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 unsigned char *sha1,
  22                                               int quiet)
  23 {
  24         struct object *obj = deref_tag(parse_object(sha1), 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 unsigned char *sha1)
  32 {
  33         return lookup_commit_reference_gently(sha1, 0);
  34 }
  35
  36 struct commit *lookup_commit_or_die(const unsigned char *sha1, const char *ref_name)
  37 {
  38         struct commit *c = lookup_commit_reference(sha1);
  39         if (!c)
  40                 die(_("could not parse %s"), ref_name);
  41         if (hashcmp(sha1, c->object.sha1)) {
  42                 warning(_("%s %s is not a commit!"),
  43                         ref_name, sha1_to_hex(sha1));
  44         }
  45         return c;
  46 }
  47
  48 struct commit *lookup_commit(const unsigned char *sha1)
  49 {
  50         struct object *obj = lookup_object(sha1);
  51         if (!obj)
  52                 return create_object(sha1, 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.hash);
  64         if (parse_commit(commit))
  65                 return NULL;
  66         return commit;
  67 }
  68
  69 static unsigned long 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 strtoul will stop at buf-1 */
  93         return strtoul(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(sizeof(*graft) + 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 unsigned char *sha1)
 220 {
 221         int pos = commit_graft_pos(sha1);
 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.sha1, &type, &size);
 266                 if (!ret)
 267                         die("cannot read commit object %s",
 268                             sha1_to_hex(commit->object.sha1));
 269                 if (type != OBJ_COMMIT)
 270                         die("expected commit for %s, got %s",
 271                             sha1_to_hex(commit->object.sha1), 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", sha1_to_hex(item->object.sha1));
 331         if (get_sha1_hex(bufptr + 5, parent.hash) < 0)
 332                 return error("bad tree pointer in commit %s",
 333                              sha1_to_hex(item->object.sha1));
 334         item->tree = lookup_tree(parent.hash);
 335         bufptr += tree_entry_len + 1; /* "tree " + "hex sha1" + "\n" */
 336         pptr = &item->parents;
 337
 338         graft = lookup_commit_graft(item->object.sha1);
 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", sha1_to_hex(item->object.sha1));
 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.hash);
 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].hash);
 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.sha1, &type, &size);
 384         if (!buffer)
 385                 return quiet_on_missing ? -1 :
 386                         error("Could not read %s",
 387                              sha1_to_hex(item->object.sha1));
 388         if (type != OBJ_COMMIT) {
 389                 free(buffer);
 390                 return error("Object %s not a commit",
 391                              sha1_to_hex(item->object.sha1));
 392         }
 393         ret = parse_commit_buffer(item, buffer, size);
 394         if (save_commit_buffer && !ret) {
 395                 set_commit_buffer(item, buffer, size);
 396                 return 0;
 397         }
 398         free(buffer);
 399         return ret;
 400 }
 401
 402 void parse_commit_or_die(struct commit *item)
 403 {
 404         if (parse_commit(item))
 405                 die("unable to parse commit %s",
 406                     item ? sha1_to_hex(item->object.sha1) : "(null)");
 407 }
 408
 409 int find_commit_subject(const char *commit_buffer, const char **subject)
 410 {
 411         const char *eol;
 412         const char *p = commit_buffer;
 413
 414         while (*p && (*p != '\n' || p[1] != '\n'))
 415                 p++;
 416         if (*p) {
 417                 p += 2;
 418                 for (eol = p; *eol && *eol != '\n'; eol++)
 419                         ; /* do nothing */
 420         } else
 421                 eol = p;
 422
 423         *subject = p;
 424
 425         return eol - p;
 426 }
 427
 428 struct commit_list *commit_list_insert(struct commit *item, struct commit_list **list_p)
 429 {
 430         struct commit_list *new_list = xmalloc(sizeof(struct commit_list));
 431         new_list->item = item;
 432         new_list->next = *list_p;
 433         *list_p = new_list;
 434         return new_list;
 435 }
 436
 437 unsigned commit_list_count(const struct commit_list *l)
 438 {
 439         unsigned c = 0;
 440         for (; l; l = l->next )
 441                 c++;
 442         return c;
 443 }
 444
 445 struct commit_list *copy_commit_list(struct commit_list *list)
 446 {
 447         struct commit_list *head = NULL;
 448         struct commit_list **pp = &head;
 449         while (list) {
 450                 pp = commit_list_append(list->item, pp);
 451                 list = list->next;
 452         }
 453         return head;
 454 }
 455
 456 void free_commit_list(struct commit_list *list)
 457 {
 458         while (list) {
 459                 struct commit_list *temp = list;
 460                 list = temp->next;
 461                 free(temp);
 462         }
 463 }
 464
 465 struct commit_list * commit_list_insert_by_date(struct commit *item, struct commit_list **list)
 466 {
 467         struct commit_list **pp = list;
 468         struct commit_list *p;
 469         while ((p = *pp) != NULL) {
 470                 if (p->item->date < item->date) {
 471                         break;
 472                 }
 473                 pp = &p->next;
 474         }
 475         return commit_list_insert(item, pp);
 476 }
 477
 478 static int commit_list_compare_by_date(const void *a, const void *b)
 479 {
 480         unsigned long a_date = ((const struct commit_list *)a)->item->date;
 481         unsigned long b_date = ((const struct commit_list *)b)->item->date;
 482         if (a_date < b_date)
 483                 return 1;
 484         if (a_date > b_date)
 485                 return -1;
 486         return 0;
 487 }
 488
 489 static void *commit_list_get_next(const void *a)
 490 {
 491         return ((const struct commit_list *)a)->next;
 492 }
 493
 494 static void commit_list_set_next(void *a, void *next)
 495 {
 496         ((struct commit_list *)a)->next = next;
 497 }
 498
 499 void commit_list_sort_by_date(struct commit_list **list)
 500 {
 501         *list = llist_mergesort(*list, commit_list_get_next, commit_list_set_next,
 502                                 commit_list_compare_by_date);
 503 }
 504
 505 struct commit *pop_most_recent_commit(struct commit_list **list,
 506                                       unsigned int mark)
 507 {
 508         struct commit *ret = (*list)->item;
 509         struct commit_list *parents = ret->parents;
 510         struct commit_list *old = *list;
 511
 512         *list = (*list)->next;
 513         free(old);
 514
 515         while (parents) {
 516                 struct commit *commit = parents->item;
 517                 if (!parse_commit(commit) && !(commit->object.flags & mark)) {
 518                         commit->object.flags |= mark;
 519                         commit_list_insert_by_date(commit, list);
 520                 }
 521                 parents = parents->next;
 522         }
 523         return ret;
 524 }
 525
 526 static void clear_commit_marks_1(struct commit_list **plist,
 527                                  struct commit *commit, unsigned int mark)
 528 {
 529         while (commit) {
 530                 struct commit_list *parents;
 531
 532                 if (!(mark & commit->object.flags))
 533                         return;
 534
 535                 commit->object.flags &= ~mark;
 536
 537                 parents = commit->parents;
 538                 if (!parents)
 539                         return;
 540
 541                 while ((parents = parents->next))
 542                         commit_list_insert(parents->item, plist);
 543
 544                 commit = commit->parents->item;
 545         }
 546 }
 547
 548 void clear_commit_marks_many(int nr, struct commit **commit, unsigned int mark)
 549 {
 550         struct commit_list *list = NULL;
 551
 552         while (nr--) {
 553                 commit_list_insert(*commit, &list);
 554                 commit++;
 555         }
 556         while (list)
 557                 clear_commit_marks_1(&list, pop_commit(&list), mark);
 558 }
 559
 560 void clear_commit_marks(struct commit *commit, unsigned int mark)
 561 {
 562         clear_commit_marks_many(1, &commit, mark);
 563 }
 564
 565 void clear_commit_marks_for_object_array(struct object_array *a, unsigned mark)
 566 {
 567         struct object *object;
 568         struct commit *commit;
 569         unsigned int i;
 570
 571         for (i = 0; i < a->nr; i++) {
 572                 object = a->objects[i].item;
 573                 commit = lookup_commit_reference_gently(object->sha1, 1);
 574                 if (commit)
 575                         clear_commit_marks(commit, mark);
 576         }
 577 }
 578
 579 struct commit *pop_commit(struct commit_list **stack)
 580 {
 581         struct commit_list *top = *stack;
 582         struct commit *item = top ? top->item : NULL;
 583
 584         if (top) {
 585                 *stack = top->next;
 586                 free(top);
 587         }
 588         return item;
 589 }
 590
 591 /*
 592  * Topological sort support
 593  */
 594
 595 /* count number of children that have not been emitted */
 596 define_commit_slab(indegree_slab, int);
 597
 598 /* record author-date for each commit object */
 599 define_commit_slab(author_date_slab, unsigned long);
 600
 601 static void record_author_date(struct author_date_slab *author_date,
 602                                struct commit *commit)
 603 {
 604         const char *buffer = get_commit_buffer(commit, NULL);
 605         struct ident_split ident;
 606         const char *ident_line;
 607         size_t ident_len;
 608         char *date_end;
 609         unsigned long date;
 610
 611         ident_line = find_commit_header(buffer, "author", &ident_len);
 612         if (!ident_line)
 613                 goto fail_exit; /* no author line */
 614         if (split_ident_line(&ident, ident_line, ident_len) ||
 615             !ident.date_begin || !ident.date_end)
 616                 goto fail_exit; /* malformed "author" line */
 617
 618         date = strtoul(ident.date_begin, &date_end, 10);
 619         if (date_end != ident.date_end)
 620                 goto fail_exit; /* malformed date */
 621         *(author_date_slab_at(author_date, commit)) = date;
 622
 623 fail_exit:
 624         unuse_commit_buffer(commit, buffer);
 625 }
 626
 627 static int compare_commits_by_author_date(const void *a_, const void *b_,
 628                                           void *cb_data)
 629 {
 630         const struct commit *a = a_, *b = b_;
 631         struct author_date_slab *author_date = cb_data;
 632         unsigned long a_date = *(author_date_slab_at(author_date, a));
 633         unsigned long b_date = *(author_date_slab_at(author_date, b));
 634
 635         /* newer commits with larger date first */
 636         if (a_date < b_date)
 637                 return 1;
 638         else if (a_date > b_date)
 639                 return -1;
 640         return 0;
 641 }
 642
 643 int compare_commits_by_commit_date(const void *a_, const void *b_, void *unused)
 644 {
 645         const struct commit *a = a_, *b = b_;
 646         /* newer commits with larger date first */
 647         if (a->date < b->date)
 648                 return 1;
 649         else if (a->date > b->date)
 650                 return -1;
 651         return 0;
 652 }
 653
 654 /*
 655  * Performs an in-place topological sort on the list supplied.
 656  */
 657 void sort_in_topological_order(struct commit_list **list, enum rev_sort_order sort_order)
 658 {
 659         struct commit_list *next, *orig = *list;
 660         struct commit_list **pptr;
 661         struct indegree_slab indegree;
 662         struct prio_queue queue;
 663         struct commit *commit;
 664         struct author_date_slab author_date;
 665
 666         if (!orig)
 667                 return;
 668         *list = NULL;
 669
 670         init_indegree_slab(&indegree);
 671         memset(&queue, '\0', sizeof(queue));
 672
 673         switch (sort_order) {
 674         default: /* REV_SORT_IN_GRAPH_ORDER */
 675                 queue.compare = NULL;
 676                 break;
 677         case REV_SORT_BY_COMMIT_DATE:
 678                 queue.compare = compare_commits_by_commit_date;
 679                 break;
 680         case REV_SORT_BY_AUTHOR_DATE:
 681                 init_author_date_slab(&author_date);
 682                 queue.compare = compare_commits_by_author_date;
 683                 queue.cb_data = &author_date;
 684                 break;
 685         }
 686
 687         /* Mark them and clear the indegree */
 688         for (next = orig; next; next = next->next) {
 689                 struct commit *commit = next->item;
 690                 *(indegree_slab_at(&indegree, commit)) = 1;
 691                 /* also record the author dates, if needed */
 692                 if (sort_order == REV_SORT_BY_AUTHOR_DATE)
 693                         record_author_date(&author_date, commit);
 694         }
 695
 696         /* update the indegree */
 697         for (next = orig; next; next = next->next) {
 698                 struct commit_list *parents = next->item->parents;
 699                 while (parents) {
 700                         struct commit *parent = parents->item;
 701                         int *pi = indegree_slab_at(&indegree, parent);
 702
 703                         if (*pi)
 704                                 (*pi)++;
 705                         parents = parents->next;
 706                 }
 707         }
 708
 709         /*
 710          * find the tips
 711          *
 712          * tips are nodes not reachable from any other node in the list
 713          *
 714          * the tips serve as a starting set for the work queue.
 715          */
 716         for (next = orig; next; next = next->next) {
 717                 struct commit *commit = next->item;
 718
 719                 if (*(indegree_slab_at(&indegree, commit)) == 1)
 720                         prio_queue_put(&queue, commit);
 721         }
 722
 723         /*
 724          * This is unfortunate; the initial tips need to be shown
 725          * in the order given from the revision traversal machinery.
 726          */
 727         if (sort_order == REV_SORT_IN_GRAPH_ORDER)
 728                 prio_queue_reverse(&queue);
 729
 730         /* We no longer need the commit list */
 731         free_commit_list(orig);
 732
 733         pptr = list;
 734         *list = NULL;
 735         while ((commit = prio_queue_get(&queue)) != NULL) {
 736                 struct commit_list *parents;
 737
 738                 for (parents = commit->parents; parents ; parents = parents->next) {
 739                         struct commit *parent = parents->item;
 740                         int *pi = indegree_slab_at(&indegree, parent);
 741
 742                         if (!*pi)
 743                                 continue;
 744
 745                         /*
 746                          * parents are only enqueued for emission
 747                          * when all their children have been emitted thereby
 748                          * guaranteeing topological order.
 749                          */
 750                         if (--(*pi) == 1)
 751                                 prio_queue_put(&queue, parent);
 752                 }
 753                 /*
 754                  * all children of commit have already been
 755                  * emitted. we can emit it now.
 756                  */
 757                 *(indegree_slab_at(&indegree, commit)) = 0;
 758
 759                 pptr = &commit_list_insert(commit, pptr)->next;
 760         }
 761
 762         clear_indegree_slab(&indegree);
 763         clear_prio_queue(&queue);
 764         if (sort_order == REV_SORT_BY_AUTHOR_DATE)
 765                 clear_author_date_slab(&author_date);
 766 }
 767
 768 /* merge-base stuff */
 769
 770 /* Remember to update object flag allocation in object.h */
 771 #define PARENT1         (1u<<16)
 772 #define PARENT2         (1u<<17)
 773 #define STALE           (1u<<18)
 774 #define RESULT          (1u<<19)
 775
 776 static const unsigned all_flags = (PARENT1 | PARENT2 | STALE | RESULT);
 777
 778 static int queue_has_nonstale(struct prio_queue *queue)
 779 {
 780         int i;
 781         for (i = 0; i < queue->nr; i++) {
 782                 struct commit *commit = queue->array[i].data;
 783                 if (!(commit->object.flags & STALE))
 784                         return 1;
 785         }
 786         return 0;
 787 }
 788
 789 /* all input commits in one and twos[] must have been parsed! */
 790 static struct commit_list *paint_down_to_common(struct commit *one, int n, struct commit **twos)
 791 {
 792         struct prio_queue queue = { compare_commits_by_commit_date };
 793         struct commit_list *result = NULL;
 794         int i;
 795
 796         one->object.flags |= PARENT1;
 797         if (!n) {
 798                 commit_list_append(one, &result);
 799                 return result;
 800         }
 801         prio_queue_put(&queue, one);
 802
 803         for (i = 0; i < n; i++) {
 804                 twos[i]->object.flags |= PARENT2;
 805                 prio_queue_put(&queue, twos[i]);
 806         }
 807
 808         while (queue_has_nonstale(&queue)) {
 809                 struct commit *commit = prio_queue_get(&queue);
 810                 struct commit_list *parents;
 811                 int flags;
 812
 813                 flags = commit->object.flags & (PARENT1 | PARENT2 | STALE);
 814                 if (flags == (PARENT1 | PARENT2)) {
 815                         if (!(commit->object.flags & RESULT)) {
 816                                 commit->object.flags |= RESULT;
 817                                 commit_list_insert_by_date(commit, &result);
 818                         }
 819                         /* Mark parents of a found merge stale */
 820                         flags |= STALE;
 821                 }
 822                 parents = commit->parents;
 823                 while (parents) {
 824                         struct commit *p = parents->item;
 825                         parents = parents->next;
 826                         if ((p->object.flags & flags) == flags)
 827                                 continue;
 828                         if (parse_commit(p))
 829                                 return NULL;
 830                         p->object.flags |= flags;
 831                         prio_queue_put(&queue, p);
 832                 }
 833         }
 834
 835         clear_prio_queue(&queue);
 836         return result;
 837 }
 838
 839 static struct commit_list *merge_bases_many(struct commit *one, int n, struct commit **twos)
 840 {
 841         struct commit_list *list = NULL;
 842         struct commit_list *result = NULL;
 843         int i;
 844
 845         for (i = 0; i < n; i++) {
 846                 if (one == twos[i])
 847                         /*
 848                          * We do not mark this even with RESULT so we do not
 849                          * have to clean it up.
 850                          */
 851                         return commit_list_insert(one, &result);
 852         }
 853
 854         if (parse_commit(one))
 855                 return NULL;
 856         for (i = 0; i < n; i++) {
 857                 if (parse_commit(twos[i]))
 858                         return NULL;
 859         }
 860
 861         list = paint_down_to_common(one, n, twos);
 862
 863         while (list) {
 864                 struct commit_list *next = list->next;
 865                 if (!(list->item->object.flags & STALE))
 866                         commit_list_insert_by_date(list->item, &result);
 867                 free(list);
 868                 list = next;
 869         }
 870         return result;
 871 }
 872
 873 struct commit_list *get_octopus_merge_bases(struct commit_list *in)
 874 {
 875         struct commit_list *i, *j, *k, *ret = NULL;
 876
 877         if (!in)
 878                 return ret;
 879
 880         commit_list_insert(in->item, &ret);
 881
 882         for (i = in->next; i; i = i->next) {
 883                 struct commit_list *new = NULL, *end = NULL;
 884
 885                 for (j = ret; j; j = j->next) {
 886                         struct commit_list *bases;
 887                         bases = get_merge_bases(i->item, j->item);
 888                         if (!new)
 889                                 new = bases;
 890                         else
 891                                 end->next = bases;
 892                         for (k = bases; k; k = k->next)
 893                                 end = k;
 894                 }
 895                 ret = new;
 896         }
 897         return ret;
 898 }
 899
 900 static int remove_redundant(struct commit **array, int cnt)
 901 {
 902         /*
 903          * Some commit in the array may be an ancestor of
 904          * another commit.  Move such commit to the end of
 905          * the array, and return the number of commits that
 906          * are independent from each other.
 907          */
 908         struct commit **work;
 909         unsigned char *redundant;
 910         int *filled_index;
 911         int i, j, filled;
 912
 913         work = xcalloc(cnt, sizeof(*work));
 914         redundant = xcalloc(cnt, 1);
 915         filled_index = xmalloc(sizeof(*filled_index) * (cnt - 1));
 916
 917         for (i = 0; i < cnt; i++)
 918                 parse_commit(array[i]);
 919         for (i = 0; i < cnt; i++) {
 920                 struct commit_list *common;
 921
 922                 if (redundant[i])
 923                         continue;
 924                 for (j = filled = 0; j < cnt; j++) {
 925                         if (i == j || redundant[j])
 926                                 continue;
 927                         filled_index[filled] = j;
 928                         work[filled++] = array[j];
 929                 }
 930                 common = paint_down_to_common(array[i], filled, work);
 931                 if (array[i]->object.flags & PARENT2)
 932                         redundant[i] = 1;
 933                 for (j = 0; j < filled; j++)
 934                         if (work[j]->object.flags & PARENT1)
 935                                 redundant[filled_index[j]] = 1;
 936                 clear_commit_marks(array[i], all_flags);
 937                 for (j = 0; j < filled; j++)
 938                         clear_commit_marks(work[j], all_flags);
 939                 free_commit_list(common);
 940         }
 941
 942         /* Now collect the result */
 943         memcpy(work, array, sizeof(*array) * cnt);
 944         for (i = filled = 0; i < cnt; i++)
 945                 if (!redundant[i])
 946                         array[filled++] = work[i];
 947         for (j = filled, i = 0; i < cnt; i++)
 948                 if (redundant[i])
 949                         array[j++] = work[i];
 950         free(work);
 951         free(redundant);
 952         free(filled_index);
 953         return filled;
 954 }
 955
 956 static struct commit_list *get_merge_bases_many_0(struct commit *one,
 957                                                   int n,
 958                                                   struct commit **twos,
 959                                                   int cleanup)
 960 {
 961         struct commit_list *list;
 962         struct commit **rslt;
 963         struct commit_list *result;
 964         int cnt, i;
 965
 966         result = merge_bases_many(one, n, twos);
 967         for (i = 0; i < n; i++) {
 968                 if (one == twos[i])
 969                         return result;
 970         }
 971         if (!result || !result->next) {
 972                 if (cleanup) {
 973                         clear_commit_marks(one, all_flags);
 974                         clear_commit_marks_many(n, twos, all_flags);
 975                 }
 976                 return result;
 977         }
 978
 979         /* There are more than one */
 980         cnt = commit_list_count(result);
 981         rslt = xcalloc(cnt, sizeof(*rslt));
 982         for (list = result, i = 0; list; list = list->next)
 983                 rslt[i++] = list->item;
 984         free_commit_list(result);
 985
 986         clear_commit_marks(one, all_flags);
 987         clear_commit_marks_many(n, twos, all_flags);
 988
 989         cnt = remove_redundant(rslt, cnt);
 990         result = NULL;
 991         for (i = 0; i < cnt; i++)
 992                 commit_list_insert_by_date(rslt[i], &result);
 993         free(rslt);
 994         return result;
 995 }
 996
 997 struct commit_list *get_merge_bases_many(struct commit *one,
 998                                          int n,
 999                                          struct commit **twos)
1000 {
1001         return get_merge_bases_many_0(one, n, twos, 1);
1002 }
1003
1004 struct commit_list *get_merge_bases_many_dirty(struct commit *one,
1005                                                int n,
1006                                                struct commit **twos)
1007 {
1008         return get_merge_bases_many_0(one, n, twos, 0);
1009 }
1010
1011 struct commit_list *get_merge_bases(struct commit *one, struct commit *two)
1012 {
1013         return get_merge_bases_many_0(one, 1, &two, 1);
1014 }
1015
1016 /*
1017  * Is "commit" a descendant of one of the elements on the "with_commit" list?
1018  */
1019 int is_descendant_of(struct commit *commit, struct commit_list *with_commit)
1020 {
1021         if (!with_commit)
1022                 return 1;
1023         while (with_commit) {
1024                 struct commit *other;
1025
1026                 other = with_commit->item;
1027                 with_commit = with_commit->next;
1028                 if (in_merge_bases(other, commit))
1029                         return 1;
1030         }
1031         return 0;
1032 }
1033
1034 /*
1035  * Is "commit" an ancestor of one of the "references"?
1036  */
1037 int in_merge_bases_many(struct commit *commit, int nr_reference, struct commit **reference)
1038 {
1039         struct commit_list *bases;
1040         int ret = 0, i;
1041
1042         if (parse_commit(commit))
1043                 return ret;
1044         for (i = 0; i < nr_reference; i++)
1045                 if (parse_commit(reference[i]))
1046                         return ret;
1047
1048         bases = paint_down_to_common(commit, nr_reference, reference);
1049         if (commit->object.flags & PARENT2)
1050                 ret = 1;
1051         clear_commit_marks(commit, all_flags);
1052         clear_commit_marks_many(nr_reference, reference, all_flags);
1053         free_commit_list(bases);
1054         return ret;
1055 }
1056
1057 /*
1058  * Is "commit" an ancestor of (i.e. reachable from) the "reference"?
1059  */
1060 int in_merge_bases(struct commit *commit, struct commit *reference)
1061 {
1062         return in_merge_bases_many(commit, 1, &reference);
1063 }
1064
1065 struct commit_list *reduce_heads(struct commit_list *heads)
1066 {
1067         struct commit_list *p;
1068         struct commit_list *result = NULL, **tail = &result;
1069         struct commit **array;
1070         int num_head, i;
1071
1072         if (!heads)
1073                 return NULL;
1074
1075         /* Uniquify */
1076         for (p = heads; p; p = p->next)
1077                 p->item->object.flags &= ~STALE;
1078         for (p = heads, num_head = 0; p; p = p->next) {
1079                 if (p->item->object.flags & STALE)
1080                         continue;
1081                 p->item->object.flags |= STALE;
1082                 num_head++;
1083         }
1084         array = xcalloc(num_head, sizeof(*array));
1085         for (p = heads, i = 0; p; p = p->next) {
1086                 if (p->item->object.flags & STALE) {
1087                         array[i++] = p->item;
1088                         p->item->object.flags &= ~STALE;
1089                 }
1090         }
1091         num_head = remove_redundant(array, num_head);
1092         for (i = 0; i < num_head; i++)
1093                 tail = &commit_list_insert(array[i], tail)->next;
1094         return result;
1095 }
1096
1097 static const char gpg_sig_header[] = "gpgsig";
1098 static const int gpg_sig_header_len = sizeof(gpg_sig_header) - 1;
1099
1100 static int do_sign_commit(struct strbuf *buf, const char *keyid)
1101 {
1102         struct strbuf sig = STRBUF_INIT;
1103         int inspos, copypos;
1104
1105         /* find the end of the header */
1106         inspos = strstr(buf->buf, "\n\n") - buf->buf + 1;
1107
1108         if (!keyid || !*keyid)
1109                 keyid = get_signing_key();
1110         if (sign_buffer(buf, &sig, keyid)) {
1111                 strbuf_release(&sig);
1112                 return -1;
1113         }
1114
1115         for (copypos = 0; sig.buf[copypos]; ) {
1116                 const char *bol = sig.buf + copypos;
1117                 const char *eol = strchrnul(bol, '\n');
1118                 int len = (eol - bol) + !!*eol;
1119
1120                 if (!copypos) {
1121                         strbuf_insert(buf, inspos, gpg_sig_header, gpg_sig_header_len);
1122                         inspos += gpg_sig_header_len;
1123                 }
1124                 strbuf_insert(buf, inspos++, " ", 1);
1125                 strbuf_insert(buf, inspos, bol, len);
1126                 inspos += len;
1127                 copypos += len;
1128         }
1129         strbuf_release(&sig);
1130         return 0;
1131 }
1132
1133 int parse_signed_commit(const struct commit *commit,
1134                         struct strbuf *payload, struct strbuf *signature)
1135 {
1136
1137         unsigned long size;
1138         const char *buffer = get_commit_buffer(commit, &size);
1139         int in_signature, saw_signature = -1;
1140         const char *line, *tail;
1141
1142         line = buffer;
1143         tail = buffer + size;
1144         in_signature = 0;
1145         saw_signature = 0;
1146         while (line < tail) {
1147                 const char *sig = NULL;
1148                 const char *next = memchr(line, '\n', tail - line);
1149
1150                 next = next ? next + 1 : tail;
1151                 if (in_signature && line[0] == ' ')
1152                         sig = line + 1;
1153                 else if (starts_with(line, gpg_sig_header) &&
1154                          line[gpg_sig_header_len] == ' ')
1155                         sig = line + gpg_sig_header_len + 1;
1156                 if (sig) {
1157                         strbuf_add(signature, sig, next - sig);
1158                         saw_signature = 1;
1159                         in_signature = 1;
1160                 } else {
1161                         if (*line == '\n')
1162                                 /* dump the whole remainder of the buffer */
1163                                 next = tail;
1164                         strbuf_add(payload, line, next - line);
1165                         in_signature = 0;
1166                 }
1167                 line = next;
1168         }
1169         unuse_commit_buffer(commit, buffer);
1170         return saw_signature;
1171 }
1172
1173 int remove_signature(struct strbuf *buf)
1174 {
1175         const char *line = buf->buf;
1176         const char *tail = buf->buf + buf->len;
1177         int in_signature = 0;
1178         const char *sig_start = NULL;
1179         const char *sig_end = NULL;
1180
1181         while (line < tail) {
1182                 const char *next = memchr(line, '\n', tail - line);
1183                 next = next ? next + 1 : tail;
1184
1185                 if (in_signature && line[0] == ' ')
1186                         sig_end = next;
1187                 else if (starts_with(line, gpg_sig_header) &&
1188                          line[gpg_sig_header_len] == ' ') {
1189                         sig_start = line;
1190                         sig_end = next;
1191                         in_signature = 1;
1192                 } else {
1193                         if (*line == '\n')
1194                                 /* dump the whole remainder of the buffer */
1195                                 next = tail;
1196                         in_signature = 0;
1197                 }
1198                 line = next;
1199         }
1200
1201         if (sig_start)
1202                 strbuf_remove(buf, sig_start - buf->buf, sig_end - sig_start);
1203
1204         return sig_start != NULL;
1205 }
1206
1207 static void handle_signed_tag(struct commit *parent, struct commit_extra_header ***tail)
1208 {
1209         struct merge_remote_desc *desc;
1210         struct commit_extra_header *mergetag;
1211         char *buf;
1212         unsigned long size, len;
1213         enum object_type type;
1214
1215         desc = merge_remote_util(parent);
1216         if (!desc || !desc->obj)
1217                 return;
1218         buf = read_sha1_file(desc->obj->sha1, &type, &size);
1219         if (!buf || type != OBJ_TAG)
1220                 goto free_return;
1221         len = parse_signature(buf, size);
1222         if (size == len)
1223                 goto free_return;
1224         /*
1225          * We could verify this signature and either omit the tag when
1226          * it does not validate, but the integrator may not have the
1227          * public key of the signer of the tag he is merging, while a
1228          * later auditor may have it while auditing, so let's not run
1229          * verify-signed-buffer here for now...
1230          *
1231          * if (verify_signed_buffer(buf, len, buf + len, size - len, ...))
1232          *      warn("warning: signed tag unverified.");
1233          */
1234         mergetag = xcalloc(1, sizeof(*mergetag));
1235         mergetag->key = xstrdup("mergetag");
1236         mergetag->value = buf;
1237         mergetag->len = size;
1238
1239         **tail = mergetag;
1240         *tail = &mergetag->next;
1241         return;
1242
1243 free_return:
1244         free(buf);
1245 }
1246
1247 int check_commit_signature(const struct commit *commit, struct signature_check *sigc)
1248 {
1249         struct strbuf payload = STRBUF_INIT;
1250         struct strbuf signature = STRBUF_INIT;
1251         int ret = 1;
1252
1253         sigc->result = 'N';
1254
1255         if (parse_signed_commit(commit, &payload, &signature) <= 0)
1256                 goto out;
1257         ret = check_signature(payload.buf, payload.len, signature.buf,
1258                 signature.len, sigc);
1259
1260  out:
1261         strbuf_release(&payload);
1262         strbuf_release(&signature);
1263
1264         return ret;
1265 }
1266
1267
1268
1269 void append_merge_tag_headers(struct commit_list *parents,
1270                               struct commit_extra_header ***tail)
1271 {
1272         while (parents) {
1273                 struct commit *parent = parents->item;
1274                 handle_signed_tag(parent, tail);
1275                 parents = parents->next;
1276         }
1277 }
1278
1279 static void add_extra_header(struct strbuf *buffer,
1280                              struct commit_extra_header *extra)
1281 {
1282         strbuf_addstr(buffer, extra->key);
1283         if (extra->len)
1284                 strbuf_add_lines(buffer, " ", extra->value, extra->len);
1285         else
1286                 strbuf_addch(buffer, '\n');
1287 }
1288
1289 struct commit_extra_header *read_commit_extra_headers(struct commit *commit,
1290                                                       const char **exclude)
1291 {
1292         struct commit_extra_header *extra = NULL;
1293         unsigned long size;
1294         const char *buffer = get_commit_buffer(commit, &size);
1295         extra = read_commit_extra_header_lines(buffer, size, exclude);
1296         unuse_commit_buffer(commit, buffer);
1297         return extra;
1298 }
1299
1300 void for_each_mergetag(each_mergetag_fn fn, struct commit *commit, void *data)
1301 {
1302         struct commit_extra_header *extra, *to_free;
1303
1304         to_free = read_commit_extra_headers(commit, NULL);
1305         for (extra = to_free; extra; extra = extra->next) {
1306                 if (strcmp(extra->key, "mergetag"))
1307                         continue; /* not a merge tag */
1308                 fn(commit, extra, data);
1309         }
1310         free_commit_extra_headers(to_free);
1311 }
1312
1313 static inline int standard_header_field(const char *field, size_t len)
1314 {
1315         return ((len == 4 && !memcmp(field, "tree ", 5)) ||
1316                 (len == 6 && !memcmp(field, "parent ", 7)) ||
1317                 (len == 6 && !memcmp(field, "author ", 7)) ||
1318                 (len == 9 && !memcmp(field, "committer ", 10)) ||
1319                 (len == 8 && !memcmp(field, "encoding ", 9)));
1320 }
1321
1322 static int excluded_header_field(const char *field, size_t len, const char **exclude)
1323 {
1324         if (!exclude)
1325                 return 0;
1326
1327         while (*exclude) {
1328                 size_t xlen = strlen(*exclude);
1329                 if (len == xlen &&
1330                     !memcmp(field, *exclude, xlen) && field[xlen] == ' ')
1331                         return 1;
1332                 exclude++;
1333         }
1334         return 0;
1335 }
1336
1337 static struct commit_extra_header *read_commit_extra_header_lines(
1338         const char *buffer, size_t size,
1339         const char **exclude)
1340 {
1341         struct commit_extra_header *extra = NULL, **tail = &extra, *it = NULL;
1342         const char *line, *next, *eof, *eob;
1343         struct strbuf buf = STRBUF_INIT;
1344
1345         for (line = buffer, eob = line + size;
1346              line < eob && *line != '\n';
1347              line = next) {
1348                 next = memchr(line, '\n', eob - line);
1349                 next = next ? next + 1 : eob;
1350                 if (*line == ' ') {
1351                         /* continuation */
1352                         if (it)
1353                                 strbuf_add(&buf, line + 1, next - (line + 1));
1354                         continue;
1355                 }
1356                 if (it)
1357                         it->value = strbuf_detach(&buf, &it->len);
1358                 strbuf_reset(&buf);
1359                 it = NULL;
1360
1361                 eof = strchr(line, ' ');
1362                 if (next <= eof)
1363                         eof = next;
1364
1365                 if (standard_header_field(line, eof - line) ||
1366                     excluded_header_field(line, eof - line, exclude))
1367                         continue;
1368
1369                 it = xcalloc(1, sizeof(*it));
1370                 it->key = xmemdupz(line, eof-line);
1371                 *tail = it;
1372                 tail = &it->next;
1373                 if (eof + 1 < next)
1374                         strbuf_add(&buf, eof + 1, next - (eof + 1));
1375         }
1376         if (it)
1377                 it->value = strbuf_detach(&buf, &it->len);
1378         return extra;
1379 }
1380
1381 void free_commit_extra_headers(struct commit_extra_header *extra)
1382 {
1383         while (extra) {
1384                 struct commit_extra_header *next = extra->next;
1385                 free(extra->key);
1386                 free(extra->value);
1387                 free(extra);
1388                 extra = next;
1389         }
1390 }
1391
1392 int commit_tree(const char *msg, size_t msg_len,
1393                 const unsigned char *tree,
1394                 struct commit_list *parents, unsigned char *ret,
1395                 const char *author, const char *sign_commit)
1396 {
1397         struct commit_extra_header *extra = NULL, **tail = &extra;
1398         int result;
1399
1400         append_merge_tag_headers(parents, &tail);
1401         result = commit_tree_extended(msg, msg_len, tree, parents, ret,
1402                                       author, sign_commit, extra);
1403         free_commit_extra_headers(extra);
1404         return result;
1405 }
1406
1407 static int find_invalid_utf8(const char *buf, int len)
1408 {
1409         int offset = 0;
1410         static const unsigned int max_codepoint[] = {
1411                 0x7f, 0x7ff, 0xffff, 0x10ffff
1412         };
1413
1414         while (len) {
1415                 unsigned char c = *buf++;
1416                 int bytes, bad_offset;
1417                 unsigned int codepoint;
1418                 unsigned int min_val, max_val;
1419
1420                 len--;
1421                 offset++;
1422
1423                 /* Simple US-ASCII? No worries. */
1424                 if (c < 0x80)
1425                         continue;
1426
1427                 bad_offset = offset-1;
1428
1429                 /*
1430                  * Count how many more high bits set: that's how
1431                  * many more bytes this sequence should have.
1432                  */
1433                 bytes = 0;
1434                 while (c & 0x40) {
1435                         c <<= 1;
1436                         bytes++;
1437                 }
1438
1439                 /*
1440                  * Must be between 1 and 3 more bytes.  Longer sequences result in
1441                  * codepoints beyond U+10FFFF, which are guaranteed never to exist.
1442                  */
1443                 if (bytes < 1 || 3 < bytes)
1444                         return bad_offset;
1445
1446                 /* Do we *have* that many bytes? */
1447                 if (len < bytes)
1448                         return bad_offset;
1449
1450                 /*
1451                  * Place the encoded bits at the bottom of the value and compute the
1452                  * valid range.
1453                  */
1454                 codepoint = (c & 0x7f) >> bytes;
1455                 min_val = max_codepoint[bytes-1] + 1;
1456                 max_val = max_codepoint[bytes];
1457
1458                 offset += bytes;
1459                 len -= bytes;
1460
1461                 /* And verify that they are good continuation bytes */
1462                 do {
1463                         codepoint <<= 6;
1464                         codepoint |= *buf & 0x3f;
1465                         if ((*buf++ & 0xc0) != 0x80)
1466                                 return bad_offset;
1467                 } while (--bytes);
1468
1469                 /* Reject codepoints that are out of range for the sequence length. */
1470                 if (codepoint < min_val || codepoint > max_val)
1471                         return bad_offset;
1472                 /* Surrogates are only for UTF-16 and cannot be encoded in UTF-8. */
1473                 if ((codepoint & 0x1ff800) == 0xd800)
1474                         return bad_offset;
1475                 /* U+xxFFFE and U+xxFFFF are guaranteed non-characters. */
1476                 if ((codepoint & 0xfffe) == 0xfffe)
1477                         return bad_offset;
1478                 /* So are anything in the range U+FDD0..U+FDEF. */
1479                 if (codepoint >= 0xfdd0 && codepoint <= 0xfdef)
1480                         return bad_offset;
1481         }
1482         return -1;
1483 }
1484
1485 /*
1486  * This verifies that the buffer is in proper utf8 format.
1487  *
1488  * If it isn't, it assumes any non-utf8 characters are Latin1,
1489  * and does the conversion.
1490  */
1491 static int verify_utf8(struct strbuf *buf)
1492 {
1493         int ok = 1;
1494         long pos = 0;
1495
1496         for (;;) {
1497                 int bad;
1498                 unsigned char c;
1499                 unsigned char replace[2];
1500
1501                 bad = find_invalid_utf8(buf->buf + pos, buf->len - pos);
1502                 if (bad < 0)
1503                         return ok;
1504                 pos += bad;
1505                 ok = 0;
1506                 c = buf->buf[pos];
1507                 strbuf_remove(buf, pos, 1);
1508
1509                 /* We know 'c' must be in the range 128-255 */
1510                 replace[0] = 0xc0 + (c >> 6);
1511                 replace[1] = 0x80 + (c & 0x3f);
1512                 strbuf_insert(buf, pos, replace, 2);
1513                 pos += 2;
1514         }
1515 }
1516
1517 static const char commit_utf8_warn[] =
1518 "Warning: commit message did not conform to UTF-8.\n"
1519 "You may want to amend it after fixing the message, or set the config\n"
1520 "variable i18n.commitencoding to the encoding your project uses.\n";
1521
1522 int commit_tree_extended(const char *msg, size_t msg_len,
1523                          const unsigned char *tree,
1524                          struct commit_list *parents, unsigned char *ret,
1525                          const char *author, const char *sign_commit,
1526                          struct commit_extra_header *extra)
1527 {
1528         int result;
1529         int encoding_is_utf8;
1530         struct strbuf buffer;
1531
1532         assert_sha1_type(tree, OBJ_TREE);
1533
1534         if (memchr(msg, '\0', msg_len))
1535                 return error("a NUL byte in commit log message not allowed.");
1536
1537         /* Not having i18n.commitencoding is the same as having utf-8 */
1538         encoding_is_utf8 = is_encoding_utf8(git_commit_encoding);
1539
1540         strbuf_init(&buffer, 8192); /* should avoid reallocs for the headers */
1541         strbuf_addf(&buffer, "tree %s\n", sha1_to_hex(tree));
1542
1543         /*
1544          * NOTE! This ordering means that the same exact tree merged with a
1545          * different order of parents will be a _different_ changeset even
1546          * if everything else stays the same.
1547          */
1548         while (parents) {
1549                 struct commit_list *next = parents->next;
1550                 struct commit *parent = parents->item;
1551
1552                 strbuf_addf(&buffer, "parent %s\n",
1553                             sha1_to_hex(parent->object.sha1));
1554                 free(parents);
1555                 parents = next;
1556         }
1557
1558         /* Person/date information */
1559         if (!author)
1560                 author = git_author_info(IDENT_STRICT);
1561         strbuf_addf(&buffer, "author %s\n", author);
1562         strbuf_addf(&buffer, "committer %s\n", git_committer_info(IDENT_STRICT));
1563         if (!encoding_is_utf8)
1564                 strbuf_addf(&buffer, "encoding %s\n", git_commit_encoding);
1565
1566         while (extra) {
1567                 add_extra_header(&buffer, extra);
1568                 extra = extra->next;
1569         }
1570         strbuf_addch(&buffer, '\n');
1571
1572         /* And add the comment */
1573         strbuf_add(&buffer, msg, msg_len);
1574
1575         /* And check the encoding */
1576         if (encoding_is_utf8 && !verify_utf8(&buffer))
1577                 fprintf(stderr, commit_utf8_warn);
1578
1579         if (sign_commit && do_sign_commit(&buffer, sign_commit))
1580                 return -1;
1581
1582         result = write_sha1_file(buffer.buf, buffer.len, commit_type, ret);
1583         strbuf_release(&buffer);
1584         return result;
1585 }
1586
1587 struct commit *get_merge_parent(const char *name)
1588 {
1589         struct object *obj;
1590         struct commit *commit;
1591         struct object_id oid;
1592         if (get_sha1(name, oid.hash))
1593                 return NULL;
1594         obj = parse_object(oid.hash);
1595         commit = (struct commit *)peel_to_type(name, 0, obj, OBJ_COMMIT);
1596         if (commit && !commit->util) {
1597                 struct merge_remote_desc *desc;
1598                 desc = xmalloc(sizeof(*desc));
1599                 desc->obj = obj;
1600                 desc->name = strdup(name);
1601                 commit->util = desc;
1602         }
1603         return commit;
1604 }
1605
1606 /*
1607  * Append a commit to the end of the commit_list.
1608  *
1609  * next starts by pointing to the variable that holds the head of an
1610  * empty commit_list, and is updated to point to the "next" field of
1611  * the last item on the list as new commits are appended.
1612  *
1613  * Usage example:
1614  *
1615  *     struct commit_list *list;
1616  *     struct commit_list **next = &list;
1617  *
1618  *     next = commit_list_append(c1, next);
1619  *     next = commit_list_append(c2, next);
1620  *     assert(commit_list_count(list) == 2);
1621  *     return list;
1622  */
1623 struct commit_list **commit_list_append(struct commit *commit,
1624                                         struct commit_list **next)
1625 {
1626         struct commit_list *new = xmalloc(sizeof(struct commit_list));
1627         new->item = commit;
1628         *next = new;
1629         new->next = NULL;
1630         return &new->next;
1631 }
1632
1633 void print_commit_list(struct commit_list *list,
1634                        const char *format_cur,
1635                        const char *format_last)
1636 {
1637         for ( ; list; list = list->next) {
1638                 const char *format = list->next ? format_cur : format_last;
1639                 printf(format, sha1_to_hex(list->item->object.sha1));
1640         }
1641 }
1642
1643 const char *find_commit_header(const char *msg, const char *key, size_t *out_len)
1644 {
1645         int key_len = strlen(key);
1646         const char *line = msg;
1647
1648         while (line) {
1649                 const char *eol = strchrnul(line, '\n');
1650
1651                 if (line == eol)
1652                         return NULL;
1653
1654                 if (eol - line > key_len &&
1655                     !strncmp(line, key, key_len) &&
1656                     line[key_len] == ' ') {
1657                         *out_len = eol - line - key_len - 1;
1658                         return line + key_len + 1;
1659                 }
1660                 line = *eol ? eol + 1 : NULL;
1661         }
1662         return NULL;
1663 }
1664
1665 /*
1666  * Inspect sb and determine the true "end" of the log message, in
1667  * order to find where to put a new Signed-off-by: line.  Ignored are
1668  * trailing comment lines and blank lines, and also the traditional
1669  * "Conflicts:" block that is not commented out, so that we can use
1670  * "git commit -s --amend" on an existing commit that forgot to remove
1671  * it.
1672  *
1673  * Returns the number of bytes from the tail to ignore, to be fed as
1674  * the second parameter to append_signoff().
1675  */
1676 int ignore_non_trailer(struct strbuf *sb)
1677 {
1678         int boc = 0;
1679         int bol = 0;
1680         int in_old_conflicts_block = 0;
1681
1682         while (bol < sb->len) {
1683                 char *next_line;
1684
1685                 if (!(next_line = memchr(sb->buf + bol, '\n', sb->len - bol)))
1686                         next_line = sb->buf + sb->len;
1687                 else
1688                         next_line++;
1689
1690                 if (sb->buf[bol] == comment_line_char || sb->buf[bol] == '\n') {
1691                         /* is this the first of the run of comments? */
1692                         if (!boc)
1693                                 boc = bol;
1694                         /* otherwise, it is just continuing */
1695                 } else if (starts_with(sb->buf + bol, "Conflicts:\n")) {
1696                         in_old_conflicts_block = 1;
1697                         if (!boc)
1698                                 boc = bol;
1699                 } else if (in_old_conflicts_block && sb->buf[bol] == '\t') {
1700                         ; /* a pathname in the conflicts block */
1701                 } else if (boc) {
1702                         /* the previous was not trailing comment */
1703                         boc = 0;
1704                         in_old_conflicts_block = 0;
1705                 }
1706                 bol = next_line - sb->buf;
1707         }
1708         return boc ? sb->len - boc : 0;
1709 }