commit.c

   1 #include "cache.h"
   2 #include "tag.h"
   3 #include "commit.h"
   4 #include "commit-graph.h"
   5 #include "repository.h"
   6 #include "object-store.h"
   7 #include "pkt-line.h"
   8 #include "utf8.h"
   9 #include "diff.h"
  10 #include "revision.h"
  11 #include "notes.h"
  12 #include "alloc.h"
  13 #include "gpg-interface.h"
  14 #include "mergesort.h"
  15 #include "commit-slab.h"
  16 #include "prio-queue.h"
  17 #include "hash-lookup.h"
  18 #include "wt-status.h"
  19 #include "advice.h"
  20 #include "refs.h"
  21 #include "commit-reach.h"
  22 #include "run-command.h"
  23 #include "shallow.h"
  24 #include "hook.h"
  25
  26 static struct commit_extra_header *read_commit_extra_header_lines(const char *buf, size_t len, const char **);
  27
  28 int save_commit_buffer = 1;
  29 int no_graft_file_deprecated_advice;
  30
  31 const char *commit_type = "commit";
  32
  33 struct commit *lookup_commit_reference_gently(struct repository *r,
  34                 const struct object_id *oid, int quiet)
  35 {
  36         struct object *obj = deref_tag(r,
  37                                        parse_object(r, oid),
  38                                        NULL, 0);
  39
  40         if (!obj)
  41                 return NULL;
  42         return object_as_type(obj, OBJ_COMMIT, quiet);
  43 }
  44
  45 struct commit *lookup_commit_reference(struct repository *r, const struct object_id *oid)
  46 {
  47         return lookup_commit_reference_gently(r, oid, 0);
  48 }
  49
  50 struct commit *lookup_commit_or_die(const struct object_id *oid, const char *ref_name)
  51 {
  52         struct commit *c = lookup_commit_reference(the_repository, oid);
  53         if (!c)
  54                 die(_("could not parse %s"), ref_name);
  55         if (!oideq(oid, &c->object.oid)) {
  56                 warning(_("%s %s is not a commit!"),
  57                         ref_name, oid_to_hex(oid));
  58         }
  59         return c;
  60 }
  61
  62 struct commit *lookup_commit(struct repository *r, const struct object_id *oid)
  63 {
  64         struct object *obj = lookup_object(r, oid);
  65         if (!obj)
  66                 return create_object(r, oid, alloc_commit_node(r));
  67         return object_as_type(obj, OBJ_COMMIT, 0);
  68 }
  69
  70 struct commit *lookup_commit_reference_by_name(const char *name)
  71 {
  72         struct object_id oid;
  73         struct commit *commit;
  74
  75         if (get_oid_committish(name, &oid))
  76                 return NULL;
  77         commit = lookup_commit_reference(the_repository, &oid);
  78         if (parse_commit(commit))
  79                 return NULL;
  80         return commit;
  81 }
  82
  83 static timestamp_t parse_commit_date(const char *buf, const char *tail)
  84 {
  85         const char *dateptr;
  86
  87         if (buf + 6 >= tail)
  88                 return 0;
  89         if (memcmp(buf, "author", 6))
  90                 return 0;
  91         while (buf < tail && *buf++ != '\n')
  92                 /* nada */;
  93         if (buf + 9 >= tail)
  94                 return 0;
  95         if (memcmp(buf, "committer", 9))
  96                 return 0;
  97         while (buf < tail && *buf++ != '>')
  98                 /* nada */;
  99         if (buf >= tail)
 100                 return 0;
 101         dateptr = buf;
 102         while (buf < tail && *buf++ != '\n')
 103                 /* nada */;
 104         if (buf >= tail)
 105                 return 0;
 106         /* dateptr < buf && buf[-1] == '\n', so parsing will stop at buf-1 */
 107         return parse_timestamp(dateptr, NULL, 10);
 108 }
 109
 110 static const struct object_id *commit_graft_oid_access(size_t index, const void *table)
 111 {
 112         const struct commit_graft * const *commit_graft_table = table;
 113         return &commit_graft_table[index]->oid;
 114 }
 115
 116 int commit_graft_pos(struct repository *r, const struct object_id *oid)
 117 {
 118         return oid_pos(oid, r->parsed_objects->grafts,
 119                        r->parsed_objects->grafts_nr,
 120                        commit_graft_oid_access);
 121 }
 122
 123 static void unparse_commit(struct repository *r, const struct object_id *oid)
 124 {
 125         struct commit *c = lookup_commit(r, oid);
 126
 127         if (!c->object.parsed)
 128                 return;
 129         free_commit_list(c->parents);
 130         c->parents = NULL;
 131         c->object.parsed = 0;
 132 }
 133
 134 int register_commit_graft(struct repository *r, struct commit_graft *graft,
 135                           int ignore_dups)
 136 {
 137         int pos = commit_graft_pos(r, &graft->oid);
 138
 139         if (0 <= pos) {
 140                 if (ignore_dups)
 141                         free(graft);
 142                 else {
 143                         free(r->parsed_objects->grafts[pos]);
 144                         r->parsed_objects->grafts[pos] = graft;
 145                 }
 146                 return 1;
 147         }
 148         pos = -pos - 1;
 149         ALLOC_GROW(r->parsed_objects->grafts,
 150                    r->parsed_objects->grafts_nr + 1,
 151                    r->parsed_objects->grafts_alloc);
 152         r->parsed_objects->grafts_nr++;
 153         if (pos < r->parsed_objects->grafts_nr)
 154                 memmove(r->parsed_objects->grafts + pos + 1,
 155                         r->parsed_objects->grafts + pos,
 156                         (r->parsed_objects->grafts_nr - pos - 1) *
 157                         sizeof(*r->parsed_objects->grafts));
 158         r->parsed_objects->grafts[pos] = graft;
 159         unparse_commit(r, &graft->oid);
 160         return 0;
 161 }
 162
 163 struct commit_graft *read_graft_line(struct strbuf *line)
 164 {
 165         /* The format is just "Commit Parent1 Parent2 ...\n" */
 166         int i, phase;
 167         const char *tail = NULL;
 168         struct commit_graft *graft = NULL;
 169         struct object_id dummy_oid, *oid;
 170
 171         strbuf_rtrim(line);
 172         if (!line->len || line->buf[0] == '#')
 173                 return NULL;
 174         /*
 175          * phase 0 verifies line, counts hashes in line and allocates graft
 176          * phase 1 fills graft
 177          */
 178         for (phase = 0; phase < 2; phase++) {
 179                 oid = graft ? &graft->oid : &dummy_oid;
 180                 if (parse_oid_hex(line->buf, oid, &tail))
 181                         goto bad_graft_data;
 182                 for (i = 0; *tail != '\0'; i++) {
 183                         oid = graft ? &graft->parent[i] : &dummy_oid;
 184                         if (!isspace(*tail++) || parse_oid_hex(tail, oid, &tail))
 185                                 goto bad_graft_data;
 186                 }
 187                 if (!graft) {
 188                         graft = xmalloc(st_add(sizeof(*graft),
 189                                                st_mult(sizeof(struct object_id), i)));
 190                         graft->nr_parent = i;
 191                 }
 192         }
 193         return graft;
 194
 195 bad_graft_data:
 196         error("bad graft data: %s", line->buf);
 197         assert(!graft);
 198         return NULL;
 199 }
 200
 201 static int read_graft_file(struct repository *r, const char *graft_file)
 202 {
 203         FILE *fp = fopen_or_warn(graft_file, "r");
 204         struct strbuf buf = STRBUF_INIT;
 205         if (!fp)
 206                 return -1;
 207         if (!no_graft_file_deprecated_advice &&
 208             advice_enabled(ADVICE_GRAFT_FILE_DEPRECATED))
 209                 advise(_("Support for <GIT_DIR>/info/grafts is deprecated\n"
 210                          "and will be removed in a future Git version.\n"
 211                          "\n"
 212                          "Please use \"git replace --convert-graft-file\"\n"
 213                          "to convert the grafts into replace refs.\n"
 214                          "\n"
 215                          "Turn this message off by running\n"
 216                          "\"git config advice.graftFileDeprecated false\""));
 217         while (!strbuf_getwholeline(&buf, fp, '\n')) {
 218                 /* The format is just "Commit Parent1 Parent2 ...\n" */
 219                 struct commit_graft *graft = read_graft_line(&buf);
 220                 if (!graft)
 221                         continue;
 222                 if (register_commit_graft(r, graft, 1))
 223                         error("duplicate graft data: %s", buf.buf);
 224         }
 225         fclose(fp);
 226         strbuf_release(&buf);
 227         return 0;
 228 }
 229
 230 void prepare_commit_graft(struct repository *r)
 231 {
 232         char *graft_file;
 233
 234         if (r->parsed_objects->commit_graft_prepared)
 235                 return;
 236         if (!startup_info->have_repository)
 237                 return;
 238
 239         graft_file = get_graft_file(r);
 240         read_graft_file(r, graft_file);
 241         /* make sure shallows are read */
 242         is_repository_shallow(r);
 243         r->parsed_objects->commit_graft_prepared = 1;
 244 }
 245
 246 struct commit_graft *lookup_commit_graft(struct repository *r, const struct object_id *oid)
 247 {
 248         int pos;
 249         prepare_commit_graft(r);
 250         pos = commit_graft_pos(r, oid);
 251         if (pos < 0)
 252                 return NULL;
 253         return r->parsed_objects->grafts[pos];
 254 }
 255
 256 int for_each_commit_graft(each_commit_graft_fn fn, void *cb_data)
 257 {
 258         int i, ret;
 259         for (i = ret = 0; i < the_repository->parsed_objects->grafts_nr && !ret; i++)
 260                 ret = fn(the_repository->parsed_objects->grafts[i], cb_data);
 261         return ret;
 262 }
 263
 264 void reset_commit_grafts(struct repository *r)
 265 {
 266         int i;
 267
 268         for (i = 0; i < r->parsed_objects->grafts_nr; i++) {
 269                 unparse_commit(r, &r->parsed_objects->grafts[i]->oid);
 270                 free(r->parsed_objects->grafts[i]);
 271         }
 272         r->parsed_objects->grafts_nr = 0;
 273         r->parsed_objects->commit_graft_prepared = 0;
 274 }
 275
 276 struct commit_buffer {
 277         void *buffer;
 278         unsigned long size;
 279 };
 280 define_commit_slab(buffer_slab, struct commit_buffer);
 281
 282 struct buffer_slab *allocate_commit_buffer_slab(void)
 283 {
 284         struct buffer_slab *bs = xmalloc(sizeof(*bs));
 285         init_buffer_slab(bs);
 286         return bs;
 287 }
 288
 289 void free_commit_buffer_slab(struct buffer_slab *bs)
 290 {
 291         clear_buffer_slab(bs);
 292         free(bs);
 293 }
 294
 295 void set_commit_buffer(struct repository *r, struct commit *commit, void *buffer, unsigned long size)
 296 {
 297         struct commit_buffer *v = buffer_slab_at(
 298                 r->parsed_objects->buffer_slab, commit);
 299         v->buffer = buffer;
 300         v->size = size;
 301 }
 302
 303 const void *get_cached_commit_buffer(struct repository *r, const struct commit *commit, unsigned long *sizep)
 304 {
 305         struct commit_buffer *v = buffer_slab_peek(
 306                 r->parsed_objects->buffer_slab, commit);
 307         if (!v) {
 308                 if (sizep)
 309                         *sizep = 0;
 310                 return NULL;
 311         }
 312         if (sizep)
 313                 *sizep = v->size;
 314         return v->buffer;
 315 }
 316
 317 const void *repo_get_commit_buffer(struct repository *r,
 318                                    const struct commit *commit,
 319                                    unsigned long *sizep)
 320 {
 321         const void *ret = get_cached_commit_buffer(r, commit, sizep);
 322         if (!ret) {
 323                 enum object_type type;
 324                 unsigned long size;
 325                 ret = repo_read_object_file(r, &commit->object.oid, &type, &size);
 326                 if (!ret)
 327                         die("cannot read commit object %s",
 328                             oid_to_hex(&commit->object.oid));
 329                 if (type != OBJ_COMMIT)
 330                         die("expected commit for %s, got %s",
 331                             oid_to_hex(&commit->object.oid), type_name(type));
 332                 if (sizep)
 333                         *sizep = size;
 334         }
 335         return ret;
 336 }
 337
 338 void repo_unuse_commit_buffer(struct repository *r,
 339                               const struct commit *commit,
 340                               const void *buffer)
 341 {
 342         struct commit_buffer *v = buffer_slab_peek(
 343                 r->parsed_objects->buffer_slab, commit);
 344         if (!(v && v->buffer == buffer))
 345                 free((void *)buffer);
 346 }
 347
 348 void free_commit_buffer(struct parsed_object_pool *pool, struct commit *commit)
 349 {
 350         struct commit_buffer *v = buffer_slab_peek(
 351                 pool->buffer_slab, commit);
 352         if (v) {
 353                 FREE_AND_NULL(v->buffer);
 354                 v->size = 0;
 355         }
 356 }
 357
 358 static inline void set_commit_tree(struct commit *c, struct tree *t)
 359 {
 360         c->maybe_tree = t;
 361 }
 362
 363 struct tree *repo_get_commit_tree(struct repository *r,
 364                                   const struct commit *commit)
 365 {
 366         if (commit->maybe_tree || !commit->object.parsed)
 367                 return commit->maybe_tree;
 368
 369         if (commit_graph_position(commit) != COMMIT_NOT_FROM_GRAPH)
 370                 return get_commit_tree_in_graph(r, commit);
 371
 372         return NULL;
 373 }
 374
 375 struct object_id *get_commit_tree_oid(const struct commit *commit)
 376 {
 377         struct tree *tree = get_commit_tree(commit);
 378         return tree ? &tree->object.oid : NULL;
 379 }
 380
 381 void release_commit_memory(struct parsed_object_pool *pool, struct commit *c)
 382 {
 383         set_commit_tree(c, NULL);
 384         free_commit_buffer(pool, c);
 385         c->index = 0;
 386         free_commit_list(c->parents);
 387
 388         c->object.parsed = 0;
 389 }
 390
 391 const void *detach_commit_buffer(struct commit *commit, unsigned long *sizep)
 392 {
 393         struct commit_buffer *v = buffer_slab_peek(
 394                 the_repository->parsed_objects->buffer_slab, commit);
 395         void *ret;
 396
 397         if (!v) {
 398                 if (sizep)
 399                         *sizep = 0;
 400                 return NULL;
 401         }
 402         ret = v->buffer;
 403         if (sizep)
 404                 *sizep = v->size;
 405
 406         v->buffer = NULL;
 407         v->size = 0;
 408         return ret;
 409 }
 410
 411 int parse_commit_buffer(struct repository *r, struct commit *item, const void *buffer, unsigned long size, int check_graph)
 412 {
 413         const char *tail = buffer;
 414         const char *bufptr = buffer;
 415         struct object_id parent;
 416         struct commit_list **pptr;
 417         struct commit_graft *graft;
 418         const int tree_entry_len = the_hash_algo->hexsz + 5;
 419         const int parent_entry_len = the_hash_algo->hexsz + 7;
 420         struct tree *tree;
 421
 422         if (item->object.parsed)
 423                 return 0;
 424         /*
 425          * Presumably this is leftover from an earlier failed parse;
 426          * clear it out in preparation for us re-parsing (we'll hit the
 427          * same error, but that's good, since it lets our caller know
 428          * the result cannot be trusted.
 429          */
 430         free_commit_list(item->parents);
 431         item->parents = NULL;
 432
 433         tail += size;
 434         if (tail <= bufptr + tree_entry_len + 1 || memcmp(bufptr, "tree ", 5) ||
 435                         bufptr[tree_entry_len] != '\n')
 436                 return error("bogus commit object %s", oid_to_hex(&item->object.oid));
 437         if (get_oid_hex(bufptr + 5, &parent) < 0)
 438                 return error("bad tree pointer in commit %s",
 439                              oid_to_hex(&item->object.oid));
 440         tree = lookup_tree(r, &parent);
 441         if (!tree)
 442                 return error("bad tree pointer %s in commit %s",
 443                              oid_to_hex(&parent),
 444                              oid_to_hex(&item->object.oid));
 445         set_commit_tree(item, tree);
 446         bufptr += tree_entry_len + 1; /* "tree " + "hex sha1" + "\n" */
 447         pptr = &item->parents;
 448
 449         graft = lookup_commit_graft(r, &item->object.oid);
 450         if (graft)
 451                 r->parsed_objects->substituted_parent = 1;
 452         while (bufptr + parent_entry_len < tail && !memcmp(bufptr, "parent ", 7)) {
 453                 struct commit *new_parent;
 454
 455                 if (tail <= bufptr + parent_entry_len + 1 ||
 456                     get_oid_hex(bufptr + 7, &parent) ||
 457                     bufptr[parent_entry_len] != '\n')
 458                         return error("bad parents in commit %s", oid_to_hex(&item->object.oid));
 459                 bufptr += parent_entry_len + 1;
 460                 /*
 461                  * The clone is shallow if nr_parent < 0, and we must
 462                  * not traverse its real parents even when we unhide them.
 463                  */
 464                 if (graft && (graft->nr_parent < 0 || grafts_replace_parents))
 465                         continue;
 466                 new_parent = lookup_commit(r, &parent);
 467                 if (!new_parent)
 468                         return error("bad parent %s in commit %s",
 469                                      oid_to_hex(&parent),
 470                                      oid_to_hex(&item->object.oid));
 471                 pptr = &commit_list_insert(new_parent, pptr)->next;
 472         }
 473         if (graft) {
 474                 int i;
 475                 struct commit *new_parent;
 476                 for (i = 0; i < graft->nr_parent; i++) {
 477                         new_parent = lookup_commit(r,
 478                                                    &graft->parent[i]);
 479                         if (!new_parent)
 480                                 return error("bad graft parent %s in commit %s",
 481                                              oid_to_hex(&graft->parent[i]),
 482                                              oid_to_hex(&item->object.oid));
 483                         pptr = &commit_list_insert(new_parent, pptr)->next;
 484                 }
 485         }
 486         item->date = parse_commit_date(bufptr, tail);
 487
 488         if (check_graph)
 489                 load_commit_graph_info(r, item);
 490
 491         item->object.parsed = 1;
 492         return 0;
 493 }
 494
 495 int repo_parse_commit_internal(struct repository *r,
 496                                struct commit *item,
 497                                int quiet_on_missing,
 498                                int use_commit_graph)
 499 {
 500         enum object_type type;
 501         void *buffer;
 502         unsigned long size;
 503         int ret;
 504
 505         if (!item)
 506                 return -1;
 507         if (item->object.parsed)
 508                 return 0;
 509         if (use_commit_graph && parse_commit_in_graph(r, item))
 510                 return 0;
 511         buffer = repo_read_object_file(r, &item->object.oid, &type, &size);
 512         if (!buffer)
 513                 return quiet_on_missing ? -1 :
 514                         error("Could not read %s",
 515                              oid_to_hex(&item->object.oid));
 516         if (type != OBJ_COMMIT) {
 517                 free(buffer);
 518                 return error("Object %s not a commit",
 519                              oid_to_hex(&item->object.oid));
 520         }
 521
 522         ret = parse_commit_buffer(r, item, buffer, size, 0);
 523         if (save_commit_buffer && !ret) {
 524                 set_commit_buffer(r, item, buffer, size);
 525                 return 0;
 526         }
 527         free(buffer);
 528         return ret;
 529 }
 530
 531 int repo_parse_commit_gently(struct repository *r,
 532                              struct commit *item, int quiet_on_missing)
 533 {
 534         return repo_parse_commit_internal(r, item, quiet_on_missing, 1);
 535 }
 536
 537 void parse_commit_or_die(struct commit *item)
 538 {
 539         if (parse_commit(item))
 540                 die("unable to parse commit %s",
 541                     item ? oid_to_hex(&item->object.oid) : "(null)");
 542 }
 543
 544 int find_commit_subject(const char *commit_buffer, const char **subject)
 545 {
 546         const char *eol;
 547         const char *p = commit_buffer;
 548
 549         while (*p && (*p != '\n' || p[1] != '\n'))
 550                 p++;
 551         if (*p) {
 552                 p = skip_blank_lines(p + 2);
 553                 eol = strchrnul(p, '\n');
 554         } else
 555                 eol = p;
 556
 557         *subject = p;
 558
 559         return eol - p;
 560 }
 561
 562 size_t commit_subject_length(const char *body)
 563 {
 564         const char *p = body;
 565         while (*p) {
 566                 const char *next = skip_blank_lines(p);
 567                 if (next != p)
 568                         break;
 569                 p = strchrnul(p, '\n');
 570                 if (*p)
 571                         p++;
 572         }
 573         return p - body;
 574 }
 575
 576 struct commit_list *commit_list_insert(struct commit *item, struct commit_list **list_p)
 577 {
 578         struct commit_list *new_list = xmalloc(sizeof(struct commit_list));
 579         new_list->item = item;
 580         new_list->next = *list_p;
 581         *list_p = new_list;
 582         return new_list;
 583 }
 584
 585 int commit_list_contains(struct commit *item, struct commit_list *list)
 586 {
 587         while (list) {
 588                 if (list->item == item)
 589                         return 1;
 590                 list = list->next;
 591         }
 592
 593         return 0;
 594 }
 595
 596 unsigned commit_list_count(const struct commit_list *l)
 597 {
 598         unsigned c = 0;
 599         for (; l; l = l->next )
 600                 c++;
 601         return c;
 602 }
 603
 604 struct commit_list *copy_commit_list(struct commit_list *list)
 605 {
 606         struct commit_list *head = NULL;
 607         struct commit_list **pp = &head;
 608         while (list) {
 609                 pp = commit_list_append(list->item, pp);
 610                 list = list->next;
 611         }
 612         return head;
 613 }
 614
 615 struct commit_list *reverse_commit_list(struct commit_list *list)
 616 {
 617         struct commit_list *next = NULL, *current, *backup;
 618         for (current = list; current; current = backup) {
 619                 backup = current->next;
 620                 current->next = next;
 621                 next = current;
 622         }
 623         return next;
 624 }
 625
 626 void free_commit_list(struct commit_list *list)
 627 {
 628         while (list)
 629                 pop_commit(&list);
 630 }
 631
 632 struct commit_list * commit_list_insert_by_date(struct commit *item, struct commit_list **list)
 633 {
 634         struct commit_list **pp = list;
 635         struct commit_list *p;
 636         while ((p = *pp) != NULL) {
 637                 if (p->item->date < item->date) {
 638                         break;
 639                 }
 640                 pp = &p->next;
 641         }
 642         return commit_list_insert(item, pp);
 643 }
 644
 645 static int commit_list_compare_by_date(const struct commit_list *a,
 646                                        const struct commit_list *b)
 647 {
 648         timestamp_t a_date = a->item->date;
 649         timestamp_t b_date = b->item->date;
 650         if (a_date < b_date)
 651                 return 1;
 652         if (a_date > b_date)
 653                 return -1;
 654         return 0;
 655 }
 656
 657 DEFINE_LIST_SORT(static, commit_list_sort, struct commit_list, next);
 658
 659 void commit_list_sort_by_date(struct commit_list **list)
 660 {
 661         commit_list_sort(list, commit_list_compare_by_date);
 662 }
 663
 664 struct commit *pop_most_recent_commit(struct commit_list **list,
 665                                       unsigned int mark)
 666 {
 667         struct commit *ret = pop_commit(list);
 668         struct commit_list *parents = ret->parents;
 669
 670         while (parents) {
 671                 struct commit *commit = parents->item;
 672                 if (!parse_commit(commit) && !(commit->object.flags & mark)) {
 673                         commit->object.flags |= mark;
 674                         commit_list_insert_by_date(commit, list);
 675                 }
 676                 parents = parents->next;
 677         }
 678         return ret;
 679 }
 680
 681 static void clear_commit_marks_1(struct commit_list **plist,
 682                                  struct commit *commit, unsigned int mark)
 683 {
 684         while (commit) {
 685                 struct commit_list *parents;
 686
 687                 if (!(mark & commit->object.flags))
 688                         return;
 689
 690                 commit->object.flags &= ~mark;
 691
 692                 parents = commit->parents;
 693                 if (!parents)
 694                         return;
 695
 696                 while ((parents = parents->next))
 697                         commit_list_insert(parents->item, plist);
 698
 699                 commit = commit->parents->item;
 700         }
 701 }
 702
 703 void clear_commit_marks_many(int nr, struct commit **commit, unsigned int mark)
 704 {
 705         struct commit_list *list = NULL;
 706
 707         while (nr--) {
 708                 clear_commit_marks_1(&list, *commit, mark);
 709                 commit++;
 710         }
 711         while (list)
 712                 clear_commit_marks_1(&list, pop_commit(&list), mark);
 713 }
 714
 715 void clear_commit_marks(struct commit *commit, unsigned int mark)
 716 {
 717         clear_commit_marks_many(1, &commit, mark);
 718 }
 719
 720 struct commit *pop_commit(struct commit_list **stack)
 721 {
 722         struct commit_list *top = *stack;
 723         struct commit *item = top ? top->item : NULL;
 724
 725         if (top) {
 726                 *stack = top->next;
 727                 free(top);
 728         }
 729         return item;
 730 }
 731
 732 /*
 733  * Topological sort support
 734  */
 735
 736 /* count number of children that have not been emitted */
 737 define_commit_slab(indegree_slab, int);
 738
 739 define_commit_slab(author_date_slab, timestamp_t);
 740
 741 void record_author_date(struct author_date_slab *author_date,
 742                         struct commit *commit)
 743 {
 744         const char *buffer = get_commit_buffer(commit, NULL);
 745         struct ident_split ident;
 746         const char *ident_line;
 747         size_t ident_len;
 748         char *date_end;
 749         timestamp_t date;
 750
 751         ident_line = find_commit_header(buffer, "author", &ident_len);
 752         if (!ident_line)
 753                 goto fail_exit; /* no author line */
 754         if (split_ident_line(&ident, ident_line, ident_len) ||
 755             !ident.date_begin || !ident.date_end)
 756                 goto fail_exit; /* malformed "author" line */
 757
 758         date = parse_timestamp(ident.date_begin, &date_end, 10);
 759         if (date_end != ident.date_end)
 760                 goto fail_exit; /* malformed date */
 761         *(author_date_slab_at(author_date, commit)) = date;
 762
 763 fail_exit:
 764         unuse_commit_buffer(commit, buffer);
 765 }
 766
 767 int compare_commits_by_author_date(const void *a_, const void *b_,
 768                                    void *cb_data)
 769 {
 770         const struct commit *a = a_, *b = b_;
 771         struct author_date_slab *author_date = cb_data;
 772         timestamp_t a_date = *(author_date_slab_at(author_date, a));
 773         timestamp_t b_date = *(author_date_slab_at(author_date, b));
 774
 775         /* newer commits with larger date first */
 776         if (a_date < b_date)
 777                 return 1;
 778         else if (a_date > b_date)
 779                 return -1;
 780         return 0;
 781 }
 782
 783 int compare_commits_by_gen_then_commit_date(const void *a_, const void *b_, void *unused)
 784 {
 785         const struct commit *a = a_, *b = b_;
 786         const timestamp_t generation_a = commit_graph_generation(a),
 787                           generation_b = commit_graph_generation(b);
 788
 789         /* newer commits first */
 790         if (generation_a < generation_b)
 791                 return 1;
 792         else if (generation_a > generation_b)
 793                 return -1;
 794
 795         /* use date as a heuristic when generations are equal */
 796         if (a->date < b->date)
 797                 return 1;
 798         else if (a->date > b->date)
 799                 return -1;
 800         return 0;
 801 }
 802
 803 int compare_commits_by_commit_date(const void *a_, const void *b_, void *unused)
 804 {
 805         const struct commit *a = a_, *b = b_;
 806         /* newer commits with larger date first */
 807         if (a->date < b->date)
 808                 return 1;
 809         else if (a->date > b->date)
 810                 return -1;
 811         return 0;
 812 }
 813
 814 /*
 815  * Performs an in-place topological sort on the list supplied.
 816  */
 817 void sort_in_topological_order(struct commit_list **list, enum rev_sort_order sort_order)
 818 {
 819         struct commit_list *next, *orig = *list;
 820         struct commit_list **pptr;
 821         struct indegree_slab indegree;
 822         struct prio_queue queue;
 823         struct commit *commit;
 824         struct author_date_slab author_date;
 825
 826         if (!orig)
 827                 return;
 828         *list = NULL;
 829
 830         init_indegree_slab(&indegree);
 831         memset(&queue, '\0', sizeof(queue));
 832
 833         switch (sort_order) {
 834         default: /* REV_SORT_IN_GRAPH_ORDER */
 835                 queue.compare = NULL;
 836                 break;
 837         case REV_SORT_BY_COMMIT_DATE:
 838                 queue.compare = compare_commits_by_commit_date;
 839                 break;
 840         case REV_SORT_BY_AUTHOR_DATE:
 841                 init_author_date_slab(&author_date);
 842                 queue.compare = compare_commits_by_author_date;
 843                 queue.cb_data = &author_date;
 844                 break;
 845         }
 846
 847         /* Mark them and clear the indegree */
 848         for (next = orig; next; next = next->next) {
 849                 struct commit *commit = next->item;
 850                 *(indegree_slab_at(&indegree, commit)) = 1;
 851                 /* also record the author dates, if needed */
 852                 if (sort_order == REV_SORT_BY_AUTHOR_DATE)
 853                         record_author_date(&author_date, commit);
 854         }
 855
 856         /* update the indegree */
 857         for (next = orig; next; next = next->next) {
 858                 struct commit_list *parents = next->item->parents;
 859                 while (parents) {
 860                         struct commit *parent = parents->item;
 861                         int *pi = indegree_slab_at(&indegree, parent);
 862
 863                         if (*pi)
 864                                 (*pi)++;
 865                         parents = parents->next;
 866                 }
 867         }
 868
 869         /*
 870          * find the tips
 871          *
 872          * tips are nodes not reachable from any other node in the list
 873          *
 874          * the tips serve as a starting set for the work queue.
 875          */
 876         for (next = orig; next; next = next->next) {
 877                 struct commit *commit = next->item;
 878
 879                 if (*(indegree_slab_at(&indegree, commit)) == 1)
 880                         prio_queue_put(&queue, commit);
 881         }
 882
 883         /*
 884          * This is unfortunate; the initial tips need to be shown
 885          * in the order given from the revision traversal machinery.
 886          */
 887         if (sort_order == REV_SORT_IN_GRAPH_ORDER)
 888                 prio_queue_reverse(&queue);
 889
 890         /* We no longer need the commit list */
 891         free_commit_list(orig);
 892
 893         pptr = list;
 894         *list = NULL;
 895         while ((commit = prio_queue_get(&queue)) != NULL) {
 896                 struct commit_list *parents;
 897
 898                 for (parents = commit->parents; parents ; parents = parents->next) {
 899                         struct commit *parent = parents->item;
 900                         int *pi = indegree_slab_at(&indegree, parent);
 901
 902                         if (!*pi)
 903                                 continue;
 904
 905                         /*
 906                          * parents are only enqueued for emission
 907                          * when all their children have been emitted thereby
 908                          * guaranteeing topological order.
 909                          */
 910                         if (--(*pi) == 1)
 911                                 prio_queue_put(&queue, parent);
 912                 }
 913                 /*
 914                  * all children of commit have already been
 915                  * emitted. we can emit it now.
 916                  */
 917                 *(indegree_slab_at(&indegree, commit)) = 0;
 918
 919                 pptr = &commit_list_insert(commit, pptr)->next;
 920         }
 921
 922         clear_indegree_slab(&indegree);
 923         clear_prio_queue(&queue);
 924         if (sort_order == REV_SORT_BY_AUTHOR_DATE)
 925                 clear_author_date_slab(&author_date);
 926 }
 927
 928 struct rev_collect {
 929         struct commit **commit;
 930         int nr;
 931         int alloc;
 932         unsigned int initial : 1;
 933 };
 934
 935 static void add_one_commit(struct object_id *oid, struct rev_collect *revs)
 936 {
 937         struct commit *commit;
 938
 939         if (is_null_oid(oid))
 940                 return;
 941
 942         commit = lookup_commit(the_repository, oid);
 943         if (!commit ||
 944             (commit->object.flags & TMP_MARK) ||
 945             parse_commit(commit))
 946                 return;
 947
 948         ALLOC_GROW(revs->commit, revs->nr + 1, revs->alloc);
 949         revs->commit[revs->nr++] = commit;
 950         commit->object.flags |= TMP_MARK;
 951 }
 952
 953 static int collect_one_reflog_ent(struct object_id *ooid, struct object_id *noid,
 954                                   const char *ident UNUSED,
 955                                   timestamp_t timestamp UNUSED, int tz UNUSED,
 956                                   const char *message UNUSED, void *cbdata)
 957 {
 958         struct rev_collect *revs = cbdata;
 959
 960         if (revs->initial) {
 961                 revs->initial = 0;
 962                 add_one_commit(ooid, revs);
 963         }
 964         add_one_commit(noid, revs);
 965         return 0;
 966 }
 967
 968 struct commit *get_fork_point(const char *refname, struct commit *commit)
 969 {
 970         struct object_id oid;
 971         struct rev_collect revs;
 972         struct commit_list *bases;
 973         int i;
 974         struct commit *ret = NULL;
 975         char *full_refname;
 976
 977         switch (dwim_ref(refname, strlen(refname), &oid, &full_refname, 0)) {
 978         case 0:
 979                 die("No such ref: '%s'", refname);
 980         case 1:
 981                 break; /* good */
 982         default:
 983                 die("Ambiguous refname: '%s'", refname);
 984         }
 985
 986         memset(&revs, 0, sizeof(revs));
 987         revs.initial = 1;
 988         for_each_reflog_ent(full_refname, collect_one_reflog_ent, &revs);
 989
 990         if (!revs.nr)
 991                 add_one_commit(&oid, &revs);
 992
 993         for (i = 0; i < revs.nr; i++)
 994                 revs.commit[i]->object.flags &= ~TMP_MARK;
 995
 996         bases = get_merge_bases_many(commit, revs.nr, revs.commit);
 997
 998         /*
 999          * There should be one and only one merge base, when we found
1000          * a common ancestor among reflog entries.
1001          */
1002         if (!bases || bases->next)
1003                 goto cleanup_return;
1004
1005         /* And the found one must be one of the reflog entries */
1006         for (i = 0; i < revs.nr; i++)
1007                 if (&bases->item->object == &revs.commit[i]->object)
1008                         break; /* found */
1009         if (revs.nr <= i)
1010                 goto cleanup_return;
1011
1012         ret = bases->item;
1013
1014 cleanup_return:
1015         free_commit_list(bases);
1016         free(full_refname);
1017         return ret;
1018 }
1019
1020 /*
1021  * Indexed by hash algorithm identifier.
1022  */
1023 static const char *gpg_sig_headers[] = {
1024         NULL,
1025         "gpgsig",
1026         "gpgsig-sha256",
1027 };
1028
1029 int sign_with_header(struct strbuf *buf, const char *keyid)
1030 {
1031         struct strbuf sig = STRBUF_INIT;
1032         int inspos, copypos;
1033         const char *eoh;
1034         const char *gpg_sig_header = gpg_sig_headers[hash_algo_by_ptr(the_hash_algo)];
1035         int gpg_sig_header_len = strlen(gpg_sig_header);
1036
1037         /* find the end of the header */
1038         eoh = strstr(buf->buf, "\n\n");
1039         if (!eoh)
1040                 inspos = buf->len;
1041         else
1042                 inspos = eoh - buf->buf + 1;
1043
1044         if (!keyid || !*keyid)
1045                 keyid = get_signing_key();
1046         if (sign_buffer(buf, &sig, keyid)) {
1047                 strbuf_release(&sig);
1048                 return -1;
1049         }
1050
1051         for (copypos = 0; sig.buf[copypos]; ) {
1052                 const char *bol = sig.buf + copypos;
1053                 const char *eol = strchrnul(bol, '\n');
1054                 int len = (eol - bol) + !!*eol;
1055
1056                 if (!copypos) {
1057                         strbuf_insert(buf, inspos, gpg_sig_header, gpg_sig_header_len);
1058                         inspos += gpg_sig_header_len;
1059                 }
1060                 strbuf_insertstr(buf, inspos++, " ");
1061                 strbuf_insert(buf, inspos, bol, len);
1062                 inspos += len;
1063                 copypos += len;
1064         }
1065         strbuf_release(&sig);
1066         return 0;
1067 }
1068
1069
1070
1071 int parse_signed_commit(const struct commit *commit,
1072                         struct strbuf *payload, struct strbuf *signature,
1073                         const struct git_hash_algo *algop)
1074 {
1075         unsigned long size;
1076         const char *buffer = get_commit_buffer(commit, &size);
1077         int ret = parse_buffer_signed_by_header(buffer, size, payload, signature, algop);
1078
1079         unuse_commit_buffer(commit, buffer);
1080         return ret;
1081 }
1082
1083 int parse_buffer_signed_by_header(const char *buffer,
1084                                   unsigned long size,
1085                                   struct strbuf *payload,
1086                                   struct strbuf *signature,
1087                                   const struct git_hash_algo *algop)
1088 {
1089         int in_signature = 0, saw_signature = 0, other_signature = 0;
1090         const char *line, *tail, *p;
1091         const char *gpg_sig_header = gpg_sig_headers[hash_algo_by_ptr(algop)];
1092
1093         line = buffer;
1094         tail = buffer + size;
1095         while (line < tail) {
1096                 const char *sig = NULL;
1097                 const char *next = memchr(line, '\n', tail - line);
1098
1099                 next = next ? next + 1 : tail;
1100                 if (in_signature && line[0] == ' ')
1101                         sig = line + 1;
1102                 else if (skip_prefix(line, gpg_sig_header, &p) &&
1103                          *p == ' ') {
1104                         sig = line + strlen(gpg_sig_header) + 1;
1105                         other_signature = 0;
1106                 }
1107                 else if (starts_with(line, "gpgsig"))
1108                         other_signature = 1;
1109                 else if (other_signature && line[0] != ' ')
1110                         other_signature = 0;
1111                 if (sig) {
1112                         strbuf_add(signature, sig, next - sig);
1113                         saw_signature = 1;
1114                         in_signature = 1;
1115                 } else {
1116                         if (*line == '\n')
1117                                 /* dump the whole remainder of the buffer */
1118                                 next = tail;
1119                         if (!other_signature)
1120                                 strbuf_add(payload, line, next - line);
1121                         in_signature = 0;
1122                 }
1123                 line = next;
1124         }
1125         return saw_signature;
1126 }
1127
1128 int remove_signature(struct strbuf *buf)
1129 {
1130         const char *line = buf->buf;
1131         const char *tail = buf->buf + buf->len;
1132         int in_signature = 0;
1133         struct sigbuf {
1134                 const char *start;
1135                 const char *end;
1136         } sigs[2], *sigp = &sigs[0];
1137         int i;
1138         const char *orig_buf = buf->buf;
1139
1140         memset(sigs, 0, sizeof(sigs));
1141
1142         while (line < tail) {
1143                 const char *next = memchr(line, '\n', tail - line);
1144                 next = next ? next + 1 : tail;
1145
1146                 if (in_signature && line[0] == ' ')
1147                         sigp->end = next;
1148                 else if (starts_with(line, "gpgsig")) {
1149                         int i;
1150                         for (i = 1; i < GIT_HASH_NALGOS; i++) {
1151                                 const char *p;
1152                                 if (skip_prefix(line, gpg_sig_headers[i], &p) &&
1153                                     *p == ' ') {
1154                                         sigp->start = line;
1155                                         sigp->end = next;
1156                                         in_signature = 1;
1157                                 }
1158                         }
1159                 } else {
1160                         if (*line == '\n')
1161                                 /* dump the whole remainder of the buffer */
1162                                 next = tail;
1163                         if (in_signature && sigp - sigs != ARRAY_SIZE(sigs))
1164                                 sigp++;
1165                         in_signature = 0;
1166                 }
1167                 line = next;
1168         }
1169
1170         for (i = ARRAY_SIZE(sigs) - 1; i >= 0; i--)
1171                 if (sigs[i].start)
1172                         strbuf_remove(buf, sigs[i].start - orig_buf, sigs[i].end - sigs[i].start);
1173
1174         return sigs[0].start != NULL;
1175 }
1176
1177 static void handle_signed_tag(struct commit *parent, struct commit_extra_header ***tail)
1178 {
1179         struct merge_remote_desc *desc;
1180         struct commit_extra_header *mergetag;
1181         char *buf;
1182         unsigned long size;
1183         enum object_type type;
1184         struct strbuf payload = STRBUF_INIT;
1185         struct strbuf signature = STRBUF_INIT;
1186
1187         desc = merge_remote_util(parent);
1188         if (!desc || !desc->obj)
1189                 return;
1190         buf = read_object_file(&desc->obj->oid, &type, &size);
1191         if (!buf || type != OBJ_TAG)
1192                 goto free_return;
1193         if (!parse_signature(buf, size, &payload, &signature))
1194                 goto free_return;
1195         /*
1196          * We could verify this signature and either omit the tag when
1197          * it does not validate, but the integrator may not have the
1198          * public key of the signer of the tag being merged, while a
1199          * later auditor may have it while auditing, so let's not run
1200          * verify-signed-buffer here for now...
1201          *
1202          * if (verify_signed_buffer(buf, len, buf + len, size - len, ...))
1203          *      warn("warning: signed tag unverified.");
1204          */
1205         CALLOC_ARRAY(mergetag, 1);
1206         mergetag->key = xstrdup("mergetag");
1207         mergetag->value = buf;
1208         mergetag->len = size;
1209
1210         **tail = mergetag;
1211         *tail = &mergetag->next;
1212         strbuf_release(&payload);
1213         strbuf_release(&signature);
1214         return;
1215
1216 free_return:
1217         free(buf);
1218 }
1219
1220 int check_commit_signature(const struct commit *commit, struct signature_check *sigc)
1221 {
1222         struct strbuf payload = STRBUF_INIT;
1223         struct strbuf signature = STRBUF_INIT;
1224         int ret = 1;
1225
1226         sigc->result = 'N';
1227
1228         if (parse_signed_commit(commit, &payload, &signature, the_hash_algo) <= 0)
1229                 goto out;
1230
1231         sigc->payload_type = SIGNATURE_PAYLOAD_COMMIT;
1232         sigc->payload = strbuf_detach(&payload, &sigc->payload_len);
1233         ret = check_signature(sigc, signature.buf, signature.len);
1234
1235  out:
1236         strbuf_release(&payload);
1237         strbuf_release(&signature);
1238
1239         return ret;
1240 }
1241
1242 void verify_merge_signature(struct commit *commit, int verbosity,
1243                             int check_trust)
1244 {
1245         char hex[GIT_MAX_HEXSZ + 1];
1246         struct signature_check signature_check;
1247         int ret;
1248         memset(&signature_check, 0, sizeof(signature_check));
1249
1250         ret = check_commit_signature(commit, &signature_check);
1251
1252         find_unique_abbrev_r(hex, &commit->object.oid, DEFAULT_ABBREV);
1253         switch (signature_check.result) {
1254         case 'G':
1255                 if (ret || (check_trust && signature_check.trust_level < TRUST_MARGINAL))
1256                         die(_("Commit %s has an untrusted GPG signature, "
1257                               "allegedly by %s."), hex, signature_check.signer);
1258                 break;
1259         case 'B':
1260                 die(_("Commit %s has a bad GPG signature "
1261                       "allegedly by %s."), hex, signature_check.signer);
1262         default: /* 'N' */
1263                 die(_("Commit %s does not have a GPG signature."), hex);
1264         }
1265         if (verbosity >= 0 && signature_check.result == 'G')
1266                 printf(_("Commit %s has a good GPG signature by %s\n"),
1267                        hex, signature_check.signer);
1268
1269         signature_check_clear(&signature_check);
1270 }
1271
1272 void append_merge_tag_headers(struct commit_list *parents,
1273                               struct commit_extra_header ***tail)
1274 {
1275         while (parents) {
1276                 struct commit *parent = parents->item;
1277                 handle_signed_tag(parent, tail);
1278                 parents = parents->next;
1279         }
1280 }
1281
1282 static void add_extra_header(struct strbuf *buffer,
1283                              struct commit_extra_header *extra)
1284 {
1285         strbuf_addstr(buffer, extra->key);
1286         if (extra->len)
1287                 strbuf_add_lines(buffer, " ", extra->value, extra->len);
1288         else
1289                 strbuf_addch(buffer, '\n');
1290 }
1291
1292 struct commit_extra_header *read_commit_extra_headers(struct commit *commit,
1293                                                       const char **exclude)
1294 {
1295         struct commit_extra_header *extra = NULL;
1296         unsigned long size;
1297         const char *buffer = get_commit_buffer(commit, &size);
1298         extra = read_commit_extra_header_lines(buffer, size, exclude);
1299         unuse_commit_buffer(commit, buffer);
1300         return extra;
1301 }
1302
1303 int for_each_mergetag(each_mergetag_fn fn, struct commit *commit, void *data)
1304 {
1305         struct commit_extra_header *extra, *to_free;
1306         int res = 0;
1307
1308         to_free = read_commit_extra_headers(commit, NULL);
1309         for (extra = to_free; !res && extra; extra = extra->next) {
1310                 if (strcmp(extra->key, "mergetag"))
1311                         continue; /* not a merge tag */
1312                 res = fn(commit, extra, data);
1313         }
1314         free_commit_extra_headers(to_free);
1315         return res;
1316 }
1317
1318 static inline int standard_header_field(const char *field, size_t len)
1319 {
1320         return ((len == 4 && !memcmp(field, "tree", 4)) ||
1321                 (len == 6 && !memcmp(field, "parent", 6)) ||
1322                 (len == 6 && !memcmp(field, "author", 6)) ||
1323                 (len == 9 && !memcmp(field, "committer", 9)) ||
1324                 (len == 8 && !memcmp(field, "encoding", 8)));
1325 }
1326
1327 static int excluded_header_field(const char *field, size_t len, const char **exclude)
1328 {
1329         if (!exclude)
1330                 return 0;
1331
1332         while (*exclude) {
1333                 size_t xlen = strlen(*exclude);
1334                 if (len == xlen && !memcmp(field, *exclude, 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 = memchr(line, ' ', next - line);
1366                 if (!eof)
1367                         eof = next;
1368                 else if (standard_header_field(line, eof - line) ||
1369                          excluded_header_field(line, eof - line, exclude))
1370                         continue;
1371
1372                 CALLOC_ARRAY(it, 1);
1373                 it->key = xmemdupz(line, eof-line);
1374                 *tail = it;
1375                 tail = &it->next;
1376                 if (eof + 1 < next)
1377                         strbuf_add(&buf, eof + 1, next - (eof + 1));
1378         }
1379         if (it)
1380                 it->value = strbuf_detach(&buf, &it->len);
1381         return extra;
1382 }
1383
1384 void free_commit_extra_headers(struct commit_extra_header *extra)
1385 {
1386         while (extra) {
1387                 struct commit_extra_header *next = extra->next;
1388                 free(extra->key);
1389                 free(extra->value);
1390                 free(extra);
1391                 extra = next;
1392         }
1393 }
1394
1395 int commit_tree(const char *msg, size_t msg_len, const struct object_id *tree,
1396                 struct commit_list *parents, struct object_id *ret,
1397                 const char *author, const char *sign_commit)
1398 {
1399         struct commit_extra_header *extra = NULL, **tail = &extra;
1400         int result;
1401
1402         append_merge_tag_headers(parents, &tail);
1403         result = commit_tree_extended(msg, msg_len, tree, parents, ret, author,
1404                                       NULL, sign_commit, extra);
1405         free_commit_extra_headers(extra);
1406         return result;
1407 }
1408
1409 static int find_invalid_utf8(const char *buf, int len)
1410 {
1411         int offset = 0;
1412         static const unsigned int max_codepoint[] = {
1413                 0x7f, 0x7ff, 0xffff, 0x10ffff
1414         };
1415
1416         while (len) {
1417                 unsigned char c = *buf++;
1418                 int bytes, bad_offset;
1419                 unsigned int codepoint;
1420                 unsigned int min_val, max_val;
1421
1422                 len--;
1423                 offset++;
1424
1425                 /* Simple US-ASCII? No worries. */
1426                 if (c < 0x80)
1427                         continue;
1428
1429                 bad_offset = offset-1;
1430
1431                 /*
1432                  * Count how many more high bits set: that's how
1433                  * many more bytes this sequence should have.
1434                  */
1435                 bytes = 0;
1436                 while (c & 0x40) {
1437                         c <<= 1;
1438                         bytes++;
1439                 }
1440
1441                 /*
1442                  * Must be between 1 and 3 more bytes.  Longer sequences result in
1443                  * codepoints beyond U+10FFFF, which are guaranteed never to exist.
1444                  */
1445                 if (bytes < 1 || 3 < bytes)
1446                         return bad_offset;
1447
1448                 /* Do we *have* that many bytes? */
1449                 if (len < bytes)
1450                         return bad_offset;
1451
1452                 /*
1453                  * Place the encoded bits at the bottom of the value and compute the
1454                  * valid range.
1455                  */
1456                 codepoint = (c & 0x7f) >> bytes;
1457                 min_val = max_codepoint[bytes-1] + 1;
1458                 max_val = max_codepoint[bytes];
1459
1460                 offset += bytes;
1461                 len -= bytes;
1462
1463                 /* And verify that they are good continuation bytes */
1464                 do {
1465                         codepoint <<= 6;
1466                         codepoint |= *buf & 0x3f;
1467                         if ((*buf++ & 0xc0) != 0x80)
1468                                 return bad_offset;
1469                 } while (--bytes);
1470
1471                 /* Reject codepoints that are out of range for the sequence length. */
1472                 if (codepoint < min_val || codepoint > max_val)
1473                         return bad_offset;
1474                 /* Surrogates are only for UTF-16 and cannot be encoded in UTF-8. */
1475                 if ((codepoint & 0x1ff800) == 0xd800)
1476                         return bad_offset;
1477                 /* U+xxFFFE and U+xxFFFF are guaranteed non-characters. */
1478                 if ((codepoint & 0xfffe) == 0xfffe)
1479                         return bad_offset;
1480                 /* So are anything in the range U+FDD0..U+FDEF. */
1481                 if (codepoint >= 0xfdd0 && codepoint <= 0xfdef)
1482                         return bad_offset;
1483         }
1484         return -1;
1485 }
1486
1487 /*
1488  * This verifies that the buffer is in proper utf8 format.
1489  *
1490  * If it isn't, it assumes any non-utf8 characters are Latin1,
1491  * and does the conversion.
1492  */
1493 static int verify_utf8(struct strbuf *buf)
1494 {
1495         int ok = 1;
1496         long pos = 0;
1497
1498         for (;;) {
1499                 int bad;
1500                 unsigned char c;
1501                 unsigned char replace[2];
1502
1503                 bad = find_invalid_utf8(buf->buf + pos, buf->len - pos);
1504                 if (bad < 0)
1505                         return ok;
1506                 pos += bad;
1507                 ok = 0;
1508                 c = buf->buf[pos];
1509                 strbuf_remove(buf, pos, 1);
1510
1511                 /* We know 'c' must be in the range 128-255 */
1512                 replace[0] = 0xc0 + (c >> 6);
1513                 replace[1] = 0x80 + (c & 0x3f);
1514                 strbuf_insert(buf, pos, replace, 2);
1515                 pos += 2;
1516         }
1517 }
1518
1519 static const char commit_utf8_warn[] =
1520 N_("Warning: commit message did not conform to UTF-8.\n"
1521    "You may want to amend it after fixing the message, or set the config\n"
1522    "variable i18n.commitEncoding to the encoding your project uses.\n");
1523
1524 int commit_tree_extended(const char *msg, size_t msg_len,
1525                          const struct object_id *tree,
1526                          struct commit_list *parents, struct object_id *ret,
1527                          const char *author, const char *committer,
1528                          const char *sign_commit,
1529                          struct commit_extra_header *extra)
1530 {
1531         int result;
1532         int encoding_is_utf8;
1533         struct strbuf buffer;
1534
1535         assert_oid_type(tree, OBJ_TREE);
1536
1537         if (memchr(msg, '\0', msg_len))
1538                 return error("a NUL byte in commit log message not allowed.");
1539
1540         /* Not having i18n.commitencoding is the same as having utf-8 */
1541         encoding_is_utf8 = is_encoding_utf8(git_commit_encoding);
1542
1543         strbuf_init(&buffer, 8192); /* should avoid reallocs for the headers */
1544         strbuf_addf(&buffer, "tree %s\n", oid_to_hex(tree));
1545
1546         /*
1547          * NOTE! This ordering means that the same exact tree merged with a
1548          * different order of parents will be a _different_ changeset even
1549          * if everything else stays the same.
1550          */
1551         while (parents) {
1552                 struct commit *parent = pop_commit(&parents);
1553                 strbuf_addf(&buffer, "parent %s\n",
1554                             oid_to_hex(&parent->object.oid));
1555         }
1556
1557         /* Person/date information */
1558         if (!author)
1559                 author = git_author_info(IDENT_STRICT);
1560         strbuf_addf(&buffer, "author %s\n", author);
1561         if (!committer)
1562                 committer = git_committer_info(IDENT_STRICT);
1563         strbuf_addf(&buffer, "committer %s\n", committer);
1564         if (!encoding_is_utf8)
1565                 strbuf_addf(&buffer, "encoding %s\n", git_commit_encoding);
1566
1567         while (extra) {
1568                 add_extra_header(&buffer, extra);
1569                 extra = extra->next;
1570         }
1571         strbuf_addch(&buffer, '\n');
1572
1573         /* And add the comment */
1574         strbuf_add(&buffer, msg, msg_len);
1575
1576         /* And check the encoding */
1577         if (encoding_is_utf8 && !verify_utf8(&buffer))
1578                 fprintf(stderr, _(commit_utf8_warn));
1579
1580         if (sign_commit && sign_with_header(&buffer, sign_commit)) {
1581                 result = -1;
1582                 goto out;
1583         }
1584
1585         result = write_object_file(buffer.buf, buffer.len, OBJ_COMMIT, ret);
1586 out:
1587         strbuf_release(&buffer);
1588         return result;
1589 }
1590
1591 define_commit_slab(merge_desc_slab, struct merge_remote_desc *);
1592 static struct merge_desc_slab merge_desc_slab = COMMIT_SLAB_INIT(1, merge_desc_slab);
1593
1594 struct merge_remote_desc *merge_remote_util(struct commit *commit)
1595 {
1596         return *merge_desc_slab_at(&merge_desc_slab, commit);
1597 }
1598
1599 void set_merge_remote_desc(struct commit *commit,
1600                            const char *name, struct object *obj)
1601 {
1602         struct merge_remote_desc *desc;
1603         FLEX_ALLOC_STR(desc, name, name);
1604         desc->obj = obj;
1605         *merge_desc_slab_at(&merge_desc_slab, commit) = desc;
1606 }
1607
1608 struct commit *get_merge_parent(const char *name)
1609 {
1610         struct object *obj;
1611         struct commit *commit;
1612         struct object_id oid;
1613         if (get_oid(name, &oid))
1614                 return NULL;
1615         obj = parse_object(the_repository, &oid);
1616         commit = (struct commit *)peel_to_type(name, 0, obj, OBJ_COMMIT);
1617         if (commit && !merge_remote_util(commit))
1618                 set_merge_remote_desc(commit, name, obj);
1619         return commit;
1620 }
1621
1622 /*
1623  * Append a commit to the end of the commit_list.
1624  *
1625  * next starts by pointing to the variable that holds the head of an
1626  * empty commit_list, and is updated to point to the "next" field of
1627  * the last item on the list as new commits are appended.
1628  *
1629  * Usage example:
1630  *
1631  *     struct commit_list *list;
1632  *     struct commit_list **next = &list;
1633  *
1634  *     next = commit_list_append(c1, next);
1635  *     next = commit_list_append(c2, next);
1636  *     assert(commit_list_count(list) == 2);
1637  *     return list;
1638  */
1639 struct commit_list **commit_list_append(struct commit *commit,
1640                                         struct commit_list **next)
1641 {
1642         struct commit_list *new_commit = xmalloc(sizeof(struct commit_list));
1643         new_commit->item = commit;
1644         *next = new_commit;
1645         new_commit->next = NULL;
1646         return &new_commit->next;
1647 }
1648
1649 const char *find_header_mem(const char *msg, size_t len,
1650                         const char *key, size_t *out_len)
1651 {
1652         int key_len = strlen(key);
1653         const char *line = msg;
1654
1655         /*
1656          * NEEDSWORK: It's possible for strchrnul() to scan beyond the range
1657          * given by len. However, current callers are safe because they compute
1658          * len by scanning a NUL-terminated block of memory starting at msg.
1659          * Nonetheless, it would be better to ensure the function does not look
1660          * at msg beyond the len provided by the caller.
1661          */
1662         while (line && line < msg + len) {
1663                 const char *eol = strchrnul(line, '\n');
1664
1665                 if (line == eol)
1666                         return NULL;
1667
1668                 if (eol - line > key_len &&
1669                     !strncmp(line, key, key_len) &&
1670                     line[key_len] == ' ') {
1671                         *out_len = eol - line - key_len - 1;
1672                         return line + key_len + 1;
1673                 }
1674                 line = *eol ? eol + 1 : NULL;
1675         }
1676         return NULL;
1677 }
1678
1679 const char *find_commit_header(const char *msg, const char *key, size_t *out_len)
1680 {
1681         return find_header_mem(msg, strlen(msg), key, out_len);
1682 }
1683 /*
1684  * Inspect the given string and determine the true "end" of the log message, in
1685  * order to find where to put a new Signed-off-by trailer.  Ignored are
1686  * trailing comment lines and blank lines.  To support "git commit -s
1687  * --amend" on an existing commit, we also ignore "Conflicts:".  To
1688  * support "git commit -v", we truncate at cut lines.
1689  *
1690  * Returns the number of bytes from the tail to ignore, to be fed as
1691  * the second parameter to append_signoff().
1692  */
1693 size_t ignore_non_trailer(const char *buf, size_t len)
1694 {
1695         size_t boc = 0;
1696         size_t bol = 0;
1697         int in_old_conflicts_block = 0;
1698         size_t cutoff = wt_status_locate_end(buf, len);
1699
1700         while (bol < cutoff) {
1701                 const char *next_line = memchr(buf + bol, '\n', len - bol);
1702
1703                 if (!next_line)
1704                         next_line = buf + len;
1705                 else
1706                         next_line++;
1707
1708                 if (buf[bol] == comment_line_char || buf[bol] == '\n') {
1709                         /* is this the first of the run of comments? */
1710                         if (!boc)
1711                                 boc = bol;
1712                         /* otherwise, it is just continuing */
1713                 } else if (starts_with(buf + bol, "Conflicts:\n")) {
1714                         in_old_conflicts_block = 1;
1715                         if (!boc)
1716                                 boc = bol;
1717                 } else if (in_old_conflicts_block && buf[bol] == '\t') {
1718                         ; /* a pathname in the conflicts block */
1719                 } else if (boc) {
1720                         /* the previous was not trailing comment */
1721                         boc = 0;
1722                         in_old_conflicts_block = 0;
1723                 }
1724                 bol = next_line - buf;
1725         }
1726         return boc ? len - boc : len - cutoff;
1727 }
1728
1729 int run_commit_hook(int editor_is_used, const char *index_file,
1730                     int *invoked_hook, const char *name, ...)
1731 {
1732         struct run_hooks_opt opt = RUN_HOOKS_OPT_INIT;
1733         va_list args;
1734         const char *arg;
1735
1736         strvec_pushf(&opt.env, "GIT_INDEX_FILE=%s", index_file);
1737
1738         /*
1739          * Let the hook know that no editor will be launched.
1740          */
1741         if (!editor_is_used)
1742                 strvec_push(&opt.env, "GIT_EDITOR=:");
1743
1744         va_start(args, name);
1745         while ((arg = va_arg(args, const char *)))
1746                 strvec_push(&opt.args, arg);
1747         va_end(args);
1748
1749         opt.invoked_hook = invoked_hook;
1750         return run_hooks_opt(name, &opt);
1751 }