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 struct commit_buffer {
 249         void *buffer;
 250         unsigned long size;
 251 };
 252 define_commit_slab(buffer_slab, struct commit_buffer);
 253 static struct buffer_slab buffer_slab = COMMIT_SLAB_INIT(1, buffer_slab);
 254
 255 void set_commit_buffer(struct commit *commit, void *buffer, unsigned long size)
 256 {
 257         struct commit_buffer *v = buffer_slab_at(&buffer_slab, commit);
 258         v->buffer = buffer;
 259         v->size = size;
 260 }
 261
 262 const void *get_cached_commit_buffer(const struct commit *commit, unsigned long *sizep)
 263 {
 264         struct commit_buffer *v = buffer_slab_at(&buffer_slab, commit);
 265         if (sizep)
 266                 *sizep = v->size;
 267         return v->buffer;
 268 }
 269
 270 const void *get_commit_buffer(const struct commit *commit, unsigned long *sizep)
 271 {
 272         const void *ret = get_cached_commit_buffer(commit, sizep);
 273         if (!ret) {
 274                 enum object_type type;
 275                 unsigned long size;
 276                 ret = read_sha1_file(commit->object.sha1, &type, &size);
 277                 if (!ret)
 278                         die("cannot read commit object %s",
 279                             sha1_to_hex(commit->object.sha1));
 280                 if (type != OBJ_COMMIT)
 281                         die("expected commit for %s, got %s",
 282                             sha1_to_hex(commit->object.sha1), typename(type));
 283                 if (sizep)
 284                         *sizep = size;
 285         }
 286         return ret;
 287 }
 288
 289 void unuse_commit_buffer(const struct commit *commit, const void *buffer)
 290 {
 291         struct commit_buffer *v = buffer_slab_at(&buffer_slab, commit);
 292         if (v->buffer != buffer)
 293                 free((void *)buffer);
 294 }
 295
 296 void free_commit_buffer(struct commit *commit)
 297 {
 298         struct commit_buffer *v = buffer_slab_at(&buffer_slab, commit);
 299         free(v->buffer);
 300         v->buffer = NULL;
 301         v->size = 0;
 302 }
 303
 304 const void *detach_commit_buffer(struct commit *commit, unsigned long *sizep)
 305 {
 306         struct commit_buffer *v = buffer_slab_at(&buffer_slab, commit);
 307         void *ret;
 308
 309         ret = v->buffer;
 310         if (sizep)
 311                 *sizep = v->size;
 312
 313         v->buffer = NULL;
 314         v->size = 0;
 315         return ret;
 316 }
 317
 318 int parse_commit_buffer(struct commit *item, const void *buffer, unsigned long size)
 319 {
 320         const char *tail = buffer;
 321         const char *bufptr = buffer;
 322         unsigned char parent[20];
 323         struct commit_list **pptr;
 324         struct commit_graft *graft;
 325
 326         if (item->object.parsed)
 327                 return 0;
 328         item->object.parsed = 1;
 329         tail += size;
 330         if (tail <= bufptr + 46 || memcmp(bufptr, "tree ", 5) || bufptr[45] != '\n')
 331                 return error("bogus commit object %s", sha1_to_hex(item->object.sha1));
 332         if (get_sha1_hex(bufptr + 5, parent) < 0)
 333                 return error("bad tree pointer in commit %s",
 334                              sha1_to_hex(item->object.sha1));
 335         item->tree = lookup_tree(parent);
 336         bufptr += 46; /* "tree " + "hex sha1" + "\n" */
 337         pptr = &item->parents;
 338
 339         graft = lookup_commit_graft(item->object.sha1);
 340         while (bufptr + 48 < tail && !memcmp(bufptr, "parent ", 7)) {
 341                 struct commit *new_parent;
 342
 343                 if (tail <= bufptr + 48 ||
 344                     get_sha1_hex(bufptr + 7, parent) ||
 345                     bufptr[47] != '\n')
 346                         return error("bad parents in commit %s", sha1_to_hex(item->object.sha1));
 347                 bufptr += 48;
 348                 /*
 349                  * The clone is shallow if nr_parent < 0, and we must
 350                  * not traverse its real parents even when we unhide them.
 351                  */
 352                 if (graft && (graft->nr_parent < 0 || grafts_replace_parents))
 353                         continue;
 354                 new_parent = lookup_commit(parent);
 355                 if (new_parent)
 356                         pptr = &commit_list_insert(new_parent, pptr)->next;
 357         }
 358         if (graft) {
 359                 int i;
 360                 struct commit *new_parent;
 361                 for (i = 0; i < graft->nr_parent; i++) {
 362                         new_parent = lookup_commit(graft->parent[i]);
 363                         if (!new_parent)
 364                                 continue;
 365                         pptr = &commit_list_insert(new_parent, pptr)->next;
 366                 }
 367         }
 368         item->date = parse_commit_date(bufptr, tail);
 369
 370         return 0;
 371 }
 372
 373 int parse_commit(struct commit *item)
 374 {
 375         enum object_type type;
 376         void *buffer;
 377         unsigned long size;
 378         int ret;
 379
 380         if (!item)
 381                 return -1;
 382         if (item->object.parsed)
 383                 return 0;
 384         buffer = read_sha1_file(item->object.sha1, &type, &size);
 385         if (!buffer)
 386                 return error("Could not read %s",
 387                              sha1_to_hex(item->object.sha1));
 388         if (type != OBJ_COMMIT) {
 389                 free(buffer);
 390                 return error("Object %s not a commit",
 391                              sha1_to_hex(item->object.sha1));
 392         }
 393         ret = parse_commit_buffer(item, buffer, size);
 394         if (save_commit_buffer && !ret) {
 395                 set_commit_buffer(item, buffer, size);
 396                 return 0;
 397         }
 398         free(buffer);
 399         return ret;
 400 }
 401
 402 void parse_commit_or_die(struct commit *item)
 403 {
 404         if (parse_commit(item))
 405                 die("unable to parse commit %s",
 406                     item ? sha1_to_hex(item->object.sha1) : "(null)");
 407 }
 408
 409 int find_commit_subject(const char *commit_buffer, const char **subject)
 410 {
 411         const char *eol;
 412         const char *p = commit_buffer;
 413
 414         while (*p && (*p != '\n' || p[1] != '\n'))
 415                 p++;
 416         if (*p) {
 417                 p += 2;
 418                 for (eol = p; *eol && *eol != '\n'; eol++)
 419                         ; /* do nothing */
 420         } else
 421                 eol = p;
 422
 423         *subject = p;
 424
 425         return eol - p;
 426 }
 427
 428 struct commit_list *commit_list_insert(struct commit *item, struct commit_list **list_p)
 429 {
 430         struct commit_list *new_list = xmalloc(sizeof(struct commit_list));
 431         new_list->item = item;
 432         new_list->next = *list_p;
 433         *list_p = new_list;
 434         return new_list;
 435 }
 436
 437 unsigned commit_list_count(const struct commit_list *l)
 438 {
 439         unsigned c = 0;
 440         for (; l; l = l->next )
 441                 c++;
 442         return c;
 443 }
 444
 445 struct commit_list *copy_commit_list(struct commit_list *list)
 446 {
 447         struct commit_list *head = NULL;
 448         struct commit_list **pp = &head;
 449         while (list) {
 450                 struct commit_list *new;
 451                 new = xmalloc(sizeof(struct commit_list));
 452                 new->item = list->item;
 453                 new->next = NULL;
 454                 *pp = new;
 455                 pp = &new->next;
 456                 list = list->next;
 457         }
 458         return head;
 459 }
 460
 461 void free_commit_list(struct commit_list *list)
 462 {
 463         while (list) {
 464                 struct commit_list *temp = list;
 465                 list = temp->next;
 466                 free(temp);
 467         }
 468 }
 469
 470 struct commit_list * commit_list_insert_by_date(struct commit *item, struct commit_list **list)
 471 {
 472         struct commit_list **pp = list;
 473         struct commit_list *p;
 474         while ((p = *pp) != NULL) {
 475                 if (p->item->date < item->date) {
 476                         break;
 477                 }
 478                 pp = &p->next;
 479         }
 480         return commit_list_insert(item, pp);
 481 }
 482
 483 static int commit_list_compare_by_date(const void *a, const void *b)
 484 {
 485         unsigned long a_date = ((const struct commit_list *)a)->item->date;
 486         unsigned long b_date = ((const struct commit_list *)b)->item->date;
 487         if (a_date < b_date)
 488                 return 1;
 489         if (a_date > b_date)
 490                 return -1;
 491         return 0;
 492 }
 493
 494 static void *commit_list_get_next(const void *a)
 495 {
 496         return ((const struct commit_list *)a)->next;
 497 }
 498
 499 static void commit_list_set_next(void *a, void *next)
 500 {
 501         ((struct commit_list *)a)->next = next;
 502 }
 503
 504 void commit_list_sort_by_date(struct commit_list **list)
 505 {
 506         *list = llist_mergesort(*list, commit_list_get_next, commit_list_set_next,
 507                                 commit_list_compare_by_date);
 508 }
 509
 510 struct commit *pop_most_recent_commit(struct commit_list **list,
 511                                       unsigned int mark)
 512 {
 513         struct commit *ret = (*list)->item;
 514         struct commit_list *parents = ret->parents;
 515         struct commit_list *old = *list;
 516
 517         *list = (*list)->next;
 518         free(old);
 519
 520         while (parents) {
 521                 struct commit *commit = parents->item;
 522                 if (!parse_commit(commit) && !(commit->object.flags & mark)) {
 523                         commit->object.flags |= mark;
 524                         commit_list_insert_by_date(commit, list);
 525                 }
 526                 parents = parents->next;
 527         }
 528         return ret;
 529 }
 530
 531 static void clear_commit_marks_1(struct commit_list **plist,
 532                                  struct commit *commit, unsigned int mark)
 533 {
 534         while (commit) {
 535                 struct commit_list *parents;
 536
 537                 if (!(mark & commit->object.flags))
 538                         return;
 539
 540                 commit->object.flags &= ~mark;
 541
 542                 parents = commit->parents;
 543                 if (!parents)
 544                         return;
 545
 546                 while ((parents = parents->next))
 547                         commit_list_insert(parents->item, plist);
 548
 549                 commit = commit->parents->item;
 550         }
 551 }
 552
 553 void clear_commit_marks_many(int nr, struct commit **commit, unsigned int mark)
 554 {
 555         struct commit_list *list = NULL;
 556
 557         while (nr--) {
 558                 commit_list_insert(*commit, &list);
 559                 commit++;
 560         }
 561         while (list)
 562                 clear_commit_marks_1(&list, pop_commit(&list), mark);
 563 }
 564
 565 void clear_commit_marks(struct commit *commit, unsigned int mark)
 566 {
 567         clear_commit_marks_many(1, &commit, mark);
 568 }
 569
 570 void clear_commit_marks_for_object_array(struct object_array *a, unsigned mark)
 571 {
 572         struct object *object;
 573         struct commit *commit;
 574         unsigned int i;
 575
 576         for (i = 0; i < a->nr; i++) {
 577                 object = a->objects[i].item;
 578                 commit = lookup_commit_reference_gently(object->sha1, 1);
 579                 if (commit)
 580                         clear_commit_marks(commit, mark);
 581         }
 582 }
 583
 584 struct commit *pop_commit(struct commit_list **stack)
 585 {
 586         struct commit_list *top = *stack;
 587         struct commit *item = top ? top->item : NULL;
 588
 589         if (top) {
 590                 *stack = top->next;
 591                 free(top);
 592         }
 593         return item;
 594 }
 595
 596 /*
 597  * Topological sort support
 598  */
 599
 600 /* count number of children that have not been emitted */
 601 define_commit_slab(indegree_slab, int);
 602
 603 /* record author-date for each commit object */
 604 define_commit_slab(author_date_slab, unsigned long);
 605
 606 static void record_author_date(struct author_date_slab *author_date,
 607                                struct commit *commit)
 608 {
 609         const char *buf, *line_end, *ident_line;
 610         const char *buffer = get_commit_buffer(commit, NULL);
 611         struct ident_split ident;
 612         char *date_end;
 613         unsigned long date;
 614
 615         for (buf = buffer; buf; buf = line_end + 1) {
 616                 line_end = strchrnul(buf, '\n');
 617                 ident_line = skip_prefix(buf, "author ");
 618                 if (!ident_line) {
 619                         if (!line_end[0] || line_end[1] == '\n')
 620                                 return; /* end of header */
 621                         continue;
 622                 }
 623                 if (split_ident_line(&ident,
 624                                      ident_line, line_end - ident_line) ||
 625                     !ident.date_begin || !ident.date_end)
 626                         goto fail_exit; /* malformed "author" line */
 627                 break;
 628         }
 629
 630         date = strtoul(ident.date_begin, &date_end, 10);
 631         if (date_end != ident.date_end)
 632                 goto fail_exit; /* malformed date */
 633         *(author_date_slab_at(author_date, commit)) = date;
 634
 635 fail_exit:
 636         unuse_commit_buffer(commit, buffer);
 637 }
 638
 639 static int compare_commits_by_author_date(const void *a_, const void *b_,
 640                                           void *cb_data)
 641 {
 642         const struct commit *a = a_, *b = b_;
 643         struct author_date_slab *author_date = cb_data;
 644         unsigned long a_date = *(author_date_slab_at(author_date, a));
 645         unsigned long b_date = *(author_date_slab_at(author_date, b));
 646
 647         /* newer commits with larger date first */
 648         if (a_date < b_date)
 649                 return 1;
 650         else if (a_date > b_date)
 651                 return -1;
 652         return 0;
 653 }
 654
 655 int compare_commits_by_commit_date(const void *a_, const void *b_, void *unused)
 656 {
 657         const struct commit *a = a_, *b = b_;
 658         /* newer commits with larger date first */
 659         if (a->date < b->date)
 660                 return 1;
 661         else if (a->date > b->date)
 662                 return -1;
 663         return 0;
 664 }
 665
 666 /*
 667  * Performs an in-place topological sort on the list supplied.
 668  */
 669 void sort_in_topological_order(struct commit_list **list, enum rev_sort_order sort_order)
 670 {
 671         struct commit_list *next, *orig = *list;
 672         struct commit_list **pptr;
 673         struct indegree_slab indegree;
 674         struct prio_queue queue;
 675         struct commit *commit;
 676         struct author_date_slab author_date;
 677
 678         if (!orig)
 679                 return;
 680         *list = NULL;
 681
 682         init_indegree_slab(&indegree);
 683         memset(&queue, '\0', sizeof(queue));
 684
 685         switch (sort_order) {
 686         default: /* REV_SORT_IN_GRAPH_ORDER */
 687                 queue.compare = NULL;
 688                 break;
 689         case REV_SORT_BY_COMMIT_DATE:
 690                 queue.compare = compare_commits_by_commit_date;
 691                 break;
 692         case REV_SORT_BY_AUTHOR_DATE:
 693                 init_author_date_slab(&author_date);
 694                 queue.compare = compare_commits_by_author_date;
 695                 queue.cb_data = &author_date;
 696                 break;
 697         }
 698
 699         /* Mark them and clear the indegree */
 700         for (next = orig; next; next = next->next) {
 701                 struct commit *commit = next->item;
 702                 *(indegree_slab_at(&indegree, commit)) = 1;
 703                 /* also record the author dates, if needed */
 704                 if (sort_order == REV_SORT_BY_AUTHOR_DATE)
 705                         record_author_date(&author_date, commit);
 706         }
 707
 708         /* update the indegree */
 709         for (next = orig; next; next = next->next) {
 710                 struct commit_list *parents = next->item->parents;
 711                 while (parents) {
 712                         struct commit *parent = parents->item;
 713                         int *pi = indegree_slab_at(&indegree, parent);
 714
 715                         if (*pi)
 716                                 (*pi)++;
 717                         parents = parents->next;
 718                 }
 719         }
 720
 721         /*
 722          * find the tips
 723          *
 724          * tips are nodes not reachable from any other node in the list
 725          *
 726          * the tips serve as a starting set for the work queue.
 727          */
 728         for (next = orig; next; next = next->next) {
 729                 struct commit *commit = next->item;
 730
 731                 if (*(indegree_slab_at(&indegree, commit)) == 1)
 732                         prio_queue_put(&queue, commit);
 733         }
 734
 735         /*
 736          * This is unfortunate; the initial tips need to be shown
 737          * in the order given from the revision traversal machinery.
 738          */
 739         if (sort_order == REV_SORT_IN_GRAPH_ORDER)
 740                 prio_queue_reverse(&queue);
 741
 742         /* We no longer need the commit list */
 743         free_commit_list(orig);
 744
 745         pptr = list;
 746         *list = NULL;
 747         while ((commit = prio_queue_get(&queue)) != NULL) {
 748                 struct commit_list *parents;
 749
 750                 for (parents = commit->parents; parents ; parents = parents->next) {
 751                         struct commit *parent = parents->item;
 752                         int *pi = indegree_slab_at(&indegree, parent);
 753
 754                         if (!*pi)
 755                                 continue;
 756
 757                         /*
 758                          * parents are only enqueued for emission
 759                          * when all their children have been emitted thereby
 760                          * guaranteeing topological order.
 761                          */
 762                         if (--(*pi) == 1)
 763                                 prio_queue_put(&queue, parent);
 764                 }
 765                 /*
 766                  * all children of commit have already been
 767                  * emitted. we can emit it now.
 768                  */
 769                 *(indegree_slab_at(&indegree, commit)) = 0;
 770
 771                 pptr = &commit_list_insert(commit, pptr)->next;
 772         }
 773
 774         clear_indegree_slab(&indegree);
 775         clear_prio_queue(&queue);
 776         if (sort_order == REV_SORT_BY_AUTHOR_DATE)
 777                 clear_author_date_slab(&author_date);
 778 }
 779
 780 /* merge-base stuff */
 781
 782 /* Remember to update object flag allocation in object.h */
 783 #define PARENT1         (1u<<16)
 784 #define PARENT2         (1u<<17)
 785 #define STALE           (1u<<18)
 786 #define RESULT          (1u<<19)
 787
 788 static const unsigned all_flags = (PARENT1 | PARENT2 | STALE | RESULT);
 789
 790 static struct commit *interesting(struct commit_list *list)
 791 {
 792         while (list) {
 793                 struct commit *commit = list->item;
 794                 list = list->next;
 795                 if (commit->object.flags & STALE)
 796                         continue;
 797                 return commit;
 798         }
 799         return NULL;
 800 }
 801
 802 /* all input commits in one and twos[] must have been parsed! */
 803 static struct commit_list *paint_down_to_common(struct commit *one, int n, struct commit **twos)
 804 {
 805         struct commit_list *list = NULL;
 806         struct commit_list *result = NULL;
 807         int i;
 808
 809         one->object.flags |= PARENT1;
 810         commit_list_insert_by_date(one, &list);
 811         if (!n)
 812                 return list;
 813         for (i = 0; i < n; i++) {
 814                 twos[i]->object.flags |= PARENT2;
 815                 commit_list_insert_by_date(twos[i], &list);
 816         }
 817
 818         while (interesting(list)) {
 819                 struct commit *commit;
 820                 struct commit_list *parents;
 821                 struct commit_list *next;
 822                 int flags;
 823
 824                 commit = list->item;
 825                 next = list->next;
 826                 free(list);
 827                 list = next;
 828
 829                 flags = commit->object.flags & (PARENT1 | PARENT2 | STALE);
 830                 if (flags == (PARENT1 | PARENT2)) {
 831                         if (!(commit->object.flags & RESULT)) {
 832                                 commit->object.flags |= RESULT;
 833                                 commit_list_insert_by_date(commit, &result);
 834                         }
 835                         /* Mark parents of a found merge stale */
 836                         flags |= STALE;
 837                 }
 838                 parents = commit->parents;
 839                 while (parents) {
 840                         struct commit *p = parents->item;
 841                         parents = parents->next;
 842                         if ((p->object.flags & flags) == flags)
 843                                 continue;
 844                         if (parse_commit(p))
 845                                 return NULL;
 846                         p->object.flags |= flags;
 847                         commit_list_insert_by_date(p, &list);
 848                 }
 849         }
 850
 851         free_commit_list(list);
 852         return result;
 853 }
 854
 855 static struct commit_list *merge_bases_many(struct commit *one, int n, struct commit **twos)
 856 {
 857         struct commit_list *list = NULL;
 858         struct commit_list *result = NULL;
 859         int i;
 860
 861         for (i = 0; i < n; i++) {
 862                 if (one == twos[i])
 863                         /*
 864                          * We do not mark this even with RESULT so we do not
 865                          * have to clean it up.
 866                          */
 867                         return commit_list_insert(one, &result);
 868         }
 869
 870         if (parse_commit(one))
 871                 return NULL;
 872         for (i = 0; i < n; i++) {
 873                 if (parse_commit(twos[i]))
 874                         return NULL;
 875         }
 876
 877         list = paint_down_to_common(one, n, twos);
 878
 879         while (list) {
 880                 struct commit_list *next = list->next;
 881                 if (!(list->item->object.flags & STALE))
 882                         commit_list_insert_by_date(list->item, &result);
 883                 free(list);
 884                 list = next;
 885         }
 886         return result;
 887 }
 888
 889 struct commit_list *get_octopus_merge_bases(struct commit_list *in)
 890 {
 891         struct commit_list *i, *j, *k, *ret = NULL;
 892
 893         if (!in)
 894                 return ret;
 895
 896         commit_list_insert(in->item, &ret);
 897
 898         for (i = in->next; i; i = i->next) {
 899                 struct commit_list *new = NULL, *end = NULL;
 900
 901                 for (j = ret; j; j = j->next) {
 902                         struct commit_list *bases;
 903                         bases = get_merge_bases(i->item, j->item, 1);
 904                         if (!new)
 905                                 new = bases;
 906                         else
 907                                 end->next = bases;
 908                         for (k = bases; k; k = k->next)
 909                                 end = k;
 910                 }
 911                 ret = new;
 912         }
 913         return ret;
 914 }
 915
 916 static int remove_redundant(struct commit **array, int cnt)
 917 {
 918         /*
 919          * Some commit in the array may be an ancestor of
 920          * another commit.  Move such commit to the end of
 921          * the array, and return the number of commits that
 922          * are independent from each other.
 923          */
 924         struct commit **work;
 925         unsigned char *redundant;
 926         int *filled_index;
 927         int i, j, filled;
 928
 929         work = xcalloc(cnt, sizeof(*work));
 930         redundant = xcalloc(cnt, 1);
 931         filled_index = xmalloc(sizeof(*filled_index) * (cnt - 1));
 932
 933         for (i = 0; i < cnt; i++)
 934                 parse_commit(array[i]);
 935         for (i = 0; i < cnt; i++) {
 936                 struct commit_list *common;
 937
 938                 if (redundant[i])
 939                         continue;
 940                 for (j = filled = 0; j < cnt; j++) {
 941                         if (i == j || redundant[j])
 942                                 continue;
 943                         filled_index[filled] = j;
 944                         work[filled++] = array[j];
 945                 }
 946                 common = paint_down_to_common(array[i], filled, work);
 947                 if (array[i]->object.flags & PARENT2)
 948                         redundant[i] = 1;
 949                 for (j = 0; j < filled; j++)
 950                         if (work[j]->object.flags & PARENT1)
 951                                 redundant[filled_index[j]] = 1;
 952                 clear_commit_marks(array[i], all_flags);
 953                 for (j = 0; j < filled; j++)
 954                         clear_commit_marks(work[j], all_flags);
 955                 free_commit_list(common);
 956         }
 957
 958         /* Now collect the result */
 959         memcpy(work, array, sizeof(*array) * cnt);
 960         for (i = filled = 0; i < cnt; i++)
 961                 if (!redundant[i])
 962                         array[filled++] = work[i];
 963         for (j = filled, i = 0; i < cnt; i++)
 964                 if (redundant[i])
 965                         array[j++] = work[i];
 966         free(work);
 967         free(redundant);
 968         free(filled_index);
 969         return filled;
 970 }
 971
 972 struct commit_list *get_merge_bases_many(struct commit *one,
 973                                          int n,
 974                                          struct commit **twos,
 975                                          int cleanup)
 976 {
 977         struct commit_list *list;
 978         struct commit **rslt;
 979         struct commit_list *result;
 980         int cnt, i;
 981
 982         result = merge_bases_many(one, n, twos);
 983         for (i = 0; i < n; i++) {
 984                 if (one == twos[i])
 985                         return result;
 986         }
 987         if (!result || !result->next) {
 988                 if (cleanup) {
 989                         clear_commit_marks(one, all_flags);
 990                         clear_commit_marks_many(n, twos, all_flags);
 991                 }
 992                 return result;
 993         }
 994
 995         /* There are more than one */
 996         cnt = 0;
 997         list = result;
 998         while (list) {
 999                 list = list->next;
1000                 cnt++;
1001         }
1002         rslt = xcalloc(cnt, sizeof(*rslt));
1003         for (list = result, i = 0; list; list = list->next)
1004                 rslt[i++] = list->item;
1005         free_commit_list(result);
1006
1007         clear_commit_marks(one, all_flags);
1008         clear_commit_marks_many(n, twos, all_flags);
1009
1010         cnt = remove_redundant(rslt, cnt);
1011         result = NULL;
1012         for (i = 0; i < cnt; i++)
1013                 commit_list_insert_by_date(rslt[i], &result);
1014         free(rslt);
1015         return result;
1016 }
1017
1018 struct commit_list *get_merge_bases(struct commit *one, struct commit *two,
1019                                     int cleanup)
1020 {
1021         return get_merge_bases_many(one, 1, &two, cleanup);
1022 }
1023
1024 /*
1025  * Is "commit" a descendant of one of the elements on the "with_commit" list?
1026  */
1027 int is_descendant_of(struct commit *commit, struct commit_list *with_commit)
1028 {
1029         if (!with_commit)
1030                 return 1;
1031         while (with_commit) {
1032                 struct commit *other;
1033
1034                 other = with_commit->item;
1035                 with_commit = with_commit->next;
1036                 if (in_merge_bases(other, commit))
1037                         return 1;
1038         }
1039         return 0;
1040 }
1041
1042 /*
1043  * Is "commit" an ancestor of one of the "references"?
1044  */
1045 int in_merge_bases_many(struct commit *commit, int nr_reference, struct commit **reference)
1046 {
1047         struct commit_list *bases;
1048         int ret = 0, i;
1049
1050         if (parse_commit(commit))
1051                 return ret;
1052         for (i = 0; i < nr_reference; i++)
1053                 if (parse_commit(reference[i]))
1054                         return ret;
1055
1056         bases = paint_down_to_common(commit, nr_reference, reference);
1057         if (commit->object.flags & PARENT2)
1058                 ret = 1;
1059         clear_commit_marks(commit, all_flags);
1060         clear_commit_marks_many(nr_reference, reference, all_flags);
1061         free_commit_list(bases);
1062         return ret;
1063 }
1064
1065 /*
1066  * Is "commit" an ancestor of (i.e. reachable from) the "reference"?
1067  */
1068 int in_merge_bases(struct commit *commit, struct commit *reference)
1069 {
1070         return in_merge_bases_many(commit, 1, &reference);
1071 }
1072
1073 struct commit_list *reduce_heads(struct commit_list *heads)
1074 {
1075         struct commit_list *p;
1076         struct commit_list *result = NULL, **tail = &result;
1077         struct commit **array;
1078         int num_head, i;
1079
1080         if (!heads)
1081                 return NULL;
1082
1083         /* Uniquify */
1084         for (p = heads; p; p = p->next)
1085                 p->item->object.flags &= ~STALE;
1086         for (p = heads, num_head = 0; p; p = p->next) {
1087                 if (p->item->object.flags & STALE)
1088                         continue;
1089                 p->item->object.flags |= STALE;
1090                 num_head++;
1091         }
1092         array = xcalloc(num_head, sizeof(*array));
1093         for (p = heads, i = 0; p; p = p->next) {
1094                 if (p->item->object.flags & STALE) {
1095                         array[i++] = p->item;
1096                         p->item->object.flags &= ~STALE;
1097                 }
1098         }
1099         num_head = remove_redundant(array, num_head);
1100         for (i = 0; i < num_head; i++)
1101                 tail = &commit_list_insert(array[i], tail)->next;
1102         return result;
1103 }
1104
1105 static const char gpg_sig_header[] = "gpgsig";
1106 static const int gpg_sig_header_len = sizeof(gpg_sig_header) - 1;
1107
1108 static int do_sign_commit(struct strbuf *buf, const char *keyid)
1109 {
1110         struct strbuf sig = STRBUF_INIT;
1111         int inspos, copypos;
1112
1113         /* find the end of the header */
1114         inspos = strstr(buf->buf, "\n\n") - buf->buf + 1;
1115
1116         if (!keyid || !*keyid)
1117                 keyid = get_signing_key();
1118         if (sign_buffer(buf, &sig, keyid)) {
1119                 strbuf_release(&sig);
1120                 return -1;
1121         }
1122
1123         for (copypos = 0; sig.buf[copypos]; ) {
1124                 const char *bol = sig.buf + copypos;
1125                 const char *eol = strchrnul(bol, '\n');
1126                 int len = (eol - bol) + !!*eol;
1127
1128                 if (!copypos) {
1129                         strbuf_insert(buf, inspos, gpg_sig_header, gpg_sig_header_len);
1130                         inspos += gpg_sig_header_len;
1131                 }
1132                 strbuf_insert(buf, inspos++, " ", 1);
1133                 strbuf_insert(buf, inspos, bol, len);
1134                 inspos += len;
1135                 copypos += len;
1136         }
1137         strbuf_release(&sig);
1138         return 0;
1139 }
1140
1141 int parse_signed_commit(const struct commit *commit,
1142                         struct strbuf *payload, struct strbuf *signature)
1143 {
1144
1145         unsigned long size;
1146         const char *buffer = get_commit_buffer(commit, &size);
1147         int in_signature, saw_signature = -1;
1148         const char *line, *tail;
1149
1150         line = buffer;
1151         tail = buffer + size;
1152         in_signature = 0;
1153         saw_signature = 0;
1154         while (line < tail) {
1155                 const char *sig = NULL;
1156                 const char *next = memchr(line, '\n', tail - line);
1157
1158                 next = next ? next + 1 : tail;
1159                 if (in_signature && line[0] == ' ')
1160                         sig = line + 1;
1161                 else if (starts_with(line, gpg_sig_header) &&
1162                          line[gpg_sig_header_len] == ' ')
1163                         sig = line + gpg_sig_header_len + 1;
1164                 if (sig) {
1165                         strbuf_add(signature, sig, next - sig);
1166                         saw_signature = 1;
1167                         in_signature = 1;
1168                 } else {
1169                         if (*line == '\n')
1170                                 /* dump the whole remainder of the buffer */
1171                                 next = tail;
1172                         strbuf_add(payload, line, next - line);
1173                         in_signature = 0;
1174                 }
1175                 line = next;
1176         }
1177         unuse_commit_buffer(commit, buffer);
1178         return saw_signature;
1179 }
1180
1181 static void handle_signed_tag(struct commit *parent, struct commit_extra_header ***tail)
1182 {
1183         struct merge_remote_desc *desc;
1184         struct commit_extra_header *mergetag;
1185         char *buf;
1186         unsigned long size, len;
1187         enum object_type type;
1188
1189         desc = merge_remote_util(parent);
1190         if (!desc || !desc->obj)
1191                 return;
1192         buf = read_sha1_file(desc->obj->sha1, &type, &size);
1193         if (!buf || type != OBJ_TAG)
1194                 goto free_return;
1195         len = parse_signature(buf, size);
1196         if (size == len)
1197                 goto free_return;
1198         /*
1199          * We could verify this signature and either omit the tag when
1200          * it does not validate, but the integrator may not have the
1201          * public key of the signer of the tag he is merging, while a
1202          * later auditor may have it while auditing, so let's not run
1203          * verify-signed-buffer here for now...
1204          *
1205          * if (verify_signed_buffer(buf, len, buf + len, size - len, ...))
1206          *      warn("warning: signed tag unverified.");
1207          */
1208         mergetag = xcalloc(1, sizeof(*mergetag));
1209         mergetag->key = xstrdup("mergetag");
1210         mergetag->value = buf;
1211         mergetag->len = size;
1212
1213         **tail = mergetag;
1214         *tail = &mergetag->next;
1215         return;
1216
1217 free_return:
1218         free(buf);
1219 }
1220
1221 static struct {
1222         char result;
1223         const char *check;
1224 } sigcheck_gpg_status[] = {
1225         { 'G', "\n[GNUPG:] GOODSIG " },
1226         { 'B', "\n[GNUPG:] BADSIG " },
1227         { 'U', "\n[GNUPG:] TRUST_NEVER" },
1228         { 'U', "\n[GNUPG:] TRUST_UNDEFINED" },
1229 };
1230
1231 static void parse_gpg_output(struct signature_check *sigc)
1232 {
1233         const char *buf = sigc->gpg_status;
1234         int i;
1235
1236         /* Iterate over all search strings */
1237         for (i = 0; i < ARRAY_SIZE(sigcheck_gpg_status); i++) {
1238                 const char *found, *next;
1239
1240                 found = skip_prefix(buf, sigcheck_gpg_status[i].check + 1);
1241                 if (!found) {
1242                         found = strstr(buf, sigcheck_gpg_status[i].check);
1243                         if (!found)
1244                                 continue;
1245                         found += strlen(sigcheck_gpg_status[i].check);
1246                 }
1247                 sigc->result = sigcheck_gpg_status[i].result;
1248                 /* The trust messages are not followed by key/signer information */
1249                 if (sigc->result != 'U') {
1250                         sigc->key = xmemdupz(found, 16);
1251                         found += 17;
1252                         next = strchrnul(found, '\n');
1253                         sigc->signer = xmemdupz(found, next - found);
1254                 }
1255         }
1256 }
1257
1258 void check_commit_signature(const struct commit* commit, struct signature_check *sigc)
1259 {
1260         struct strbuf payload = STRBUF_INIT;
1261         struct strbuf signature = STRBUF_INIT;
1262         struct strbuf gpg_output = STRBUF_INIT;
1263         struct strbuf gpg_status = STRBUF_INIT;
1264         int status;
1265
1266         sigc->result = 'N';
1267
1268         if (parse_signed_commit(commit, &payload, &signature) <= 0)
1269                 goto out;
1270         status = verify_signed_buffer(payload.buf, payload.len,
1271                                       signature.buf, signature.len,
1272                                       &gpg_output, &gpg_status);
1273         if (status && !gpg_output.len)
1274                 goto out;
1275         sigc->gpg_output = strbuf_detach(&gpg_output, NULL);
1276         sigc->gpg_status = strbuf_detach(&gpg_status, NULL);
1277         parse_gpg_output(sigc);
1278
1279  out:
1280         strbuf_release(&gpg_status);
1281         strbuf_release(&gpg_output);
1282         strbuf_release(&payload);
1283         strbuf_release(&signature);
1284 }
1285
1286
1287
1288 void append_merge_tag_headers(struct commit_list *parents,
1289                               struct commit_extra_header ***tail)
1290 {
1291         while (parents) {
1292                 struct commit *parent = parents->item;
1293                 handle_signed_tag(parent, tail);
1294                 parents = parents->next;
1295         }
1296 }
1297
1298 static void add_extra_header(struct strbuf *buffer,
1299                              struct commit_extra_header *extra)
1300 {
1301         strbuf_addstr(buffer, extra->key);
1302         if (extra->len)
1303                 strbuf_add_lines(buffer, " ", extra->value, extra->len);
1304         else
1305                 strbuf_addch(buffer, '\n');
1306 }
1307
1308 struct commit_extra_header *read_commit_extra_headers(struct commit *commit,
1309                                                       const char **exclude)
1310 {
1311         struct commit_extra_header *extra = NULL;
1312         unsigned long size;
1313         const char *buffer = get_commit_buffer(commit, &size);
1314         extra = read_commit_extra_header_lines(buffer, size, exclude);
1315         unuse_commit_buffer(commit, buffer);
1316         return extra;
1317 }
1318
1319 static inline int standard_header_field(const char *field, size_t len)
1320 {
1321         return ((len == 4 && !memcmp(field, "tree ", 5)) ||
1322                 (len == 6 && !memcmp(field, "parent ", 7)) ||
1323                 (len == 6 && !memcmp(field, "author ", 7)) ||
1324                 (len == 9 && !memcmp(field, "committer ", 10)) ||
1325                 (len == 8 && !memcmp(field, "encoding ", 9)));
1326 }
1327
1328 static int excluded_header_field(const char *field, size_t len, const char **exclude)
1329 {
1330         if (!exclude)
1331                 return 0;
1332
1333         while (*exclude) {
1334                 size_t xlen = strlen(*exclude);
1335                 if (len == xlen &&
1336                     !memcmp(field, *exclude, xlen) && field[xlen] == ' ')
1337                         return 1;
1338                 exclude++;
1339         }
1340         return 0;
1341 }
1342
1343 static struct commit_extra_header *read_commit_extra_header_lines(
1344         const char *buffer, size_t size,
1345         const char **exclude)
1346 {
1347         struct commit_extra_header *extra = NULL, **tail = &extra, *it = NULL;
1348         const char *line, *next, *eof, *eob;
1349         struct strbuf buf = STRBUF_INIT;
1350
1351         for (line = buffer, eob = line + size;
1352              line < eob && *line != '\n';
1353              line = next) {
1354                 next = memchr(line, '\n', eob - line);
1355                 next = next ? next + 1 : eob;
1356                 if (*line == ' ') {
1357                         /* continuation */
1358                         if (it)
1359                                 strbuf_add(&buf, line + 1, next - (line + 1));
1360                         continue;
1361                 }
1362                 if (it)
1363                         it->value = strbuf_detach(&buf, &it->len);
1364                 strbuf_reset(&buf);
1365                 it = NULL;
1366
1367                 eof = strchr(line, ' ');
1368                 if (next <= eof)
1369                         eof = next;
1370
1371                 if (standard_header_field(line, eof - line) ||
1372                     excluded_header_field(line, eof - line, exclude))
1373                         continue;
1374
1375                 it = xcalloc(1, sizeof(*it));
1376                 it->key = xmemdupz(line, eof-line);
1377                 *tail = it;
1378                 tail = &it->next;
1379                 if (eof + 1 < next)
1380                         strbuf_add(&buf, eof + 1, next - (eof + 1));
1381         }
1382         if (it)
1383                 it->value = strbuf_detach(&buf, &it->len);
1384         return extra;
1385 }
1386
1387 void free_commit_extra_headers(struct commit_extra_header *extra)
1388 {
1389         while (extra) {
1390                 struct commit_extra_header *next = extra->next;
1391                 free(extra->key);
1392                 free(extra->value);
1393                 free(extra);
1394                 extra = next;
1395         }
1396 }
1397
1398 int commit_tree(const char *msg, size_t msg_len,
1399                 const unsigned char *tree,
1400                 struct commit_list *parents, unsigned char *ret,
1401                 const char *author, const char *sign_commit)
1402 {
1403         struct commit_extra_header *extra = NULL, **tail = &extra;
1404         int result;
1405
1406         append_merge_tag_headers(parents, &tail);
1407         result = commit_tree_extended(msg, msg_len, tree, parents, ret,
1408                                       author, sign_commit, extra);
1409         free_commit_extra_headers(extra);
1410         return result;
1411 }
1412
1413 static int find_invalid_utf8(const char *buf, int len)
1414 {
1415         int offset = 0;
1416         static const unsigned int max_codepoint[] = {
1417                 0x7f, 0x7ff, 0xffff, 0x10ffff
1418         };
1419
1420         while (len) {
1421                 unsigned char c = *buf++;
1422                 int bytes, bad_offset;
1423                 unsigned int codepoint;
1424                 unsigned int min_val, max_val;
1425
1426                 len--;
1427                 offset++;
1428
1429                 /* Simple US-ASCII? No worries. */
1430                 if (c < 0x80)
1431                         continue;
1432
1433                 bad_offset = offset-1;
1434
1435                 /*
1436                  * Count how many more high bits set: that's how
1437                  * many more bytes this sequence should have.
1438                  */
1439                 bytes = 0;
1440                 while (c & 0x40) {
1441                         c <<= 1;
1442                         bytes++;
1443                 }
1444
1445                 /*
1446                  * Must be between 1 and 3 more bytes.  Longer sequences result in
1447                  * codepoints beyond U+10FFFF, which are guaranteed never to exist.
1448                  */
1449                 if (bytes < 1 || 3 < bytes)
1450                         return bad_offset;
1451
1452                 /* Do we *have* that many bytes? */
1453                 if (len < bytes)
1454                         return bad_offset;
1455
1456                 /*
1457                  * Place the encoded bits at the bottom of the value and compute the
1458                  * valid range.
1459                  */
1460                 codepoint = (c & 0x7f) >> bytes;
1461                 min_val = max_codepoint[bytes-1] + 1;
1462                 max_val = max_codepoint[bytes];
1463
1464                 offset += bytes;
1465                 len -= bytes;
1466
1467                 /* And verify that they are good continuation bytes */
1468                 do {
1469                         codepoint <<= 6;
1470                         codepoint |= *buf & 0x3f;
1471                         if ((*buf++ & 0xc0) != 0x80)
1472                                 return bad_offset;
1473                 } while (--bytes);
1474
1475                 /* Reject codepoints that are out of range for the sequence length. */
1476                 if (codepoint < min_val || codepoint > max_val)
1477                         return bad_offset;
1478                 /* Surrogates are only for UTF-16 and cannot be encoded in UTF-8. */
1479                 if ((codepoint & 0x1ff800) == 0xd800)
1480                         return bad_offset;
1481                 /* U+xxFFFE and U+xxFFFF are guaranteed non-characters. */
1482                 if ((codepoint & 0xfffe) == 0xfffe)
1483                         return bad_offset;
1484                 /* So are anything in the range U+FDD0..U+FDEF. */
1485                 if (codepoint >= 0xfdd0 && codepoint <= 0xfdef)
1486                         return bad_offset;
1487         }
1488         return -1;
1489 }
1490
1491 /*
1492  * This verifies that the buffer is in proper utf8 format.
1493  *
1494  * If it isn't, it assumes any non-utf8 characters are Latin1,
1495  * and does the conversion.
1496  */
1497 static int verify_utf8(struct strbuf *buf)
1498 {
1499         int ok = 1;
1500         long pos = 0;
1501
1502         for (;;) {
1503                 int bad;
1504                 unsigned char c;
1505                 unsigned char replace[2];
1506
1507                 bad = find_invalid_utf8(buf->buf + pos, buf->len - pos);
1508                 if (bad < 0)
1509                         return ok;
1510                 pos += bad;
1511                 ok = 0;
1512                 c = buf->buf[pos];
1513                 strbuf_remove(buf, pos, 1);
1514
1515                 /* We know 'c' must be in the range 128-255 */
1516                 replace[0] = 0xc0 + (c >> 6);
1517                 replace[1] = 0x80 + (c & 0x3f);
1518                 strbuf_insert(buf, pos, replace, 2);
1519                 pos += 2;
1520         }
1521 }
1522
1523 static const char commit_utf8_warn[] =
1524 "Warning: commit message did not conform to UTF-8.\n"
1525 "You may want to amend it after fixing the message, or set the config\n"
1526 "variable i18n.commitencoding to the encoding your project uses.\n";
1527
1528 int commit_tree_extended(const char *msg, size_t msg_len,
1529                          const unsigned char *tree,
1530                          struct commit_list *parents, unsigned char *ret,
1531                          const char *author, const char *sign_commit,
1532                          struct commit_extra_header *extra)
1533 {
1534         int result;
1535         int encoding_is_utf8;
1536         struct strbuf buffer;
1537
1538         assert_sha1_type(tree, OBJ_TREE);
1539
1540         if (memchr(msg, '\0', msg_len))
1541                 return error("a NUL byte in commit log message not allowed.");
1542
1543         /* Not having i18n.commitencoding is the same as having utf-8 */
1544         encoding_is_utf8 = is_encoding_utf8(git_commit_encoding);
1545
1546         strbuf_init(&buffer, 8192); /* should avoid reallocs for the headers */
1547         strbuf_addf(&buffer, "tree %s\n", sha1_to_hex(tree));
1548
1549         /*
1550          * NOTE! This ordering means that the same exact tree merged with a
1551          * different order of parents will be a _different_ changeset even
1552          * if everything else stays the same.
1553          */
1554         while (parents) {
1555                 struct commit_list *next = parents->next;
1556                 struct commit *parent = parents->item;
1557
1558                 strbuf_addf(&buffer, "parent %s\n",
1559                             sha1_to_hex(parent->object.sha1));
1560                 free(parents);
1561                 parents = next;
1562         }
1563
1564         /* Person/date information */
1565         if (!author)
1566                 author = git_author_info(IDENT_STRICT);
1567         strbuf_addf(&buffer, "author %s\n", author);
1568         strbuf_addf(&buffer, "committer %s\n", git_committer_info(IDENT_STRICT));
1569         if (!encoding_is_utf8)
1570                 strbuf_addf(&buffer, "encoding %s\n", git_commit_encoding);
1571
1572         while (extra) {
1573                 add_extra_header(&buffer, extra);
1574                 extra = extra->next;
1575         }
1576         strbuf_addch(&buffer, '\n');
1577
1578         /* And add the comment */
1579         strbuf_add(&buffer, msg, msg_len);
1580
1581         /* And check the encoding */
1582         if (encoding_is_utf8 && !verify_utf8(&buffer))
1583                 fprintf(stderr, commit_utf8_warn);
1584
1585         if (sign_commit && do_sign_commit(&buffer, sign_commit))
1586                 return -1;
1587
1588         result = write_sha1_file(buffer.buf, buffer.len, commit_type, ret);
1589         strbuf_release(&buffer);
1590         return result;
1591 }
1592
1593 struct commit *get_merge_parent(const char *name)
1594 {
1595         struct object *obj;
1596         struct commit *commit;
1597         unsigned char sha1[20];
1598         if (get_sha1(name, sha1))
1599                 return NULL;
1600         obj = parse_object(sha1);
1601         commit = (struct commit *)peel_to_type(name, 0, obj, OBJ_COMMIT);
1602         if (commit && !commit->util) {
1603                 struct merge_remote_desc *desc;
1604                 desc = xmalloc(sizeof(*desc));
1605                 desc->obj = obj;
1606                 desc->name = strdup(name);
1607                 commit->util = desc;
1608         }
1609         return commit;
1610 }
1611
1612 /*
1613  * Append a commit to the end of the commit_list.
1614  *
1615  * next starts by pointing to the variable that holds the head of an
1616  * empty commit_list, and is updated to point to the "next" field of
1617  * the last item on the list as new commits are appended.
1618  *
1619  * Usage example:
1620  *
1621  *     struct commit_list *list;
1622  *     struct commit_list **next = &list;
1623  *
1624  *     next = commit_list_append(c1, next);
1625  *     next = commit_list_append(c2, next);
1626  *     assert(commit_list_count(list) == 2);
1627  *     return list;
1628  */
1629 struct commit_list **commit_list_append(struct commit *commit,
1630                                         struct commit_list **next)
1631 {
1632         struct commit_list *new = xmalloc(sizeof(struct commit_list));
1633         new->item = commit;
1634         *next = new;
1635         new->next = NULL;
1636         return &new->next;
1637 }
1638
1639 void print_commit_list(struct commit_list *list,
1640                        const char *format_cur,
1641                        const char *format_last)
1642 {
1643         for ( ; list; list = list->next) {
1644                 const char *format = list->next ? format_cur : format_last;
1645                 printf(format, sha1_to_hex(list->item->object.sha1));
1646         }
1647 }