775                 clear_author_date_slab(&author_date);
 776 }
 777
 778 /* merge-base stuff */
 779
 780 /* Remember to update object flag allocation in object.h */
 781 #define PARENT1         (1u<<16)
 782 #define PARENT2         (1u<<17)
 783 #define STALE           (1u<<18)
 784 #define RESULT          (1u<<19)
 785
 786 static const unsigned all_flags = (PARENT1 | PARENT2 | STALE | RESULT);
 787
 788 static int queue_has_nonstale(struct prio_queue *queue)
 789 {
 790         int i;
 791         for (i = 0; i < queue->nr; i++) {
 792                 struct commit *commit = queue->array[i].data;
 793                 if (!(commit->object.flags & STALE))
 794                         return 1;
 795         }
 796         return 0;
 797 }
 798
 799 /* all input commits in one and twos[] must have been parsed! */
 800 static struct commit_list *paint_down_to_common(struct commit *one, int n, struct commit **twos)
 801 {
 802         struct prio_queue queue = { compare_commits_by_commit_date };
 803         struct commit_list *result = NULL;
 804         int i;
 805
 806         one->object.flags |= PARENT1;
 807         if (!n) {
 808                 commit_list_append(one, &result);
 809                 return result;
 810         }
 811         prio_queue_put(&queue, one);
 812
 813         for (i = 0; i < n; i++) {
 814                 twos[i]->object.flags |= PARENT2;
 815                 prio_queue_put(&queue, twos[i]);
 816         }
 817
 818         while (queue_has_nonstale(&queue)) {
 819                 struct commit *commit = prio_queue_get(&queue);
 820                 struct commit_list *parents;
 821                 int flags;
 822
 823                 flags = commit->object.flags & (PARENT1 | PARENT2 | STALE);
 824                 if (flags == (PARENT1 | PARENT2)) {
 825                         if (!(commit->object.flags & RESULT)) {
 826                                 commit->object.flags |= RESULT;
 827                                 commit_list_insert_by_date(commit, &result);
 828                         }
 829                         /* Mark parents of a found merge stale */
 830                         flags |= STALE;
 831                 }
 832                 parents = commit->parents;
 833                 while (parents) {
 834                         struct commit *p = parents->item;
 835                         parents = parents->next;
 836                         if ((p->object.flags & flags) == flags)
 837                                 continue;
 838                         if (parse_commit(p))
 839                                 return NULL;
 840                         p->object.flags |= flags;
 841                         prio_queue_put(&queue, p);
 842                 }
 843         }
 844
 845         clear_prio_queue(&queue);
 846         return result;
 847 }
 848
 849 static struct commit_list *merge_bases_many(struct commit *one, int n, struct commit **twos)
 850 {
 851         struct commit_list *list = NULL;
 852         struct commit_list *result = NULL;
 853         int i;
 854
 855         for (i = 0; i < n; i++) {
 856                 if (one == twos[i])
 857                         /*
 858                          * We do not mark this even with RESULT so we do not
 859                          * have to clean it up.
 860                          */
 861                         return commit_list_insert(one, &result);
 862         }
 863
 864         if (parse_commit(one))
 865                 return NULL;
 866         for (i = 0; i < n; i++) {
 867                 if (parse_commit(twos[i]))
 868                         return NULL;
 869         }
 870
 871         list = paint_down_to_common(one, n, twos);
 872
 873         while (list) {
 874                 struct commit *commit = pop_commit(&list);
 875                 if (!(commit->object.flags & STALE))
 876                         commit_list_insert_by_date(commit, &result);
 877         }
 878         return result;
 879 }
 880
 881 struct commit_list *get_octopus_merge_bases(struct commit_list *in)
 882 {
 883         struct commit_list *i, *j, *k, *ret = NULL;
 884
 885         if (!in)
 886                 return ret;
 887
 888         commit_list_insert(in->item, &ret);
 889
 890         for (i = in->next; i; i = i->next) {
 891                 struct commit_list *new_commits = NULL, *end = NULL;
 892
 893                 for (j = ret; j; j = j->next) {
 894                         struct commit_list *bases;
 895                         bases = get_merge_bases(i->item, j->item);
 896                         if (!new_commits)
 897                                 new_commits = bases;
 898                         else
 899                                 end->next = bases;
 900                         for (k = bases; k; k = k->next)
 901                                 end = k;
 902                 }
 903                 ret = new_commits;
 904         }
 905         return ret;
 906 }
 907
 908 static int remove_redundant(struct commit **array, int cnt)
 909 {
 910         /*
 911          * Some commit in the array may be an ancestor of
 912          * another commit.  Move such commit to the end of
 913          * the array, and return the number of commits that
 914          * are independent from each other.
 915          */
 916         struct commit **work;
 917         unsigned char *redundant;
 918         int *filled_index;
 919         int i, j, filled;
 920
 921         work = xcalloc(cnt, sizeof(*work));
 922         redundant = xcalloc(cnt, 1);
 923         ALLOC_ARRAY(filled_index, cnt - 1);
 924
 925         for (i = 0; i < cnt; i++)
 926                 parse_commit(array[i]);
 927         for (i = 0; i < cnt; i++) {
 928                 struct commit_list *common;
 929
 930                 if (redundant[i])
 931                         continue;
 932                 for (j = filled = 0; j < cnt; j++) {
 933                         if (i == j || redundant[j])
 934                                 continue;
 935                         filled_index[filled] = j;
 936                         work[filled++] = array[j];
 937                 }
 938                 common = paint_down_to_common(array[i], filled, work);
 939                 if (array[i]->object.flags & PARENT2)
 940                         redundant[i] = 1;
 941                 for (j = 0; j < filled; j++)
 942                         if (work[j]->object.flags & PARENT1)
 943                                 redundant[filled_index[j]] = 1;
 944                 clear_commit_marks(array[i], all_flags);
 945                 clear_commit_marks_many(filled, work, all_flags);
 946                 free_commit_list(common);
 947         }
 948
 949         /* Now collect the result */
 950         COPY_ARRAY(work, array, cnt);
 951         for (i = filled = 0; i < cnt; i++)
 952                 if (!redundant[i])
 953                         array[filled++] = work[i];
 954         for (j = filled, i = 0; i < cnt; i++)
 955                 if (redundant[i])
 956                         array[j++] = work[i];
 957         free(work);
 958         free(redundant);
 959         free(filled_index);
 960         return filled;
 961 }
 962
 963 static struct commit_list *get_merge_bases_many_0(struct commit *one,
 964                                                   int n,
 965                                                   struct commit **twos,
 966                                                   int cleanup)
 967 {
 968         struct commit_list *list;
 969         struct commit **rslt;
 970         struct commit_list *result;
 971         int cnt, i;
 972
 973         result = merge_bases_many(one, n, twos);
 974         for (i = 0; i < n; i++) {
 975                 if (one == twos[i])
 976                         return result;
 977         }
 978         if (!result || !result->next) {
 979                 if (cleanup) {
 980                         clear_commit_marks(one, all_flags);
 981                         clear_commit_marks_many(n, twos, all_flags);
 982                 }
 983                 return result;
 984         }
 985
 986         /* There are more than one */
 987         cnt = commit_list_count(result);
 988         rslt = xcalloc(cnt, sizeof(*rslt));
 989         for (list = result, i = 0; list; list = list->next)
 990                 rslt[i++] = list->item;
 991         free_commit_list(result);
 992
 993         clear_commit_marks(one, all_flags);
 994         clear_commit_marks_many(n, twos, all_flags);
 995
 996         cnt = remove_redundant(rslt, cnt);
 997         result = NULL;
 998         for (i = 0; i < cnt; i++)
 999                 commit_list_insert_by_date(rslt[i], &result);
1000         free(rslt);
1001         return result;
1002 }
1003
1004 struct commit_list *get_merge_bases_many(struct commit *one,
1005                                          int n,
1006                                          struct commit **twos)
1007 {
1008         return get_merge_bases_many_0(one, n, twos, 1);
1009 }
1010
1011 struct commit_list *get_merge_bases_many_dirty(struct commit *one,
1012                                                int n,
1013                                                struct commit **twos)
1014 {
1015         return get_merge_bases_many_0(one, n, twos, 0);
1016 }
1017
1018 struct commit_list *get_merge_bases(struct commit *one, struct commit *two)
1019 {
1020         return get_merge_bases_many_0(one, 1, &two, 1);
1021 }
1022
1023 /*
1024  * Is "commit" a descendant of one of the elements on the "with_commit" list?
1025  */
1026 int is_descendant_of(struct commit *commit, struct commit_list *with_commit)
1027 {
1028         if (!with_commit)
1029                 return 1;
1030         while (with_commit) {
1031                 struct commit *other;
1032
1033                 other = with_commit->item;
1034                 with_commit = with_commit->next;
1035                 if (in_merge_bases(other, commit))
1036                         return 1;
1037         }
1038         return 0;
1039 }
1040
1041 /*
1042  * Is "commit" an ancestor of one of the "references"?
1043  */
1044 int in_merge_bases_many(struct commit *commit, int nr_reference, struct commit **reference)
1045 {
1046         struct commit_list *bases;
1047         int ret = 0, i;
1048
1049         if (parse_commit(commit))
1050                 return ret;
1051         for (i = 0; i < nr_reference; i++)
1052                 if (parse_commit(reference[i]))
1053                         return ret;
1054
1055         bases = paint_down_to_common(commit, nr_reference, reference);
1056         if (commit->object.flags & PARENT2)
1057                 ret = 1;
1058         clear_commit_marks(commit, all_flags);
1059         clear_commit_marks_many(nr_reference, reference, all_flags);
1060         free_commit_list(bases);
1061         return ret;
1062 }
1063
1064 /*
1065  * Is "commit" an ancestor of (i.e. reachable from) the "reference"?
1066  */
1067 int in_merge_bases(struct commit *commit, struct commit *reference)
1068 {
1069         return in_merge_bases_many(commit, 1, &reference);
1070 }
1071
1072 struct commit_list *reduce_heads(struct commit_list *heads)
1073 {
1074         struct commit_list *p;
1075         struct commit_list *result = NULL, **tail = &result;
1076         struct commit **array;
1077         int num_head, i;
1078
1079         if (!heads)
1080                 return NULL;
1081
1082         /* Uniquify */
1083         for (p = heads; p; p = p->next)
1084                 p->item->object.flags &= ~STALE;
1085         for (p = heads, num_head = 0; p; p = p->next) {
1086                 if (p->item->object.flags & STALE)
1087                         continue;
1088                 p->item->object.flags |= STALE;
1089                 num_head++;
1090         }
1091         array = xcalloc(num_head, sizeof(*array));
1092         for (p = heads, i = 0; p; p = p->next) {
1093                 if (p->item->object.flags & STALE) {
1094                         array[i++] = p->item;
1095                         p->item->object.flags &= ~STALE;
1096                 }
1097         }
1098         num_head = remove_redundant(array, num_head);
1099         for (i = 0; i < num_head; i++)
1100                 tail = &commit_list_insert(array[i], tail)->next;
1101         free(array);
1102         return result;
1103 }
1104
1105 void reduce_heads_replace(struct commit_list **heads)
1106 {
1107         struct commit_list *result = reduce_heads(*heads);
1108         free_commit_list(*heads);
1109         *heads = result;
1110 }
1111
1112 static const char gpg_sig_header[] = "gpgsig";
1113 static const int gpg_sig_header_len = sizeof(gpg_sig_header) - 1;
1114
1115 static int do_sign_commit(struct strbuf *buf, const char *keyid)
1116 {
1117         struct strbuf sig = STRBUF_INIT;
1118         int inspos, copypos;
1119         const char *eoh;
1120
1121         /* find the end of the header */
1122         eoh = strstr(buf->buf, "\n\n");
1123         if (!eoh)
1124                 inspos = buf->len;
1125         else
1126                 inspos = eoh - buf->buf + 1;
1127
1128         if (!keyid || !*keyid)
1129                 keyid = get_signing_key();
1130         if (sign_buffer(buf, &sig, keyid)) {
1131                 strbuf_release(&sig);
1132                 return -1;
1133         }
1134
1135         for (copypos = 0; sig.buf[copypos]; ) {
1136                 const char *bol = sig.buf + copypos;
1137                 const char *eol = strchrnul(bol, '\n');
1138                 int len = (eol - bol) + !!*eol;
1139
1140                 if (!copypos) {
1141                         strbuf_insert(buf, inspos, gpg_sig_header, gpg_sig_header_len);
1142                         inspos += gpg_sig_header_len;
1143                 }
1144                 strbuf_insert(buf, inspos++, " ", 1);
1145                 strbuf_insert(buf, inspos, bol, len);
1146                 inspos += len;
1147                 copypos += len;
1148         }
1149         strbuf_release(&sig);
1150         return 0;
1151 }
1152
1153 int parse_signed_commit(const struct commit *commit,
1154                         struct strbuf *payload, struct strbuf *signature)
1155 {
1156
1157         unsigned long size;
1158         const char *buffer = get_commit_buffer(commit, &size);
1159         int in_signature, saw_signature = -1;
1160         const char *line, *tail;
1161
1162         line = buffer;
1163         tail = buffer + size;
1164         in_signature = 0;
1165         saw_signature = 0;
1166         while (line < tail) {
1167                 const char *sig = NULL;
1168                 const char *next = memchr(line, '\n', tail - line);
1169
1170                 next = next ? next + 1 : tail;
1171                 if (in_signature && line[0] == ' ')
1172                         sig = line + 1;
1173                 else if (starts_with(line, gpg_sig_header) &&
1174                          line[gpg_sig_header_len] == ' ')
1175                         sig = line + gpg_sig_header_len + 1;
1176                 if (sig) {
1177                         strbuf_add(signature, sig, next - sig);
1178                         saw_signature = 1;
1179                         in_signature = 1;
1180                 } else {
1181                         if (*line == '\n')
1182                                 /* dump the whole remainder of the buffer */
1183                                 next = tail;
1184                         strbuf_add(payload, line, next - line);
1185                         in_signature = 0;
1186                 }
1187                 line = next;
1188         }
1189         unuse_commit_buffer(commit, buffer);
1190         return saw_signature;
1191 }
1192
1193 int remove_signature(struct strbuf *buf)
1194 {
1195         const char *line = buf->buf;
1196         const char *tail = buf->buf + buf->len;
1197         int in_signature = 0;
1198         const char *sig_start = NULL;
1199         const char *sig_end = NULL;
1200
1201         while (line < tail) {
1202                 const char *next = memchr(line, '\n', tail - line);
1203                 next = next ? next + 1 : tail;
1204
1205                 if (in_signature && line[0] == ' ')
1206                         sig_end = next;
1207                 else if (starts_with(line, gpg_sig_header) &&
1208                          line[gpg_sig_header_len] == ' ') {
1209                         sig_start = line;
1210                         sig_end = next;
1211                         in_signature = 1;
1212                 } else {
1213                         if (*line == '\n')
1214                                 /* dump the whole remainder of the buffer */
1215                                 next = tail;
1216                         in_signature = 0;
1217                 }
1218                 line = next;
1219         }
1220
1221         if (sig_start)
1222                 strbuf_remove(buf, sig_start - buf->buf, sig_end - sig_start);
1223
1224         return sig_start != NULL;
1225 }
1226
1227 static void handle_signed_tag(struct commit *parent, struct commit_extra_header ***tail)
1228 {
1229         struct merge_remote_desc *desc;
1230         struct commit_extra_header *mergetag;
1231         char *buf;
1232         unsigned long size, len;
1233         enum object_type type;
1234
1235         desc = merge_remote_util(parent);
1236         if (!desc || !desc->obj)
1237                 return;
1238         buf = read_object_file(&desc->obj->oid, &type, &size);
1239         if (!buf || type != OBJ_TAG)
1240                 goto free_return;
1241         len = parse_signature(buf, size);
1242         if (size == len)
1243                 goto free_return;
1244         /*
1245          * We could verify this signature and either omit the tag when
1246          * it does not validate, but the integrator may not have the
1247          * public key of the signer of the tag he is merging, while a
1248          * later auditor may have it while auditing, so let's not run
1249          * verify-signed-buffer here for now...
1250          *
1251          * if (verify_signed_buffer(buf, len, buf + len, size - len, ...))
1252          *      warn("warning: signed tag unverified.");
1253          */
1254         mergetag = xcalloc(1, sizeof(*mergetag));
1255         mergetag->key = xstrdup("mergetag");
1256         mergetag->value = buf;
1257         mergetag->len = size;
1258
1259         **tail = mergetag;
1260         *tail = &mergetag->next;
1261         return;
1262
1263 free_return:
1264         free(buf);
1265 }
1266
1267 int check_commit_signature(const struct commit *commit, struct signature_check *sigc)
1268 {
1269         struct strbuf payload = STRBUF_INIT;
1270         struct strbuf signature = STRBUF_INIT;
1271         int ret = 1;
1272
1273         sigc->result = 'N';
1274
1275         if (parse_signed_commit(commit, &payload, &signature) <= 0)
1276                 goto out;
1277         ret = check_signature(payload.buf, payload.len, signature.buf,
1278                 signature.len, sigc);
1279
1280  out:
1281         strbuf_release(&payload);
1282         strbuf_release(&signature);
1283
1284         return ret;
1285 }
1286
1287
1288
1289 void append_merge_tag_headers(struct commit_list *parents,
1290                               struct commit_extra_header ***tail)
1291 {
1292         while (parents) {
1293                 struct commit *parent = parents->item;
1294                 handle_signed_tag(parent, tail);
1295                 parents = parents->next;
1296         }
1297 }
1298
1299 static void add_extra_header(struct strbuf *buffer,
1300                              struct commit_extra_header *extra)
1301 {
1302         strbuf_addstr(buffer, extra->key);
1303         if (extra->len)
1304                 strbuf_add_lines(buffer, " ", extra->value, extra->len);
1305         else
1306                 strbuf_addch(buffer, '\n');
1307 }
1308
1309 struct commit_extra_header *read_commit_extra_headers(struct commit *commit,
1310                                                       const char **exclude)
1311 {
1312         struct commit_extra_header *extra = NULL;
1313         unsigned long size;
1314         const char *buffer = get_commit_buffer(commit, &size);
1315         extra = read_commit_extra_header_lines(buffer, size, exclude);
1316         unuse_commit_buffer(commit, buffer);
1317         return extra;
1318 }
1319
1320 int for_each_mergetag(each_mergetag_fn fn, struct commit *commit, void *data)
1321 {
1322         struct commit_extra_header *extra, *to_free;
1323         int res = 0;
1324
1325         to_free = read_commit_extra_headers(commit, NULL);
1326         for (extra = to_free; !res && extra; extra = extra->next) {
1327                 if (strcmp(extra->key, "mergetag"))
1328                         continue; /* not a merge tag */
1329                 res = fn(commit, extra, data);
1330         }
1331         free_commit_extra_headers(to_free);
1332         return res;
1333 }
1334
1335 static inline int standard_header_field(const char *field, size_t len)
1336 {
1337         return ((len == 4 && !memcmp(field, "tree", 4)) ||
1338                 (len == 6 && !memcmp(field, "parent", 6)) ||
1339                 (len == 6 && !memcmp(field, "author", 6)) ||
1340                 (len == 9 && !memcmp(field, "committer", 9)) ||
1341                 (len == 8 && !memcmp(field, "encoding", 8)));
1342 }
1343
1344 static int excluded_header_field(const char *field, size_t len, const char **exclude)
1345 {
1346         if (!exclude)
1347                 return 0;
1348
1349         while (*exclude) {
1350                 size_t xlen = strlen(*exclude);
1351                 if (len == xlen && !memcmp(field, *exclude, xlen))
1352                         return 1;
1353                 exclude++;
1354         }
1355         return 0;
1356 }
1357
1358 static struct commit_extra_header *read_commit_extra_header_lines(
1359         const char *buffer, size_t size,
1360         const char **exclude)
1361 {
1362         struct commit_extra_header *extra = NULL, **tail = &extra, *it = NULL;
1363         const char *line, *next, *eof, *eob;
1364         struct strbuf buf = STRBUF_INIT;
1365
1366         for (line = buffer, eob = line + size;
1367              line < eob && *line != '\n';
1368              line = next) {
1369                 next = memchr(line, '\n', eob - line);
1370                 next = next ? next + 1 : eob;
1371                 if (*line == ' ') {
1372                         /* continuation */
1373                         if (it)
1374                                 strbuf_add(&buf, line + 1, next - (line + 1));
1375                         continue;
1376                 }
1377                 if (it)
1378                         it->value = strbuf_detach(&buf, &it->len);
1379                 strbuf_reset(&buf);
1380                 it = NULL;
1381
1382                 eof = memchr(line, ' ', next - line);
1383                 if (!eof)
1384                         eof = next;
1385                 else if (standard_header_field(line, eof - line) ||
1386                          excluded_header_field(line, eof - line, exclude))
1387                         continue;
1388
1389                 it = xcalloc(1, sizeof(*it));
1390                 it->key = xmemdupz(line, eof-line);
1391                 *tail = it;
1392                 tail = &it->next;
1393                 if (eof + 1 < next)
1394                         strbuf_add(&buf, eof + 1, next - (eof + 1));
1395         }
1396         if (it)
1397                 it->value = strbuf_detach(&buf, &it->len);
1398         return extra;
1399 }
1400
1401 void free_commit_extra_headers(struct commit_extra_header *extra)
1402 {
1403         while (extra) {
1404                 struct commit_extra_header *next = extra->next;
1405                 free(extra->key);
1406                 free(extra->value);
1407                 free(extra);
1408                 extra = next;
1409         }
1410 }
1411
1412 int commit_tree(const char *msg, size_t msg_len, const struct object_id *tree,
1413                 struct commit_list *parents, struct object_id *ret,
1414                 const char *author, const char *sign_commit)
1415 {
1416         struct commit_extra_header *extra = NULL, **tail = &extra;
1417         int result;
1418
1419         append_merge_tag_headers(parents, &tail);
1420         result = commit_tree_extended(msg, msg_len, tree, parents, ret,
1421                                       author, sign_commit, extra);
1422         free_commit_extra_headers(extra);
1423         return result;
1424 }
1425
1426 static int find_invalid_utf8(const char *buf, int len)
1427 {
1428         int offset = 0;
1429         static const unsigned int max_codepoint[] = {
1430                 0x7f, 0x7ff, 0xffff, 0x10ffff
1431         };
1432
1433         while (len) {
1434                 unsigned char c = *buf++;
1435                 int bytes, bad_offset;
1436                 unsigned int codepoint;
1437                 unsigned int min_val, max_val;
1438
1439                 len--;
1440                 offset++;
1441
1442                 /* Simple US-ASCII? No worries. */
1443                 if (c < 0x80)
1444                         continue;
1445
1446                 bad_offset = offset-1;
1447
1448                 /*
1449                  * Count how many more high bits set: that's how
1450                  * many more bytes this sequence should have.
1451                  */
1452                 bytes = 0;
1453                 while (c & 0x40) {
1454                         c <<= 1;
1455                         bytes++;
1456                 }
1457
1458                 /*
1459                  * Must be between 1 and 3 more bytes.  Longer sequences result in
1460                  * codepoints beyond U+10FFFF, which are guaranteed never to exist.
1461                  */
1462                 if (bytes < 1 || 3 < bytes)
1463                         return bad_offset;
1464
1465                 /* Do we *have* that many bytes? */
1466                 if (len < bytes)
1467                         return bad_offset;
1468
1469                 /*
1470                  * Place the encoded bits at the bottom of the value and compute the
1471                  * valid range.
1472                  */
1473                 codepoint = (c & 0x7f) >> bytes;
1474                 min_val = max_codepoint[bytes-1] + 1;
1475                 max_val = max_codepoint[bytes];
1476
1477                 offset += bytes;
1478                 len -= bytes;
1479
1480                 /* And verify that they are good continuation bytes */
1481                 do {
1482                         codepoint <<= 6;
1483                         codepoint |= *buf & 0x3f;
1484                         if ((*buf++ & 0xc0) != 0x80)
1485                                 return bad_offset;
1486                 } while (--bytes);
1487
1488                 /* Reject codepoints that are out of range for the sequence length. */
1489                 if (codepoint < min_val || codepoint > max_val)
1490                         return bad_offset;
1491                 /* Surrogates are only for UTF-16 and cannot be encoded in UTF-8. */
1492                 if ((codepoint & 0x1ff800) == 0xd800)
1493                         return bad_offset;
1494                 /* U+xxFFFE and U+xxFFFF are guaranteed non-characters. */
1495                 if ((codepoint & 0xfffe) == 0xfffe)
1496                         return bad_offset;
1497                 /* So are anything in the range U+FDD0..U+FDEF. */
1498                 if (codepoint >= 0xfdd0 && codepoint <= 0xfdef)
1499                         return bad_offset;
1500         }
1501         return -1;
1502 }
1503
1504 /*
1505  * This verifies that the buffer is in proper utf8 format.
1506  *
1507  * If it isn't, it assumes any non-utf8 characters are Latin1,
1508  * and does the conversion.
1509  */
1510 static int verify_utf8(struct strbuf *buf)
1511 {
1512         int ok = 1;
1513         long pos = 0;
1514
1515         for (;;) {
1516                 int bad;
1517                 unsigned char c;
1518                 unsigned char replace[2];
1519
1520                 bad = find_invalid_utf8(buf->buf + pos, buf->len - pos);
1521                 if (bad < 0)
1522                         return ok;
1523                 pos += bad;
1524                 ok = 0;
1525                 c = buf->buf[pos];
1526                 strbuf_remove(buf, pos, 1);
1527
1528                 /* We know 'c' must be in the range 128-255 */
1529                 replace[0] = 0xc0 + (c >> 6);
1530                 replace[1] = 0x80 + (c & 0x3f);
1531                 strbuf_insert(buf, pos, replace, 2);
1532                 pos += 2;
1533         }
1534 }
1535
1536 static const char commit_utf8_warn[] =
1537 N_("Warning: commit message did not conform to UTF-8.\n"
1538    "You may want to amend it after fixing the message, or set the config\n"
1539    "variable i18n.commitencoding to the encoding your project uses.\n");
1540
1541 int commit_tree_extended(const char *msg, size_t msg_len,
1542                          const struct object_id *tree,
1543                          struct commit_list *parents, struct object_id *ret,
1544                          const char *author, const char *sign_commit,
1545                          struct commit_extra_header *extra)
1546 {
1547         int result;
1548         int encoding_is_utf8;
1549         struct strbuf buffer;
1550
1551         assert_oid_type(tree, OBJ_TREE);
1552
1553         if (memchr(msg, '\0', msg_len))
1554                 return error("a NUL byte in commit log message not allowed.");
1555
1556         /* Not having i18n.commitencoding is the same as having utf-8 */
1557         encoding_is_utf8 = is_encoding_utf8(git_commit_encoding);
1558
1559         strbuf_init(&buffer, 8192); /* should avoid reallocs for the headers */
1560         strbuf_addf(&buffer, "tree %s\n", oid_to_hex(tree));
1561
1562         /*
1563          * NOTE! This ordering means that the same exact tree merged with a
1564          * different order of parents will be a _different_ changeset even
1565          * if everything else stays the same.
1566          */
1567         while (parents) {
1568                 struct commit *parent = pop_commit(&parents);
1569                 strbuf_addf(&buffer, "parent %s\n",
1570                             oid_to_hex(&parent->object.oid));
1571         }
1572
1573         /* Person/date information */
1574         if (!author)
1575                 author = git_author_info(IDENT_STRICT);
1576         strbuf_addf(&buffer, "author %s\n", author);
1577         strbuf_addf(&buffer, "committer %s\n", git_committer_info(IDENT_STRICT));
1578         if (!encoding_is_utf8)
1579                 strbuf_addf(&buffer, "encoding %s\n", git_commit_encoding);
1580
1581         while (extra) {
1582                 add_extra_header(&buffer, extra);
1583                 extra = extra->next;
1584         }
1585         strbuf_addch(&buffer, '\n');
1586
1587         /* And add the comment */
1588         strbuf_add(&buffer, msg, msg_len);
1589
1590         /* And check the encoding */
1591         if (encoding_is_utf8 && !verify_utf8(&buffer))
1592                 fprintf(stderr, _(commit_utf8_warn));
1593
1594         if (sign_commit && do_sign_commit(&buffer, sign_commit)) {
1595                 result = -1;
1596                 goto out;
1597         }
1598
1599         result = write_object_file(buffer.buf, buffer.len, commit_type, ret);
1600 out:
1601         strbuf_release(&buffer);
1602         return result;
1603 }
1604
1605 void set_merge_remote_desc(struct commit *commit,
1606                            const char *name, struct object *obj)
1607 {
1608         struct merge_remote_desc *desc;
1609         FLEX_ALLOC_STR(desc, name, name);
1610         desc->obj = obj;
1611         commit->util = desc;
1612 }
1613
1614 struct commit *get_merge_parent(const char *name)
1615 {
1616         struct object *obj;
1617         struct commit *commit;
1618         struct object_id oid;
1619         if (get_oid(name, &oid))
1620                 return NULL;
1621         obj = parse_object(&oid);
1622         commit = (struct commit *)peel_to_type(name, 0, obj, OBJ_COMMIT);
1623         if (commit && !commit->util)
1624                 set_merge_remote_desc(commit, name, obj);
1625         return commit;
1626 }
1627
1628 /*
1629  * Append a commit to the end of the commit_list.
1630  *
1631  * next starts by pointing to the variable that holds the head of an
1632  * empty commit_list, and is updated to point to the "next" field of
1633  * the last item on the list as new commits are appended.
1634  *
1635  * Usage example:
1636  *
1637  *     struct commit_list *list;
1638  *     struct commit_list **next = &list;
1639  *
1640  *     next = commit_list_append(c1, next);
1641  *     next = commit_list_append(c2, next);
1642  *     assert(commit_list_count(list) == 2);
1643  *     return list;
1644  */
1645 struct commit_list **commit_list_append(struct commit *commit,
1646                                         struct commit_list **next)
1647 {
1648         struct commit_list *new_commit = xmalloc(sizeof(struct commit_list));
1649         new_commit->item = commit;
1650         *next = new_commit;
1651         new_commit->next = NULL;
1652         return &new_commit->next;
1653 }
1654
1655 const char *find_commit_header(const char *msg, const char *key, size_t *out_len)
1656 {
1657         int key_len = strlen(key);
1658         const char *line = msg;
1659
1660         while (line) {
1661                 const char *eol = strchrnul(line, '\n');
1662
1663                 if (line == eol)
1664                         return NULL;
1665
1666                 if (eol - line > key_len &&
1667                     !strncmp(line, key, key_len) &&
1668                     line[key_len] == ' ') {
1669                         *out_len = eol - line - key_len - 1;
1670                         return line + key_len + 1;
1671                 }
1672                 line = *eol ? eol + 1 : NULL;
1673         }
1674         return NULL;
1675 }
1676
1677 /*
1678  * Inspect the given string and determine the true "end" of the log message, in
1679  * order to find where to put a new Signed-off-by: line.  Ignored are
1680  * trailing comment lines and blank lines.  To support "git commit -s
1681  * --amend" on an existing commit, we also ignore "Conflicts:".  To
1682  * support "git commit -v", we truncate at cut lines.
1683  *
1684  * Returns the number of bytes from the tail to ignore, to be fed as
1685  * the second parameter to append_signoff().
1686  */
1687 int ignore_non_trailer(const char *buf, size_t len)
1688 {
1689         int boc = 0;
1690         int bol = 0;
1691         int in_old_conflicts_block = 0;
1692         size_t cutoff = wt_status_locate_end(buf, len);
1693
1694         while (bol < cutoff) {
1695                 const char *next_line = memchr(buf + bol, '\n', len - bol);
1696
1697                 if (!next_line)
1698                         next_line = buf + len;
1699                 else
1700                         next_line++;
1701
1702                 if (buf[bol] == comment_line_char || buf[bol] == '\n') {
1703                         /* is this the first of the run of comments? */
1704                         if (!boc)
1705                                 boc = bol;
1706                         /* otherwise, it is just continuing */
1707                 } else if (starts_with(buf + bol, "Conflicts:\n")) {
1708                         in_old_conflicts_block = 1;
1709                         if (!boc)
1710                                 boc = bol;
1711                 } else if (in_old_conflicts_block && buf[bol] == '\t') {
1712                         ; /* a pathname in the conflicts block */
1713                 } else if (boc) {
1714                         /* the previous was not trailing comment */
1715                         boc = 0;
1716                         in_old_conflicts_block = 0;
1717                 }
1718                 bol = next_line - buf;
1719         }
1720         return boc ? len - boc : len - cutoff;
1721 }