commit.c

   1 #include "cache.h"
   2 #include "tag.h"
   3 #include "commit.h"
   4 #include "pkt-line.h"
   5 #include "utf8.h"
   6 #include "diff.h"
   7 #include "revision.h"
   8 #include "notes.h"
   9 #include "gpg-interface.h"
  10 #include "mergesort.h"
  11 #include "commit-slab.h"
  12 #include "prio-queue.h"
  13 #include "sha1-lookup.h"
  14
  15 static struct commit_extra_header *read_commit_extra_header_lines(const char *buf, size_t len, const char **);
  16
  17 int save_commit_buffer = 1;
  18
  19 const char *commit_type = "commit";
  20
  21 static struct commit *check_commit(struct object *obj,
  22                                    const unsigned char *sha1,
  23                                    int quiet)
  24 {
  25         if (obj->type != OBJ_COMMIT) {
  26                 if (!quiet)
  27                         error("Object %s is a %s, not a commit",
  28                               sha1_to_hex(sha1), typename(obj->type));
  29                 return NULL;
  30         }
  31         return (struct commit *) obj;
  32 }
  33
  34 struct commit *lookup_commit_reference_gently(const unsigned char *sha1,
  35                                               int quiet)
  36 {
  37         struct object *obj = deref_tag(parse_object(sha1), NULL, 0);
  38
  39         if (!obj)
  40                 return NULL;
  41         return check_commit(obj, sha1, quiet);
  42 }
  43
  44 struct commit *lookup_commit_reference(const unsigned char *sha1)
  45 {
  46         return lookup_commit_reference_gently(sha1, 0);
  47 }
  48
  49 struct commit *lookup_commit_or_die(const unsigned char *sha1, const char *ref_name)
  50 {
  51         struct commit *c = lookup_commit_reference(sha1);
  52         if (!c)
  53                 die(_("could not parse %s"), ref_name);
  54         if (hashcmp(sha1, c->object.sha1)) {
  55                 warning(_("%s %s is not a commit!"),
  56                         ref_name, sha1_to_hex(sha1));
  57         }
  58         return c;
  59 }
  60
  61 struct commit *lookup_commit(const unsigned char *sha1)
  62 {
  63         struct object *obj = lookup_object(sha1);
  64         if (!obj) {
  65                 struct commit *c = alloc_commit_node();
  66                 return create_object(sha1, OBJ_COMMIT, c);
  67         }
  68         if (!obj->type)
  69                 obj->type = OBJ_COMMIT;
  70         return check_commit(obj, sha1, 0);
  71 }
  72
  73 struct commit *lookup_commit_reference_by_name(const char *name)
  74 {
  75         unsigned char sha1[20];
  76         struct commit *commit;
  77
  78         if (get_sha1_committish(name, sha1))
  79                 return NULL;
  80         commit = lookup_commit_reference(sha1);
  81         if (parse_commit(commit))
  82                 return NULL;
  83         return commit;
  84 }
  85
  86 static unsigned long parse_commit_date(const char *buf, const char *tail)
  87 {
  88         const char *dateptr;
  89
  90         if (buf + 6 >= tail)
  91                 return 0;
  92         if (memcmp(buf, "author", 6))
  93                 return 0;
  94         while (buf < tail && *buf++ != '\n')
  95                 /* nada */;
  96         if (buf + 9 >= tail)
  97                 return 0;
  98         if (memcmp(buf, "committer", 9))
  99                 return 0;
 100         while (buf < tail && *buf++ != '>')
 101                 /* nada */;
 102         if (buf >= tail)
 103                 return 0;
 104         dateptr = buf;
 105         while (buf < tail && *buf++ != '\n')
 106                 /* nada */;
 107         if (buf >= tail)
 108                 return 0;
 109         /* dateptr < buf && buf[-1] == '\n', so strtoul will stop at buf-1 */
 110         return strtoul(dateptr, NULL, 10);
 111 }
 112
 113 static struct commit_graft **commit_graft;
 114 static int commit_graft_alloc, commit_graft_nr;
 115
 116 static const unsigned char *commit_graft_sha1_access(size_t index, void *table)
 117 {
 118         struct commit_graft **commit_graft_table = table;
 119         return commit_graft_table[index]->sha1;
 120 }
 121
 122 static int commit_graft_pos(const unsigned char *sha1)
 123 {
 124         return sha1_pos(sha1, commit_graft, commit_graft_nr,
 125                         commit_graft_sha1_access);
 126 }
 127
 128 int register_commit_graft(struct commit_graft *graft, int ignore_dups)
 129 {
 130         int pos = commit_graft_pos(graft->sha1);
 131
 132         if (0 <= pos) {
 133                 if (ignore_dups)
 134                         free(graft);
 135                 else {
 136                         free(commit_graft[pos]);
 137                         commit_graft[pos] = graft;
 138                 }
 139                 return 1;
 140         }
 141         pos = -pos - 1;
 142         ALLOC_GROW(commit_graft, commit_graft_nr + 1, commit_graft_alloc);
 143         commit_graft_nr++;
 144         if (pos < commit_graft_nr)
 145                 memmove(commit_graft + pos + 1,
 146                         commit_graft + pos,
 147                         (commit_graft_nr - pos - 1) *
 148                         sizeof(*commit_graft));
 149         commit_graft[pos] = graft;
 150         return 0;
 151 }
 152
 153 struct commit_graft *read_graft_line(char *buf, int len)
 154 {
 155         /* The format is just "Commit Parent1 Parent2 ...\n" */
 156         int i;
 157         struct commit_graft *graft = NULL;
 158
 159         while (len && isspace(buf[len-1]))
 160                 buf[--len] = '\0';
 161         if (buf[0] == '#' || buf[0] == '\0')
 162                 return NULL;
 163         if ((len + 1) % 41)
 164                 goto bad_graft_data;
 165         i = (len + 1) / 41 - 1;
 166         graft = xmalloc(sizeof(*graft) + 20 * i);
 167         graft->nr_parent = i;
 168         if (get_sha1_hex(buf, graft->sha1))
 169                 goto bad_graft_data;
 170         for (i = 40; i < len; i += 41) {
 171                 if (buf[i] != ' ')
 172                         goto bad_graft_data;
 173                 if (get_sha1_hex(buf + i + 1, graft->parent[i/41]))
 174                         goto bad_graft_data;
 175         }
 176         return graft;
 177
 178 bad_graft_data:
 179         error("bad graft data: %s", buf);
 180         free(graft);
 181         return NULL;
 182 }
 183
 184 static int read_graft_file(const char *graft_file)
 185 {
 186         FILE *fp = fopen(graft_file, "r");
 187         struct strbuf buf = STRBUF_INIT;
 188         if (!fp)
 189                 return -1;
 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.buf, buf.len);
 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 unsigned char *sha1)
 218 {
 219         int pos;
 220         prepare_commit_graft();
 221         pos = commit_graft_pos(sha1);
 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 unsigned char *sha1)
 236 {
 237         int pos = commit_graft_pos(sha1);
 238         if (pos < 0)
 239                 return -1;
 240         if (pos + 1 < commit_graft_nr)
 241                 memmove(commit_graft + pos, commit_graft + pos + 1,
 242                                 sizeof(struct commit_graft *)
 243                                 * (commit_graft_nr - pos - 1));
 244         commit_graft_nr--;
 245         return 0;
 246 }
 247
 248 void set_commit_buffer(struct commit *commit, void *buffer)
 249 {
 250         commit->buffer = buffer;
 251 }
 252
 253 const void *get_cached_commit_buffer(const struct commit *commit)
 254 {
 255         return commit->buffer;
 256 }
 257
 258 const void *get_commit_buffer(const struct commit *commit)
 259 {
 260         const void *ret = get_cached_commit_buffer(commit);
 261         if (!ret) {
 262                 enum object_type type;
 263                 unsigned long size;
 264                 ret = read_sha1_file(commit->object.sha1, &type, &size);
 265                 if (!ret)
 266                         die("cannot read commit object %s",
 267                             sha1_to_hex(commit->object.sha1));
 268                 if (type != OBJ_COMMIT)
 269                         die("expected commit for %s, got %s",
 270                             sha1_to_hex(commit->object.sha1), typename(type));
 271         }
 272         return ret;
 273 }
 274
 275 void unuse_commit_buffer(const struct commit *commit, const void *buffer)
 276 {
 277         if (commit->buffer != buffer)
 278                 free((void *)buffer);
 279 }
 280
 281 void free_commit_buffer(struct commit *commit)
 282 {
 283         free(commit->buffer);
 284         commit->buffer = NULL;
 285 }
 286
 287 const void *detach_commit_buffer(struct commit *commit)
 288 {
 289         void *ret = commit->buffer;
 290         commit->buffer = NULL;
 291         return ret;
 292 }
 293
 294 int parse_commit_buffer(struct commit *item, const void *buffer, unsigned long size)
 295 {
 296         const char *tail = buffer;
 297         const char *bufptr = buffer;
 298         unsigned char parent[20];
 299         struct commit_list **pptr;
 300         struct commit_graft *graft;
 301
 302         if (item->object.parsed)
 303                 return 0;
 304         item->object.parsed = 1;
 305         tail += size;
 306         if (tail <= bufptr + 46 || memcmp(bufptr, "tree ", 5) || bufptr[45] != '\n')
 307                 return error("bogus commit object %s", sha1_to_hex(item->object.sha1));
 308         if (get_sha1_hex(bufptr + 5, parent) < 0)
 309                 return error("bad tree pointer in commit %s",
 310                              sha1_to_hex(item->object.sha1));
 311         item->tree = lookup_tree(parent);
 312         bufptr += 46; /* "tree " + "hex sha1" + "\n" */
 313         pptr = &item->parents;
 314
 315         graft = lookup_commit_graft(item->object.sha1);
 316         while (bufptr + 48 < tail && !memcmp(bufptr, "parent ", 7)) {
 317                 struct commit *new_parent;
 318
 319                 if (tail <= bufptr + 48 ||
 320                     get_sha1_hex(bufptr + 7, parent) ||
 321                     bufptr[47] != '\n')
 322                         return error("bad parents in commit %s", sha1_to_hex(item->object.sha1));
 323                 bufptr += 48;
 324                 /*
 325                  * The clone is shallow if nr_parent < 0, and we must
 326                  * not traverse its real parents even when we unhide them.
 327                  */
 328                 if (graft && (graft->nr_parent < 0 || grafts_replace_parents))
 329                         continue;
 330                 new_parent = lookup_commit(parent);
 331                 if (new_parent)
 332                         pptr = &commit_list_insert(new_parent, pptr)->next;
 333         }
 334         if (graft) {
 335                 int i;
 336                 struct commit *new_parent;
 337                 for (i = 0; i < graft->nr_parent; i++) {
 338                         new_parent = lookup_commit(graft->parent[i]);
 339                         if (!new_parent)
 340                                 continue;
 341                         pptr = &commit_list_insert(new_parent, pptr)->next;
 342                 }
 343         }
 344         item->date = parse_commit_date(bufptr, tail);
 345
 346         return 0;
 347 }
 348
 349 int parse_commit(struct commit *item)
 350 {
 351         enum object_type type;
 352         void *buffer;
 353         unsigned long size;
 354         int ret;
 355
 356         if (!item)
 357                 return -1;
 358         if (item->object.parsed)
 359                 return 0;
 360         buffer = read_sha1_file(item->object.sha1, &type, &size);
 361         if (!buffer)
 362                 return error("Could not read %s",
 363                              sha1_to_hex(item->object.sha1));
 364         if (type != OBJ_COMMIT) {
 365                 free(buffer);
 366                 return error("Object %s not a commit",
 367                              sha1_to_hex(item->object.sha1));
 368         }
 369         ret = parse_commit_buffer(item, buffer, size);
 370         if (save_commit_buffer && !ret) {
 371                 set_commit_buffer(item, buffer);
 372                 return 0;
 373         }
 374         free(buffer);
 375         return ret;
 376 }
 377
 378 void parse_commit_or_die(struct commit *item)
 379 {
 380         if (parse_commit(item))
 381                 die("unable to parse commit %s",
 382                     item ? sha1_to_hex(item->object.sha1) : "(null)");
 383 }
 384
 385 int find_commit_subject(const char *commit_buffer, const char **subject)
 386 {
 387         const char *eol;
 388         const char *p = commit_buffer;
 389
 390         while (*p && (*p != '\n' || p[1] != '\n'))
 391                 p++;
 392         if (*p) {
 393                 p += 2;
 394                 for (eol = p; *eol && *eol != '\n'; eol++)
 395                         ; /* do nothing */
 396         } else
 397                 eol = p;
 398
 399         *subject = p;
 400
 401         return eol - p;
 402 }
 403
 404 struct commit_list *commit_list_insert(struct commit *item, struct commit_list **list_p)
 405 {
 406         struct commit_list *new_list = xmalloc(sizeof(struct commit_list));
 407         new_list->item = item;
 408         new_list->next = *list_p;
 409         *list_p = new_list;
 410         return new_list;
 411 }
 412
 413 unsigned commit_list_count(const struct commit_list *l)
 414 {
 415         unsigned c = 0;
 416         for (; l; l = l->next )
 417                 c++;
 418         return c;
 419 }
 420
 421 struct commit_list *copy_commit_list(struct commit_list *list)
 422 {
 423         struct commit_list *head = NULL;
 424         struct commit_list **pp = &head;
 425         while (list) {
 426                 struct commit_list *new;
 427                 new = xmalloc(sizeof(struct commit_list));
 428                 new->item = list->item;
 429                 new->next = NULL;
 430                 *pp = new;
 431                 pp = &new->next;
 432                 list = list->next;
 433         }
 434         return head;
 435 }
 436
 437 void free_commit_list(struct commit_list *list)
 438 {
 439         while (list) {
 440                 struct commit_list *temp = list;
 441                 list = temp->next;
 442                 free(temp);
 443         }
 444 }
 445
 446 struct commit_list * commit_list_insert_by_date(struct commit *item, struct commit_list **list)
 447 {
 448         struct commit_list **pp = list;
 449         struct commit_list *p;
 450         while ((p = *pp) != NULL) {
 451                 if (p->item->date < item->date) {
 452                         break;
 453                 }
 454                 pp = &p->next;
 455         }
 456         return commit_list_insert(item, pp);
 457 }
 458
 459 static int commit_list_compare_by_date(const void *a, const void *b)
 460 {
 461         unsigned long a_date = ((const struct commit_list *)a)->item->date;
 462         unsigned long b_date = ((const struct commit_list *)b)->item->date;
 463         if (a_date < b_date)
 464                 return 1;
 465         if (a_date > b_date)
 466                 return -1;
 467         return 0;
 468 }
 469
 470 static void *commit_list_get_next(const void *a)
 471 {
 472         return ((const struct commit_list *)a)->next;
 473 }
 474
 475 static void commit_list_set_next(void *a, void *next)
 476 {
 477         ((struct commit_list *)a)->next = next;
 478 }
 479
 480 void commit_list_sort_by_date(struct commit_list **list)
 481 {
 482         *list = llist_mergesort(*list, commit_list_get_next, commit_list_set_next,
 483                                 commit_list_compare_by_date);
 484 }
 485
 486 struct commit *pop_most_recent_commit(struct commit_list **list,
 487                                       unsigned int mark)
 488 {
 489         struct commit *ret = (*list)->item;
 490         struct commit_list *parents = ret->parents;
 491         struct commit_list *old = *list;
 492
 493         *list = (*list)->next;
 494         free(old);
 495
 496         while (parents) {
 497                 struct commit *commit = parents->item;
 498                 if (!parse_commit(commit) && !(commit->object.flags & mark)) {
 499                         commit->object.flags |= mark;
 500                         commit_list_insert_by_date(commit, list);
 501                 }
 502                 parents = parents->next;
 503         }
 504         return ret;
 505 }
 506
 507 static void clear_commit_marks_1(struct commit_list **plist,
 508                                  struct commit *commit, unsigned int mark)
 509 {
 510         while (commit) {
 511                 struct commit_list *parents;
 512
 513                 if (!(mark & commit->object.flags))
 514                         return;
 515
 516                 commit->object.flags &= ~mark;
 517
 518                 parents = commit->parents;
 519                 if (!parents)
 520                         return;
 521
 522                 while ((parents = parents->next))
 523                         commit_list_insert(parents->item, plist);
 524
 525                 commit = commit->parents->item;
 526         }
 527 }
 528
 529 void clear_commit_marks_many(int nr, struct commit **commit, unsigned int mark)
 530 {
 531         struct commit_list *list = NULL;
 532
 533         while (nr--) {
 534                 commit_list_insert(*commit, &list);
 535                 commit++;
 536         }
 537         while (list)
 538                 clear_commit_marks_1(&list, pop_commit(&list), mark);
 539 }
 540
 541 void clear_commit_marks(struct commit *commit, unsigned int mark)
 542 {
 543         clear_commit_marks_many(1, &commit, mark);
 544 }
 545
 546 void clear_commit_marks_for_object_array(struct object_array *a, unsigned mark)
 547 {
 548         struct object *object;
 549         struct commit *commit;
 550         unsigned int i;
 551
 552         for (i = 0; i < a->nr; i++) {
 553                 object = a->objects[i].item;
 554                 commit = lookup_commit_reference_gently(object->sha1, 1);
 555                 if (commit)
 556                         clear_commit_marks(commit, mark);
 557         }
 558 }
 559
 560 struct commit *pop_commit(struct commit_list **stack)
 561 {
 562         struct commit_list *top = *stack;
 563         struct commit *item = top ? top->item : NULL;
 564
 565         if (top) {
 566                 *stack = top->next;
 567                 free(top);
 568         }
 569         return item;
 570 }
 571
 572 /*
 573  * Topological sort support
 574  */
 575
 576 /* count number of children that have not been emitted */
 577 define_commit_slab(indegree_slab, int);
 578
 579 /* record author-date for each commit object */
 580 define_commit_slab(author_date_slab, unsigned long);
 581
 582 static void record_author_date(struct author_date_slab *author_date,
 583                                struct commit *commit)
 584 {
 585         const char *buf, *line_end, *ident_line;
 586         char *buffer = NULL;
 587         struct ident_split ident;
 588         char *date_end;
 589         unsigned long date;
 590
 591         if (!commit->buffer) {
 592                 unsigned long size;
 593                 enum object_type type;
 594                 buffer = read_sha1_file(commit->object.sha1, &type, &size);
 595                 if (!buffer)
 596                         return;
 597         }
 598
 599         for (buf = commit->buffer ? commit->buffer : buffer;
 600              buf;
 601              buf = line_end + 1) {
 602                 line_end = strchrnul(buf, '\n');
 603                 ident_line = skip_prefix(buf, "author ");
 604                 if (!ident_line) {
 605                         if (!line_end[0] || line_end[1] == '\n')
 606                                 return; /* end of header */
 607                         continue;
 608                 }
 609                 if (split_ident_line(&ident,
 610                                      ident_line, line_end - ident_line) ||
 611                     !ident.date_begin || !ident.date_end)
 612                         goto fail_exit; /* malformed "author" line */
 613                 break;
 614         }
 615
 616         date = strtoul(ident.date_begin, &date_end, 10);
 617         if (date_end != ident.date_end)
 618                 goto fail_exit; /* malformed date */
 619         *(author_date_slab_at(author_date, commit)) = date;
 620
 621 fail_exit:
 622         free(buffer);
 623 }
 624
 625 static int compare_commits_by_author_date(const void *a_, const void *b_,
 626                                           void *cb_data)
 627 {
 628         const struct commit *a = a_, *b = b_;
 629         struct author_date_slab *author_date = cb_data;
 630         unsigned long a_date = *(author_date_slab_at(author_date, a));
 631         unsigned long b_date = *(author_date_slab_at(author_date, b));
 632
 633         /* newer commits with larger date first */
 634         if (a_date < b_date)
 635                 return 1;
 636         else if (a_date > b_date)
 637                 return -1;
 638         return 0;
 639 }
 640
 641 int compare_commits_by_commit_date(const void *a_, const void *b_, void *unused)
 642 {
 643         const struct commit *a = a_, *b = b_;
 644         /* newer commits with larger date first */
 645         if (a->date < b->date)
 646                 return 1;
 647         else if (a->date > b->date)
 648                 return -1;
 649         return 0;
 650 }
 651
 652 /*
 653  * Performs an in-place topological sort on the list supplied.
 654  */
 655 void sort_in_topological_order(struct commit_list **list, enum rev_sort_order sort_order)
 656 {
 657         struct commit_list *next, *orig = *list;
 658         struct commit_list **pptr;
 659         struct indegree_slab indegree;
 660         struct prio_queue queue;
 661         struct commit *commit;
 662         struct author_date_slab author_date;
 663
 664         if (!orig)
 665                 return;
 666         *list = NULL;
 667
 668         init_indegree_slab(&indegree);
 669         memset(&queue, '\0', sizeof(queue));
 670
 671         switch (sort_order) {
 672         default: /* REV_SORT_IN_GRAPH_ORDER */
 673                 queue.compare = NULL;
 674                 break;
 675         case REV_SORT_BY_COMMIT_DATE:
 676                 queue.compare = compare_commits_by_commit_date;
 677                 break;
 678         case REV_SORT_BY_AUTHOR_DATE:
 679                 init_author_date_slab(&author_date);
 680                 queue.compare = compare_commits_by_author_date;
 681                 queue.cb_data = &author_date;
 682                 break;
 683         }
 684
 685         /* Mark them and clear the indegree */
 686         for (next = orig; next; next = next->next) {
 687                 struct commit *commit = next->item;
 688                 *(indegree_slab_at(&indegree, commit)) = 1;
 689                 /* also record the author dates, if needed */
 690                 if (sort_order == REV_SORT_BY_AUTHOR_DATE)
 691                         record_author_date(&author_date, commit);
 692         }
 693
 694         /* update the indegree */
 695         for (next = orig; next; next = next->next) {
 696                 struct commit_list *parents = next->item->parents;
 697                 while (parents) {
 698                         struct commit *parent = parents->item;
 699                         int *pi = indegree_slab_at(&indegree, parent);
 700
 701                         if (*pi)
 702                                 (*pi)++;
 703                         parents = parents->next;
 704                 }
 705         }
 706
 707         /*
 708          * find the tips
 709          *
 710          * tips are nodes not reachable from any other node in the list
 711          *
 712          * the tips serve as a starting set for the work queue.
 713          */
 714         for (next = orig; next; next = next->next) {
 715                 struct commit *commit = next->item;
 716
 717                 if (*(indegree_slab_at(&indegree, commit)) == 1)
 718                         prio_queue_put(&queue, commit);
 719         }
 720
 721         /*
 722          * This is unfortunate; the initial tips need to be shown
 723          * in the order given from the revision traversal machinery.
 724          */
 725         if (sort_order == REV_SORT_IN_GRAPH_ORDER)
 726                 prio_queue_reverse(&queue);
 727
 728         /* We no longer need the commit list */
 729         free_commit_list(orig);
 730
 731         pptr = list;
 732         *list = NULL;
 733         while ((commit = prio_queue_get(&queue)) != NULL) {
 734                 struct commit_list *parents;
 735
 736                 for (parents = commit->parents; parents ; parents = parents->next) {
 737                         struct commit *parent = parents->item;
 738                         int *pi = indegree_slab_at(&indegree, parent);
 739
 740                         if (!*pi)
 741                                 continue;
 742
 743                         /*
 744                          * parents are only enqueued for emission
 745                          * when all their children have been emitted thereby
 746                          * guaranteeing topological order.
 747                          */
 748                         if (--(*pi) == 1)
 749                                 prio_queue_put(&queue, parent);
 750                 }
 751                 /*
 752                  * all children of commit have already been
 753                  * emitted. we can emit it now.
 754                  */
 755                 *(indegree_slab_at(&indegree, commit)) = 0;
 756
 757                 pptr = &commit_list_insert(commit, pptr)->next;
 758         }
 759
 760         clear_indegree_slab(&indegree);
 761         clear_prio_queue(&queue);
 762         if (sort_order == REV_SORT_BY_AUTHOR_DATE)
 763                 clear_author_date_slab(&author_date);
 764 }
 765
 766 /* merge-base stuff */
 767
 768 /* Remember to update object flag allocation in object.h */
 769 #define PARENT1         (1u<<16)
 770 #define PARENT2         (1u<<17)
 771 #define STALE           (1u<<18)
 772 #define RESULT          (1u<<19)
 773
 774 static const unsigned all_flags = (PARENT1 | PARENT2 | STALE | RESULT);
 775
 776 static struct commit *interesting(struct commit_list *list)
 777 {
 778         while (list) {
 779                 struct commit *commit = list->item;
 780                 list = list->next;
 781                 if (commit->object.flags & STALE)
 782                         continue;
 783                 return commit;
 784         }
 785         return NULL;
 786 }
 787
 788 /* all input commits in one and twos[] must have been parsed! */
 789 static struct commit_list *paint_down_to_common(struct commit *one, int n, struct commit **twos)
 790 {
 791         struct commit_list *list = NULL;
 792         struct commit_list *result = NULL;
 793         int i;
 794
 795         one->object.flags |= PARENT1;
 796         commit_list_insert_by_date(one, &list);
 797         if (!n)
 798                 return list;
 799         for (i = 0; i < n; i++) {
 800                 twos[i]->object.flags |= PARENT2;
 801                 commit_list_insert_by_date(twos[i], &list);
 802         }
 803
 804         while (interesting(list)) {
 805                 struct commit *commit;
 806                 struct commit_list *parents;
 807                 struct commit_list *next;
 808                 int flags;
 809
 810                 commit = list->item;
 811                 next = list->next;
 812                 free(list);
 813                 list = next;
 814
 815                 flags = commit->object.flags & (PARENT1 | PARENT2 | STALE);
 816                 if (flags == (PARENT1 | PARENT2)) {
 817                         if (!(commit->object.flags & RESULT)) {
 818                                 commit->object.flags |= RESULT;
 819                                 commit_list_insert_by_date(commit, &result);
 820                         }
 821                         /* Mark parents of a found merge stale */
 822                         flags |= STALE;
 823                 }
 824                 parents = commit->parents;
 825                 while (parents) {
 826                         struct commit *p = parents->item;
 827                         parents = parents->next;
 828                         if ((p->object.flags & flags) == flags)
 829                                 continue;
 830                         if (parse_commit(p))
 831                                 return NULL;
 832                         p->object.flags |= flags;
 833                         commit_list_insert_by_date(p, &list);
 834                 }
 835         }
 836
 837         free_commit_list(list);
 838         return result;
 839 }
 840
 841 static struct commit_list *merge_bases_many(struct commit *one, int n, struct commit **twos)
 842 {
 843         struct commit_list *list = NULL;
 844         struct commit_list *result = NULL;
 845         int i;
 846
 847         for (i = 0; i < n; i++) {
 848                 if (one == twos[i])
 849                         /*
 850                          * We do not mark this even with RESULT so we do not
 851                          * have to clean it up.
 852                          */
 853                         return commit_list_insert(one, &result);
 854         }
 855
 856         if (parse_commit(one))
 857                 return NULL;
 858         for (i = 0; i < n; i++) {
 859                 if (parse_commit(twos[i]))
 860                         return NULL;
 861         }
 862
 863         list = paint_down_to_common(one, n, twos);
 864
 865         while (list) {
 866                 struct commit_list *next = list->next;
 867                 if (!(list->item->object.flags & STALE))
 868                         commit_list_insert_by_date(list->item, &result);
 869                 free(list);
 870                 list = next;
 871         }
 872         return result;
 873 }
 874
 875 struct commit_list *get_octopus_merge_bases(struct commit_list *in)
 876 {
 877         struct commit_list *i, *j, *k, *ret = NULL;
 878
 879         if (!in)
 880                 return ret;
 881
 882         commit_list_insert(in->item, &ret);
 883
 884         for (i = in->next; i; i = i->next) {
 885                 struct commit_list *new = NULL, *end = NULL;
 886
 887                 for (j = ret; j; j = j->next) {
 888                         struct commit_list *bases;
 889                         bases = get_merge_bases(i->item, j->item, 1);
 890                         if (!new)
 891                                 new = bases;
 892                         else
 893                                 end->next = bases;
 894                         for (k = bases; k; k = k->next)
 895                                 end = k;
 896                 }
 897                 ret = new;
 898         }
 899         return ret;
 900 }
 901
 902 static int remove_redundant(struct commit **array, int cnt)
 903 {
 904         /*
 905          * Some commit in the array may be an ancestor of
 906          * another commit.  Move such commit to the end of
 907          * the array, and return the number of commits that
 908          * are independent from each other.
 909          */
 910         struct commit **work;
 911         unsigned char *redundant;
 912         int *filled_index;
 913         int i, j, filled;
 914
 915         work = xcalloc(cnt, sizeof(*work));
 916         redundant = xcalloc(cnt, 1);
 917         filled_index = xmalloc(sizeof(*filled_index) * (cnt - 1));
 918
 919         for (i = 0; i < cnt; i++)
 920                 parse_commit(array[i]);
 921         for (i = 0; i < cnt; i++) {
 922                 struct commit_list *common;
 923
 924                 if (redundant[i])
 925                         continue;
 926                 for (j = filled = 0; j < cnt; j++) {
 927                         if (i == j || redundant[j])
 928                                 continue;
 929                         filled_index[filled] = j;
 930                         work[filled++] = array[j];
 931                 }
 932                 common = paint_down_to_common(array[i], filled, work);
 933                 if (array[i]->object.flags & PARENT2)
 934                         redundant[i] = 1;
 935                 for (j = 0; j < filled; j++)
 936                         if (work[j]->object.flags & PARENT1)
 937                                 redundant[filled_index[j]] = 1;
 938                 clear_commit_marks(array[i], all_flags);
 939                 for (j = 0; j < filled; j++)
 940                         clear_commit_marks(work[j], all_flags);
 941                 free_commit_list(common);
 942         }
 943
 944         /* Now collect the result */
 945         memcpy(work, array, sizeof(*array) * cnt);
 946         for (i = filled = 0; i < cnt; i++)
 947                 if (!redundant[i])
 948                         array[filled++] = work[i];
 949         for (j = filled, i = 0; i < cnt; i++)
 950                 if (redundant[i])
 951                         array[j++] = work[i];
 952         free(work);
 953         free(redundant);
 954         free(filled_index);
 955         return filled;
 956 }
 957
 958 struct commit_list *get_merge_bases_many(struct commit *one,
 959                                          int n,
 960                                          struct commit **twos,
 961                                          int cleanup)
 962 {
 963         struct commit_list *list;
 964         struct commit **rslt;
 965         struct commit_list *result;
 966         int cnt, i;
 967
 968         result = merge_bases_many(one, n, twos);
 969         for (i = 0; i < n; i++) {
 970                 if (one == twos[i])
 971                         return result;
 972         }
 973         if (!result || !result->next) {
 974                 if (cleanup) {
 975                         clear_commit_marks(one, all_flags);
 976                         clear_commit_marks_many(n, twos, all_flags);
 977                 }
 978                 return result;
 979         }
 980
 981         /* There are more than one */
 982         cnt = 0;
 983         list = result;
 984         while (list) {
 985                 list = list->next;
 986                 cnt++;
 987         }
 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(struct commit *one, struct commit *two,
1005                                     int cleanup)
1006 {
1007         return get_merge_bases_many(one, 1, &two, cleanup);
1008 }
1009
1010 /*
1011  * Is "commit" a descendant of one of the elements on the "with_commit" list?
1012  */
1013 int is_descendant_of(struct commit *commit, struct commit_list *with_commit)
1014 {
1015         if (!with_commit)
1016                 return 1;
1017         while (with_commit) {
1018                 struct commit *other;
1019
1020                 other = with_commit->item;
1021                 with_commit = with_commit->next;
1022                 if (in_merge_bases(other, commit))
1023                         return 1;
1024         }
1025         return 0;
1026 }
1027
1028 /*
1029  * Is "commit" an ancestor of one of the "references"?
1030  */
1031 int in_merge_bases_many(struct commit *commit, int nr_reference, struct commit **reference)
1032 {
1033         struct commit_list *bases;
1034         int ret = 0, i;
1035
1036         if (parse_commit(commit))
1037                 return ret;
1038         for (i = 0; i < nr_reference; i++)
1039                 if (parse_commit(reference[i]))
1040                         return ret;
1041
1042         bases = paint_down_to_common(commit, nr_reference, reference);
1043         if (commit->object.flags & PARENT2)
1044                 ret = 1;
1045         clear_commit_marks(commit, all_flags);
1046         clear_commit_marks_many(nr_reference, reference, all_flags);
1047         free_commit_list(bases);
1048         return ret;
1049 }
1050
1051 /*
1052  * Is "commit" an ancestor of (i.e. reachable from) the "reference"?
1053  */
1054 int in_merge_bases(struct commit *commit, struct commit *reference)
1055 {
1056         return in_merge_bases_many(commit, 1, &reference);
1057 }
1058
1059 struct commit_list *reduce_heads(struct commit_list *heads)
1060 {
1061         struct commit_list *p;
1062         struct commit_list *result = NULL, **tail = &result;
1063         struct commit **array;
1064         int num_head, i;
1065
1066         if (!heads)
1067                 return NULL;
1068
1069         /* Uniquify */
1070         for (p = heads; p; p = p->next)
1071                 p->item->object.flags &= ~STALE;
1072         for (p = heads, num_head = 0; p; p = p->next) {
1073                 if (p->item->object.flags & STALE)
1074                         continue;
1075                 p->item->object.flags |= STALE;
1076                 num_head++;
1077         }
1078         array = xcalloc(sizeof(*array), num_head);
1079         for (p = heads, i = 0; p; p = p->next) {
1080                 if (p->item->object.flags & STALE) {
1081                         array[i++] = p->item;
1082                         p->item->object.flags &= ~STALE;
1083                 }
1084         }
1085         num_head = remove_redundant(array, num_head);
1086         for (i = 0; i < num_head; i++)
1087                 tail = &commit_list_insert(array[i], tail)->next;
1088         return result;
1089 }
1090
1091 static const char gpg_sig_header[] = "gpgsig";
1092 static const int gpg_sig_header_len = sizeof(gpg_sig_header) - 1;
1093
1094 static int do_sign_commit(struct strbuf *buf, const char *keyid)
1095 {
1096         struct strbuf sig = STRBUF_INIT;
1097         int inspos, copypos;
1098
1099         /* find the end of the header */
1100         inspos = strstr(buf->buf, "\n\n") - buf->buf + 1;
1101
1102         if (!keyid || !*keyid)
1103                 keyid = get_signing_key();
1104         if (sign_buffer(buf, &sig, keyid)) {
1105                 strbuf_release(&sig);
1106                 return -1;
1107         }
1108
1109         for (copypos = 0; sig.buf[copypos]; ) {
1110                 const char *bol = sig.buf + copypos;
1111                 const char *eol = strchrnul(bol, '\n');
1112                 int len = (eol - bol) + !!*eol;
1113
1114                 if (!copypos) {
1115                         strbuf_insert(buf, inspos, gpg_sig_header, gpg_sig_header_len);
1116                         inspos += gpg_sig_header_len;
1117                 }
1118                 strbuf_insert(buf, inspos++, " ", 1);
1119                 strbuf_insert(buf, inspos, bol, len);
1120                 inspos += len;
1121                 copypos += len;
1122         }
1123         strbuf_release(&sig);
1124         return 0;
1125 }
1126
1127 int parse_signed_commit(const unsigned char *sha1,
1128                         struct strbuf *payload, struct strbuf *signature)
1129 {
1130         unsigned long size;
1131         enum object_type type;
1132         char *buffer = read_sha1_file(sha1, &type, &size);
1133         int in_signature, saw_signature = -1;
1134         char *line, *tail;
1135
1136         if (!buffer || type != OBJ_COMMIT)
1137                 goto cleanup;
1138
1139         line = buffer;
1140         tail = buffer + size;
1141         in_signature = 0;
1142         saw_signature = 0;
1143         while (line < tail) {
1144                 const char *sig = NULL;
1145                 char *next = memchr(line, '\n', tail - line);
1146
1147                 next = next ? next + 1 : tail;
1148                 if (in_signature && line[0] == ' ')
1149                         sig = line + 1;
1150                 else if (starts_with(line, gpg_sig_header) &&
1151                          line[gpg_sig_header_len] == ' ')
1152                         sig = line + gpg_sig_header_len + 1;
1153                 if (sig) {
1154                         strbuf_add(signature, sig, next - sig);
1155                         saw_signature = 1;
1156                         in_signature = 1;
1157                 } else {
1158                         if (*line == '\n')
1159                                 /* dump the whole remainder of the buffer */
1160                                 next = tail;
1161                         strbuf_add(payload, line, next - line);
1162                         in_signature = 0;
1163                 }
1164                 line = next;
1165         }
1166  cleanup:
1167         free(buffer);
1168         return saw_signature;
1169 }
1170
1171 static void handle_signed_tag(struct commit *parent, struct commit_extra_header ***tail)
1172 {
1173         struct merge_remote_desc *desc;
1174         struct commit_extra_header *mergetag;
1175         char *buf;
1176         unsigned long size, len;
1177         enum object_type type;
1178
1179         desc = merge_remote_util(parent);
1180         if (!desc || !desc->obj)
1181                 return;
1182         buf = read_sha1_file(desc->obj->sha1, &type, &size);
1183         if (!buf || type != OBJ_TAG)
1184                 goto free_return;
1185         len = parse_signature(buf, size);
1186         if (size == len)
1187                 goto free_return;
1188         /*
1189          * We could verify this signature and either omit the tag when
1190          * it does not validate, but the integrator may not have the
1191          * public key of the signer of the tag he is merging, while a
1192          * later auditor may have it while auditing, so let's not run
1193          * verify-signed-buffer here for now...
1194          *
1195          * if (verify_signed_buffer(buf, len, buf + len, size - len, ...))
1196          *      warn("warning: signed tag unverified.");
1197          */
1198         mergetag = xcalloc(1, sizeof(*mergetag));
1199         mergetag->key = xstrdup("mergetag");
1200         mergetag->value = buf;
1201         mergetag->len = size;
1202
1203         **tail = mergetag;
1204         *tail = &mergetag->next;
1205         return;
1206
1207 free_return:
1208         free(buf);
1209 }
1210
1211 static struct {
1212         char result;
1213         const char *check;
1214 } sigcheck_gpg_status[] = {
1215         { 'G', "\n[GNUPG:] GOODSIG " },
1216         { 'B', "\n[GNUPG:] BADSIG " },
1217         { 'U', "\n[GNUPG:] TRUST_NEVER" },
1218         { 'U', "\n[GNUPG:] TRUST_UNDEFINED" },
1219 };
1220
1221 static void parse_gpg_output(struct signature_check *sigc)
1222 {
1223         const char *buf = sigc->gpg_status;
1224         int i;
1225
1226         /* Iterate over all search strings */
1227         for (i = 0; i < ARRAY_SIZE(sigcheck_gpg_status); i++) {
1228                 const char *found, *next;
1229
1230                 found = skip_prefix(buf, sigcheck_gpg_status[i].check + 1);
1231                 if (!found) {
1232                         found = strstr(buf, sigcheck_gpg_status[i].check);
1233                         if (!found)
1234                                 continue;
1235                         found += strlen(sigcheck_gpg_status[i].check);
1236                 }
1237                 sigc->result = sigcheck_gpg_status[i].result;
1238                 /* The trust messages are not followed by key/signer information */
1239                 if (sigc->result != 'U') {
1240                         sigc->key = xmemdupz(found, 16);
1241                         found += 17;
1242                         next = strchrnul(found, '\n');
1243                         sigc->signer = xmemdupz(found, next - found);
1244                 }
1245         }
1246 }
1247
1248 void check_commit_signature(const struct commit* commit, struct signature_check *sigc)
1249 {
1250         struct strbuf payload = STRBUF_INIT;
1251         struct strbuf signature = STRBUF_INIT;
1252         struct strbuf gpg_output = STRBUF_INIT;
1253         struct strbuf gpg_status = STRBUF_INIT;
1254         int status;
1255
1256         sigc->result = 'N';
1257
1258         if (parse_signed_commit(commit->object.sha1,
1259                                 &payload, &signature) <= 0)
1260                 goto out;
1261         status = verify_signed_buffer(payload.buf, payload.len,
1262                                       signature.buf, signature.len,
1263                                       &gpg_output, &gpg_status);
1264         if (status && !gpg_output.len)
1265                 goto out;
1266         sigc->gpg_output = strbuf_detach(&gpg_output, NULL);
1267         sigc->gpg_status = strbuf_detach(&gpg_status, NULL);
1268         parse_gpg_output(sigc);
1269
1270  out:
1271         strbuf_release(&gpg_status);
1272         strbuf_release(&gpg_output);
1273         strbuf_release(&payload);
1274         strbuf_release(&signature);
1275 }
1276
1277
1278
1279 void append_merge_tag_headers(struct commit_list *parents,
1280                               struct commit_extra_header ***tail)
1281 {
1282         while (parents) {
1283                 struct commit *parent = parents->item;
1284                 handle_signed_tag(parent, tail);
1285                 parents = parents->next;
1286         }
1287 }
1288
1289 static void add_extra_header(struct strbuf *buffer,
1290                              struct commit_extra_header *extra)
1291 {
1292         strbuf_addstr(buffer, extra->key);
1293         if (extra->len)
1294                 strbuf_add_lines(buffer, " ", extra->value, extra->len);
1295         else
1296                 strbuf_addch(buffer, '\n');
1297 }
1298
1299 struct commit_extra_header *read_commit_extra_headers(struct commit *commit,
1300                                                       const char **exclude)
1301 {
1302         struct commit_extra_header *extra = NULL;
1303         unsigned long size;
1304         enum object_type type;
1305         char *buffer = read_sha1_file(commit->object.sha1, &type, &size);
1306         if (buffer && type == OBJ_COMMIT)
1307                 extra = read_commit_extra_header_lines(buffer, size, exclude);
1308         free(buffer);
1309         return extra;
1310 }
1311
1312 static inline int standard_header_field(const char *field, size_t len)
1313 {
1314         return ((len == 4 && !memcmp(field, "tree ", 5)) ||
1315                 (len == 6 && !memcmp(field, "parent ", 7)) ||
1316                 (len == 6 && !memcmp(field, "author ", 7)) ||
1317                 (len == 9 && !memcmp(field, "committer ", 10)) ||
1318                 (len == 8 && !memcmp(field, "encoding ", 9)));
1319 }
1320
1321 static int excluded_header_field(const char *field, size_t len, const char **exclude)
1322 {
1323         if (!exclude)
1324                 return 0;
1325
1326         while (*exclude) {
1327                 size_t xlen = strlen(*exclude);
1328                 if (len == xlen &&
1329                     !memcmp(field, *exclude, xlen) && field[xlen] == ' ')
1330                         return 1;
1331                 exclude++;
1332         }
1333         return 0;
1334 }
1335
1336 static struct commit_extra_header *read_commit_extra_header_lines(
1337         const char *buffer, size_t size,
1338         const char **exclude)
1339 {
1340         struct commit_extra_header *extra = NULL, **tail = &extra, *it = NULL;
1341         const char *line, *next, *eof, *eob;
1342         struct strbuf buf = STRBUF_INIT;
1343
1344         for (line = buffer, eob = line + size;
1345              line < eob && *line != '\n';
1346              line = next) {
1347                 next = memchr(line, '\n', eob - line);
1348                 next = next ? next + 1 : eob;
1349                 if (*line == ' ') {
1350                         /* continuation */
1351                         if (it)
1352                                 strbuf_add(&buf, line + 1, next - (line + 1));
1353                         continue;
1354                 }
1355                 if (it)
1356                         it->value = strbuf_detach(&buf, &it->len);
1357                 strbuf_reset(&buf);
1358                 it = NULL;
1359
1360                 eof = strchr(line, ' ');
1361                 if (next <= eof)
1362                         eof = next;
1363
1364                 if (standard_header_field(line, eof - line) ||
1365                     excluded_header_field(line, eof - line, exclude))
1366                         continue;
1367
1368                 it = xcalloc(1, sizeof(*it));
1369                 it->key = xmemdupz(line, eof-line);
1370                 *tail = it;
1371                 tail = &it->next;
1372                 if (eof + 1 < next)
1373                         strbuf_add(&buf, eof + 1, next - (eof + 1));
1374         }
1375         if (it)
1376                 it->value = strbuf_detach(&buf, &it->len);
1377         return extra;
1378 }
1379
1380 void free_commit_extra_headers(struct commit_extra_header *extra)
1381 {
1382         while (extra) {
1383                 struct commit_extra_header *next = extra->next;
1384                 free(extra->key);
1385                 free(extra->value);
1386                 free(extra);
1387                 extra = next;
1388         }
1389 }
1390
1391 int commit_tree(const char *msg, size_t msg_len,
1392                 const unsigned char *tree,
1393                 struct commit_list *parents, unsigned char *ret,
1394                 const char *author, const char *sign_commit)
1395 {
1396         struct commit_extra_header *extra = NULL, **tail = &extra;
1397         int result;
1398
1399         append_merge_tag_headers(parents, &tail);
1400         result = commit_tree_extended(msg, msg_len, tree, parents, ret,
1401                                       author, sign_commit, extra);
1402         free_commit_extra_headers(extra);
1403         return result;
1404 }
1405
1406 static int find_invalid_utf8(const char *buf, int len)
1407 {
1408         int offset = 0;
1409         static const unsigned int max_codepoint[] = {
1410                 0x7f, 0x7ff, 0xffff, 0x10ffff
1411         };
1412
1413         while (len) {
1414                 unsigned char c = *buf++;
1415                 int bytes, bad_offset;
1416                 unsigned int codepoint;
1417                 unsigned int min_val, max_val;
1418
1419                 len--;
1420                 offset++;
1421
1422                 /* Simple US-ASCII? No worries. */
1423                 if (c < 0x80)
1424                         continue;
1425
1426                 bad_offset = offset-1;
1427
1428                 /*
1429                  * Count how many more high bits set: that's how
1430                  * many more bytes this sequence should have.
1431                  */
1432                 bytes = 0;
1433                 while (c & 0x40) {
1434                         c <<= 1;
1435                         bytes++;
1436                 }
1437
1438                 /*
1439                  * Must be between 1 and 3 more bytes.  Longer sequences result in
1440                  * codepoints beyond U+10FFFF, which are guaranteed never to exist.
1441                  */
1442                 if (bytes < 1 || 3 < bytes)
1443                         return bad_offset;
1444
1445                 /* Do we *have* that many bytes? */
1446                 if (len < bytes)
1447                         return bad_offset;
1448
1449                 /*
1450                  * Place the encoded bits at the bottom of the value and compute the
1451                  * valid range.
1452                  */
1453                 codepoint = (c & 0x7f) >> bytes;
1454                 min_val = max_codepoint[bytes-1] + 1;
1455                 max_val = max_codepoint[bytes];
1456
1457                 offset += bytes;
1458                 len -= bytes;
1459
1460                 /* And verify that they are good continuation bytes */
1461                 do {
1462                         codepoint <<= 6;
1463                         codepoint |= *buf & 0x3f;
1464                         if ((*buf++ & 0xc0) != 0x80)
1465                                 return bad_offset;
1466                 } while (--bytes);
1467
1468                 /* Reject codepoints that are out of range for the sequence length. */
1469                 if (codepoint < min_val || codepoint > max_val)
1470                         return bad_offset;
1471                 /* Surrogates are only for UTF-16 and cannot be encoded in UTF-8. */
1472                 if ((codepoint & 0x1ff800) == 0xd800)
1473                         return bad_offset;
1474                 /* U+xxFFFE and U+xxFFFF are guaranteed non-characters. */
1475                 if ((codepoint & 0xfffe) == 0xfffe)
1476                         return bad_offset;
1477                 /* So are anything in the range U+FDD0..U+FDEF. */
1478                 if (codepoint >= 0xfdd0 && codepoint <= 0xfdef)
1479                         return bad_offset;
1480         }
1481         return -1;
1482 }
1483
1484 /*
1485  * This verifies that the buffer is in proper utf8 format.
1486  *
1487  * If it isn't, it assumes any non-utf8 characters are Latin1,
1488  * and does the conversion.
1489  */
1490 static int verify_utf8(struct strbuf *buf)
1491 {
1492         int ok = 1;
1493         long pos = 0;
1494
1495         for (;;) {
1496                 int bad;
1497                 unsigned char c;
1498                 unsigned char replace[2];
1499
1500                 bad = find_invalid_utf8(buf->buf + pos, buf->len - pos);
1501                 if (bad < 0)
1502                         return ok;
1503                 pos += bad;
1504                 ok = 0;
1505                 c = buf->buf[pos];
1506                 strbuf_remove(buf, pos, 1);
1507
1508                 /* We know 'c' must be in the range 128-255 */
1509                 replace[0] = 0xc0 + (c >> 6);
1510                 replace[1] = 0x80 + (c & 0x3f);
1511                 strbuf_insert(buf, pos, replace, 2);
1512                 pos += 2;
1513         }
1514 }
1515
1516 static const char commit_utf8_warn[] =
1517 "Warning: commit message did not conform to UTF-8.\n"
1518 "You may want to amend it after fixing the message, or set the config\n"
1519 "variable i18n.commitencoding to the encoding your project uses.\n";
1520
1521 int commit_tree_extended(const char *msg, size_t msg_len,
1522                          const unsigned char *tree,
1523                          struct commit_list *parents, unsigned char *ret,
1524                          const char *author, const char *sign_commit,
1525                          struct commit_extra_header *extra)
1526 {
1527         int result;
1528         int encoding_is_utf8;
1529         struct strbuf buffer;
1530
1531         assert_sha1_type(tree, OBJ_TREE);
1532
1533         if (memchr(msg, '\0', msg_len))
1534                 return error("a NUL byte in commit log message not allowed.");
1535
1536         /* Not having i18n.commitencoding is the same as having utf-8 */
1537         encoding_is_utf8 = is_encoding_utf8(git_commit_encoding);
1538
1539         strbuf_init(&buffer, 8192); /* should avoid reallocs for the headers */
1540         strbuf_addf(&buffer, "tree %s\n", sha1_to_hex(tree));
1541
1542         /*
1543          * NOTE! This ordering means that the same exact tree merged with a
1544          * different order of parents will be a _different_ changeset even
1545          * if everything else stays the same.
1546          */
1547         while (parents) {
1548                 struct commit_list *next = parents->next;
1549                 struct commit *parent = parents->item;
1550
1551                 strbuf_addf(&buffer, "parent %s\n",
1552                             sha1_to_hex(parent->object.sha1));
1553                 free(parents);
1554                 parents = next;
1555         }
1556
1557         /* Person/date information */
1558         if (!author)
1559                 author = git_author_info(IDENT_STRICT);
1560         strbuf_addf(&buffer, "author %s\n", author);
1561         strbuf_addf(&buffer, "committer %s\n", git_committer_info(IDENT_STRICT));
1562         if (!encoding_is_utf8)
1563                 strbuf_addf(&buffer, "encoding %s\n", git_commit_encoding);
1564
1565         while (extra) {
1566                 add_extra_header(&buffer, extra);
1567                 extra = extra->next;
1568         }
1569         strbuf_addch(&buffer, '\n');
1570
1571         /* And add the comment */
1572         strbuf_add(&buffer, msg, msg_len);
1573
1574         /* And check the encoding */
1575         if (encoding_is_utf8 && !verify_utf8(&buffer))
1576                 fprintf(stderr, commit_utf8_warn);
1577
1578         if (sign_commit && do_sign_commit(&buffer, sign_commit))
1579                 return -1;
1580
1581         result = write_sha1_file(buffer.buf, buffer.len, commit_type, ret);
1582         strbuf_release(&buffer);
1583         return result;
1584 }
1585
1586 struct commit *get_merge_parent(const char *name)
1587 {
1588         struct object *obj;
1589         struct commit *commit;
1590         unsigned char sha1[20];
1591         if (get_sha1(name, sha1))
1592                 return NULL;
1593         obj = parse_object(sha1);
1594         commit = (struct commit *)peel_to_type(name, 0, obj, OBJ_COMMIT);
1595         if (commit && !commit->util) {
1596                 struct merge_remote_desc *desc;
1597                 desc = xmalloc(sizeof(*desc));
1598                 desc->obj = obj;
1599                 desc->name = strdup(name);
1600                 commit->util = desc;
1601         }
1602         return commit;
1603 }
1604
1605 /*
1606  * Append a commit to the end of the commit_list.
1607  *
1608  * next starts by pointing to the variable that holds the head of an
1609  * empty commit_list, and is updated to point to the "next" field of
1610  * the last item on the list as new commits are appended.
1611  *
1612  * Usage example:
1613  *
1614  *     struct commit_list *list;
1615  *     struct commit_list **next = &list;
1616  *
1617  *     next = commit_list_append(c1, next);
1618  *     next = commit_list_append(c2, next);
1619  *     assert(commit_list_count(list) == 2);
1620  *     return list;
1621  */
1622 struct commit_list **commit_list_append(struct commit *commit,
1623                                         struct commit_list **next)
1624 {
1625         struct commit_list *new = xmalloc(sizeof(struct commit_list));
1626         new->item = commit;
1627         *next = new;
1628         new->next = NULL;
1629         return &new->next;
1630 }
1631
1632 void print_commit_list(struct commit_list *list,
1633                        const char *format_cur,
1634                        const char *format_last)
1635 {
1636         for ( ; list; list = list->next) {
1637                 const char *format = list->next ? format_cur : format_last;
1638                 printf(format, sha1_to_hex(list->item->object.sha1));
1639         }
1640 }