commit.c

   1 #include "cache.h"
   2 #include "tag.h"
   3 #include "commit.h"
   4 #include "pkt-line.h"
   5 #include "utf8.h"
   6 #include "diff.h"
   7 #include "revision.h"
   8 #include "notes.h"
   9 #include "gpg-interface.h"
  10 #include "mergesort.h"
  11 #include "commit-slab.h"
  12 #include "prio-queue.h"
  13 #include "sha1-lookup.h"
  14
  15 static struct commit_extra_header *read_commit_extra_header_lines(const char *buf, size_t len, const char **);
  16
  17 int save_commit_buffer = 1;
  18
  19 const char *commit_type = "commit";
  20
  21 static struct commit *check_commit(struct object *obj,
  22                                    const unsigned char *sha1,
  23                                    int quiet)
  24 {
  25         if (obj->type != OBJ_COMMIT) {
  26                 if (!quiet)
  27                         error("Object %s is a %s, not a commit",
  28                               sha1_to_hex(sha1), typename(obj->type));
  29                 return NULL;
  30         }
  31         return (struct commit *) obj;
  32 }
  33
  34 struct commit *lookup_commit_reference_gently(const unsigned char *sha1,
  35                                               int quiet)
  36 {
  37         struct object *obj = deref_tag(parse_object(sha1), NULL, 0);
  38
  39         if (!obj)
  40                 return NULL;
  41         return check_commit(obj, sha1, quiet);
  42 }
  43
  44 struct commit *lookup_commit_reference(const unsigned char *sha1)
  45 {
  46         return lookup_commit_reference_gently(sha1, 0);
  47 }
  48
  49 struct commit *lookup_commit_or_die(const unsigned char *sha1, const char *ref_name)
  50 {
  51         struct commit *c = lookup_commit_reference(sha1);
  52         if (!c)
  53                 die(_("could not parse %s"), ref_name);
  54         if (hashcmp(sha1, c->object.sha1)) {
  55                 warning(_("%s %s is not a commit!"),
  56                         ref_name, sha1_to_hex(sha1));
  57         }
  58         return c;
  59 }
  60
  61 struct commit *lookup_commit(const unsigned char *sha1)
  62 {
  63         struct object *obj = lookup_object(sha1);
  64         if (!obj)
  65                 return create_object(sha1, alloc_commit_node());
  66         if (!obj->type)
  67                 obj->type = OBJ_COMMIT;
  68         return check_commit(obj, sha1, 0);
  69 }
  70
  71 struct commit *lookup_commit_reference_by_name(const char *name)
  72 {
  73         unsigned char sha1[20];
  74         struct commit *commit;
  75
  76         if (get_sha1_committish(name, sha1))
  77                 return NULL;
  78         commit = lookup_commit_reference(sha1);
  79         if (parse_commit(commit))
  80                 return NULL;
  81         return commit;
  82 }
  83
  84 static unsigned long parse_commit_date(const char *buf, const char *tail)
  85 {
  86         const char *dateptr;
  87
  88         if (buf + 6 >= tail)
  89                 return 0;
  90         if (memcmp(buf, "author", 6))
  91                 return 0;
  92         while (buf < tail && *buf++ != '\n')
  93                 /* nada */;
  94         if (buf + 9 >= tail)
  95                 return 0;
  96         if (memcmp(buf, "committer", 9))
  97                 return 0;
  98         while (buf < tail && *buf++ != '>')
  99                 /* nada */;
 100         if (buf >= tail)
 101                 return 0;
 102         dateptr = buf;
 103         while (buf < tail && *buf++ != '\n')
 104                 /* nada */;
 105         if (buf >= tail)
 106                 return 0;
 107         /* dateptr < buf && buf[-1] == '\n', so strtoul will stop at buf-1 */
 108         return strtoul(dateptr, NULL, 10);
 109 }
 110
 111 static struct commit_graft **commit_graft;
 112 static int commit_graft_alloc, commit_graft_nr;
 113
 114 static const unsigned char *commit_graft_sha1_access(size_t index, void *table)
 115 {
 116         struct commit_graft **commit_graft_table = table;
 117         return commit_graft_table[index]->sha1;
 118 }
 119
 120 static int commit_graft_pos(const unsigned char *sha1)
 121 {
 122         return sha1_pos(sha1, commit_graft, commit_graft_nr,
 123                         commit_graft_sha1_access);
 124 }
 125
 126 int register_commit_graft(struct commit_graft *graft, int ignore_dups)
 127 {
 128         int pos = commit_graft_pos(graft->sha1);
 129
 130         if (0 <= pos) {
 131                 if (ignore_dups)
 132                         free(graft);
 133                 else {
 134                         free(commit_graft[pos]);
 135                         commit_graft[pos] = graft;
 136                 }
 137                 return 1;
 138         }
 139         pos = -pos - 1;
 140         ALLOC_GROW(commit_graft, commit_graft_nr + 1, commit_graft_alloc);
 141         commit_graft_nr++;
 142         if (pos < commit_graft_nr)
 143                 memmove(commit_graft + pos + 1,
 144                         commit_graft + pos,
 145                         (commit_graft_nr - pos - 1) *
 146                         sizeof(*commit_graft));
 147         commit_graft[pos] = graft;
 148         return 0;
 149 }
 150
 151 struct commit_graft *read_graft_line(char *buf, int len)
 152 {
 153         /* The format is just "Commit Parent1 Parent2 ...\n" */
 154         int i;
 155         struct commit_graft *graft = NULL;
 156
 157         while (len && isspace(buf[len-1]))
 158                 buf[--len] = '\0';
 159         if (buf[0] == '#' || buf[0] == '\0')
 160                 return NULL;
 161         if ((len + 1) % 41)
 162                 goto bad_graft_data;
 163         i = (len + 1) / 41 - 1;
 164         graft = xmalloc(sizeof(*graft) + 20 * i);
 165         graft->nr_parent = i;
 166         if (get_sha1_hex(buf, graft->sha1))
 167                 goto bad_graft_data;
 168         for (i = 40; i < len; i += 41) {
 169                 if (buf[i] != ' ')
 170                         goto bad_graft_data;
 171                 if (get_sha1_hex(buf + i + 1, graft->parent[i/41]))
 172                         goto bad_graft_data;
 173         }
 174         return graft;
 175
 176 bad_graft_data:
 177         error("bad graft data: %s", buf);
 178         free(graft);
 179         return NULL;
 180 }
 181
 182 static int read_graft_file(const char *graft_file)
 183 {
 184         FILE *fp = fopen(graft_file, "r");
 185         struct strbuf buf = STRBUF_INIT;
 186         if (!fp)
 187                 return -1;
 188         while (!strbuf_getwholeline(&buf, fp, '\n')) {
 189                 /* The format is just "Commit Parent1 Parent2 ...\n" */
 190                 struct commit_graft *graft = read_graft_line(buf.buf, buf.len);
 191                 if (!graft)
 192                         continue;
 193                 if (register_commit_graft(graft, 1))
 194                         error("duplicate graft data: %s", buf.buf);
 195         }
 196         fclose(fp);
 197         strbuf_release(&buf);
 198         return 0;
 199 }
 200
 201 static void prepare_commit_graft(void)
 202 {
 203         static int commit_graft_prepared;
 204         char *graft_file;
 205
 206         if (commit_graft_prepared)
 207                 return;
 208         graft_file = get_graft_file();
 209         read_graft_file(graft_file);
 210         /* make sure shallows are read */
 211         is_repository_shallow();
 212         commit_graft_prepared = 1;
 213 }
 214
 215 struct commit_graft *lookup_commit_graft(const unsigned char *sha1)
 216 {
 217         int pos;
 218         prepare_commit_graft();
 219         pos = commit_graft_pos(sha1);
 220         if (pos < 0)
 221                 return NULL;
 222         return commit_graft[pos];
 223 }
 224
 225 int for_each_commit_graft(each_commit_graft_fn fn, void *cb_data)
 226 {
 227         int i, ret;
 228         for (i = ret = 0; i < commit_graft_nr && !ret; i++)
 229                 ret = fn(commit_graft[i], cb_data);
 230         return ret;
 231 }
 232
 233 int unregister_shallow(const unsigned char *sha1)
 234 {
 235         int pos = commit_graft_pos(sha1);
 236         if (pos < 0)
 237                 return -1;
 238         if (pos + 1 < commit_graft_nr)
 239                 memmove(commit_graft + pos, commit_graft + pos + 1,
 240                                 sizeof(struct commit_graft *)
 241                                 * (commit_graft_nr - pos - 1));
 242         commit_graft_nr--;
 243         return 0;
 244 }
 245
 246 struct commit_buffer {
 247         void *buffer;
 248         unsigned long size;
 249 };
 250 define_commit_slab(buffer_slab, struct commit_buffer);
 251 static struct buffer_slab buffer_slab = COMMIT_SLAB_INIT(1, buffer_slab);
 252
 253 void set_commit_buffer(struct commit *commit, void *buffer, unsigned long size)
 254 {
 255         struct commit_buffer *v = buffer_slab_at(&buffer_slab, commit);
 256         v->buffer = buffer;
 257         v->size = size;
 258 }
 259
 260 const void *get_cached_commit_buffer(const struct commit *commit, unsigned long *sizep)
 261 {
 262         struct commit_buffer *v = buffer_slab_at(&buffer_slab, commit);
 263         if (sizep)
 264                 *sizep = v->size;
 265         return v->buffer;
 266 }
 267
 268 const void *get_commit_buffer(const struct commit *commit, unsigned long *sizep)
 269 {
 270         const void *ret = get_cached_commit_buffer(commit, sizep);
 271         if (!ret) {
 272                 enum object_type type;
 273                 unsigned long size;
 274                 ret = read_sha1_file(commit->object.sha1, &type, &size);
 275                 if (!ret)
 276                         die("cannot read commit object %s",
 277                             sha1_to_hex(commit->object.sha1));
 278                 if (type != OBJ_COMMIT)
 279                         die("expected commit for %s, got %s",
 280                             sha1_to_hex(commit->object.sha1), typename(type));
 281                 if (sizep)
 282                         *sizep = size;
 283         }
 284         return ret;
 285 }
 286
 287 void unuse_commit_buffer(const struct commit *commit, const void *buffer)
 288 {
 289         struct commit_buffer *v = buffer_slab_at(&buffer_slab, commit);
 290         if (v->buffer != buffer)
 291                 free((void *)buffer);
 292 }
 293
 294 void free_commit_buffer(struct commit *commit)
 295 {
 296         struct commit_buffer *v = buffer_slab_at(&buffer_slab, commit);
 297         free(v->buffer);
 298         v->buffer = NULL;
 299         v->size = 0;
 300 }
 301
 302 const void *detach_commit_buffer(struct commit *commit, unsigned long *sizep)
 303 {
 304         struct commit_buffer *v = buffer_slab_at(&buffer_slab, commit);
 305         void *ret;
 306
 307         ret = v->buffer;
 308         if (sizep)
 309                 *sizep = v->size;
 310
 311         v->buffer = NULL;
 312         v->size = 0;
 313         return ret;
 314 }
 315
 316 int parse_commit_buffer(struct commit *item, const void *buffer, unsigned long size)
 317 {
 318         const char *tail = buffer;
 319         const char *bufptr = buffer;
 320         unsigned char parent[20];
 321         struct commit_list **pptr;
 322         struct commit_graft *graft;
 323
 324         if (item->object.parsed)
 325                 return 0;
 326         item->object.parsed = 1;
 327         tail += size;
 328         if (tail <= bufptr + 46 || memcmp(bufptr, "tree ", 5) || bufptr[45] != '\n')
 329                 return error("bogus commit object %s", sha1_to_hex(item->object.sha1));
 330         if (get_sha1_hex(bufptr + 5, parent) < 0)
 331                 return error("bad tree pointer in commit %s",
 332                              sha1_to_hex(item->object.sha1));
 333         item->tree = lookup_tree(parent);
 334         bufptr += 46; /* "tree " + "hex sha1" + "\n" */
 335         pptr = &item->parents;
 336
 337         graft = lookup_commit_graft(item->object.sha1);
 338         while (bufptr + 48 < tail && !memcmp(bufptr, "parent ", 7)) {
 339                 struct commit *new_parent;
 340
 341                 if (tail <= bufptr + 48 ||
 342                     get_sha1_hex(bufptr + 7, parent) ||
 343                     bufptr[47] != '\n')
 344                         return error("bad parents in commit %s", sha1_to_hex(item->object.sha1));
 345                 bufptr += 48;
 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(struct commit *item)
 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.sha1, &type, &size);
 383         if (!buffer)
 384                 return error("Could not read %s",
 385                              sha1_to_hex(item->object.sha1));
 386         if (type != OBJ_COMMIT) {
 387                 free(buffer);
 388                 return error("Object %s not a commit",
 389                              sha1_to_hex(item->object.sha1));
 390         }
 391         ret = parse_commit_buffer(item, buffer, size);
 392         if (save_commit_buffer && !ret) {
 393                 set_commit_buffer(item, buffer, size);
 394                 return 0;
 395         }
 396         free(buffer);
 397         return ret;
 398 }
 399
 400 void parse_commit_or_die(struct commit *item)
 401 {
 402         if (parse_commit(item))
 403                 die("unable to parse commit %s",
 404                     item ? sha1_to_hex(item->object.sha1) : "(null)");
 405 }
 406
 407 int find_commit_subject(const char *commit_buffer, const char **subject)
 408 {
 409         const char *eol;
 410         const char *p = commit_buffer;
 411
 412         while (*p && (*p != '\n' || p[1] != '\n'))
 413                 p++;
 414         if (*p) {
 415                 p += 2;
 416                 for (eol = p; *eol && *eol != '\n'; eol++)
 417                         ; /* do nothing */
 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                 struct commit_list *new;
 449                 new = xmalloc(sizeof(struct commit_list));
 450                 new->item = list->item;
 451                 new->next = NULL;
 452                 *pp = new;
 453                 pp = &new->next;
 454                 list = list->next;
 455         }
 456         return head;
 457 }
 458
 459 void free_commit_list(struct commit_list *list)
 460 {
 461         while (list) {
 462                 struct commit_list *temp = list;
 463                 list = temp->next;
 464                 free(temp);
 465         }
 466 }
 467
 468 struct commit_list * commit_list_insert_by_date(struct commit *item, struct commit_list **list)
 469 {
 470         struct commit_list **pp = list;
 471         struct commit_list *p;
 472         while ((p = *pp) != NULL) {
 473                 if (p->item->date < item->date) {
 474                         break;
 475                 }
 476                 pp = &p->next;
 477         }
 478         return commit_list_insert(item, pp);
 479 }
 480
 481 static int commit_list_compare_by_date(const void *a, const void *b)
 482 {
 483         unsigned long a_date = ((const struct commit_list *)a)->item->date;
 484         unsigned long b_date = ((const struct commit_list *)b)->item->date;
 485         if (a_date < b_date)
 486                 return 1;
 487         if (a_date > b_date)
 488                 return -1;
 489         return 0;
 490 }
 491
 492 static void *commit_list_get_next(const void *a)
 493 {
 494         return ((const struct commit_list *)a)->next;
 495 }
 496
 497 static void commit_list_set_next(void *a, void *next)
 498 {
 499         ((struct commit_list *)a)->next = next;
 500 }
 501
 502 void commit_list_sort_by_date(struct commit_list **list)
 503 {
 504         *list = llist_mergesort(*list, commit_list_get_next, commit_list_set_next,
 505                                 commit_list_compare_by_date);
 506 }
 507
 508 struct commit *pop_most_recent_commit(struct commit_list **list,
 509                                       unsigned int mark)
 510 {
 511         struct commit *ret = (*list)->item;
 512         struct commit_list *parents = ret->parents;
 513         struct commit_list *old = *list;
 514
 515         *list = (*list)->next;
 516         free(old);
 517
 518         while (parents) {
 519                 struct commit *commit = parents->item;
 520                 if (!parse_commit(commit) && !(commit->object.flags & mark)) {
 521                         commit->object.flags |= mark;
 522                         commit_list_insert_by_date(commit, list);
 523                 }
 524                 parents = parents->next;
 525         }
 526         return ret;
 527 }
 528
 529 static void clear_commit_marks_1(struct commit_list **plist,
 530                                  struct commit *commit, unsigned int mark)
 531 {
 532         while (commit) {
 533                 struct commit_list *parents;
 534
 535                 if (!(mark & commit->object.flags))
 536                         return;
 537
 538                 commit->object.flags &= ~mark;
 539
 540                 parents = commit->parents;
 541                 if (!parents)
 542                         return;
 543
 544                 while ((parents = parents->next))
 545                         commit_list_insert(parents->item, plist);
 546
 547                 commit = commit->parents->item;
 548         }
 549 }
 550
 551 void clear_commit_marks_many(int nr, struct commit **commit, unsigned int mark)
 552 {
 553         struct commit_list *list = NULL;
 554
 555         while (nr--) {
 556                 commit_list_insert(*commit, &list);
 557                 commit++;
 558         }
 559         while (list)
 560                 clear_commit_marks_1(&list, pop_commit(&list), mark);
 561 }
 562
 563 void clear_commit_marks(struct commit *commit, unsigned int mark)
 564 {
 565         clear_commit_marks_many(1, &commit, mark);
 566 }
 567
 568 void clear_commit_marks_for_object_array(struct object_array *a, unsigned mark)
 569 {
 570         struct object *object;
 571         struct commit *commit;
 572         unsigned int i;
 573
 574         for (i = 0; i < a->nr; i++) {
 575                 object = a->objects[i].item;
 576                 commit = lookup_commit_reference_gently(object->sha1, 1);
 577                 if (commit)
 578                         clear_commit_marks(commit, mark);
 579         }
 580 }
 581
 582 struct commit *pop_commit(struct commit_list **stack)
 583 {
 584         struct commit_list *top = *stack;
 585         struct commit *item = top ? top->item : NULL;
 586
 587         if (top) {
 588                 *stack = top->next;
 589                 free(top);
 590         }
 591         return item;
 592 }
 593
 594 /*
 595  * Topological sort support
 596  */
 597
 598 /* count number of children that have not been emitted */
 599 define_commit_slab(indegree_slab, int);
 600
 601 /* record author-date for each commit object */
 602 define_commit_slab(author_date_slab, unsigned long);
 603
 604 static void record_author_date(struct author_date_slab *author_date,
 605                                struct commit *commit)
 606 {
 607         const char *buf, *line_end, *ident_line;
 608         const char *buffer = get_commit_buffer(commit, NULL);
 609         struct ident_split ident;
 610         char *date_end;
 611         unsigned long date;
 612
 613         for (buf = buffer; buf; buf = line_end + 1) {
 614                 line_end = strchrnul(buf, '\n');
 615                 ident_line = skip_prefix(buf, "author ");
 616                 if (!ident_line) {
 617                         if (!line_end[0] || line_end[1] == '\n')
 618                                 return; /* end of header */
 619                         continue;
 620                 }
 621                 if (split_ident_line(&ident,
 622                                      ident_line, line_end - ident_line) ||
 623                     !ident.date_begin || !ident.date_end)
 624                         goto fail_exit; /* malformed "author" line */
 625                 break;
 626         }
 627
 628         date = strtoul(ident.date_begin, &date_end, 10);
 629         if (date_end != ident.date_end)
 630                 goto fail_exit; /* malformed date */
 631         *(author_date_slab_at(author_date, commit)) = date;
 632
 633 fail_exit:
 634         unuse_commit_buffer(commit, buffer);
 635 }
 636
 637 static int compare_commits_by_author_date(const void *a_, const void *b_,
 638                                           void *cb_data)
 639 {
 640         const struct commit *a = a_, *b = b_;
 641         struct author_date_slab *author_date = cb_data;
 642         unsigned long a_date = *(author_date_slab_at(author_date, a));
 643         unsigned long b_date = *(author_date_slab_at(author_date, b));
 644
 645         /* newer commits with larger date first */
 646         if (a_date < b_date)
 647                 return 1;
 648         else if (a_date > b_date)
 649                 return -1;
 650         return 0;
 651 }
 652
 653 int compare_commits_by_commit_date(const void *a_, const void *b_, void *unused)
 654 {
 655         const struct commit *a = a_, *b = b_;
 656         /* newer commits with larger date first */
 657         if (a->date < b->date)
 658                 return 1;
 659         else if (a->date > b->date)
 660                 return -1;
 661         return 0;
 662 }
 663
 664 /*
 665  * Performs an in-place topological sort on the list supplied.
 666  */
 667 void sort_in_topological_order(struct commit_list **list, enum rev_sort_order sort_order)
 668 {
 669         struct commit_list *next, *orig = *list;
 670         struct commit_list **pptr;
 671         struct indegree_slab indegree;
 672         struct prio_queue queue;
 673         struct commit *commit;
 674         struct author_date_slab author_date;
 675
 676         if (!orig)
 677                 return;
 678         *list = NULL;
 679
 680         init_indegree_slab(&indegree);
 681         memset(&queue, '\0', sizeof(queue));
 682
 683         switch (sort_order) {
 684         default: /* REV_SORT_IN_GRAPH_ORDER */
 685                 queue.compare = NULL;
 686                 break;
 687         case REV_SORT_BY_COMMIT_DATE:
 688                 queue.compare = compare_commits_by_commit_date;
 689                 break;
 690         case REV_SORT_BY_AUTHOR_DATE:
 691                 init_author_date_slab(&author_date);
 692                 queue.compare = compare_commits_by_author_date;
 693                 queue.cb_data = &author_date;
 694                 break;
 695         }
 696
 697         /* Mark them and clear the indegree */
 698         for (next = orig; next; next = next->next) {
 699                 struct commit *commit = next->item;
 700                 *(indegree_slab_at(&indegree, commit)) = 1;
 701                 /* also record the author dates, if needed */
 702                 if (sort_order == REV_SORT_BY_AUTHOR_DATE)
 703                         record_author_date(&author_date, commit);
 704         }
 705
 706         /* update the indegree */
 707         for (next = orig; next; next = next->next) {
 708                 struct commit_list *parents = next->item->parents;
 709                 while (parents) {
 710                         struct commit *parent = parents->item;
 711                         int *pi = indegree_slab_at(&indegree, parent);
 712
 713                         if (*pi)
 714                                 (*pi)++;
 715                         parents = parents->next;
 716                 }
 717         }
 718
 719         /*
 720          * find the tips
 721          *
 722          * tips are nodes not reachable from any other node in the list
 723          *
 724          * the tips serve as a starting set for the work queue.
 725          */
 726         for (next = orig; next; next = next->next) {
 727                 struct commit *commit = next->item;
 728
 729                 if (*(indegree_slab_at(&indegree, commit)) == 1)
 730                         prio_queue_put(&queue, commit);
 731         }
 732
 733         /*
 734          * This is unfortunate; the initial tips need to be shown
 735          * in the order given from the revision traversal machinery.
 736          */
 737         if (sort_order == REV_SORT_IN_GRAPH_ORDER)
 738                 prio_queue_reverse(&queue);
 739
 740         /* We no longer need the commit list */
 741         free_commit_list(orig);
 742
 743         pptr = list;
 744         *list = NULL;
 745         while ((commit = prio_queue_get(&queue)) != NULL) {
 746                 struct commit_list *parents;
 747
 748                 for (parents = commit->parents; parents ; parents = parents->next) {
 749                         struct commit *parent = parents->item;
 750                         int *pi = indegree_slab_at(&indegree, parent);
 751
 752                         if (!*pi)
 753                                 continue;
 754
 755                         /*
 756                          * parents are only enqueued for emission
 757                          * when all their children have been emitted thereby
 758                          * guaranteeing topological order.
 759                          */
 760                         if (--(*pi) == 1)
 761                                 prio_queue_put(&queue, parent);
 762                 }
 763                 /*
 764                  * all children of commit have already been
 765                  * emitted. we can emit it now.
 766                  */
 767                 *(indegree_slab_at(&indegree, commit)) = 0;
 768
 769                 pptr = &commit_list_insert(commit, pptr)->next;
 770         }
 771
 772         clear_indegree_slab(&indegree);
 773         clear_prio_queue(&queue);
 774         if (sort_order == REV_SORT_BY_AUTHOR_DATE)
 775                 clear_author_date_slab(&author_date);
 776 }
 777
 778 /* merge-base stuff */
 779
 780 /* Remember to update object flag allocation in object.h */
 781 #define PARENT1         (1u<<16)
 782 #define PARENT2         (1u<<17)
 783 #define STALE           (1u<<18)
 784 #define RESULT          (1u<<19)
 785
 786 static const unsigned all_flags = (PARENT1 | PARENT2 | STALE | RESULT);
 787
 788 static struct commit *interesting(struct commit_list *list)
 789 {
 790         while (list) {
 791                 struct commit *commit = list->item;
 792                 list = list->next;
 793                 if (commit->object.flags & STALE)
 794                         continue;
 795                 return commit;
 796         }
 797         return NULL;
 798 }
 799
 800 /* all input commits in one and twos[] must have been parsed! */
 801 static struct commit_list *paint_down_to_common(struct commit *one, int n, struct commit **twos)
 802 {
 803         struct commit_list *list = NULL;
 804         struct commit_list *result = NULL;
 805         int i;
 806
 807         one->object.flags |= PARENT1;
 808         commit_list_insert_by_date(one, &list);
 809         if (!n)
 810                 return list;
 811         for (i = 0; i < n; i++) {
 812                 twos[i]->object.flags |= PARENT2;
 813                 commit_list_insert_by_date(twos[i], &list);
 814         }
 815
 816         while (interesting(list)) {
 817                 struct commit *commit;
 818                 struct commit_list *parents;
 819                 struct commit_list *next;
 820                 int flags;
 821
 822                 commit = list->item;
 823                 next = list->next;
 824                 free(list);
 825                 list = next;
 826
 827                 flags = commit->object.flags & (PARENT1 | PARENT2 | STALE);
 828                 if (flags == (PARENT1 | PARENT2)) {
 829                         if (!(commit->object.flags & RESULT)) {
 830                                 commit->object.flags |= RESULT;
 831                                 commit_list_insert_by_date(commit, &result);
 832                         }
 833                         /* Mark parents of a found merge stale */
 834                         flags |= STALE;
 835                 }
 836                 parents = commit->parents;
 837                 while (parents) {
 838                         struct commit *p = parents->item;
 839                         parents = parents->next;
 840                         if ((p->object.flags & flags) == flags)
 841                                 continue;
 842                         if (parse_commit(p))
 843                                 return NULL;
 844                         p->object.flags |= flags;
 845                         commit_list_insert_by_date(p, &list);
 846                 }
 847         }
 848
 849         free_commit_list(list);
 850         return result;
 851 }
 852
 853 static struct commit_list *merge_bases_many(struct commit *one, int n, struct commit **twos)
 854 {
 855         struct commit_list *list = NULL;
 856         struct commit_list *result = NULL;
 857         int i;
 858
 859         for (i = 0; i < n; i++) {
 860                 if (one == twos[i])
 861                         /*
 862                          * We do not mark this even with RESULT so we do not
 863                          * have to clean it up.
 864                          */
 865                         return commit_list_insert(one, &result);
 866         }
 867
 868         if (parse_commit(one))
 869                 return NULL;
 870         for (i = 0; i < n; i++) {
 871                 if (parse_commit(twos[i]))
 872                         return NULL;
 873         }
 874
 875         list = paint_down_to_common(one, n, twos);
 876
 877         while (list) {
 878                 struct commit_list *next = list->next;
 879                 if (!(list->item->object.flags & STALE))
 880                         commit_list_insert_by_date(list->item, &result);
 881                 free(list);
 882                 list = next;
 883         }
 884         return result;
 885 }
 886
 887 struct commit_list *get_octopus_merge_bases(struct commit_list *in)
 888 {
 889         struct commit_list *i, *j, *k, *ret = NULL;
 890
 891         if (!in)
 892                 return ret;
 893
 894         commit_list_insert(in->item, &ret);
 895
 896         for (i = in->next; i; i = i->next) {
 897                 struct commit_list *new = NULL, *end = NULL;
 898
 899                 for (j = ret; j; j = j->next) {
 900                         struct commit_list *bases;
 901                         bases = get_merge_bases(i->item, j->item, 1);
 902                         if (!new)
 903                                 new = bases;
 904                         else
 905                                 end->next = bases;
 906                         for (k = bases; k; k = k->next)
 907                                 end = k;
 908                 }
 909                 ret = new;
 910         }
 911         return ret;
 912 }
 913
 914 static int remove_redundant(struct commit **array, int cnt)
 915 {
 916         /*
 917          * Some commit in the array may be an ancestor of
 918          * another commit.  Move such commit to the end of
 919          * the array, and return the number of commits that
 920          * are independent from each other.
 921          */
 922         struct commit **work;
 923         unsigned char *redundant;
 924         int *filled_index;
 925         int i, j, filled;
 926
 927         work = xcalloc(cnt, sizeof(*work));
 928         redundant = xcalloc(cnt, 1);
 929         filled_index = xmalloc(sizeof(*filled_index) * (cnt - 1));
 930
 931         for (i = 0; i < cnt; i++)
 932                 parse_commit(array[i]);
 933         for (i = 0; i < cnt; i++) {
 934                 struct commit_list *common;
 935
 936                 if (redundant[i])
 937                         continue;
 938                 for (j = filled = 0; j < cnt; j++) {
 939                         if (i == j || redundant[j])
 940                                 continue;
 941                         filled_index[filled] = j;
 942                         work[filled++] = array[j];
 943                 }
 944                 common = paint_down_to_common(array[i], filled, work);
 945                 if (array[i]->object.flags & PARENT2)
 946                         redundant[i] = 1;
 947                 for (j = 0; j < filled; j++)
 948                         if (work[j]->object.flags & PARENT1)
 949                                 redundant[filled_index[j]] = 1;
 950                 clear_commit_marks(array[i], all_flags);
 951                 for (j = 0; j < filled; j++)
 952                         clear_commit_marks(work[j], all_flags);
 953                 free_commit_list(common);
 954         }
 955
 956         /* Now collect the result */
 957         memcpy(work, array, sizeof(*array) * cnt);
 958         for (i = filled = 0; i < cnt; i++)
 959                 if (!redundant[i])
 960                         array[filled++] = work[i];
 961         for (j = filled, i = 0; i < cnt; i++)
 962                 if (redundant[i])
 963                         array[j++] = work[i];
 964         free(work);
 965         free(redundant);
 966         free(filled_index);
 967         return filled;
 968 }
 969
 970 struct commit_list *get_merge_bases_many(struct commit *one,
 971                                          int n,
 972                                          struct commit **twos,
 973                                          int cleanup)
 974 {
 975         struct commit_list *list;
 976         struct commit **rslt;
 977         struct commit_list *result;
 978         int cnt, i;
 979
 980         result = merge_bases_many(one, n, twos);
 981         for (i = 0; i < n; i++) {
 982                 if (one == twos[i])
 983                         return result;
 984         }
 985         if (!result || !result->next) {
 986                 if (cleanup) {
 987                         clear_commit_marks(one, all_flags);
 988                         clear_commit_marks_many(n, twos, all_flags);
 989                 }
 990                 return result;
 991         }
 992
 993         /* There are more than one */
 994         cnt = 0;
 995         list = result;
 996         while (list) {
 997                 list = list->next;
 998                 cnt++;
 999         }
1000         rslt = xcalloc(cnt, sizeof(*rslt));
1001         for (list = result, i = 0; list; list = list->next)
1002                 rslt[i++] = list->item;
1003         free_commit_list(result);
1004
1005         clear_commit_marks(one, all_flags);
1006         clear_commit_marks_many(n, twos, all_flags);
1007
1008         cnt = remove_redundant(rslt, cnt);
1009         result = NULL;
1010         for (i = 0; i < cnt; i++)
1011                 commit_list_insert_by_date(rslt[i], &result);
1012         free(rslt);
1013         return result;
1014 }
1015
1016 struct commit_list *get_merge_bases(struct commit *one, struct commit *two,
1017                                     int cleanup)
1018 {
1019         return get_merge_bases_many(one, 1, &two, cleanup);
1020 }
1021
1022 /*
1023  * Is "commit" a descendant of one of the elements on the "with_commit" list?
1024  */
1025 int is_descendant_of(struct commit *commit, struct commit_list *with_commit)
1026 {
1027         if (!with_commit)
1028                 return 1;
1029         while (with_commit) {
1030                 struct commit *other;
1031
1032                 other = with_commit->item;
1033                 with_commit = with_commit->next;
1034                 if (in_merge_bases(other, commit))
1035                         return 1;
1036         }
1037         return 0;
1038 }
1039
1040 /*
1041  * Is "commit" an ancestor of one of the "references"?
1042  */
1043 int in_merge_bases_many(struct commit *commit, int nr_reference, struct commit **reference)
1044 {
1045         struct commit_list *bases;
1046         int ret = 0, i;
1047
1048         if (parse_commit(commit))
1049                 return ret;
1050         for (i = 0; i < nr_reference; i++)
1051                 if (parse_commit(reference[i]))
1052                         return ret;
1053
1054         bases = paint_down_to_common(commit, nr_reference, reference);
1055         if (commit->object.flags & PARENT2)
1056                 ret = 1;
1057         clear_commit_marks(commit, all_flags);
1058         clear_commit_marks_many(nr_reference, reference, all_flags);
1059         free_commit_list(bases);
1060         return ret;
1061 }
1062
1063 /*
1064  * Is "commit" an ancestor of (i.e. reachable from) the "reference"?
1065  */
1066 int in_merge_bases(struct commit *commit, struct commit *reference)
1067 {
1068         return in_merge_bases_many(commit, 1, &reference);
1069 }
1070
1071 struct commit_list *reduce_heads(struct commit_list *heads)
1072 {
1073         struct commit_list *p;
1074         struct commit_list *result = NULL, **tail = &result;
1075         struct commit **array;
1076         int num_head, i;
1077
1078         if (!heads)
1079                 return NULL;
1080
1081         /* Uniquify */
1082         for (p = heads; p; p = p->next)
1083                 p->item->object.flags &= ~STALE;
1084         for (p = heads, num_head = 0; p; p = p->next) {
1085                 if (p->item->object.flags & STALE)
1086                         continue;
1087                 p->item->object.flags |= STALE;
1088                 num_head++;
1089         }
1090         array = xcalloc(num_head, sizeof(*array));
1091         for (p = heads, i = 0; p; p = p->next) {
1092                 if (p->item->object.flags & STALE) {
1093                         array[i++] = p->item;
1094                         p->item->object.flags &= ~STALE;
1095                 }
1096         }
1097         num_head = remove_redundant(array, num_head);
1098         for (i = 0; i < num_head; i++)
1099                 tail = &commit_list_insert(array[i], tail)->next;
1100         return result;
1101 }
1102
1103 static const char gpg_sig_header[] = "gpgsig";
1104 static const int gpg_sig_header_len = sizeof(gpg_sig_header) - 1;
1105
1106 static int do_sign_commit(struct strbuf *buf, const char *keyid)
1107 {
1108         struct strbuf sig = STRBUF_INIT;
1109         int inspos, copypos;
1110
1111         /* find the end of the header */
1112         inspos = strstr(buf->buf, "\n\n") - buf->buf + 1;
1113
1114         if (!keyid || !*keyid)
1115                 keyid = get_signing_key();
1116         if (sign_buffer(buf, &sig, keyid)) {
1117                 strbuf_release(&sig);
1118                 return -1;
1119         }
1120
1121         for (copypos = 0; sig.buf[copypos]; ) {
1122                 const char *bol = sig.buf + copypos;
1123                 const char *eol = strchrnul(bol, '\n');
1124                 int len = (eol - bol) + !!*eol;
1125
1126                 if (!copypos) {
1127                         strbuf_insert(buf, inspos, gpg_sig_header, gpg_sig_header_len);
1128                         inspos += gpg_sig_header_len;
1129                 }
1130                 strbuf_insert(buf, inspos++, " ", 1);
1131                 strbuf_insert(buf, inspos, bol, len);
1132                 inspos += len;
1133                 copypos += len;
1134         }
1135         strbuf_release(&sig);
1136         return 0;
1137 }
1138
1139 int parse_signed_commit(const struct commit *commit,
1140                         struct strbuf *payload, struct strbuf *signature)
1141 {
1142
1143         unsigned long size;
1144         const char *buffer = get_commit_buffer(commit, &size);
1145         int in_signature, saw_signature = -1;
1146         const char *line, *tail;
1147
1148         line = buffer;
1149         tail = buffer + size;
1150         in_signature = 0;
1151         saw_signature = 0;
1152         while (line < tail) {
1153                 const char *sig = NULL;
1154                 const char *next = memchr(line, '\n', tail - line);
1155
1156                 next = next ? next + 1 : tail;
1157                 if (in_signature && line[0] == ' ')
1158                         sig = line + 1;
1159                 else if (starts_with(line, gpg_sig_header) &&
1160                          line[gpg_sig_header_len] == ' ')
1161                         sig = line + gpg_sig_header_len + 1;
1162                 if (sig) {
1163                         strbuf_add(signature, sig, next - sig);
1164                         saw_signature = 1;
1165                         in_signature = 1;
1166                 } else {
1167                         if (*line == '\n')
1168                                 /* dump the whole remainder of the buffer */
1169                                 next = tail;
1170                         strbuf_add(payload, line, next - line);
1171                         in_signature = 0;
1172                 }
1173                 line = next;
1174         }
1175         unuse_commit_buffer(commit, buffer);
1176         return saw_signature;
1177 }
1178
1179 static void handle_signed_tag(struct commit *parent, struct commit_extra_header ***tail)
1180 {
1181         struct merge_remote_desc *desc;
1182         struct commit_extra_header *mergetag;
1183         char *buf;
1184         unsigned long size, len;
1185         enum object_type type;
1186
1187         desc = merge_remote_util(parent);
1188         if (!desc || !desc->obj)
1189                 return;
1190         buf = read_sha1_file(desc->obj->sha1, &type, &size);
1191         if (!buf || type != OBJ_TAG)
1192                 goto free_return;
1193         len = parse_signature(buf, size);
1194         if (size == len)
1195                 goto free_return;
1196         /*
1197          * We could verify this signature and either omit the tag when
1198          * it does not validate, but the integrator may not have the
1199          * public key of the signer of the tag he is merging, while a
1200          * later auditor may have it while auditing, so let's not run
1201          * verify-signed-buffer here for now...
1202          *
1203          * if (verify_signed_buffer(buf, len, buf + len, size - len, ...))
1204          *      warn("warning: signed tag unverified.");
1205          */
1206         mergetag = xcalloc(1, sizeof(*mergetag));
1207         mergetag->key = xstrdup("mergetag");
1208         mergetag->value = buf;
1209         mergetag->len = size;
1210
1211         **tail = mergetag;
1212         *tail = &mergetag->next;
1213         return;
1214
1215 free_return:
1216         free(buf);
1217 }
1218
1219 static struct {
1220         char result;
1221         const char *check;
1222 } sigcheck_gpg_status[] = {
1223         { 'G', "\n[GNUPG:] GOODSIG " },
1224         { 'B', "\n[GNUPG:] BADSIG " },
1225         { 'U', "\n[GNUPG:] TRUST_NEVER" },
1226         { 'U', "\n[GNUPG:] TRUST_UNDEFINED" },
1227 };
1228
1229 static void parse_gpg_output(struct signature_check *sigc)
1230 {
1231         const char *buf = sigc->gpg_status;
1232         int i;
1233
1234         /* Iterate over all search strings */
1235         for (i = 0; i < ARRAY_SIZE(sigcheck_gpg_status); i++) {
1236                 const char *found, *next;
1237
1238                 found = skip_prefix(buf, sigcheck_gpg_status[i].check + 1);
1239                 if (!found) {
1240                         found = strstr(buf, sigcheck_gpg_status[i].check);
1241                         if (!found)
1242                                 continue;
1243                         found += strlen(sigcheck_gpg_status[i].check);
1244                 }
1245                 sigc->result = sigcheck_gpg_status[i].result;
1246                 /* The trust messages are not followed by key/signer information */
1247                 if (sigc->result != 'U') {
1248                         sigc->key = xmemdupz(found, 16);
1249                         found += 17;
1250                         next = strchrnul(found, '\n');
1251                         sigc->signer = xmemdupz(found, next - found);
1252                 }
1253         }
1254 }
1255
1256 void check_commit_signature(const struct commit* commit, struct signature_check *sigc)
1257 {
1258         struct strbuf payload = STRBUF_INIT;
1259         struct strbuf signature = STRBUF_INIT;
1260         struct strbuf gpg_output = STRBUF_INIT;
1261         struct strbuf gpg_status = STRBUF_INIT;
1262         int status;
1263
1264         sigc->result = 'N';
1265
1266         if (parse_signed_commit(commit, &payload, &signature) <= 0)
1267                 goto out;
1268         status = verify_signed_buffer(payload.buf, payload.len,
1269                                       signature.buf, signature.len,
1270                                       &gpg_output, &gpg_status);
1271         if (status && !gpg_output.len)
1272                 goto out;
1273         sigc->gpg_output = strbuf_detach(&gpg_output, NULL);
1274         sigc->gpg_status = strbuf_detach(&gpg_status, NULL);
1275         parse_gpg_output(sigc);
1276
1277  out:
1278         strbuf_release(&gpg_status);
1279         strbuf_release(&gpg_output);
1280         strbuf_release(&payload);
1281         strbuf_release(&signature);
1282 }
1283
1284
1285
1286 void append_merge_tag_headers(struct commit_list *parents,
1287                               struct commit_extra_header ***tail)
1288 {
1289         while (parents) {
1290                 struct commit *parent = parents->item;
1291                 handle_signed_tag(parent, tail);
1292                 parents = parents->next;
1293         }
1294 }
1295
1296 static void add_extra_header(struct strbuf *buffer,
1297                              struct commit_extra_header *extra)
1298 {
1299         strbuf_addstr(buffer, extra->key);
1300         if (extra->len)
1301                 strbuf_add_lines(buffer, " ", extra->value, extra->len);
1302         else
1303                 strbuf_addch(buffer, '\n');
1304 }
1305
1306 struct commit_extra_header *read_commit_extra_headers(struct commit *commit,
1307                                                       const char **exclude)
1308 {
1309         struct commit_extra_header *extra = NULL;
1310         unsigned long size;
1311         const char *buffer = get_commit_buffer(commit, &size);
1312         extra = read_commit_extra_header_lines(buffer, size, exclude);
1313         unuse_commit_buffer(commit, buffer);
1314         return extra;
1315 }
1316
1317 static inline int standard_header_field(const char *field, size_t len)
1318 {
1319         return ((len == 4 && !memcmp(field, "tree ", 5)) ||
1320                 (len == 6 && !memcmp(field, "parent ", 7)) ||
1321                 (len == 6 && !memcmp(field, "author ", 7)) ||
1322                 (len == 9 && !memcmp(field, "committer ", 10)) ||
1323                 (len == 8 && !memcmp(field, "encoding ", 9)));
1324 }
1325
1326 static int excluded_header_field(const char *field, size_t len, const char **exclude)
1327 {
1328         if (!exclude)
1329                 return 0;
1330
1331         while (*exclude) {
1332                 size_t xlen = strlen(*exclude);
1333                 if (len == xlen &&
1334                     !memcmp(field, *exclude, xlen) && field[xlen] == ' ')
1335                         return 1;
1336                 exclude++;
1337         }
1338         return 0;
1339 }
1340
1341 static struct commit_extra_header *read_commit_extra_header_lines(
1342         const char *buffer, size_t size,
1343         const char **exclude)
1344 {
1345         struct commit_extra_header *extra = NULL, **tail = &extra, *it = NULL;
1346         const char *line, *next, *eof, *eob;
1347         struct strbuf buf = STRBUF_INIT;
1348
1349         for (line = buffer, eob = line + size;
1350              line < eob && *line != '\n';
1351              line = next) {
1352                 next = memchr(line, '\n', eob - line);
1353                 next = next ? next + 1 : eob;
1354                 if (*line == ' ') {
1355                         /* continuation */
1356                         if (it)
1357                                 strbuf_add(&buf, line + 1, next - (line + 1));
1358                         continue;
1359                 }
1360                 if (it)
1361                         it->value = strbuf_detach(&buf, &it->len);
1362                 strbuf_reset(&buf);
1363                 it = NULL;
1364
1365                 eof = strchr(line, ' ');
1366                 if (next <= eof)
1367                         eof = next;
1368
1369                 if (standard_header_field(line, eof - line) ||
1370                     excluded_header_field(line, eof - line, exclude))
1371                         continue;
1372
1373                 it = xcalloc(1, sizeof(*it));
1374                 it->key = xmemdupz(line, eof-line);
1375                 *tail = it;
1376                 tail = &it->next;
1377                 if (eof + 1 < next)
1378                         strbuf_add(&buf, eof + 1, next - (eof + 1));
1379         }
1380         if (it)
1381                 it->value = strbuf_detach(&buf, &it->len);
1382         return extra;
1383 }
1384
1385 void free_commit_extra_headers(struct commit_extra_header *extra)
1386 {
1387         while (extra) {
1388                 struct commit_extra_header *next = extra->next;
1389                 free(extra->key);
1390                 free(extra->value);
1391                 free(extra);
1392                 extra = next;
1393         }
1394 }
1395
1396 int commit_tree(const char *msg, size_t msg_len,
1397                 const unsigned char *tree,
1398                 struct commit_list *parents, unsigned char *ret,
1399                 const char *author, const char *sign_commit)
1400 {
1401         struct commit_extra_header *extra = NULL, **tail = &extra;
1402         int result;
1403
1404         append_merge_tag_headers(parents, &tail);
1405         result = commit_tree_extended(msg, msg_len, tree, parents, ret,
1406                                       author, sign_commit, extra);
1407         free_commit_extra_headers(extra);
1408         return result;
1409 }
1410
1411 static int find_invalid_utf8(const char *buf, int len)
1412 {
1413         int offset = 0;
1414         static const unsigned int max_codepoint[] = {
1415                 0x7f, 0x7ff, 0xffff, 0x10ffff
1416         };
1417
1418         while (len) {
1419                 unsigned char c = *buf++;
1420                 int bytes, bad_offset;
1421                 unsigned int codepoint;
1422                 unsigned int min_val, max_val;
1423
1424                 len--;
1425                 offset++;
1426
1427                 /* Simple US-ASCII? No worries. */
1428                 if (c < 0x80)
1429                         continue;
1430
1431                 bad_offset = offset-1;
1432
1433                 /*
1434                  * Count how many more high bits set: that's how
1435                  * many more bytes this sequence should have.
1436                  */
1437                 bytes = 0;
1438                 while (c & 0x40) {
1439                         c <<= 1;
1440                         bytes++;
1441                 }
1442
1443                 /*
1444                  * Must be between 1 and 3 more bytes.  Longer sequences result in
1445                  * codepoints beyond U+10FFFF, which are guaranteed never to exist.
1446                  */
1447                 if (bytes < 1 || 3 < bytes)
1448                         return bad_offset;
1449
1450                 /* Do we *have* that many bytes? */
1451                 if (len < bytes)
1452                         return bad_offset;
1453
1454                 /*
1455                  * Place the encoded bits at the bottom of the value and compute the
1456                  * valid range.
1457                  */
1458                 codepoint = (c & 0x7f) >> bytes;
1459                 min_val = max_codepoint[bytes-1] + 1;
1460                 max_val = max_codepoint[bytes];
1461
1462                 offset += bytes;
1463                 len -= bytes;
1464
1465                 /* And verify that they are good continuation bytes */
1466                 do {
1467                         codepoint <<= 6;
1468                         codepoint |= *buf & 0x3f;
1469                         if ((*buf++ & 0xc0) != 0x80)
1470                                 return bad_offset;
1471                 } while (--bytes);
1472
1473                 /* Reject codepoints that are out of range for the sequence length. */
1474                 if (codepoint < min_val || codepoint > max_val)
1475                         return bad_offset;
1476                 /* Surrogates are only for UTF-16 and cannot be encoded in UTF-8. */
1477                 if ((codepoint & 0x1ff800) == 0xd800)
1478                         return bad_offset;
1479                 /* U+xxFFFE and U+xxFFFF are guaranteed non-characters. */
1480                 if ((codepoint & 0xfffe) == 0xfffe)
1481                         return bad_offset;
1482                 /* So are anything in the range U+FDD0..U+FDEF. */
1483                 if (codepoint >= 0xfdd0 && codepoint <= 0xfdef)
1484                         return bad_offset;
1485         }
1486         return -1;
1487 }
1488
1489 /*
1490  * This verifies that the buffer is in proper utf8 format.
1491  *
1492  * If it isn't, it assumes any non-utf8 characters are Latin1,
1493  * and does the conversion.
1494  */
1495 static int verify_utf8(struct strbuf *buf)
1496 {
1497         int ok = 1;
1498         long pos = 0;
1499
1500         for (;;) {
1501                 int bad;
1502                 unsigned char c;
1503                 unsigned char replace[2];
1504
1505                 bad = find_invalid_utf8(buf->buf + pos, buf->len - pos);
1506                 if (bad < 0)
1507                         return ok;
1508                 pos += bad;
1509                 ok = 0;
1510                 c = buf->buf[pos];
1511                 strbuf_remove(buf, pos, 1);
1512
1513                 /* We know 'c' must be in the range 128-255 */
1514                 replace[0] = 0xc0 + (c >> 6);
1515                 replace[1] = 0x80 + (c & 0x3f);
1516                 strbuf_insert(buf, pos, replace, 2);
1517                 pos += 2;
1518         }
1519 }
1520
1521 static const char commit_utf8_warn[] =
1522 "Warning: commit message did not conform to UTF-8.\n"
1523 "You may want to amend it after fixing the message, or set the config\n"
1524 "variable i18n.commitencoding to the encoding your project uses.\n";
1525
1526 int commit_tree_extended(const char *msg, size_t msg_len,
1527                          const unsigned char *tree,
1528                          struct commit_list *parents, unsigned char *ret,
1529                          const char *author, const char *sign_commit,
1530                          struct commit_extra_header *extra)
1531 {
1532         int result;
1533         int encoding_is_utf8;
1534         struct strbuf buffer;
1535
1536         assert_sha1_type(tree, OBJ_TREE);
1537
1538         if (memchr(msg, '\0', msg_len))
1539                 return error("a NUL byte in commit log message not allowed.");
1540
1541         /* Not having i18n.commitencoding is the same as having utf-8 */
1542         encoding_is_utf8 = is_encoding_utf8(git_commit_encoding);
1543
1544         strbuf_init(&buffer, 8192); /* should avoid reallocs for the headers */
1545         strbuf_addf(&buffer, "tree %s\n", sha1_to_hex(tree));
1546
1547         /*
1548          * NOTE! This ordering means that the same exact tree merged with a
1549          * different order of parents will be a _different_ changeset even
1550          * if everything else stays the same.
1551          */
1552         while (parents) {
1553                 struct commit_list *next = parents->next;
1554                 struct commit *parent = parents->item;
1555
1556                 strbuf_addf(&buffer, "parent %s\n",
1557                             sha1_to_hex(parent->object.sha1));
1558                 free(parents);
1559                 parents = next;
1560         }
1561
1562         /* Person/date information */
1563         if (!author)
1564                 author = git_author_info(IDENT_STRICT);
1565         strbuf_addf(&buffer, "author %s\n", author);
1566         strbuf_addf(&buffer, "committer %s\n", git_committer_info(IDENT_STRICT));
1567         if (!encoding_is_utf8)
1568                 strbuf_addf(&buffer, "encoding %s\n", git_commit_encoding);
1569
1570         while (extra) {
1571                 add_extra_header(&buffer, extra);
1572                 extra = extra->next;
1573         }
1574         strbuf_addch(&buffer, '\n');
1575
1576         /* And add the comment */
1577         strbuf_add(&buffer, msg, msg_len);
1578
1579         /* And check the encoding */
1580         if (encoding_is_utf8 && !verify_utf8(&buffer))
1581                 fprintf(stderr, commit_utf8_warn);
1582
1583         if (sign_commit && do_sign_commit(&buffer, sign_commit))
1584                 return -1;
1585
1586         result = write_sha1_file(buffer.buf, buffer.len, commit_type, ret);
1587         strbuf_release(&buffer);
1588         return result;
1589 }
1590
1591 struct commit *get_merge_parent(const char *name)
1592 {
1593         struct object *obj;
1594         struct commit *commit;
1595         unsigned char sha1[20];
1596         if (get_sha1(name, sha1))
1597                 return NULL;
1598         obj = parse_object(sha1);
1599         commit = (struct commit *)peel_to_type(name, 0, obj, OBJ_COMMIT);
1600         if (commit && !commit->util) {
1601                 struct merge_remote_desc *desc;
1602                 desc = xmalloc(sizeof(*desc));
1603                 desc->obj = obj;
1604                 desc->name = strdup(name);
1605                 commit->util = desc;
1606         }
1607         return commit;
1608 }
1609
1610 /*
1611  * Append a commit to the end of the commit_list.
1612  *
1613  * next starts by pointing to the variable that holds the head of an
1614  * empty commit_list, and is updated to point to the "next" field of
1615  * the last item on the list as new commits are appended.
1616  *
1617  * Usage example:
1618  *
1619  *     struct commit_list *list;
1620  *     struct commit_list **next = &list;
1621  *
1622  *     next = commit_list_append(c1, next);
1623  *     next = commit_list_append(c2, next);
1624  *     assert(commit_list_count(list) == 2);
1625  *     return list;
1626  */
1627 struct commit_list **commit_list_append(struct commit *commit,
1628                                         struct commit_list **next)
1629 {
1630         struct commit_list *new = xmalloc(sizeof(struct commit_list));
1631         new->item = commit;
1632         *next = new;
1633         new->next = NULL;
1634         return &new->next;
1635 }
1636
1637 void print_commit_list(struct commit_list *list,
1638                        const char *format_cur,
1639                        const char *format_last)
1640 {
1641         for ( ; list; list = list->next) {
1642                 const char *format = list->next ? format_cur : format_last;
1643                 printf(format, sha1_to_hex(list->item->object.sha1));
1644         }
1645 }