commit.c

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