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 struct commit *lookup_commit_reference_gently(const struct object_id *oid,
  22                                               int quiet)
  23 {
  24         struct object *obj = deref_tag(parse_object(oid), NULL, 0);
  25
  26         if (!obj)
  27                 return NULL;
  28         return object_as_type(obj, OBJ_COMMIT, quiet);
  29 }
  30
  31 struct commit *lookup_commit_reference(const struct object_id *oid)
  32 {
  33         return lookup_commit_reference_gently(oid, 0);
  34 }
  35
  36 struct commit *lookup_commit_or_die(const struct object_id *oid, const char *ref_name)
  37 {
  38         struct commit *c = lookup_commit_reference(oid);
  39         if (!c)
  40                 die(_("could not parse %s"), ref_name);
  41         if (oidcmp(oid, &c->object.oid)) {
  42                 warning(_("%s %s is not a commit!"),
  43                         ref_name, oid_to_hex(oid));
  44         }
  45         return c;
  46 }
  47
  48 struct commit *lookup_commit(const struct object_id *oid)
  49 {
  50         struct object *obj = lookup_object(oid->hash);
  51         if (!obj)
  52                 return create_object(oid->hash, alloc_commit_node());
  53         return object_as_type(obj, OBJ_COMMIT, 0);
  54 }
  55
  56 struct commit *lookup_commit_reference_by_name(const char *name)
  57 {
  58         struct object_id oid;
  59         struct commit *commit;
  60
  61         if (get_sha1_committish(name, oid.hash))
  62                 return NULL;
  63         commit = lookup_commit_reference(&oid);
  64         if (parse_commit(commit))
  65                 return NULL;
  66         return commit;
  67 }
  68
  69 static timestamp_t parse_commit_date(const char *buf, const char *tail)
  70 {
  71         const char *dateptr;
  72
  73         if (buf + 6 >= tail)
  74                 return 0;
  75         if (memcmp(buf, "author", 6))
  76                 return 0;
  77         while (buf < tail && *buf++ != '\n')
  78                 /* nada */;
  79         if (buf + 9 >= tail)
  80                 return 0;
  81         if (memcmp(buf, "committer", 9))
  82                 return 0;
  83         while (buf < tail && *buf++ != '>')
  84                 /* nada */;
  85         if (buf >= tail)
  86                 return 0;
  87         dateptr = buf;
  88         while (buf < tail && *buf++ != '\n')
  89                 /* nada */;
  90         if (buf >= tail)
  91                 return 0;
  92         /* dateptr < buf && buf[-1] == '\n', so parsing will stop at buf-1 */
  93         return parse_timestamp(dateptr, NULL, 10);
  94 }
  95
  96 static struct commit_graft **commit_graft;
  97 static int commit_graft_alloc, commit_graft_nr;
  98
  99 static const unsigned char *commit_graft_sha1_access(size_t index, void *table)
 100 {
 101         struct commit_graft **commit_graft_table = table;
 102         return commit_graft_table[index]->oid.hash;
 103 }
 104
 105 static int commit_graft_pos(const unsigned char *sha1)
 106 {
 107         return sha1_pos(sha1, commit_graft, commit_graft_nr,
 108                         commit_graft_sha1_access);
 109 }
 110
 111 int register_commit_graft(struct commit_graft *graft, int ignore_dups)
 112 {
 113         int pos = commit_graft_pos(graft->oid.hash);
 114
 115         if (0 <= pos) {
 116                 if (ignore_dups)
 117                         free(graft);
 118                 else {
 119                         free(commit_graft[pos]);
 120                         commit_graft[pos] = graft;
 121                 }
 122                 return 1;
 123         }
 124         pos = -pos - 1;
 125         ALLOC_GROW(commit_graft, commit_graft_nr + 1, commit_graft_alloc);
 126         commit_graft_nr++;
 127         if (pos < commit_graft_nr)
 128                 memmove(commit_graft + pos + 1,
 129                         commit_graft + pos,
 130                         (commit_graft_nr - pos - 1) *
 131                         sizeof(*commit_graft));
 132         commit_graft[pos] = graft;
 133         return 0;
 134 }
 135
 136 struct commit_graft *read_graft_line(char *buf, int len)
 137 {
 138         /* The format is just "Commit Parent1 Parent2 ...\n" */
 139         int i;
 140         struct commit_graft *graft = NULL;
 141         const int entry_size = GIT_SHA1_HEXSZ + 1;
 142
 143         while (len && isspace(buf[len-1]))
 144                 buf[--len] = '\0';
 145         if (buf[0] == '#' || buf[0] == '\0')
 146                 return NULL;
 147         if ((len + 1) % entry_size)
 148                 goto bad_graft_data;
 149         i = (len + 1) / entry_size - 1;
 150         graft = xmalloc(st_add(sizeof(*graft), st_mult(GIT_SHA1_RAWSZ, i)));
 151         graft->nr_parent = i;
 152         if (get_oid_hex(buf, &graft->oid))
 153                 goto bad_graft_data;
 154         for (i = GIT_SHA1_HEXSZ; i < len; i += entry_size) {
 155                 if (buf[i] != ' ')
 156                         goto bad_graft_data;
 157                 if (get_sha1_hex(buf + i + 1, graft->parent[i/entry_size].hash))
 158                         goto bad_graft_data;
 159         }
 160         return graft;
 161
 162 bad_graft_data:
 163         error("bad graft data: %s", buf);
 164         free(graft);
 165         return NULL;
 166 }
 167
 168 static int read_graft_file(const char *graft_file)
 169 {
 170         FILE *fp = fopen(graft_file, "r");
 171         struct strbuf buf = STRBUF_INIT;
 172         if (!fp)
 173                 return -1;
 174         while (!strbuf_getwholeline(&buf, fp, '\n')) {
 175                 /* The format is just "Commit Parent1 Parent2 ...\n" */
 176                 struct commit_graft *graft = read_graft_line(buf.buf, buf.len);
 177                 if (!graft)
 178                         continue;
 179                 if (register_commit_graft(graft, 1))
 180                         error("duplicate graft data: %s", buf.buf);
 181         }
 182         fclose(fp);
 183         strbuf_release(&buf);
 184         return 0;
 185 }
 186
 187 static void prepare_commit_graft(void)
 188 {
 189         static int commit_graft_prepared;
 190         char *graft_file;
 191
 192         if (commit_graft_prepared)
 193                 return;
 194         graft_file = get_graft_file();
 195         read_graft_file(graft_file);
 196         /* make sure shallows are read */
 197         is_repository_shallow();
 198         commit_graft_prepared = 1;
 199 }
 200
 201 struct commit_graft *lookup_commit_graft(const unsigned char *sha1)
 202 {
 203         int pos;
 204         prepare_commit_graft();
 205         pos = commit_graft_pos(sha1);
 206         if (pos < 0)
 207                 return NULL;
 208         return commit_graft[pos];
 209 }
 210
 211 int for_each_commit_graft(each_commit_graft_fn fn, void *cb_data)
 212 {
 213         int i, ret;
 214         for (i = ret = 0; i < commit_graft_nr && !ret; i++)
 215                 ret = fn(commit_graft[i], cb_data);
 216         return ret;
 217 }
 218
 219 int unregister_shallow(const struct object_id *oid)
 220 {
 221         int pos = commit_graft_pos(oid->hash);
 222         if (pos < 0)
 223                 return -1;
 224         if (pos + 1 < commit_graft_nr)
 225                 memmove(commit_graft + pos, commit_graft + pos + 1,
 226                                 sizeof(struct commit_graft *)
 227                                 * (commit_graft_nr - pos - 1));
 228         commit_graft_nr--;
 229         return 0;
 230 }
 231
 232 struct commit_buffer {
 233         void *buffer;
 234         unsigned long size;
 235 };
 236 define_commit_slab(buffer_slab, struct commit_buffer);
 237 static struct buffer_slab buffer_slab = COMMIT_SLAB_INIT(1, buffer_slab);
 238
 239 void set_commit_buffer(struct commit *commit, void *buffer, unsigned long size)
 240 {
 241         struct commit_buffer *v = buffer_slab_at(&buffer_slab, commit);
 242         v->buffer = buffer;
 243         v->size = size;
 244 }
 245
 246 const void *get_cached_commit_buffer(const struct commit *commit, unsigned long *sizep)
 247 {
 248         struct commit_buffer *v = buffer_slab_peek(&buffer_slab, commit);
 249         if (!v) {
 250                 if (sizep)
 251                         *sizep = 0;
 252                 return NULL;
 253         }
 254         if (sizep)
 255                 *sizep = v->size;
 256         return v->buffer;
 257 }
 258
 259 const void *get_commit_buffer(const struct commit *commit, unsigned long *sizep)
 260 {
 261         const void *ret = get_cached_commit_buffer(commit, sizep);
 262         if (!ret) {
 263                 enum object_type type;
 264                 unsigned long size;
 265                 ret = read_sha1_file(commit->object.oid.hash, &type, &size);
 266                 if (!ret)
 267                         die("cannot read commit object %s",
 268                             oid_to_hex(&commit->object.oid));
 269                 if (type != OBJ_COMMIT)
 270                         die("expected commit for %s, got %s",
 271                             oid_to_hex(&commit->object.oid), typename(type));
 272                 if (sizep)
 273                         *sizep = size;
 274         }
 275         return ret;
 276 }
 277
 278 void unuse_commit_buffer(const struct commit *commit, const void *buffer)
 279 {
 280         struct commit_buffer *v = buffer_slab_peek(&buffer_slab, commit);
 281         if (!(v && v->buffer == buffer))
 282                 free((void *)buffer);
 283 }
 284
 285 void free_commit_buffer(struct commit *commit)
 286 {
 287         struct commit_buffer *v = buffer_slab_peek(&buffer_slab, commit);
 288         if (v) {
 289                 free(v->buffer);
 290                 v->buffer = NULL;
 291                 v->size = 0;
 292         }
 293 }
 294
 295 const void *detach_commit_buffer(struct commit *commit, unsigned long *sizep)
 296 {
 297         struct commit_buffer *v = buffer_slab_peek(&buffer_slab, commit);
 298         void *ret;
 299
 300         if (!v) {
 301                 if (sizep)
 302                         *sizep = 0;
 303                 return NULL;
 304         }
 305         ret = v->buffer;
 306         if (sizep)
 307                 *sizep = v->size;
 308
 309         v->buffer = NULL;
 310         v->size = 0;
 311         return ret;
 312 }
 313
 314 int parse_commit_buffer(struct commit *item, const void *buffer, unsigned long size)
 315 {
 316         const char *tail = buffer;
 317         const char *bufptr = buffer;
 318         struct object_id parent;
 319         struct commit_list **pptr;
 320         struct commit_graft *graft;
 321         const int tree_entry_len = GIT_SHA1_HEXSZ + 5;
 322         const int parent_entry_len = GIT_SHA1_HEXSZ + 7;
 323
 324         if (item->object.parsed)
 325                 return 0;
 326         item->object.parsed = 1;
 327         tail += size;
 328         if (tail <= bufptr + tree_entry_len + 1 || memcmp(bufptr, "tree ", 5) ||
 329                         bufptr[tree_entry_len] != '\n')
 330                 return error("bogus commit object %s", oid_to_hex(&item->object.oid));
 331         if (get_sha1_hex(bufptr + 5, parent.hash) < 0)
 332                 return error("bad tree pointer in commit %s",
 333                              oid_to_hex(&item->object.oid));
 334         item->tree = lookup_tree(&parent);
 335         bufptr += tree_entry_len + 1; /* "tree " + "hex sha1" + "\n" */
 336         pptr = &item->parents;
 337
 338         graft = lookup_commit_graft(item->object.oid.hash);
 339         while (bufptr + parent_entry_len < tail && !memcmp(bufptr, "parent ", 7)) {
 340                 struct commit *new_parent;
 341
 342                 if (tail <= bufptr + parent_entry_len + 1 ||
 343                     get_sha1_hex(bufptr + 7, parent.hash) ||
 344                     bufptr[parent_entry_len] != '\n')
 345                         return error("bad parents in commit %s", oid_to_hex(&item->object.oid));
 346                 bufptr += parent_entry_len + 1;
 347                 /*
 348                  * The clone is shallow if nr_parent < 0, and we must
 349                  * not traverse its real parents even when we unhide them.
 350                  */
 351                 if (graft && (graft->nr_parent < 0 || grafts_replace_parents))
 352                         continue;
 353                 new_parent = lookup_commit(&parent);
 354                 if (new_parent)
 355                         pptr = &commit_list_insert(new_parent, pptr)->next;
 356         }
 357         if (graft) {
 358                 int i;
 359                 struct commit *new_parent;
 360                 for (i = 0; i < graft->nr_parent; i++) {
 361                         new_parent = lookup_commit(&graft->parent[i]);
 362                         if (!new_parent)
 363                                 continue;
 364                         pptr = &commit_list_insert(new_parent, pptr)->next;
 365                 }
 366         }
 367         item->date = parse_commit_date(bufptr, tail);
 368
 369         return 0;
 370 }
 371
 372 int parse_commit_gently(struct commit *item, int quiet_on_missing)
 373 {
 374         enum object_type type;
 375         void *buffer;
 376         unsigned long size;
 377         int ret;
 378
 379         if (!item)
 380                 return -1;
 381         if (item->object.parsed)
 382                 return 0;
 383         buffer = read_sha1_file(item->object.oid.hash, &type, &size);
 384         if (!buffer)
 385                 return quiet_on_missing ? -1 :
 386                         error("Could not read %s",
 387                              oid_to_hex(&item->object.oid));
 388         if (type != OBJ_COMMIT) {
 389                 free(buffer);
 390                 return error("Object %s not a commit",
 391                              oid_to_hex(&item->object.oid));
 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 ? oid_to_hex(&item->object.oid) : "(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 = skip_blank_lines(p + 2);
 418                 eol = strchrnul(p, '\n');
 419         } else
 420                 eol = p;
 421
 422         *subject = p;
 423
 424         return eol - p;
 425 }
 426
 427 struct commit_list *commit_list_insert(struct commit *item, struct commit_list **list_p)
 428 {
 429         struct commit_list *new_list = xmalloc(sizeof(struct commit_list));
 430         new_list->item = item;
 431         new_list->next = *list_p;
 432         *list_p = new_list;
 433         return new_list;
 434 }
 435
 436 unsigned commit_list_count(const struct commit_list *l)
 437 {
 438         unsigned c = 0;
 439         for (; l; l = l->next )
 440                 c++;
 441         return c;
 442 }
 443
 444 struct commit_list *copy_commit_list(struct commit_list *list)
 445 {
 446         struct commit_list *head = NULL;
 447         struct commit_list **pp = &head;
 448         while (list) {
 449                 pp = commit_list_append(list->item, pp);
 450                 list = list->next;
 451         }
 452         return head;
 453 }
 454
 455 void free_commit_list(struct commit_list *list)
 456 {
 457         while (list)
 458                 pop_commit(&list);
 459 }
 460
 461 struct commit_list * commit_list_insert_by_date(struct commit *item, struct commit_list **list)
 462 {
 463         struct commit_list **pp = list;
 464         struct commit_list *p;
 465         while ((p = *pp) != NULL) {
 466                 if (p->item->date < item->date) {
 467                         break;
 468                 }
 469                 pp = &p->next;
 470         }
 471         return commit_list_insert(item, pp);
 472 }
 473
 474 static int commit_list_compare_by_date(const void *a, const void *b)
 475 {
 476         timestamp_t a_date = ((const struct commit_list *)a)->item->date;
 477         timestamp_t b_date = ((const struct commit_list *)b)->item->date;
 478         if (a_date < b_date)
 479                 return 1;
 480         if (a_date > b_date)
 481                 return -1;
 482         return 0;
 483 }
 484
 485 static void *commit_list_get_next(const void *a)
 486 {
 487         return ((const struct commit_list *)a)->next;
 488 }
 489
 490 static void commit_list_set_next(void *a, void *next)
 491 {
 492         ((struct commit_list *)a)->next = next;
 493 }
 494
 495 void commit_list_sort_by_date(struct commit_list **list)
 496 {
 497         *list = llist_mergesort(*list, commit_list_get_next, commit_list_set_next,
 498                                 commit_list_compare_by_date);
 499 }
 500
 501 struct commit *pop_most_recent_commit(struct commit_list **list,
 502                                       unsigned int mark)
 503 {
 504         struct commit *ret = pop_commit(list);
 505         struct commit_list *parents = ret->parents;
 506
 507         while (parents) {
 508                 struct commit *commit = parents->item;
 509                 if (!parse_commit(commit) && !(commit->object.flags & mark)) {
 510                         commit->object.flags |= mark;
 511                         commit_list_insert_by_date(commit, list);
 512                 }
 513                 parents = parents->next;
 514         }
 515         return ret;
 516 }
 517
 518 static void clear_commit_marks_1(struct commit_list **plist,
 519                                  struct commit *commit, unsigned int mark)
 520 {
 521         while (commit) {
 522                 struct commit_list *parents;
 523
 524                 if (!(mark & commit->object.flags))
 525                         return;
 526
 527                 commit->object.flags &= ~mark;
 528
 529                 parents = commit->parents;
 530                 if (!parents)
 531                         return;
 532
 533                 while ((parents = parents->next))
 534                         commit_list_insert(parents->item, plist);
 535
 536                 commit = commit->parents->item;
 537         }
 538 }
 539
 540 void clear_commit_marks_many(int nr, struct commit **commit, unsigned int mark)
 541 {
 542         struct commit_list *list = NULL;
 543
 544         while (nr--) {
 545                 commit_list_insert(*commit, &list);
 546                 commit++;
 547         }
 548         while (list)
 549                 clear_commit_marks_1(&list, pop_commit(&list), mark);
 550 }
 551
 552 void clear_commit_marks(struct commit *commit, unsigned int mark)
 553 {
 554         clear_commit_marks_many(1, &commit, mark);
 555 }
 556
 557 void clear_commit_marks_for_object_array(struct object_array *a, unsigned mark)
 558 {
 559         struct object *object;
 560         struct commit *commit;
 561         unsigned int i;
 562
 563         for (i = 0; i < a->nr; i++) {
 564                 object = a->objects[i].item;
 565                 commit = lookup_commit_reference_gently(&object->oid, 1);
 566                 if (commit)
 567                         clear_commit_marks(commit, mark);
 568         }
 569 }
 570
 571 struct commit *pop_commit(struct commit_list **stack)
 572 {
 573         struct commit_list *top = *stack;
 574         struct commit *item = top ? top->item : NULL;
 575
 576         if (top) {
 577                 *stack = top->next;
 578                 free(top);
 579         }
 580         return item;
 581 }
 582
 583 /*
 584  * Topological sort support
 585  */
 586
 587 /* count number of children that have not been emitted */
 588 define_commit_slab(indegree_slab, int);
 589
 590 /* record author-date for each commit object */
 591 define_commit_slab(author_date_slab, unsigned long);
 592
 593 static void record_author_date(struct author_date_slab *author_date,
 594                                struct commit *commit)
 595 {
 596         const char *buffer = get_commit_buffer(commit, NULL);
 597         struct ident_split ident;
 598         const char *ident_line;
 599         size_t ident_len;
 600         char *date_end;
 601         timestamp_t date;
 602
 603         ident_line = find_commit_header(buffer, "author", &ident_len);
 604         if (!ident_line)
 605                 goto fail_exit; /* no author line */
 606         if (split_ident_line(&ident, ident_line, ident_len) ||
 607             !ident.date_begin || !ident.date_end)
 608                 goto fail_exit; /* malformed "author" line */
 609
 610         date = parse_timestamp(ident.date_begin, &date_end, 10);
 611         if (date_end != ident.date_end)
 612                 goto fail_exit; /* malformed date */
 613         *(author_date_slab_at(author_date, commit)) = date;
 614
 615 fail_exit:
 616         unuse_commit_buffer(commit, buffer);
 617 }
 618
 619 static int compare_commits_by_author_date(const void *a_, const void *b_,
 620                                           void *cb_data)
 621 {
 622         const struct commit *a = a_, *b = b_;
 623         struct author_date_slab *author_date = cb_data;
 624         timestamp_t a_date = *(author_date_slab_at(author_date, a));
 625         timestamp_t b_date = *(author_date_slab_at(author_date, b));
 626
 627         /* newer commits with larger date first */
 628         if (a_date < b_date)
 629                 return 1;
 630         else if (a_date > b_date)
 631                 return -1;
 632         return 0;
 633 }
 634
 635 int compare_commits_by_commit_date(const void *a_, const void *b_, void *unused)
 636 {
 637         const struct commit *a = a_, *b = b_;
 638         /* newer commits with larger date first */
 639         if (a->date < b->date)
 640                 return 1;
 641         else if (a->date > b->date)
 642                 return -1;
 643         return 0;
 644 }
 645
 646 /*
 647  * Performs an in-place topological sort on the list supplied.
 648  */
 649 void sort_in_topological_order(struct commit_list **list, enum rev_sort_order sort_order)
 650 {
 651         struct commit_list *next, *orig = *list;
 652         struct commit_list **pptr;
 653         struct indegree_slab indegree;
 654         struct prio_queue queue;
 655         struct commit *commit;
 656         struct author_date_slab author_date;
 657
 658         if (!orig)
 659                 return;
 660         *list = NULL;
 661
 662         init_indegree_slab(&indegree);
 663         memset(&queue, '\0', sizeof(queue));
 664
 665         switch (sort_order) {
 666         default: /* REV_SORT_IN_GRAPH_ORDER */
 667                 queue.compare = NULL;
 668                 break;
 669         case REV_SORT_BY_COMMIT_DATE:
 670                 queue.compare = compare_commits_by_commit_date;
 671                 break;
 672         case REV_SORT_BY_AUTHOR_DATE:
 673                 init_author_date_slab(&author_date);
 674                 queue.compare = compare_commits_by_author_date;
 675                 queue.cb_data = &author_date;
 676                 break;
 677         }
 678
 679         /* Mark them and clear the indegree */
 680         for (next = orig; next; next = next->next) {
 681                 struct commit *commit = next->item;
 682                 *(indegree_slab_at(&indegree, commit)) = 1;
 683                 /* also record the author dates, if needed */
 684                 if (sort_order == REV_SORT_BY_AUTHOR_DATE)
 685                         record_author_date(&author_date, commit);
 686         }
 687
 688         /* update the indegree */
 689         for (next = orig; next; next = next->next) {
 690                 struct commit_list *parents = next->item->parents;
 691                 while (parents) {
 692                         struct commit *parent = parents->item;
 693                         int *pi = indegree_slab_at(&indegree, parent);
 694
 695                         if (*pi)
 696                                 (*pi)++;
 697                         parents = parents->next;
 698                 }
 699         }
 700
 701         /*
 702          * find the tips
 703          *
 704          * tips are nodes not reachable from any other node in the list
 705          *
 706          * the tips serve as a starting set for the work queue.
 707          */
 708         for (next = orig; next; next = next->next) {
 709                 struct commit *commit = next->item;
 710
 711                 if (*(indegree_slab_at(&indegree, commit)) == 1)
 712                         prio_queue_put(&queue, commit);
 713         }
 714
 715         /*
 716          * This is unfortunate; the initial tips need to be shown
 717          * in the order given from the revision traversal machinery.
 718          */
 719         if (sort_order == REV_SORT_IN_GRAPH_ORDER)
 720                 prio_queue_reverse(&queue);
 721
 722         /* We no longer need the commit list */
 723         free_commit_list(orig);
 724
 725         pptr = list;
 726         *list = NULL;
 727         while ((commit = prio_queue_get(&queue)) != NULL) {
 728                 struct commit_list *parents;
 729
 730                 for (parents = commit->parents; parents ; parents = parents->next) {
 731                         struct commit *parent = parents->item;
 732                         int *pi = indegree_slab_at(&indegree, parent);
 733
 734                         if (!*pi)
 735                                 continue;
 736
 737                         /*
 738                          * parents are only enqueued for emission
 739                          * when all their children have been emitted thereby
 740                          * guaranteeing topological order.
 741                          */
 742                         if (--(*pi) == 1)
 743                                 prio_queue_put(&queue, parent);
 744                 }
 745                 /*
 746                  * all children of commit have already been
 747                  * emitted. we can emit it now.
 748                  */
 749                 *(indegree_slab_at(&indegree, commit)) = 0;
 750
 751                 pptr = &commit_list_insert(commit, pptr)->next;
 752         }
 753
 754         clear_indegree_slab(&indegree);
 755         clear_prio_queue(&queue);
 756         if (sort_order == REV_SORT_BY_AUTHOR_DATE)
 757                 clear_author_date_slab(&author_date);
 758 }
 759
 760 /* merge-base stuff */
 761
 762 /* Remember to update object flag allocation in object.h */
 763 #define PARENT1         (1u<<16)
 764 #define PARENT2         (1u<<17)
 765 #define STALE           (1u<<18)
 766 #define RESULT          (1u<<19)
 767
 768 static const unsigned all_flags = (PARENT1 | PARENT2 | STALE | RESULT);
 769
 770 static int queue_has_nonstale(struct prio_queue *queue)
 771 {
 772         int i;
 773         for (i = 0; i < queue->nr; i++) {
 774                 struct commit *commit = queue->array[i].data;
 775                 if (!(commit->object.flags & STALE))
 776                         return 1;
 777         }
 778         return 0;
 779 }
 780
 781 /* all input commits in one and twos[] must have been parsed! */
 782 static struct commit_list *paint_down_to_common(struct commit *one, int n, struct commit **twos)
 783 {
 784         struct prio_queue queue = { compare_commits_by_commit_date };
 785         struct commit_list *result = NULL;
 786         int i;
 787
 788         one->object.flags |= PARENT1;
 789         if (!n) {
 790                 commit_list_append(one, &result);
 791                 return result;
 792         }
 793         prio_queue_put(&queue, one);
 794
 795         for (i = 0; i < n; i++) {
 796                 twos[i]->object.flags |= PARENT2;
 797                 prio_queue_put(&queue, twos[i]);
 798         }
 799
 800         while (queue_has_nonstale(&queue)) {
 801                 struct commit *commit = prio_queue_get(&queue);
 802                 struct commit_list *parents;
 803                 int flags;
 804
 805                 flags = commit->object.flags & (PARENT1 | PARENT2 | STALE);
 806                 if (flags == (PARENT1 | PARENT2)) {
 807                         if (!(commit->object.flags & RESULT)) {
 808                                 commit->object.flags |= RESULT;
 809                                 commit_list_insert_by_date(commit, &result);
 810                         }
 811                         /* Mark parents of a found merge stale */
 812                         flags |= STALE;
 813                 }
 814                 parents = commit->parents;
 815                 while (parents) {
 816                         struct commit *p = parents->item;
 817                         parents = parents->next;
 818                         if ((p->object.flags & flags) == flags)
 819                                 continue;
 820                         if (parse_commit(p))
 821                                 return NULL;
 822                         p->object.flags |= flags;
 823                         prio_queue_put(&queue, p);
 824                 }
 825         }
 826
 827         clear_prio_queue(&queue);
 828         return result;
 829 }
 830
 831 static struct commit_list *merge_bases_many(struct commit *one, int n, struct commit **twos)
 832 {
 833         struct commit_list *list = NULL;
 834         struct commit_list *result = NULL;
 835         int i;
 836
 837         for (i = 0; i < n; i++) {
 838                 if (one == twos[i])
 839                         /*
 840                          * We do not mark this even with RESULT so we do not
 841                          * have to clean it up.
 842                          */
 843                         return commit_list_insert(one, &result);
 844         }
 845
 846         if (parse_commit(one))
 847                 return NULL;
 848         for (i = 0; i < n; i++) {
 849                 if (parse_commit(twos[i]))
 850                         return NULL;
 851         }
 852
 853         list = paint_down_to_common(one, n, twos);
 854
 855         while (list) {
 856                 struct commit *commit = pop_commit(&list);
 857                 if (!(commit->object.flags & STALE))
 858                         commit_list_insert_by_date(commit, &result);
 859         }
 860         return result;
 861 }
 862
 863 struct commit_list *get_octopus_merge_bases(struct commit_list *in)
 864 {
 865         struct commit_list *i, *j, *k, *ret = NULL;
 866
 867         if (!in)
 868                 return ret;
 869
 870         commit_list_insert(in->item, &ret);
 871
 872         for (i = in->next; i; i = i->next) {
 873                 struct commit_list *new = NULL, *end = NULL;
 874
 875                 for (j = ret; j; j = j->next) {
 876                         struct commit_list *bases;
 877                         bases = get_merge_bases(i->item, j->item);
 878                         if (!new)
 879                                 new = bases;
 880                         else
 881                                 end->next = bases;
 882                         for (k = bases; k; k = k->next)
 883                                 end = k;
 884                 }
 885                 ret = new;
 886         }
 887         return ret;
 888 }
 889
 890 static int remove_redundant(struct commit **array, int cnt)
 891 {
 892         /*
 893          * Some commit in the array may be an ancestor of
 894          * another commit.  Move such commit to the end of
 895          * the array, and return the number of commits that
 896          * are independent from each other.
 897          */
 898         struct commit **work;
 899         unsigned char *redundant;
 900         int *filled_index;
 901         int i, j, filled;
 902
 903         work = xcalloc(cnt, sizeof(*work));
 904         redundant = xcalloc(cnt, 1);
 905         ALLOC_ARRAY(filled_index, cnt - 1);
 906
 907         for (i = 0; i < cnt; i++)
 908                 parse_commit(array[i]);
 909         for (i = 0; i < cnt; i++) {
 910                 struct commit_list *common;
 911
 912                 if (redundant[i])
 913                         continue;
 914                 for (j = filled = 0; j < cnt; j++) {
 915                         if (i == j || redundant[j])
 916                                 continue;
 917                         filled_index[filled] = j;
 918                         work[filled++] = array[j];
 919                 }
 920                 common = paint_down_to_common(array[i], filled, work);
 921                 if (array[i]->object.flags & PARENT2)
 922                         redundant[i] = 1;
 923                 for (j = 0; j < filled; j++)
 924                         if (work[j]->object.flags & PARENT1)
 925                                 redundant[filled_index[j]] = 1;
 926                 clear_commit_marks(array[i], all_flags);
 927                 for (j = 0; j < filled; j++)
 928                         clear_commit_marks(work[j], all_flags);
 929                 free_commit_list(common);
 930         }
 931
 932         /* Now collect the result */
 933         COPY_ARRAY(work, array, cnt);
 934         for (i = filled = 0; i < cnt; i++)
 935                 if (!redundant[i])
 936                         array[filled++] = work[i];
 937         for (j = filled, i = 0; i < cnt; i++)
 938                 if (redundant[i])
 939                         array[j++] = work[i];
 940         free(work);
 941         free(redundant);
 942         free(filled_index);
 943         return filled;
 944 }
 945
 946 static struct commit_list *get_merge_bases_many_0(struct commit *one,
 947                                                   int n,
 948                                                   struct commit **twos,
 949                                                   int cleanup)
 950 {
 951         struct commit_list *list;
 952         struct commit **rslt;
 953         struct commit_list *result;
 954         int cnt, i;
 955
 956         result = merge_bases_many(one, n, twos);
 957         for (i = 0; i < n; i++) {
 958                 if (one == twos[i])
 959                         return result;
 960         }
 961         if (!result || !result->next) {
 962                 if (cleanup) {
 963                         clear_commit_marks(one, all_flags);
 964                         clear_commit_marks_many(n, twos, all_flags);
 965                 }
 966                 return result;
 967         }
 968
 969         /* There are more than one */
 970         cnt = commit_list_count(result);
 971         rslt = xcalloc(cnt, sizeof(*rslt));
 972         for (list = result, i = 0; list; list = list->next)
 973                 rslt[i++] = list->item;
 974         free_commit_list(result);
 975
 976         clear_commit_marks(one, all_flags);
 977         clear_commit_marks_many(n, twos, all_flags);
 978
 979         cnt = remove_redundant(rslt, cnt);
 980         result = NULL;
 981         for (i = 0; i < cnt; i++)
 982                 commit_list_insert_by_date(rslt[i], &result);
 983         free(rslt);
 984         return result;
 985 }
 986
 987 struct commit_list *get_merge_bases_many(struct commit *one,
 988                                          int n,
 989                                          struct commit **twos)
 990 {
 991         return get_merge_bases_many_0(one, n, twos, 1);
 992 }
 993
 994 struct commit_list *get_merge_bases_many_dirty(struct commit *one,
 995                                                int n,
 996                                                struct commit **twos)
 997 {
 998         return get_merge_bases_many_0(one, n, twos, 0);
 999 }
1000
1001 struct commit_list *get_merge_bases(struct commit *one, struct commit *two)
1002 {
1003         return get_merge_bases_many_0(one, 1, &two, 1);
1004 }
1005
1006 /*
1007  * Is "commit" a descendant of one of the elements on the "with_commit" list?
1008  */
1009 int is_descendant_of(struct commit *commit, struct commit_list *with_commit)
1010 {
1011         if (!with_commit)
1012                 return 1;
1013         while (with_commit) {
1014                 struct commit *other;
1015
1016                 other = with_commit->item;
1017                 with_commit = with_commit->next;
1018                 if (in_merge_bases(other, commit))
1019                         return 1;
1020         }
1021         return 0;
1022 }
1023
1024 /*
1025  * Is "commit" an ancestor of one of the "references"?
1026  */
1027 int in_merge_bases_many(struct commit *commit, int nr_reference, struct commit **reference)
1028 {
1029         struct commit_list *bases;
1030         int ret = 0, i;
1031
1032         if (parse_commit(commit))
1033                 return ret;
1034         for (i = 0; i < nr_reference; i++)
1035                 if (parse_commit(reference[i]))
1036                         return ret;
1037
1038         bases = paint_down_to_common(commit, nr_reference, reference);
1039         if (commit->object.flags & PARENT2)
1040                 ret = 1;
1041         clear_commit_marks(commit, all_flags);
1042         clear_commit_marks_many(nr_reference, reference, all_flags);
1043         free_commit_list(bases);
1044         return ret;
1045 }
1046
1047 /*
1048  * Is "commit" an ancestor of (i.e. reachable from) the "reference"?
1049  */
1050 int in_merge_bases(struct commit *commit, struct commit *reference)
1051 {
1052         return in_merge_bases_many(commit, 1, &reference);
1053 }
1054
1055 struct commit_list *reduce_heads(struct commit_list *heads)
1056 {
1057         struct commit_list *p;
1058         struct commit_list *result = NULL, **tail = &result;
1059         struct commit **array;
1060         int num_head, i;
1061
1062         if (!heads)
1063                 return NULL;
1064
1065         /* Uniquify */
1066         for (p = heads; p; p = p->next)
1067                 p->item->object.flags &= ~STALE;
1068         for (p = heads, num_head = 0; p; p = p->next) {
1069                 if (p->item->object.flags & STALE)
1070                         continue;
1071                 p->item->object.flags |= STALE;
1072                 num_head++;
1073         }
1074         array = xcalloc(num_head, sizeof(*array));
1075         for (p = heads, i = 0; p; p = p->next) {
1076                 if (p->item->object.flags & STALE) {
1077                         array[i++] = p->item;
1078                         p->item->object.flags &= ~STALE;
1079                 }
1080         }
1081         num_head = remove_redundant(array, num_head);
1082         for (i = 0; i < num_head; i++)
1083                 tail = &commit_list_insert(array[i], tail)->next;
1084         return result;
1085 }
1086
1087 static const char gpg_sig_header[] = "gpgsig";
1088 static const int gpg_sig_header_len = sizeof(gpg_sig_header) - 1;
1089
1090 static int do_sign_commit(struct strbuf *buf, const char *keyid)
1091 {
1092         struct strbuf sig = STRBUF_INIT;
1093         int inspos, copypos;
1094         const char *eoh;
1095
1096         /* find the end of the header */
1097         eoh = strstr(buf->buf, "\n\n");
1098         if (!eoh)
1099                 inspos = buf->len;
1100         else
1101                 inspos = eoh - buf->buf + 1;
1102
1103         if (!keyid || !*keyid)
1104                 keyid = get_signing_key();
1105         if (sign_buffer(buf, &sig, keyid)) {
1106                 strbuf_release(&sig);
1107                 return -1;
1108         }
1109
1110         for (copypos = 0; sig.buf[copypos]; ) {
1111                 const char *bol = sig.buf + copypos;
1112                 const char *eol = strchrnul(bol, '\n');
1113                 int len = (eol - bol) + !!*eol;
1114
1115                 if (!copypos) {
1116                         strbuf_insert(buf, inspos, gpg_sig_header, gpg_sig_header_len);
1117                         inspos += gpg_sig_header_len;
1118                 }
1119                 strbuf_insert(buf, inspos++, " ", 1);
1120                 strbuf_insert(buf, inspos, bol, len);
1121                 inspos += len;
1122                 copypos += len;
1123         }
1124         strbuf_release(&sig);
1125         return 0;
1126 }
1127
1128 int parse_signed_commit(const struct commit *commit,
1129                         struct strbuf *payload, struct strbuf *signature)
1130 {
1131
1132         unsigned long size;
1133         const char *buffer = get_commit_buffer(commit, &size);
1134         int in_signature, saw_signature = -1;
1135         const char *line, *tail;
1136
1137         line = buffer;
1138         tail = buffer + size;
1139         in_signature = 0;
1140         saw_signature = 0;
1141         while (line < tail) {
1142                 const char *sig = NULL;
1143                 const char *next = memchr(line, '\n', tail - line);
1144
1145                 next = next ? next + 1 : tail;
1146                 if (in_signature && line[0] == ' ')
1147                         sig = line + 1;
1148                 else if (starts_with(line, gpg_sig_header) &&
1149                          line[gpg_sig_header_len] == ' ')
1150                         sig = line + gpg_sig_header_len + 1;
1151                 if (sig) {
1152                         strbuf_add(signature, sig, next - sig);
1153                         saw_signature = 1;
1154                         in_signature = 1;
1155                 } else {
1156                         if (*line == '\n')
1157                                 /* dump the whole remainder of the buffer */
1158                                 next = tail;
1159                         strbuf_add(payload, line, next - line);
1160                         in_signature = 0;
1161                 }
1162                 line = next;
1163         }
1164         unuse_commit_buffer(commit, buffer);
1165         return saw_signature;
1166 }
1167
1168 int remove_signature(struct strbuf *buf)
1169 {
1170         const char *line = buf->buf;
1171         const char *tail = buf->buf + buf->len;
1172         int in_signature = 0;
1173         const char *sig_start = NULL;
1174         const char *sig_end = NULL;
1175
1176         while (line < tail) {
1177                 const char *next = memchr(line, '\n', tail - line);
1178                 next = next ? next + 1 : tail;
1179
1180                 if (in_signature && line[0] == ' ')
1181                         sig_end = next;
1182                 else if (starts_with(line, gpg_sig_header) &&
1183                          line[gpg_sig_header_len] == ' ') {
1184                         sig_start = line;
1185                         sig_end = next;
1186                         in_signature = 1;
1187                 } else {
1188                         if (*line == '\n')
1189                                 /* dump the whole remainder of the buffer */
1190                                 next = tail;
1191                         in_signature = 0;
1192                 }
1193                 line = next;
1194         }
1195
1196         if (sig_start)
1197                 strbuf_remove(buf, sig_start - buf->buf, sig_end - sig_start);
1198
1199         return sig_start != NULL;
1200 }
1201
1202 static void handle_signed_tag(struct commit *parent, struct commit_extra_header ***tail)
1203 {
1204         struct merge_remote_desc *desc;
1205         struct commit_extra_header *mergetag;
1206         char *buf;
1207         unsigned long size, len;
1208         enum object_type type;
1209
1210         desc = merge_remote_util(parent);
1211         if (!desc || !desc->obj)
1212                 return;
1213         buf = read_sha1_file(desc->obj->oid.hash, &type, &size);
1214         if (!buf || type != OBJ_TAG)
1215                 goto free_return;
1216         len = parse_signature(buf, size);
1217         if (size == len)
1218                 goto free_return;
1219         /*
1220          * We could verify this signature and either omit the tag when
1221          * it does not validate, but the integrator may not have the
1222          * public key of the signer of the tag he is merging, while a
1223          * later auditor may have it while auditing, so let's not run
1224          * verify-signed-buffer here for now...
1225          *
1226          * if (verify_signed_buffer(buf, len, buf + len, size - len, ...))
1227          *      warn("warning: signed tag unverified.");
1228          */
1229         mergetag = xcalloc(1, sizeof(*mergetag));
1230         mergetag->key = xstrdup("mergetag");
1231         mergetag->value = buf;
1232         mergetag->len = size;
1233
1234         **tail = mergetag;
1235         *tail = &mergetag->next;
1236         return;
1237
1238 free_return:
1239         free(buf);
1240 }
1241
1242 int check_commit_signature(const struct commit *commit, struct signature_check *sigc)
1243 {
1244         struct strbuf payload = STRBUF_INIT;
1245         struct strbuf signature = STRBUF_INIT;
1246         int ret = 1;
1247
1248         sigc->result = 'N';
1249
1250         if (parse_signed_commit(commit, &payload, &signature) <= 0)
1251                 goto out;
1252         ret = check_signature(payload.buf, payload.len, signature.buf,
1253                 signature.len, sigc);
1254
1255  out:
1256         strbuf_release(&payload);
1257         strbuf_release(&signature);
1258
1259         return ret;
1260 }
1261
1262
1263
1264 void append_merge_tag_headers(struct commit_list *parents,
1265                               struct commit_extra_header ***tail)
1266 {
1267         while (parents) {
1268                 struct commit *parent = parents->item;
1269                 handle_signed_tag(parent, tail);
1270                 parents = parents->next;
1271         }
1272 }
1273
1274 static void add_extra_header(struct strbuf *buffer,
1275                              struct commit_extra_header *extra)
1276 {
1277         strbuf_addstr(buffer, extra->key);
1278         if (extra->len)
1279                 strbuf_add_lines(buffer, " ", extra->value, extra->len);
1280         else
1281                 strbuf_addch(buffer, '\n');
1282 }
1283
1284 struct commit_extra_header *read_commit_extra_headers(struct commit *commit,
1285                                                       const char **exclude)
1286 {
1287         struct commit_extra_header *extra = NULL;
1288         unsigned long size;
1289         const char *buffer = get_commit_buffer(commit, &size);
1290         extra = read_commit_extra_header_lines(buffer, size, exclude);
1291         unuse_commit_buffer(commit, buffer);
1292         return extra;
1293 }
1294
1295 void for_each_mergetag(each_mergetag_fn fn, struct commit *commit, void *data)
1296 {
1297         struct commit_extra_header *extra, *to_free;
1298
1299         to_free = read_commit_extra_headers(commit, NULL);
1300         for (extra = to_free; extra; extra = extra->next) {
1301                 if (strcmp(extra->key, "mergetag"))
1302                         continue; /* not a merge tag */
1303                 fn(commit, extra, data);
1304         }
1305         free_commit_extra_headers(to_free);
1306 }
1307
1308 static inline int standard_header_field(const char *field, size_t len)
1309 {
1310         return ((len == 4 && !memcmp(field, "tree", 4)) ||
1311                 (len == 6 && !memcmp(field, "parent", 6)) ||
1312                 (len == 6 && !memcmp(field, "author", 6)) ||
1313                 (len == 9 && !memcmp(field, "committer", 9)) ||
1314                 (len == 8 && !memcmp(field, "encoding", 8)));
1315 }
1316
1317 static int excluded_header_field(const char *field, size_t len, const char **exclude)
1318 {
1319         if (!exclude)
1320                 return 0;
1321
1322         while (*exclude) {
1323                 size_t xlen = strlen(*exclude);
1324                 if (len == xlen && !memcmp(field, *exclude, xlen))
1325                         return 1;
1326                 exclude++;
1327         }
1328         return 0;
1329 }
1330
1331 static struct commit_extra_header *read_commit_extra_header_lines(
1332         const char *buffer, size_t size,
1333         const char **exclude)
1334 {
1335         struct commit_extra_header *extra = NULL, **tail = &extra, *it = NULL;
1336         const char *line, *next, *eof, *eob;
1337         struct strbuf buf = STRBUF_INIT;
1338
1339         for (line = buffer, eob = line + size;
1340              line < eob && *line != '\n';
1341              line = next) {
1342                 next = memchr(line, '\n', eob - line);
1343                 next = next ? next + 1 : eob;
1344                 if (*line == ' ') {
1345                         /* continuation */
1346                         if (it)
1347                                 strbuf_add(&buf, line + 1, next - (line + 1));
1348                         continue;
1349                 }
1350                 if (it)
1351                         it->value = strbuf_detach(&buf, &it->len);
1352                 strbuf_reset(&buf);
1353                 it = NULL;
1354
1355                 eof = memchr(line, ' ', next - line);
1356                 if (!eof)
1357                         eof = next;
1358                 else if (standard_header_field(line, eof - line) ||
1359                          excluded_header_field(line, eof - line, exclude))
1360                         continue;
1361
1362                 it = xcalloc(1, sizeof(*it));
1363                 it->key = xmemdupz(line, eof-line);
1364                 *tail = it;
1365                 tail = &it->next;
1366                 if (eof + 1 < next)
1367                         strbuf_add(&buf, eof + 1, next - (eof + 1));
1368         }
1369         if (it)
1370                 it->value = strbuf_detach(&buf, &it->len);
1371         return extra;
1372 }
1373
1374 void free_commit_extra_headers(struct commit_extra_header *extra)
1375 {
1376         while (extra) {
1377                 struct commit_extra_header *next = extra->next;
1378                 free(extra->key);
1379                 free(extra->value);
1380                 free(extra);
1381                 extra = next;
1382         }
1383 }
1384
1385 int commit_tree(const char *msg, size_t msg_len,
1386                 const unsigned char *tree,
1387                 struct commit_list *parents, unsigned char *ret,
1388                 const char *author, const char *sign_commit)
1389 {
1390         struct commit_extra_header *extra = NULL, **tail = &extra;
1391         int result;
1392
1393         append_merge_tag_headers(parents, &tail);
1394         result = commit_tree_extended(msg, msg_len, tree, parents, ret,
1395                                       author, sign_commit, extra);
1396         free_commit_extra_headers(extra);
1397         return result;
1398 }
1399
1400 static int find_invalid_utf8(const char *buf, int len)
1401 {
1402         int offset = 0;
1403         static const unsigned int max_codepoint[] = {
1404                 0x7f, 0x7ff, 0xffff, 0x10ffff
1405         };
1406
1407         while (len) {
1408                 unsigned char c = *buf++;
1409                 int bytes, bad_offset;
1410                 unsigned int codepoint;
1411                 unsigned int min_val, max_val;
1412
1413                 len--;
1414                 offset++;
1415
1416                 /* Simple US-ASCII? No worries. */
1417                 if (c < 0x80)
1418                         continue;
1419
1420                 bad_offset = offset-1;
1421
1422                 /*
1423                  * Count how many more high bits set: that's how
1424                  * many more bytes this sequence should have.
1425                  */
1426                 bytes = 0;
1427                 while (c & 0x40) {
1428                         c <<= 1;
1429                         bytes++;
1430                 }
1431
1432                 /*
1433                  * Must be between 1 and 3 more bytes.  Longer sequences result in
1434                  * codepoints beyond U+10FFFF, which are guaranteed never to exist.
1435                  */
1436                 if (bytes < 1 || 3 < bytes)
1437                         return bad_offset;
1438
1439                 /* Do we *have* that many bytes? */
1440                 if (len < bytes)
1441                         return bad_offset;
1442
1443                 /*
1444                  * Place the encoded bits at the bottom of the value and compute the
1445                  * valid range.
1446                  */
1447                 codepoint = (c & 0x7f) >> bytes;
1448                 min_val = max_codepoint[bytes-1] + 1;
1449                 max_val = max_codepoint[bytes];
1450
1451                 offset += bytes;
1452                 len -= bytes;
1453
1454                 /* And verify that they are good continuation bytes */
1455                 do {
1456                         codepoint <<= 6;
1457                         codepoint |= *buf & 0x3f;
1458                         if ((*buf++ & 0xc0) != 0x80)
1459                                 return bad_offset;
1460                 } while (--bytes);
1461
1462                 /* Reject codepoints that are out of range for the sequence length. */
1463                 if (codepoint < min_val || codepoint > max_val)
1464                         return bad_offset;
1465                 /* Surrogates are only for UTF-16 and cannot be encoded in UTF-8. */
1466                 if ((codepoint & 0x1ff800) == 0xd800)
1467                         return bad_offset;
1468                 /* U+xxFFFE and U+xxFFFF are guaranteed non-characters. */
1469                 if ((codepoint & 0xfffe) == 0xfffe)
1470                         return bad_offset;
1471                 /* So are anything in the range U+FDD0..U+FDEF. */
1472                 if (codepoint >= 0xfdd0 && codepoint <= 0xfdef)
1473                         return bad_offset;
1474         }
1475         return -1;
1476 }
1477
1478 /*
1479  * This verifies that the buffer is in proper utf8 format.
1480  *
1481  * If it isn't, it assumes any non-utf8 characters are Latin1,
1482  * and does the conversion.
1483  */
1484 static int verify_utf8(struct strbuf *buf)
1485 {
1486         int ok = 1;
1487         long pos = 0;
1488
1489         for (;;) {
1490                 int bad;
1491                 unsigned char c;
1492                 unsigned char replace[2];
1493
1494                 bad = find_invalid_utf8(buf->buf + pos, buf->len - pos);
1495                 if (bad < 0)
1496                         return ok;
1497                 pos += bad;
1498                 ok = 0;
1499                 c = buf->buf[pos];
1500                 strbuf_remove(buf, pos, 1);
1501
1502                 /* We know 'c' must be in the range 128-255 */
1503                 replace[0] = 0xc0 + (c >> 6);
1504                 replace[1] = 0x80 + (c & 0x3f);
1505                 strbuf_insert(buf, pos, replace, 2);
1506                 pos += 2;
1507         }
1508 }
1509
1510 static const char commit_utf8_warn[] =
1511 N_("Warning: commit message did not conform to UTF-8.\n"
1512    "You may want to amend it after fixing the message, or set the config\n"
1513    "variable i18n.commitencoding to the encoding your project uses.\n");
1514
1515 int commit_tree_extended(const char *msg, size_t msg_len,
1516                          const unsigned char *tree,
1517                          struct commit_list *parents, unsigned char *ret,
1518                          const char *author, const char *sign_commit,
1519                          struct commit_extra_header *extra)
1520 {
1521         int result;
1522         int encoding_is_utf8;
1523         struct strbuf buffer;
1524
1525         assert_sha1_type(tree, OBJ_TREE);
1526
1527         if (memchr(msg, '\0', msg_len))
1528                 return error("a NUL byte in commit log message not allowed.");
1529
1530         /* Not having i18n.commitencoding is the same as having utf-8 */
1531         encoding_is_utf8 = is_encoding_utf8(git_commit_encoding);
1532
1533         strbuf_init(&buffer, 8192); /* should avoid reallocs for the headers */
1534         strbuf_addf(&buffer, "tree %s\n", sha1_to_hex(tree));
1535
1536         /*
1537          * NOTE! This ordering means that the same exact tree merged with a
1538          * different order of parents will be a _different_ changeset even
1539          * if everything else stays the same.
1540          */
1541         while (parents) {
1542                 struct commit *parent = pop_commit(&parents);
1543                 strbuf_addf(&buffer, "parent %s\n",
1544                             oid_to_hex(&parent->object.oid));
1545         }
1546
1547         /* Person/date information */
1548         if (!author)
1549                 author = git_author_info(IDENT_STRICT);
1550         strbuf_addf(&buffer, "author %s\n", author);
1551         strbuf_addf(&buffer, "committer %s\n", git_committer_info(IDENT_STRICT));
1552         if (!encoding_is_utf8)
1553                 strbuf_addf(&buffer, "encoding %s\n", git_commit_encoding);
1554
1555         while (extra) {
1556                 add_extra_header(&buffer, extra);
1557                 extra = extra->next;
1558         }
1559         strbuf_addch(&buffer, '\n');
1560
1561         /* And add the comment */
1562         strbuf_add(&buffer, msg, msg_len);
1563
1564         /* And check the encoding */
1565         if (encoding_is_utf8 && !verify_utf8(&buffer))
1566                 fprintf(stderr, _(commit_utf8_warn));
1567
1568         if (sign_commit && do_sign_commit(&buffer, sign_commit))
1569                 return -1;
1570
1571         result = write_sha1_file(buffer.buf, buffer.len, commit_type, ret);
1572         strbuf_release(&buffer);
1573         return result;
1574 }
1575
1576 void set_merge_remote_desc(struct commit *commit,
1577                            const char *name, struct object *obj)
1578 {
1579         struct merge_remote_desc *desc;
1580         FLEX_ALLOC_STR(desc, name, name);
1581         desc->obj = obj;
1582         commit->util = desc;
1583 }
1584
1585 struct commit *get_merge_parent(const char *name)
1586 {
1587         struct object *obj;
1588         struct commit *commit;
1589         struct object_id oid;
1590         if (get_sha1(name, oid.hash))
1591                 return NULL;
1592         obj = parse_object(&oid);
1593         commit = (struct commit *)peel_to_type(name, 0, obj, OBJ_COMMIT);
1594         if (commit && !commit->util)
1595                 set_merge_remote_desc(commit, name, obj);
1596         return commit;
1597 }
1598
1599 /*
1600  * Append a commit to the end of the commit_list.
1601  *
1602  * next starts by pointing to the variable that holds the head of an
1603  * empty commit_list, and is updated to point to the "next" field of
1604  * the last item on the list as new commits are appended.
1605  *
1606  * Usage example:
1607  *
1608  *     struct commit_list *list;
1609  *     struct commit_list **next = &list;
1610  *
1611  *     next = commit_list_append(c1, next);
1612  *     next = commit_list_append(c2, next);
1613  *     assert(commit_list_count(list) == 2);
1614  *     return list;
1615  */
1616 struct commit_list **commit_list_append(struct commit *commit,
1617                                         struct commit_list **next)
1618 {
1619         struct commit_list *new = xmalloc(sizeof(struct commit_list));
1620         new->item = commit;
1621         *next = new;
1622         new->next = NULL;
1623         return &new->next;
1624 }
1625
1626 const char *find_commit_header(const char *msg, const char *key, size_t *out_len)
1627 {
1628         int key_len = strlen(key);
1629         const char *line = msg;
1630
1631         while (line) {
1632                 const char *eol = strchrnul(line, '\n');
1633
1634                 if (line == eol)
1635                         return NULL;
1636
1637                 if (eol - line > key_len &&
1638                     !strncmp(line, key, key_len) &&
1639                     line[key_len] == ' ') {
1640                         *out_len = eol - line - key_len - 1;
1641                         return line + key_len + 1;
1642                 }
1643                 line = *eol ? eol + 1 : NULL;
1644         }
1645         return NULL;
1646 }
1647
1648 /*
1649  * Inspect the given string and determine the true "end" of the log message, in
1650  * order to find where to put a new Signed-off-by: line.  Ignored are
1651  * trailing comment lines and blank lines, and also the traditional
1652  * "Conflicts:" block that is not commented out, so that we can use
1653  * "git commit -s --amend" on an existing commit that forgot to remove
1654  * it.
1655  *
1656  * Returns the number of bytes from the tail to ignore, to be fed as
1657  * the second parameter to append_signoff().
1658  */
1659 int ignore_non_trailer(const char *buf, size_t len)
1660 {
1661         int boc = 0;
1662         int bol = 0;
1663         int in_old_conflicts_block = 0;
1664
1665         while (bol < len) {
1666                 const char *next_line = memchr(buf + bol, '\n', len - bol);
1667
1668                 if (!next_line)
1669                         next_line = buf + len;
1670                 else
1671                         next_line++;
1672
1673                 if (buf[bol] == comment_line_char || buf[bol] == '\n') {
1674                         /* is this the first of the run of comments? */
1675                         if (!boc)
1676                                 boc = bol;
1677                         /* otherwise, it is just continuing */
1678                 } else if (starts_with(buf + bol, "Conflicts:\n")) {
1679                         in_old_conflicts_block = 1;
1680                         if (!boc)
1681                                 boc = bol;
1682                 } else if (in_old_conflicts_block && buf[bol] == '\t') {
1683                         ; /* a pathname in the conflicts block */
1684                 } else if (boc) {
1685                         /* the previous was not trailing comment */
1686                         boc = 0;
1687                         in_old_conflicts_block = 0;
1688                 }
1689                 bol = next_line - buf;
1690         }
1691         return boc ? len - boc : 0;
1692 }