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 #include "wt-status.h"
  15
  16 static struct commit_extra_header *read_commit_extra_header_lines(const char *buf, size_t len, const char **);
  17
  18 int save_commit_buffer = 1;
  19
  20 const char *commit_type = "commit";
  21
  22 struct commit *lookup_commit_reference_gently(const struct object_id *oid,
  23                                               int quiet)
  24 {
  25         struct object *obj = deref_tag(parse_object(oid), NULL, 0);
  26
  27         if (!obj)
  28                 return NULL;
  29         return object_as_type(obj, OBJ_COMMIT, quiet);
  30 }
  31
  32 struct commit *lookup_commit_reference(const struct object_id *oid)
  33 {
  34         return lookup_commit_reference_gently(oid, 0);
  35 }
  36
  37 struct commit *lookup_commit_or_die(const struct object_id *oid, const char *ref_name)
  38 {
  39         struct commit *c = lookup_commit_reference(oid);
  40         if (!c)
  41                 die(_("could not parse %s"), ref_name);
  42         if (oidcmp(oid, &c->object.oid)) {
  43                 warning(_("%s %s is not a commit!"),
  44                         ref_name, oid_to_hex(oid));
  45         }
  46         return c;
  47 }
  48
  49 struct commit *lookup_commit(const struct object_id *oid)
  50 {
  51         struct object *obj = lookup_object(oid->hash);
  52         if (!obj)
  53                 return create_object(oid->hash, alloc_commit_node());
  54         return object_as_type(obj, OBJ_COMMIT, 0);
  55 }
  56
  57 struct commit *lookup_commit_reference_by_name(const char *name)
  58 {
  59         struct object_id oid;
  60         struct commit *commit;
  61
  62         if (get_sha1_committish(name, oid.hash))
  63                 return NULL;
  64         commit = lookup_commit_reference(&oid);
  65         if (parse_commit(commit))
  66                 return NULL;
  67         return commit;
  68 }
  69
  70 static timestamp_t parse_commit_date(const char *buf, const char *tail)
  71 {
  72         const char *dateptr;
  73
  74         if (buf + 6 >= tail)
  75                 return 0;
  76         if (memcmp(buf, "author", 6))
  77                 return 0;
  78         while (buf < tail && *buf++ != '\n')
  79                 /* nada */;
  80         if (buf + 9 >= tail)
  81                 return 0;
  82         if (memcmp(buf, "committer", 9))
  83                 return 0;
  84         while (buf < tail && *buf++ != '>')
  85                 /* nada */;
  86         if (buf >= tail)
  87                 return 0;
  88         dateptr = buf;
  89         while (buf < tail && *buf++ != '\n')
  90                 /* nada */;
  91         if (buf >= tail)
  92                 return 0;
  93         /* dateptr < buf && buf[-1] == '\n', so parsing will stop at buf-1 */
  94         return parse_timestamp(dateptr, NULL, 10);
  95 }
  96
  97 static struct commit_graft **commit_graft;
  98 static int commit_graft_alloc, commit_graft_nr;
  99
 100 static const unsigned char *commit_graft_sha1_access(size_t index, void *table)
 101 {
 102         struct commit_graft **commit_graft_table = table;
 103         return commit_graft_table[index]->oid.hash;
 104 }
 105
 106 static int commit_graft_pos(const unsigned char *sha1)
 107 {
 108         return sha1_pos(sha1, commit_graft, commit_graft_nr,
 109                         commit_graft_sha1_access);
 110 }
 111
 112 int register_commit_graft(struct commit_graft *graft, int ignore_dups)
 113 {
 114         int pos = commit_graft_pos(graft->oid.hash);
 115
 116         if (0 <= pos) {
 117                 if (ignore_dups)
 118                         free(graft);
 119                 else {
 120                         free(commit_graft[pos]);
 121                         commit_graft[pos] = graft;
 122                 }
 123                 return 1;
 124         }
 125         pos = -pos - 1;
 126         ALLOC_GROW(commit_graft, commit_graft_nr + 1, commit_graft_alloc);
 127         commit_graft_nr++;
 128         if (pos < commit_graft_nr)
 129                 memmove(commit_graft + pos + 1,
 130                         commit_graft + pos,
 131                         (commit_graft_nr - pos - 1) *
 132                         sizeof(*commit_graft));
 133         commit_graft[pos] = graft;
 134         return 0;
 135 }
 136
 137 struct commit_graft *read_graft_line(char *buf, int len)
 138 {
 139         /* The format is just "Commit Parent1 Parent2 ...\n" */
 140         int i;
 141         struct commit_graft *graft = NULL;
 142         const int entry_size = GIT_SHA1_HEXSZ + 1;
 143
 144         while (len && isspace(buf[len-1]))
 145                 buf[--len] = '\0';
 146         if (buf[0] == '#' || buf[0] == '\0')
 147                 return NULL;
 148         if ((len + 1) % entry_size)
 149                 goto bad_graft_data;
 150         i = (len + 1) / entry_size - 1;
 151         graft = xmalloc(st_add(sizeof(*graft), st_mult(GIT_SHA1_RAWSZ, i)));
 152         graft->nr_parent = i;
 153         if (get_oid_hex(buf, &graft->oid))
 154                 goto bad_graft_data;
 155         for (i = GIT_SHA1_HEXSZ; i < len; i += entry_size) {
 156                 if (buf[i] != ' ')
 157                         goto bad_graft_data;
 158                 if (get_sha1_hex(buf + i + 1, graft->parent[i/entry_size].hash))
 159                         goto bad_graft_data;
 160         }
 161         return graft;
 162
 163 bad_graft_data:
 164         error("bad graft data: %s", buf);
 165         free(graft);
 166         return NULL;
 167 }
 168
 169 static int read_graft_file(const char *graft_file)
 170 {
 171         FILE *fp = fopen_or_warn(graft_file, "r");
 172         struct strbuf buf = STRBUF_INIT;
 173         if (!fp)
 174                 return -1;
 175         while (!strbuf_getwholeline(&buf, fp, '\n')) {
 176                 /* The format is just "Commit Parent1 Parent2 ...\n" */
 177                 struct commit_graft *graft = read_graft_line(buf.buf, buf.len);
 178                 if (!graft)
 179                         continue;
 180                 if (register_commit_graft(graft, 1))
 181                         error("duplicate graft data: %s", buf.buf);
 182         }
 183         fclose(fp);
 184         strbuf_release(&buf);
 185         return 0;
 186 }
 187
 188 static void prepare_commit_graft(void)
 189 {
 190         static int commit_graft_prepared;
 191         char *graft_file;
 192
 193         if (commit_graft_prepared)
 194                 return;
 195         graft_file = get_graft_file();
 196         read_graft_file(graft_file);
 197         /* make sure shallows are read */
 198         is_repository_shallow();
 199         commit_graft_prepared = 1;
 200 }
 201
 202 struct commit_graft *lookup_commit_graft(const unsigned char *sha1)
 203 {
 204         int pos;
 205         prepare_commit_graft();
 206         pos = commit_graft_pos(sha1);
 207         if (pos < 0)
 208                 return NULL;
 209         return commit_graft[pos];
 210 }
 211
 212 int for_each_commit_graft(each_commit_graft_fn fn, void *cb_data)
 213 {
 214         int i, ret;
 215         for (i = ret = 0; i < commit_graft_nr && !ret; i++)
 216                 ret = fn(commit_graft[i], cb_data);
 217         return ret;
 218 }
 219
 220 int unregister_shallow(const struct object_id *oid)
 221 {
 222         int pos = commit_graft_pos(oid->hash);
 223         if (pos < 0)
 224                 return -1;
 225         if (pos + 1 < commit_graft_nr)
 226                 MOVE_ARRAY(commit_graft + pos, commit_graft + pos + 1,
 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_AND_NULL(v->buffer);
 290                 v->size = 0;
 291         }
 292 }
 293
 294 const void *detach_commit_buffer(struct commit *commit, unsigned long *sizep)
 295 {
 296         struct commit_buffer *v = buffer_slab_peek(&buffer_slab, commit);
 297         void *ret;
 298
 299         if (!v) {
 300                 if (sizep)
 301                         *sizep = 0;
 302                 return NULL;
 303         }
 304         ret = v->buffer;
 305         if (sizep)
 306                 *sizep = v->size;
 307
 308         v->buffer = NULL;
 309         v->size = 0;
 310         return ret;
 311 }
 312
 313 int parse_commit_buffer(struct commit *item, const void *buffer, unsigned long size)
 314 {
 315         const char *tail = buffer;
 316         const char *bufptr = buffer;
 317         struct object_id parent;
 318         struct commit_list **pptr;
 319         struct commit_graft *graft;
 320         const int tree_entry_len = GIT_SHA1_HEXSZ + 5;
 321         const int parent_entry_len = GIT_SHA1_HEXSZ + 7;
 322
 323         if (item->object.parsed)
 324                 return 0;
 325         item->object.parsed = 1;
 326         tail += size;
 327         if (tail <= bufptr + tree_entry_len + 1 || memcmp(bufptr, "tree ", 5) ||
 328                         bufptr[tree_entry_len] != '\n')
 329                 return error("bogus commit object %s", oid_to_hex(&item->object.oid));
 330         if (get_sha1_hex(bufptr + 5, parent.hash) < 0)
 331                 return error("bad tree pointer in commit %s",
 332                              oid_to_hex(&item->object.oid));
 333         item->tree = lookup_tree(&parent);
 334         bufptr += tree_entry_len + 1; /* "tree " + "hex sha1" + "\n" */
 335         pptr = &item->parents;
 336
 337         graft = lookup_commit_graft(item->object.oid.hash);
 338         while (bufptr + parent_entry_len < tail && !memcmp(bufptr, "parent ", 7)) {
 339                 struct commit *new_parent;
 340
 341                 if (tail <= bufptr + parent_entry_len + 1 ||
 342                     get_sha1_hex(bufptr + 7, parent.hash) ||
 343                     bufptr[parent_entry_len] != '\n')
 344                         return error("bad parents in commit %s", oid_to_hex(&item->object.oid));
 345                 bufptr += parent_entry_len + 1;
 346                 /*
 347                  * The clone is shallow if nr_parent < 0, and we must
 348                  * not traverse its real parents even when we unhide them.
 349                  */
 350                 if (graft && (graft->nr_parent < 0 || grafts_replace_parents))
 351                         continue;
 352                 new_parent = lookup_commit(&parent);
 353                 if (new_parent)
 354                         pptr = &commit_list_insert(new_parent, pptr)->next;
 355         }
 356         if (graft) {
 357                 int i;
 358                 struct commit *new_parent;
 359                 for (i = 0; i < graft->nr_parent; i++) {
 360                         new_parent = lookup_commit(&graft->parent[i]);
 361                         if (!new_parent)
 362                                 continue;
 363                         pptr = &commit_list_insert(new_parent, pptr)->next;
 364                 }
 365         }
 366         item->date = parse_commit_date(bufptr, tail);
 367
 368         return 0;
 369 }
 370
 371 int parse_commit_gently(struct commit *item, int quiet_on_missing)
 372 {
 373         enum object_type type;
 374         void *buffer;
 375         unsigned long size;
 376         int ret;
 377
 378         if (!item)
 379                 return -1;
 380         if (item->object.parsed)
 381                 return 0;
 382         buffer = read_sha1_file(item->object.oid.hash, &type, &size);
 383         if (!buffer)
 384                 return quiet_on_missing ? -1 :
 385                         error("Could not read %s",
 386                              oid_to_hex(&item->object.oid));
 387         if (type != OBJ_COMMIT) {
 388                 free(buffer);
 389                 return error("Object %s not a commit",
 390                              oid_to_hex(&item->object.oid));
 391         }
 392         ret = parse_commit_buffer(item, buffer, size);
 393         if (save_commit_buffer && !ret) {
 394                 set_commit_buffer(item, buffer, size);
 395                 return 0;
 396         }
 397         free(buffer);
 398         return ret;
 399 }
 400
 401 void parse_commit_or_die(struct commit *item)
 402 {
 403         if (parse_commit(item))
 404                 die("unable to parse commit %s",
 405                     item ? oid_to_hex(&item->object.oid) : "(null)");
 406 }
 407
 408 int find_commit_subject(const char *commit_buffer, const char **subject)
 409 {
 410         const char *eol;
 411         const char *p = commit_buffer;
 412
 413         while (*p && (*p != '\n' || p[1] != '\n'))
 414                 p++;
 415         if (*p) {
 416                 p = skip_blank_lines(p + 2);
 417                 eol = strchrnul(p, '\n');
 418         } else
 419                 eol = p;
 420
 421         *subject = p;
 422
 423         return eol - p;
 424 }
 425
 426 struct commit_list *commit_list_insert(struct commit *item, struct commit_list **list_p)
 427 {
 428         struct commit_list *new_list = xmalloc(sizeof(struct commit_list));
 429         new_list->item = item;
 430         new_list->next = *list_p;
 431         *list_p = new_list;
 432         return new_list;
 433 }
 434
 435 unsigned commit_list_count(const struct commit_list *l)
 436 {
 437         unsigned c = 0;
 438         for (; l; l = l->next )
 439                 c++;
 440         return c;
 441 }
 442
 443 struct commit_list *copy_commit_list(struct commit_list *list)
 444 {
 445         struct commit_list *head = NULL;
 446         struct commit_list **pp = &head;
 447         while (list) {
 448                 pp = commit_list_append(list->item, pp);
 449                 list = list->next;
 450         }
 451         return head;
 452 }
 453
 454 void free_commit_list(struct commit_list *list)
 455 {
 456         while (list)
 457                 pop_commit(&list);
 458 }
 459
 460 struct commit_list * commit_list_insert_by_date(struct commit *item, struct commit_list **list)
 461 {
 462         struct commit_list **pp = list;
 463         struct commit_list *p;
 464         while ((p = *pp) != NULL) {
 465                 if (p->item->date < item->date) {
 466                         break;
 467                 }
 468                 pp = &p->next;
 469         }
 470         return commit_list_insert(item, pp);
 471 }
 472
 473 static int commit_list_compare_by_date(const void *a, const void *b)
 474 {
 475         timestamp_t a_date = ((const struct commit_list *)a)->item->date;
 476         timestamp_t b_date = ((const struct commit_list *)b)->item->date;
 477         if (a_date < b_date)
 478                 return 1;
 479         if (a_date > b_date)
 480                 return -1;
 481         return 0;
 482 }
 483
 484 static void *commit_list_get_next(const void *a)
 485 {
 486         return ((const struct commit_list *)a)->next;
 487 }
 488
 489 static void commit_list_set_next(void *a, void *next)
 490 {
 491         ((struct commit_list *)a)->next = next;
 492 }
 493
 494 void commit_list_sort_by_date(struct commit_list **list)
 495 {
 496         *list = llist_mergesort(*list, commit_list_get_next, commit_list_set_next,
 497                                 commit_list_compare_by_date);
 498 }
 499
 500 struct commit *pop_most_recent_commit(struct commit_list **list,
 501                                       unsigned int mark)
 502 {
 503         struct commit *ret = pop_commit(list);
 504         struct commit_list *parents = ret->parents;
 505
 506         while (parents) {
 507                 struct commit *commit = parents->item;
 508                 if (!parse_commit(commit) && !(commit->object.flags & mark)) {
 509                         commit->object.flags |= mark;
 510                         commit_list_insert_by_date(commit, list);
 511                 }
 512                 parents = parents->next;
 513         }
 514         return ret;
 515 }
 516
 517 static void clear_commit_marks_1(struct commit_list **plist,
 518                                  struct commit *commit, unsigned int mark)
 519 {
 520         while (commit) {
 521                 struct commit_list *parents;
 522
 523                 if (!(mark & commit->object.flags))
 524                         return;
 525
 526                 commit->object.flags &= ~mark;
 527
 528                 parents = commit->parents;
 529                 if (!parents)
 530                         return;
 531
 532                 while ((parents = parents->next))
 533                         commit_list_insert(parents->item, plist);
 534
 535                 commit = commit->parents->item;
 536         }
 537 }
 538
 539 void clear_commit_marks_many(int nr, struct commit **commit, unsigned int mark)
 540 {
 541         struct commit_list *list = NULL;
 542
 543         while (nr--) {
 544                 commit_list_insert(*commit, &list);
 545                 commit++;
 546         }
 547         while (list)
 548                 clear_commit_marks_1(&list, pop_commit(&list), mark);
 549 }
 550
 551 void clear_commit_marks(struct commit *commit, unsigned int mark)
 552 {
 553         clear_commit_marks_many(1, &commit, mark);
 554 }
 555
 556 void clear_commit_marks_for_object_array(struct object_array *a, unsigned mark)
 557 {
 558         struct object *object;
 559         struct commit *commit;
 560         unsigned int i;
 561
 562         for (i = 0; i < a->nr; i++) {
 563                 object = a->objects[i].item;
 564                 commit = lookup_commit_reference_gently(&object->oid, 1);
 565                 if (commit)
 566                         clear_commit_marks(commit, mark);
 567         }
 568 }
 569
 570 struct commit *pop_commit(struct commit_list **stack)
 571 {
 572         struct commit_list *top = *stack;
 573         struct commit *item = top ? top->item : NULL;
 574
 575         if (top) {
 576                 *stack = top->next;
 577                 free(top);
 578         }
 579         return item;
 580 }
 581
 582 /*
 583  * Topological sort support
 584  */
 585
 586 /* count number of children that have not been emitted */
 587 define_commit_slab(indegree_slab, int);
 588
 589 /* record author-date for each commit object */
 590 define_commit_slab(author_date_slab, unsigned long);
 591
 592 static void record_author_date(struct author_date_slab *author_date,
 593                                struct commit *commit)
 594 {
 595         const char *buffer = get_commit_buffer(commit, NULL);
 596         struct ident_split ident;
 597         const char *ident_line;
 598         size_t ident_len;
 599         char *date_end;
 600         timestamp_t date;
 601
 602         ident_line = find_commit_header(buffer, "author", &ident_len);
 603         if (!ident_line)
 604                 goto fail_exit; /* no author line */
 605         if (split_ident_line(&ident, ident_line, ident_len) ||
 606             !ident.date_begin || !ident.date_end)
 607                 goto fail_exit; /* malformed "author" line */
 608
 609         date = parse_timestamp(ident.date_begin, &date_end, 10);
 610         if (date_end != ident.date_end)
 611                 goto fail_exit; /* malformed date */
 612         *(author_date_slab_at(author_date, commit)) = date;
 613
 614 fail_exit:
 615         unuse_commit_buffer(commit, buffer);
 616 }
 617
 618 static int compare_commits_by_author_date(const void *a_, const void *b_,
 619                                           void *cb_data)
 620 {
 621         const struct commit *a = a_, *b = b_;
 622         struct author_date_slab *author_date = cb_data;
 623         timestamp_t a_date = *(author_date_slab_at(author_date, a));
 624         timestamp_t b_date = *(author_date_slab_at(author_date, b));
 625
 626         /* newer commits with larger date first */
 627         if (a_date < b_date)
 628                 return 1;
 629         else if (a_date > b_date)
 630                 return -1;
 631         return 0;
 632 }
 633
 634 int compare_commits_by_commit_date(const void *a_, const void *b_, void *unused)
 635 {
 636         const struct commit *a = a_, *b = b_;
 637         /* newer commits with larger date first */
 638         if (a->date < b->date)
 639                 return 1;
 640         else if (a->date > b->date)
 641                 return -1;
 642         return 0;
 643 }
 644
 645 /*
 646  * Performs an in-place topological sort on the list supplied.
 647  */
 648 void sort_in_topological_order(struct commit_list **list, enum rev_sort_order sort_order)
 649 {
 650         struct commit_list *next, *orig = *list;
 651         struct commit_list **pptr;
 652         struct indegree_slab indegree;
 653         struct prio_queue queue;
 654         struct commit *commit;
 655         struct author_date_slab author_date;
 656
 657         if (!orig)
 658                 return;
 659         *list = NULL;
 660
 661         init_indegree_slab(&indegree);
 662         memset(&queue, '\0', sizeof(queue));
 663
 664         switch (sort_order) {
 665         default: /* REV_SORT_IN_GRAPH_ORDER */
 666                 queue.compare = NULL;
 667                 break;
 668         case REV_SORT_BY_COMMIT_DATE:
 669                 queue.compare = compare_commits_by_commit_date;
 670                 break;
 671         case REV_SORT_BY_AUTHOR_DATE:
 672                 init_author_date_slab(&author_date);
 673                 queue.compare = compare_commits_by_author_date;
 674                 queue.cb_data = &author_date;
 675                 break;
 676         }
 677
 678         /* Mark them and clear the indegree */
 679         for (next = orig; next; next = next->next) {
 680                 struct commit *commit = next->item;
 681                 *(indegree_slab_at(&indegree, commit)) = 1;
 682                 /* also record the author dates, if needed */
 683                 if (sort_order == REV_SORT_BY_AUTHOR_DATE)
 684                         record_author_date(&author_date, commit);
 685         }
 686
 687         /* update the indegree */
 688         for (next = orig; next; next = next->next) {
 689                 struct commit_list *parents = next->item->parents;
 690                 while (parents) {
 691                         struct commit *parent = parents->item;
 692                         int *pi = indegree_slab_at(&indegree, parent);
 693
 694                         if (*pi)
 695                                 (*pi)++;
 696                         parents = parents->next;
 697                 }
 698         }
 699
 700         /*
 701          * find the tips
 702          *
 703          * tips are nodes not reachable from any other node in the list
 704          *
 705          * the tips serve as a starting set for the work queue.
 706          */
 707         for (next = orig; next; next = next->next) {
 708                 struct commit *commit = next->item;
 709
 710                 if (*(indegree_slab_at(&indegree, commit)) == 1)
 711                         prio_queue_put(&queue, commit);
 712         }
 713
 714         /*
 715          * This is unfortunate; the initial tips need to be shown
 716          * in the order given from the revision traversal machinery.
 717          */
 718         if (sort_order == REV_SORT_IN_GRAPH_ORDER)
 719                 prio_queue_reverse(&queue);
 720
 721         /* We no longer need the commit list */
 722         free_commit_list(orig);
 723
 724         pptr = list;
 725         *list = NULL;
 726         while ((commit = prio_queue_get(&queue)) != NULL) {
 727                 struct commit_list *parents;
 728
 729                 for (parents = commit->parents; parents ; parents = parents->next) {
 730                         struct commit *parent = parents->item;
 731                         int *pi = indegree_slab_at(&indegree, parent);
 732
 733                         if (!*pi)
 734                                 continue;
 735
 736                         /*
 737                          * parents are only enqueued for emission
 738                          * when all their children have been emitted thereby
 739                          * guaranteeing topological order.
 740                          */
 741                         if (--(*pi) == 1)
 742                                 prio_queue_put(&queue, parent);
 743                 }
 744                 /*
 745                  * all children of commit have already been
 746                  * emitted. we can emit it now.
 747                  */
 748                 *(indegree_slab_at(&indegree, commit)) = 0;
 749
 750                 pptr = &commit_list_insert(commit, pptr)->next;
 751         }
 752
 753         clear_indegree_slab(&indegree);
 754         clear_prio_queue(&queue);
 755         if (sort_order == REV_SORT_BY_AUTHOR_DATE)
 756                 clear_author_date_slab(&author_date);
 757 }
 758
 759 /* merge-base stuff */
 760
 761 /* Remember to update object flag allocation in object.h */
 762 #define PARENT1         (1u<<16)
 763 #define PARENT2         (1u<<17)
 764 #define STALE           (1u<<18)
 765 #define RESULT          (1u<<19)
 766
 767 static const unsigned all_flags = (PARENT1 | PARENT2 | STALE | RESULT);
 768
 769 static int queue_has_nonstale(struct prio_queue *queue)
 770 {
 771         int i;
 772         for (i = 0; i < queue->nr; i++) {
 773                 struct commit *commit = queue->array[i].data;
 774                 if (!(commit->object.flags & STALE))
 775                         return 1;
 776         }
 777         return 0;
 778 }
 779
 780 /* all input commits in one and twos[] must have been parsed! */
 781 static struct commit_list *paint_down_to_common(struct commit *one, int n, struct commit **twos)
 782 {
 783         struct prio_queue queue = { compare_commits_by_commit_date };
 784         struct commit_list *result = NULL;
 785         int i;
 786
 787         one->object.flags |= PARENT1;
 788         if (!n) {
 789                 commit_list_append(one, &result);
 790                 return result;
 791         }
 792         prio_queue_put(&queue, one);
 793
 794         for (i = 0; i < n; i++) {
 795                 twos[i]->object.flags |= PARENT2;
 796                 prio_queue_put(&queue, twos[i]);
 797         }
 798
 799         while (queue_has_nonstale(&queue)) {
 800                 struct commit *commit = prio_queue_get(&queue);
 801                 struct commit_list *parents;
 802                 int flags;
 803
 804                 flags = commit->object.flags & (PARENT1 | PARENT2 | STALE);
 805                 if (flags == (PARENT1 | PARENT2)) {
 806                         if (!(commit->object.flags & RESULT)) {
 807                                 commit->object.flags |= RESULT;
 808                                 commit_list_insert_by_date(commit, &result);
 809                         }
 810                         /* Mark parents of a found merge stale */
 811                         flags |= STALE;
 812                 }
 813                 parents = commit->parents;
 814                 while (parents) {
 815                         struct commit *p = parents->item;
 816                         parents = parents->next;
 817                         if ((p->object.flags & flags) == flags)
 818                                 continue;
 819                         if (parse_commit(p))
 820                                 return NULL;
 821                         p->object.flags |= flags;
 822                         prio_queue_put(&queue, p);
 823                 }
 824         }
 825
 826         clear_prio_queue(&queue);
 827         return result;
 828 }
 829
 830 static struct commit_list *merge_bases_many(struct commit *one, int n, struct commit **twos)
 831 {
 832         struct commit_list *list = NULL;
 833         struct commit_list *result = NULL;
 834         int i;
 835
 836         for (i = 0; i < n; i++) {
 837                 if (one == twos[i])
 838                         /*
 839                          * We do not mark this even with RESULT so we do not
 840                          * have to clean it up.
 841                          */
 842                         return commit_list_insert(one, &result);
 843         }
 844
 845         if (parse_commit(one))
 846                 return NULL;
 847         for (i = 0; i < n; i++) {
 848                 if (parse_commit(twos[i]))
 849                         return NULL;
 850         }
 851
 852         list = paint_down_to_common(one, n, twos);
 853
 854         while (list) {
 855                 struct commit *commit = pop_commit(&list);
 856                 if (!(commit->object.flags & STALE))
 857                         commit_list_insert_by_date(commit, &result);
 858         }
 859         return result;
 860 }
 861
 862 struct commit_list *get_octopus_merge_bases(struct commit_list *in)
 863 {
 864         struct commit_list *i, *j, *k, *ret = NULL;
 865
 866         if (!in)
 867                 return ret;
 868
 869         commit_list_insert(in->item, &ret);
 870
 871         for (i = in->next; i; i = i->next) {
 872                 struct commit_list *new = NULL, *end = NULL;
 873
 874                 for (j = ret; j; j = j->next) {
 875                         struct commit_list *bases;
 876                         bases = get_merge_bases(i->item, j->item);
 877                         if (!new)
 878                                 new = bases;
 879                         else
 880                                 end->next = bases;
 881                         for (k = bases; k; k = k->next)
 882                                 end = k;
 883                 }
 884                 ret = new;
 885         }
 886         return ret;
 887 }
 888
 889 static int remove_redundant(struct commit **array, int cnt)
 890 {
 891         /*
 892          * Some commit in the array may be an ancestor of
 893          * another commit.  Move such commit to the end of
 894          * the array, and return the number of commits that
 895          * are independent from each other.
 896          */
 897         struct commit **work;
 898         unsigned char *redundant;
 899         int *filled_index;
 900         int i, j, filled;
 901
 902         work = xcalloc(cnt, sizeof(*work));
 903         redundant = xcalloc(cnt, 1);
 904         ALLOC_ARRAY(filled_index, cnt - 1);
 905
 906         for (i = 0; i < cnt; i++)
 907                 parse_commit(array[i]);
 908         for (i = 0; i < cnt; i++) {
 909                 struct commit_list *common;
 910
 911                 if (redundant[i])
 912                         continue;
 913                 for (j = filled = 0; j < cnt; j++) {
 914                         if (i == j || redundant[j])
 915                                 continue;
 916                         filled_index[filled] = j;
 917                         work[filled++] = array[j];
 918                 }
 919                 common = paint_down_to_common(array[i], filled, work);
 920                 if (array[i]->object.flags & PARENT2)
 921                         redundant[i] = 1;
 922                 for (j = 0; j < filled; j++)
 923                         if (work[j]->object.flags & PARENT1)
 924                                 redundant[filled_index[j]] = 1;
 925                 clear_commit_marks(array[i], all_flags);
 926                 for (j = 0; j < filled; j++)
 927                         clear_commit_marks(work[j], all_flags);
 928                 free_commit_list(common);
 929         }
 930
 931         /* Now collect the result */
 932         COPY_ARRAY(work, array, cnt);
 933         for (i = filled = 0; i < cnt; i++)
 934                 if (!redundant[i])
 935                         array[filled++] = work[i];
 936         for (j = filled, i = 0; i < cnt; i++)
 937                 if (redundant[i])
 938                         array[j++] = work[i];
 939         free(work);
 940         free(redundant);
 941         free(filled_index);
 942         return filled;
 943 }
 944
 945 static struct commit_list *get_merge_bases_many_0(struct commit *one,
 946                                                   int n,
 947                                                   struct commit **twos,
 948                                                   int cleanup)
 949 {
 950         struct commit_list *list;
 951         struct commit **rslt;
 952         struct commit_list *result;
 953         int cnt, i;
 954
 955         result = merge_bases_many(one, n, twos);
 956         for (i = 0; i < n; i++) {
 957                 if (one == twos[i])
 958                         return result;
 959         }
 960         if (!result || !result->next) {
 961                 if (cleanup) {
 962                         clear_commit_marks(one, all_flags);
 963                         clear_commit_marks_many(n, twos, all_flags);
 964                 }
 965                 return result;
 966         }
 967
 968         /* There are more than one */
 969         cnt = commit_list_count(result);
 970         rslt = xcalloc(cnt, sizeof(*rslt));
 971         for (list = result, i = 0; list; list = list->next)
 972                 rslt[i++] = list->item;
 973         free_commit_list(result);
 974
 975         clear_commit_marks(one, all_flags);
 976         clear_commit_marks_many(n, twos, all_flags);
 977
 978         cnt = remove_redundant(rslt, cnt);
 979         result = NULL;
 980         for (i = 0; i < cnt; i++)
 981                 commit_list_insert_by_date(rslt[i], &result);
 982         free(rslt);
 983         return result;
 984 }
 985
 986 struct commit_list *get_merge_bases_many(struct commit *one,
 987                                          int n,
 988                                          struct commit **twos)
 989 {
 990         return get_merge_bases_many_0(one, n, twos, 1);
 991 }
 992
 993 struct commit_list *get_merge_bases_many_dirty(struct commit *one,
 994                                                int n,
 995                                                struct commit **twos)
 996 {
 997         return get_merge_bases_many_0(one, n, twos, 0);
 998 }
 999
1000 struct commit_list *get_merge_bases(struct commit *one, struct commit *two)
1001 {
1002         return get_merge_bases_many_0(one, 1, &two, 1);
1003 }
1004
1005 /*
1006  * Is "commit" a descendant of one of the elements on the "with_commit" list?
1007  */
1008 int is_descendant_of(struct commit *commit, struct commit_list *with_commit)
1009 {
1010         if (!with_commit)
1011                 return 1;
1012         while (with_commit) {
1013                 struct commit *other;
1014
1015                 other = with_commit->item;
1016                 with_commit = with_commit->next;
1017                 if (in_merge_bases(other, commit))
1018                         return 1;
1019         }
1020         return 0;
1021 }
1022
1023 /*
1024  * Is "commit" an ancestor of one of the "references"?
1025  */
1026 int in_merge_bases_many(struct commit *commit, int nr_reference, struct commit **reference)
1027 {
1028         struct commit_list *bases;
1029         int ret = 0, i;
1030
1031         if (parse_commit(commit))
1032                 return ret;
1033         for (i = 0; i < nr_reference; i++)
1034                 if (parse_commit(reference[i]))
1035                         return ret;
1036
1037         bases = paint_down_to_common(commit, nr_reference, reference);
1038         if (commit->object.flags & PARENT2)
1039                 ret = 1;
1040         clear_commit_marks(commit, all_flags);
1041         clear_commit_marks_many(nr_reference, reference, all_flags);
1042         free_commit_list(bases);
1043         return ret;
1044 }
1045
1046 /*
1047  * Is "commit" an ancestor of (i.e. reachable from) the "reference"?
1048  */
1049 int in_merge_bases(struct commit *commit, struct commit *reference)
1050 {
1051         return in_merge_bases_many(commit, 1, &reference);
1052 }
1053
1054 struct commit_list *reduce_heads(struct commit_list *heads)
1055 {
1056         struct commit_list *p;
1057         struct commit_list *result = NULL, **tail = &result;
1058         struct commit **array;
1059         int num_head, i;
1060
1061         if (!heads)
1062                 return NULL;
1063
1064         /* Uniquify */
1065         for (p = heads; p; p = p->next)
1066                 p->item->object.flags &= ~STALE;
1067         for (p = heads, num_head = 0; p; p = p->next) {
1068                 if (p->item->object.flags & STALE)
1069                         continue;
1070                 p->item->object.flags |= STALE;
1071                 num_head++;
1072         }
1073         array = xcalloc(num_head, sizeof(*array));
1074         for (p = heads, i = 0; p; p = p->next) {
1075                 if (p->item->object.flags & STALE) {
1076                         array[i++] = p->item;
1077                         p->item->object.flags &= ~STALE;
1078                 }
1079         }
1080         num_head = remove_redundant(array, num_head);
1081         for (i = 0; i < num_head; i++)
1082                 tail = &commit_list_insert(array[i], tail)->next;
1083         return result;
1084 }
1085
1086 static const char gpg_sig_header[] = "gpgsig";
1087 static const int gpg_sig_header_len = sizeof(gpg_sig_header) - 1;
1088
1089 static int do_sign_commit(struct strbuf *buf, const char *keyid)
1090 {
1091         struct strbuf sig = STRBUF_INIT;
1092         int inspos, copypos;
1093         const char *eoh;
1094
1095         /* find the end of the header */
1096         eoh = strstr(buf->buf, "\n\n");
1097         if (!eoh)
1098                 inspos = buf->len;
1099         else
1100                 inspos = eoh - buf->buf + 1;
1101
1102         if (!keyid || !*keyid)
1103                 keyid = get_signing_key();
1104         if (sign_buffer(buf, &sig, keyid)) {
1105                 strbuf_release(&sig);
1106                 return -1;
1107         }
1108
1109         for (copypos = 0; sig.buf[copypos]; ) {
1110                 const char *bol = sig.buf + copypos;
1111                 const char *eol = strchrnul(bol, '\n');
1112                 int len = (eol - bol) + !!*eol;
1113
1114                 if (!copypos) {
1115                         strbuf_insert(buf, inspos, gpg_sig_header, gpg_sig_header_len);
1116                         inspos += gpg_sig_header_len;
1117                 }
1118                 strbuf_insert(buf, inspos++, " ", 1);
1119                 strbuf_insert(buf, inspos, bol, len);
1120                 inspos += len;
1121                 copypos += len;
1122         }
1123         strbuf_release(&sig);
1124         return 0;
1125 }
1126
1127 int parse_signed_commit(const struct commit *commit,
1128                         struct strbuf *payload, struct strbuf *signature)
1129 {
1130
1131         unsigned long size;
1132         const char *buffer = get_commit_buffer(commit, &size);
1133         int in_signature, saw_signature = -1;
1134         const char *line, *tail;
1135
1136         line = buffer;
1137         tail = buffer + size;
1138         in_signature = 0;
1139         saw_signature = 0;
1140         while (line < tail) {
1141                 const char *sig = NULL;
1142                 const char *next = memchr(line, '\n', tail - line);
1143
1144                 next = next ? next + 1 : tail;
1145                 if (in_signature && line[0] == ' ')
1146                         sig = line + 1;
1147                 else if (starts_with(line, gpg_sig_header) &&
1148                          line[gpg_sig_header_len] == ' ')
1149                         sig = line + gpg_sig_header_len + 1;
1150                 if (sig) {
1151                         strbuf_add(signature, sig, next - sig);
1152                         saw_signature = 1;
1153                         in_signature = 1;
1154                 } else {
1155                         if (*line == '\n')
1156                                 /* dump the whole remainder of the buffer */
1157                                 next = tail;
1158                         strbuf_add(payload, line, next - line);
1159                         in_signature = 0;
1160                 }
1161                 line = next;
1162         }
1163         unuse_commit_buffer(commit, buffer);
1164         return saw_signature;
1165 }
1166
1167 int remove_signature(struct strbuf *buf)
1168 {
1169         const char *line = buf->buf;
1170         const char *tail = buf->buf + buf->len;
1171         int in_signature = 0;
1172         const char *sig_start = NULL;
1173         const char *sig_end = NULL;
1174
1175         while (line < tail) {
1176                 const char *next = memchr(line, '\n', tail - line);
1177                 next = next ? next + 1 : tail;
1178
1179                 if (in_signature && line[0] == ' ')
1180                         sig_end = next;
1181                 else if (starts_with(line, gpg_sig_header) &&
1182                          line[gpg_sig_header_len] == ' ') {
1183                         sig_start = line;
1184                         sig_end = next;
1185                         in_signature = 1;
1186                 } else {
1187                         if (*line == '\n')
1188                                 /* dump the whole remainder of the buffer */
1189                                 next = tail;
1190                         in_signature = 0;
1191                 }
1192                 line = next;
1193         }
1194
1195         if (sig_start)
1196                 strbuf_remove(buf, sig_start - buf->buf, sig_end - sig_start);
1197
1198         return sig_start != NULL;
1199 }
1200
1201 static void handle_signed_tag(struct commit *parent, struct commit_extra_header ***tail)
1202 {
1203         struct merge_remote_desc *desc;
1204         struct commit_extra_header *mergetag;
1205         char *buf;
1206         unsigned long size, len;
1207         enum object_type type;
1208
1209         desc = merge_remote_util(parent);
1210         if (!desc || !desc->obj)
1211                 return;
1212         buf = read_sha1_file(desc->obj->oid.hash, &type, &size);
1213         if (!buf || type != OBJ_TAG)
1214                 goto free_return;
1215         len = parse_signature(buf, size);
1216         if (size == len)
1217                 goto free_return;
1218         /*
1219          * We could verify this signature and either omit the tag when
1220          * it does not validate, but the integrator may not have the
1221          * public key of the signer of the tag he is merging, while a
1222          * later auditor may have it while auditing, so let's not run
1223          * verify-signed-buffer here for now...
1224          *
1225          * if (verify_signed_buffer(buf, len, buf + len, size - len, ...))
1226          *      warn("warning: signed tag unverified.");
1227          */
1228         mergetag = xcalloc(1, sizeof(*mergetag));
1229         mergetag->key = xstrdup("mergetag");
1230         mergetag->value = buf;
1231         mergetag->len = size;
1232
1233         **tail = mergetag;
1234         *tail = &mergetag->next;
1235         return;
1236
1237 free_return:
1238         free(buf);
1239 }
1240
1241 int check_commit_signature(const struct commit *commit, struct signature_check *sigc)
1242 {
1243         struct strbuf payload = STRBUF_INIT;
1244         struct strbuf signature = STRBUF_INIT;
1245         int ret = 1;
1246
1247         sigc->result = 'N';
1248
1249         if (parse_signed_commit(commit, &payload, &signature) <= 0)
1250                 goto out;
1251         ret = check_signature(payload.buf, payload.len, signature.buf,
1252                 signature.len, sigc);
1253
1254  out:
1255         strbuf_release(&payload);
1256         strbuf_release(&signature);
1257
1258         return ret;
1259 }
1260
1261
1262
1263 void append_merge_tag_headers(struct commit_list *parents,
1264                               struct commit_extra_header ***tail)
1265 {
1266         while (parents) {
1267                 struct commit *parent = parents->item;
1268                 handle_signed_tag(parent, tail);
1269                 parents = parents->next;
1270         }
1271 }
1272
1273 static void add_extra_header(struct strbuf *buffer,
1274                              struct commit_extra_header *extra)
1275 {
1276         strbuf_addstr(buffer, extra->key);
1277         if (extra->len)
1278                 strbuf_add_lines(buffer, " ", extra->value, extra->len);
1279         else
1280                 strbuf_addch(buffer, '\n');
1281 }
1282
1283 struct commit_extra_header *read_commit_extra_headers(struct commit *commit,
1284                                                       const char **exclude)
1285 {
1286         struct commit_extra_header *extra = NULL;
1287         unsigned long size;
1288         const char *buffer = get_commit_buffer(commit, &size);
1289         extra = read_commit_extra_header_lines(buffer, size, exclude);
1290         unuse_commit_buffer(commit, buffer);
1291         return extra;
1292 }
1293
1294 void for_each_mergetag(each_mergetag_fn fn, struct commit *commit, void *data)
1295 {
1296         struct commit_extra_header *extra, *to_free;
1297
1298         to_free = read_commit_extra_headers(commit, NULL);
1299         for (extra = to_free; extra; extra = extra->next) {
1300                 if (strcmp(extra->key, "mergetag"))
1301                         continue; /* not a merge tag */
1302                 fn(commit, extra, data);
1303         }
1304         free_commit_extra_headers(to_free);
1305 }
1306
1307 static inline int standard_header_field(const char *field, size_t len)
1308 {
1309         return ((len == 4 && !memcmp(field, "tree", 4)) ||
1310                 (len == 6 && !memcmp(field, "parent", 6)) ||
1311                 (len == 6 && !memcmp(field, "author", 6)) ||
1312                 (len == 9 && !memcmp(field, "committer", 9)) ||
1313                 (len == 8 && !memcmp(field, "encoding", 8)));
1314 }
1315
1316 static int excluded_header_field(const char *field, size_t len, const char **exclude)
1317 {
1318         if (!exclude)
1319                 return 0;
1320
1321         while (*exclude) {
1322                 size_t xlen = strlen(*exclude);
1323                 if (len == xlen && !memcmp(field, *exclude, xlen))
1324                         return 1;
1325                 exclude++;
1326         }
1327         return 0;
1328 }
1329
1330 static struct commit_extra_header *read_commit_extra_header_lines(
1331         const char *buffer, size_t size,
1332         const char **exclude)
1333 {
1334         struct commit_extra_header *extra = NULL, **tail = &extra, *it = NULL;
1335         const char *line, *next, *eof, *eob;
1336         struct strbuf buf = STRBUF_INIT;
1337
1338         for (line = buffer, eob = line + size;
1339              line < eob && *line != '\n';
1340              line = next) {
1341                 next = memchr(line, '\n', eob - line);
1342                 next = next ? next + 1 : eob;
1343                 if (*line == ' ') {
1344                         /* continuation */
1345                         if (it)
1346                                 strbuf_add(&buf, line + 1, next - (line + 1));
1347                         continue;
1348                 }
1349                 if (it)
1350                         it->value = strbuf_detach(&buf, &it->len);
1351                 strbuf_reset(&buf);
1352                 it = NULL;
1353
1354                 eof = memchr(line, ' ', next - line);
1355                 if (!eof)
1356                         eof = next;
1357                 else if (standard_header_field(line, eof - line) ||
1358                          excluded_header_field(line, eof - line, exclude))
1359                         continue;
1360
1361                 it = xcalloc(1, sizeof(*it));
1362                 it->key = xmemdupz(line, eof-line);
1363                 *tail = it;
1364                 tail = &it->next;
1365                 if (eof + 1 < next)
1366                         strbuf_add(&buf, eof + 1, next - (eof + 1));
1367         }
1368         if (it)
1369                 it->value = strbuf_detach(&buf, &it->len);
1370         return extra;
1371 }
1372
1373 void free_commit_extra_headers(struct commit_extra_header *extra)
1374 {
1375         while (extra) {
1376                 struct commit_extra_header *next = extra->next;
1377                 free(extra->key);
1378                 free(extra->value);
1379                 free(extra);
1380                 extra = next;
1381         }
1382 }
1383
1384 int commit_tree(const char *msg, size_t msg_len,
1385                 const unsigned char *tree,
1386                 struct commit_list *parents, unsigned char *ret,
1387                 const char *author, const char *sign_commit)
1388 {
1389         struct commit_extra_header *extra = NULL, **tail = &extra;
1390         int result;
1391
1392         append_merge_tag_headers(parents, &tail);
1393         result = commit_tree_extended(msg, msg_len, tree, parents, ret,
1394                                       author, sign_commit, extra);
1395         free_commit_extra_headers(extra);
1396         return result;
1397 }
1398
1399 static int find_invalid_utf8(const char *buf, int len)
1400 {
1401         int offset = 0;
1402         static const unsigned int max_codepoint[] = {
1403                 0x7f, 0x7ff, 0xffff, 0x10ffff
1404         };
1405
1406         while (len) {
1407                 unsigned char c = *buf++;
1408                 int bytes, bad_offset;
1409                 unsigned int codepoint;
1410                 unsigned int min_val, max_val;
1411
1412                 len--;
1413                 offset++;
1414
1415                 /* Simple US-ASCII? No worries. */
1416                 if (c < 0x80)
1417                         continue;
1418
1419                 bad_offset = offset-1;
1420
1421                 /*
1422                  * Count how many more high bits set: that's how
1423                  * many more bytes this sequence should have.
1424                  */
1425                 bytes = 0;
1426                 while (c & 0x40) {
1427                         c <<= 1;
1428                         bytes++;
1429                 }
1430
1431                 /*
1432                  * Must be between 1 and 3 more bytes.  Longer sequences result in
1433                  * codepoints beyond U+10FFFF, which are guaranteed never to exist.
1434                  */
1435                 if (bytes < 1 || 3 < bytes)
1436                         return bad_offset;
1437
1438                 /* Do we *have* that many bytes? */
1439                 if (len < bytes)
1440                         return bad_offset;
1441
1442                 /*
1443                  * Place the encoded bits at the bottom of the value and compute the
1444                  * valid range.
1445                  */
1446                 codepoint = (c & 0x7f) >> bytes;
1447                 min_val = max_codepoint[bytes-1] + 1;
1448                 max_val = max_codepoint[bytes];
1449
1450                 offset += bytes;
1451                 len -= bytes;
1452
1453                 /* And verify that they are good continuation bytes */
1454                 do {
1455                         codepoint <<= 6;
1456                         codepoint |= *buf & 0x3f;
1457                         if ((*buf++ & 0xc0) != 0x80)
1458                                 return bad_offset;
1459                 } while (--bytes);
1460
1461                 /* Reject codepoints that are out of range for the sequence length. */
1462                 if (codepoint < min_val || codepoint > max_val)
1463                         return bad_offset;
1464                 /* Surrogates are only for UTF-16 and cannot be encoded in UTF-8. */
1465                 if ((codepoint & 0x1ff800) == 0xd800)
1466                         return bad_offset;
1467                 /* U+xxFFFE and U+xxFFFF are guaranteed non-characters. */
1468                 if ((codepoint & 0xfffe) == 0xfffe)
1469                         return bad_offset;
1470                 /* So are anything in the range U+FDD0..U+FDEF. */
1471                 if (codepoint >= 0xfdd0 && codepoint <= 0xfdef)
1472                         return bad_offset;
1473         }
1474         return -1;
1475 }
1476
1477 /*
1478  * This verifies that the buffer is in proper utf8 format.
1479  *
1480  * If it isn't, it assumes any non-utf8 characters are Latin1,
1481  * and does the conversion.
1482  */
1483 static int verify_utf8(struct strbuf *buf)
1484 {
1485         int ok = 1;
1486         long pos = 0;
1487
1488         for (;;) {
1489                 int bad;
1490                 unsigned char c;
1491                 unsigned char replace[2];
1492
1493                 bad = find_invalid_utf8(buf->buf + pos, buf->len - pos);
1494                 if (bad < 0)
1495                         return ok;
1496                 pos += bad;
1497                 ok = 0;
1498                 c = buf->buf[pos];
1499                 strbuf_remove(buf, pos, 1);
1500
1501                 /* We know 'c' must be in the range 128-255 */
1502                 replace[0] = 0xc0 + (c >> 6);
1503                 replace[1] = 0x80 + (c & 0x3f);
1504                 strbuf_insert(buf, pos, replace, 2);
1505                 pos += 2;
1506         }
1507 }
1508
1509 static const char commit_utf8_warn[] =
1510 N_("Warning: commit message did not conform to UTF-8.\n"
1511    "You may want to amend it after fixing the message, or set the config\n"
1512    "variable i18n.commitencoding to the encoding your project uses.\n");
1513
1514 int commit_tree_extended(const char *msg, size_t msg_len,
1515                          const unsigned char *tree,
1516                          struct commit_list *parents, unsigned char *ret,
1517                          const char *author, const char *sign_commit,
1518                          struct commit_extra_header *extra)
1519 {
1520         int result;
1521         int encoding_is_utf8;
1522         struct strbuf buffer;
1523
1524         assert_sha1_type(tree, OBJ_TREE);
1525
1526         if (memchr(msg, '\0', msg_len))
1527                 return error("a NUL byte in commit log message not allowed.");
1528
1529         /* Not having i18n.commitencoding is the same as having utf-8 */
1530         encoding_is_utf8 = is_encoding_utf8(git_commit_encoding);
1531
1532         strbuf_init(&buffer, 8192); /* should avoid reallocs for the headers */
1533         strbuf_addf(&buffer, "tree %s\n", sha1_to_hex(tree));
1534
1535         /*
1536          * NOTE! This ordering means that the same exact tree merged with a
1537          * different order of parents will be a _different_ changeset even
1538          * if everything else stays the same.
1539          */
1540         while (parents) {
1541                 struct commit *parent = pop_commit(&parents);
1542                 strbuf_addf(&buffer, "parent %s\n",
1543                             oid_to_hex(&parent->object.oid));
1544         }
1545
1546         /* Person/date information */
1547         if (!author)
1548                 author = git_author_info(IDENT_STRICT);
1549         strbuf_addf(&buffer, "author %s\n", author);
1550         strbuf_addf(&buffer, "committer %s\n", git_committer_info(IDENT_STRICT));
1551         if (!encoding_is_utf8)
1552                 strbuf_addf(&buffer, "encoding %s\n", git_commit_encoding);
1553
1554         while (extra) {
1555                 add_extra_header(&buffer, extra);
1556                 extra = extra->next;
1557         }
1558         strbuf_addch(&buffer, '\n');
1559
1560         /* And add the comment */
1561         strbuf_add(&buffer, msg, msg_len);
1562
1563         /* And check the encoding */
1564         if (encoding_is_utf8 && !verify_utf8(&buffer))
1565                 fprintf(stderr, _(commit_utf8_warn));
1566
1567         if (sign_commit && do_sign_commit(&buffer, sign_commit))
1568                 return -1;
1569
1570         result = write_sha1_file(buffer.buf, buffer.len, commit_type, ret);
1571         strbuf_release(&buffer);
1572         return result;
1573 }
1574
1575 void set_merge_remote_desc(struct commit *commit,
1576                            const char *name, struct object *obj)
1577 {
1578         struct merge_remote_desc *desc;
1579         FLEX_ALLOC_STR(desc, name, name);
1580         desc->obj = obj;
1581         commit->util = desc;
1582 }
1583
1584 struct commit *get_merge_parent(const char *name)
1585 {
1586         struct object *obj;
1587         struct commit *commit;
1588         struct object_id oid;
1589         if (get_sha1(name, oid.hash))
1590                 return NULL;
1591         obj = parse_object(&oid);
1592         commit = (struct commit *)peel_to_type(name, 0, obj, OBJ_COMMIT);
1593         if (commit && !commit->util)
1594                 set_merge_remote_desc(commit, name, obj);
1595         return commit;
1596 }
1597
1598 /*
1599  * Append a commit to the end of the commit_list.
1600  *
1601  * next starts by pointing to the variable that holds the head of an
1602  * empty commit_list, and is updated to point to the "next" field of
1603  * the last item on the list as new commits are appended.
1604  *
1605  * Usage example:
1606  *
1607  *     struct commit_list *list;
1608  *     struct commit_list **next = &list;
1609  *
1610  *     next = commit_list_append(c1, next);
1611  *     next = commit_list_append(c2, next);
1612  *     assert(commit_list_count(list) == 2);
1613  *     return list;
1614  */
1615 struct commit_list **commit_list_append(struct commit *commit,
1616                                         struct commit_list **next)
1617 {
1618         struct commit_list *new = xmalloc(sizeof(struct commit_list));
1619         new->item = commit;
1620         *next = new;
1621         new->next = NULL;
1622         return &new->next;
1623 }
1624
1625 const char *find_commit_header(const char *msg, const char *key, size_t *out_len)
1626 {
1627         int key_len = strlen(key);
1628         const char *line = msg;
1629
1630         while (line) {
1631                 const char *eol = strchrnul(line, '\n');
1632
1633                 if (line == eol)
1634                         return NULL;
1635
1636                 if (eol - line > key_len &&
1637                     !strncmp(line, key, key_len) &&
1638                     line[key_len] == ' ') {
1639                         *out_len = eol - line - key_len - 1;
1640                         return line + key_len + 1;
1641                 }
1642                 line = *eol ? eol + 1 : NULL;
1643         }
1644         return NULL;
1645 }
1646
1647 /*
1648  * Inspect the given string and determine the true "end" of the log message, in
1649  * order to find where to put a new Signed-off-by: line.  Ignored are
1650  * trailing comment lines and blank lines.  To support "git commit -s
1651  * --amend" on an existing commit, we also ignore "Conflicts:".  To
1652  * support "git commit -v", we truncate at cut lines.
1653  *
1654  * Returns the number of bytes from the tail to ignore, to be fed as
1655  * the second parameter to append_signoff().
1656  */
1657 int ignore_non_trailer(const char *buf, size_t len)
1658 {
1659         int boc = 0;
1660         int bol = 0;
1661         int in_old_conflicts_block = 0;
1662         size_t cutoff = wt_status_locate_end(buf, len);
1663
1664         while (bol < cutoff) {
1665                 const char *next_line = memchr(buf + bol, '\n', len - bol);
1666
1667                 if (!next_line)
1668                         next_line = buf + len;
1669                 else
1670                         next_line++;
1671
1672                 if (buf[bol] == comment_line_char || buf[bol] == '\n') {
1673                         /* is this the first of the run of comments? */
1674                         if (!boc)
1675                                 boc = bol;
1676                         /* otherwise, it is just continuing */
1677                 } else if (starts_with(buf + bol, "Conflicts:\n")) {
1678                         in_old_conflicts_block = 1;
1679                         if (!boc)
1680                                 boc = bol;
1681                 } else if (in_old_conflicts_block && buf[bol] == '\t') {
1682                         ; /* a pathname in the conflicts block */
1683                 } else if (boc) {
1684                         /* the previous was not trailing comment */
1685                         boc = 0;
1686                         in_old_conflicts_block = 0;
1687                 }
1688                 bol = next_line - buf;
1689         }
1690         return boc ? len - boc : len - cutoff;
1691 }