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 int register_commit_graft(struct repository *r, struct commit_graft *graft,
 124                           int ignore_dups)
 125 {
 126         int pos = commit_graft_pos(r, &graft->oid);
 127
 128         if (0 <= pos) {
 129                 if (ignore_dups)
 130                         free(graft);
 131                 else {
 132                         free(r->parsed_objects->grafts[pos]);
 133                         r->parsed_objects->grafts[pos] = graft;
 134                 }
 135                 return 1;
 136         }
 137         pos = -pos - 1;
 138         ALLOC_GROW(r->parsed_objects->grafts,
 139                    r->parsed_objects->grafts_nr + 1,
 140                    r->parsed_objects->grafts_alloc);
 141         r->parsed_objects->grafts_nr++;
 142         if (pos < r->parsed_objects->grafts_nr)
 143                 memmove(r->parsed_objects->grafts + pos + 1,
 144                         r->parsed_objects->grafts + pos,
 145                         (r->parsed_objects->grafts_nr - pos - 1) *
 146                         sizeof(*r->parsed_objects->grafts));
 147         r->parsed_objects->grafts[pos] = graft;
 148         return 0;
 149 }
 150
 151 struct commit_graft *read_graft_line(struct strbuf *line)
 152 {
 153         /* The format is just "Commit Parent1 Parent2 ...\n" */
 154         int i, phase;
 155         const char *tail = NULL;
 156         struct commit_graft *graft = NULL;
 157         struct object_id dummy_oid, *oid;
 158
 159         strbuf_rtrim(line);
 160         if (!line->len || line->buf[0] == '#')
 161                 return NULL;
 162         /*
 163          * phase 0 verifies line, counts hashes in line and allocates graft
 164          * phase 1 fills graft
 165          */
 166         for (phase = 0; phase < 2; phase++) {
 167                 oid = graft ? &graft->oid : &dummy_oid;
 168                 if (parse_oid_hex(line->buf, oid, &tail))
 169                         goto bad_graft_data;
 170                 for (i = 0; *tail != '\0'; i++) {
 171                         oid = graft ? &graft->parent[i] : &dummy_oid;
 172                         if (!isspace(*tail++) || parse_oid_hex(tail, oid, &tail))
 173                                 goto bad_graft_data;
 174                 }
 175                 if (!graft) {
 176                         graft = xmalloc(st_add(sizeof(*graft),
 177                                                st_mult(sizeof(struct object_id), i)));
 178                         graft->nr_parent = i;
 179                 }
 180         }
 181         return graft;
 182
 183 bad_graft_data:
 184         error("bad graft data: %s", line->buf);
 185         assert(!graft);
 186         return NULL;
 187 }
 188
 189 static int read_graft_file(struct repository *r, const char *graft_file)
 190 {
 191         FILE *fp = fopen_or_warn(graft_file, "r");
 192         struct strbuf buf = STRBUF_INIT;
 193         if (!fp)
 194                 return -1;
 195         if (!no_graft_file_deprecated_advice &&
 196             advice_enabled(ADVICE_GRAFT_FILE_DEPRECATED))
 197                 advise(_("Support for <GIT_DIR>/info/grafts is deprecated\n"
 198                          "and will be removed in a future Git version.\n"
 199                          "\n"
 200                          "Please use \"git replace --convert-graft-file\"\n"
 201                          "to convert the grafts into replace refs.\n"
 202                          "\n"
 203                          "Turn this message off by running\n"
 204                          "\"git config advice.graftFileDeprecated false\""));
 205         while (!strbuf_getwholeline(&buf, fp, '\n')) {
 206                 /* The format is just "Commit Parent1 Parent2 ...\n" */
 207                 struct commit_graft *graft = read_graft_line(&buf);
 208                 if (!graft)
 209                         continue;
 210                 if (register_commit_graft(r, graft, 1))
 211                         error("duplicate graft data: %s", buf.buf);
 212         }
 213         fclose(fp);
 214         strbuf_release(&buf);
 215         return 0;
 216 }
 217
 218 void prepare_commit_graft(struct repository *r)
 219 {
 220         char *graft_file;
 221
 222         if (r->parsed_objects->commit_graft_prepared)
 223                 return;
 224         if (!startup_info->have_repository)
 225                 return;
 226
 227         graft_file = get_graft_file(r);
 228         read_graft_file(r, graft_file);
 229         /* make sure shallows are read */
 230         is_repository_shallow(r);
 231         r->parsed_objects->commit_graft_prepared = 1;
 232 }
 233
 234 struct commit_graft *lookup_commit_graft(struct repository *r, const struct object_id *oid)
 235 {
 236         int pos;
 237         prepare_commit_graft(r);
 238         pos = commit_graft_pos(r, oid);
 239         if (pos < 0)
 240                 return NULL;
 241         return r->parsed_objects->grafts[pos];
 242 }
 243
 244 int for_each_commit_graft(each_commit_graft_fn fn, void *cb_data)
 245 {
 246         int i, ret;
 247         for (i = ret = 0; i < the_repository->parsed_objects->grafts_nr && !ret; i++)
 248                 ret = fn(the_repository->parsed_objects->grafts[i], cb_data);
 249         return ret;
 250 }
 251
 252 void reset_commit_grafts(struct repository *r)
 253 {
 254         int i;
 255
 256         for (i = 0; i < r->parsed_objects->grafts_nr; i++)
 257                 free(r->parsed_objects->grafts[i]);
 258         r->parsed_objects->grafts_nr = 0;
 259         r->parsed_objects->commit_graft_prepared = 0;
 260 }
 261
 262 struct commit_buffer {
 263         void *buffer;
 264         unsigned long size;
 265 };
 266 define_commit_slab(buffer_slab, struct commit_buffer);
 267
 268 struct buffer_slab *allocate_commit_buffer_slab(void)
 269 {
 270         struct buffer_slab *bs = xmalloc(sizeof(*bs));
 271         init_buffer_slab(bs);
 272         return bs;
 273 }
 274
 275 void free_commit_buffer_slab(struct buffer_slab *bs)
 276 {
 277         clear_buffer_slab(bs);
 278         free(bs);
 279 }
 280
 281 void set_commit_buffer(struct repository *r, struct commit *commit, void *buffer, unsigned long size)
 282 {
 283         struct commit_buffer *v = buffer_slab_at(
 284                 r->parsed_objects->buffer_slab, commit);
 285         v->buffer = buffer;
 286         v->size = size;
 287 }
 288
 289 const void *get_cached_commit_buffer(struct repository *r, const struct commit *commit, unsigned long *sizep)
 290 {
 291         struct commit_buffer *v = buffer_slab_peek(
 292                 r->parsed_objects->buffer_slab, commit);
 293         if (!v) {
 294                 if (sizep)
 295                         *sizep = 0;
 296                 return NULL;
 297         }
 298         if (sizep)
 299                 *sizep = v->size;
 300         return v->buffer;
 301 }
 302
 303 const void *repo_get_commit_buffer(struct repository *r,
 304                                    const struct commit *commit,
 305                                    unsigned long *sizep)
 306 {
 307         const void *ret = get_cached_commit_buffer(r, commit, sizep);
 308         if (!ret) {
 309                 enum object_type type;
 310                 unsigned long size;
 311                 ret = repo_read_object_file(r, &commit->object.oid, &type, &size);
 312                 if (!ret)
 313                         die("cannot read commit object %s",
 314                             oid_to_hex(&commit->object.oid));
 315                 if (type != OBJ_COMMIT)
 316                         die("expected commit for %s, got %s",
 317                             oid_to_hex(&commit->object.oid), type_name(type));
 318                 if (sizep)
 319                         *sizep = size;
 320         }
 321         return ret;
 322 }
 323
 324 void repo_unuse_commit_buffer(struct repository *r,
 325                               const struct commit *commit,
 326                               const void *buffer)
 327 {
 328         struct commit_buffer *v = buffer_slab_peek(
 329                 r->parsed_objects->buffer_slab, commit);
 330         if (!(v && v->buffer == buffer))
 331                 free((void *)buffer);
 332 }
 333
 334 void free_commit_buffer(struct parsed_object_pool *pool, struct commit *commit)
 335 {
 336         struct commit_buffer *v = buffer_slab_peek(
 337                 pool->buffer_slab, commit);
 338         if (v) {
 339                 FREE_AND_NULL(v->buffer);
 340                 v->size = 0;
 341         }
 342 }
 343
 344 static inline void set_commit_tree(struct commit *c, struct tree *t)
 345 {
 346         c->maybe_tree = t;
 347 }
 348
 349 struct tree *repo_get_commit_tree(struct repository *r,
 350                                   const struct commit *commit)
 351 {
 352         if (commit->maybe_tree || !commit->object.parsed)
 353                 return commit->maybe_tree;
 354
 355         if (commit_graph_position(commit) != COMMIT_NOT_FROM_GRAPH)
 356                 return get_commit_tree_in_graph(r, commit);
 357
 358         return NULL;
 359 }
 360
 361 struct object_id *get_commit_tree_oid(const struct commit *commit)
 362 {
 363         struct tree *tree = get_commit_tree(commit);
 364         return tree ? &tree->object.oid : NULL;
 365 }
 366
 367 void release_commit_memory(struct parsed_object_pool *pool, struct commit *c)
 368 {
 369         set_commit_tree(c, NULL);
 370         free_commit_buffer(pool, c);
 371         c->index = 0;
 372         free_commit_list(c->parents);
 373
 374         c->object.parsed = 0;
 375 }
 376
 377 const void *detach_commit_buffer(struct commit *commit, unsigned long *sizep)
 378 {
 379         struct commit_buffer *v = buffer_slab_peek(
 380                 the_repository->parsed_objects->buffer_slab, commit);
 381         void *ret;
 382
 383         if (!v) {
 384                 if (sizep)
 385                         *sizep = 0;
 386                 return NULL;
 387         }
 388         ret = v->buffer;
 389         if (sizep)
 390                 *sizep = v->size;
 391
 392         v->buffer = NULL;
 393         v->size = 0;
 394         return ret;
 395 }
 396
 397 int parse_commit_buffer(struct repository *r, struct commit *item, const void *buffer, unsigned long size, int check_graph)
 398 {
 399         const char *tail = buffer;
 400         const char *bufptr = buffer;
 401         struct object_id parent;
 402         struct commit_list **pptr;
 403         struct commit_graft *graft;
 404         const int tree_entry_len = the_hash_algo->hexsz + 5;
 405         const int parent_entry_len = the_hash_algo->hexsz + 7;
 406         struct tree *tree;
 407
 408         if (item->object.parsed)
 409                 return 0;
 410
 411         if (item->parents) {
 412                 /*
 413                  * Presumably this is leftover from an earlier failed parse;
 414                  * clear it out in preparation for us re-parsing (we'll hit the
 415                  * same error, but that's good, since it lets our caller know
 416                  * the result cannot be trusted.
 417                  */
 418                 free_commit_list(item->parents);
 419                 item->parents = NULL;
 420         }
 421
 422         tail += size;
 423         if (tail <= bufptr + tree_entry_len + 1 || memcmp(bufptr, "tree ", 5) ||
 424                         bufptr[tree_entry_len] != '\n')
 425                 return error("bogus commit object %s", oid_to_hex(&item->object.oid));
 426         if (get_oid_hex(bufptr + 5, &parent) < 0)
 427                 return error("bad tree pointer in commit %s",
 428                              oid_to_hex(&item->object.oid));
 429         tree = lookup_tree(r, &parent);
 430         if (!tree)
 431                 return error("bad tree pointer %s in commit %s",
 432                              oid_to_hex(&parent),
 433                              oid_to_hex(&item->object.oid));
 434         set_commit_tree(item, tree);
 435         bufptr += tree_entry_len + 1; /* "tree " + "hex sha1" + "\n" */
 436         pptr = &item->parents;
 437
 438         graft = lookup_commit_graft(r, &item->object.oid);
 439         if (graft)
 440                 r->parsed_objects->substituted_parent = 1;
 441         while (bufptr + parent_entry_len < tail && !memcmp(bufptr, "parent ", 7)) {
 442                 struct commit *new_parent;
 443
 444                 if (tail <= bufptr + parent_entry_len + 1 ||
 445                     get_oid_hex(bufptr + 7, &parent) ||
 446                     bufptr[parent_entry_len] != '\n')
 447                         return error("bad parents in commit %s", oid_to_hex(&item->object.oid));
 448                 bufptr += parent_entry_len + 1;
 449                 /*
 450                  * The clone is shallow if nr_parent < 0, and we must
 451                  * not traverse its real parents even when we unhide them.
 452                  */
 453                 if (graft && (graft->nr_parent < 0 || grafts_replace_parents))
 454                         continue;
 455                 new_parent = lookup_commit(r, &parent);
 456                 if (!new_parent)
 457                         return error("bad parent %s in commit %s",
 458                                      oid_to_hex(&parent),
 459                                      oid_to_hex(&item->object.oid));
 460                 pptr = &commit_list_insert(new_parent, pptr)->next;
 461         }
 462         if (graft) {
 463                 int i;
 464                 struct commit *new_parent;
 465                 for (i = 0; i < graft->nr_parent; i++) {
 466                         new_parent = lookup_commit(r,
 467                                                    &graft->parent[i]);
 468                         if (!new_parent)
 469                                 return error("bad graft parent %s in commit %s",
 470                                              oid_to_hex(&graft->parent[i]),
 471                                              oid_to_hex(&item->object.oid));
 472                         pptr = &commit_list_insert(new_parent, pptr)->next;
 473                 }
 474         }
 475         item->date = parse_commit_date(bufptr, tail);
 476
 477         if (check_graph)
 478                 load_commit_graph_info(r, item);
 479
 480         item->object.parsed = 1;
 481         return 0;
 482 }
 483
 484 int repo_parse_commit_internal(struct repository *r,
 485                                struct commit *item,
 486                                int quiet_on_missing,
 487                                int use_commit_graph)
 488 {
 489         enum object_type type;
 490         void *buffer;
 491         unsigned long size;
 492         int ret;
 493
 494         if (!item)
 495                 return -1;
 496         if (item->object.parsed)
 497                 return 0;
 498         if (use_commit_graph && parse_commit_in_graph(r, item))
 499                 return 0;
 500         buffer = repo_read_object_file(r, &item->object.oid, &type, &size);
 501         if (!buffer)
 502                 return quiet_on_missing ? -1 :
 503                         error("Could not read %s",
 504                              oid_to_hex(&item->object.oid));
 505         if (type != OBJ_COMMIT) {
 506                 free(buffer);
 507                 return error("Object %s not a commit",
 508                              oid_to_hex(&item->object.oid));
 509         }
 510
 511         ret = parse_commit_buffer(r, item, buffer, size, 0);
 512         if (save_commit_buffer && !ret) {
 513                 set_commit_buffer(r, item, buffer, size);
 514                 return 0;
 515         }
 516         free(buffer);
 517         return ret;
 518 }
 519
 520 int repo_parse_commit_gently(struct repository *r,
 521                              struct commit *item, int quiet_on_missing)
 522 {
 523         return repo_parse_commit_internal(r, item, quiet_on_missing, 1);
 524 }
 525
 526 void parse_commit_or_die(struct commit *item)
 527 {
 528         if (parse_commit(item))
 529                 die("unable to parse commit %s",
 530                     item ? oid_to_hex(&item->object.oid) : "(null)");
 531 }
 532
 533 int find_commit_subject(const char *commit_buffer, const char **subject)
 534 {
 535         const char *eol;
 536         const char *p = commit_buffer;
 537
 538         while (*p && (*p != '\n' || p[1] != '\n'))
 539                 p++;
 540         if (*p) {
 541                 p = skip_blank_lines(p + 2);
 542                 eol = strchrnul(p, '\n');
 543         } else
 544                 eol = p;
 545
 546         *subject = p;
 547
 548         return eol - p;
 549 }
 550
 551 size_t commit_subject_length(const char *body)
 552 {
 553         const char *p = body;
 554         while (*p) {
 555                 const char *next = skip_blank_lines(p);
 556                 if (next != p)
 557                         break;
 558                 p = strchrnul(p, '\n');
 559                 if (*p)
 560                         p++;
 561         }
 562         return p - body;
 563 }
 564
 565 struct commit_list *commit_list_insert(struct commit *item, struct commit_list **list_p)
 566 {
 567         struct commit_list *new_list = xmalloc(sizeof(struct commit_list));
 568         new_list->item = item;
 569         new_list->next = *list_p;
 570         *list_p = new_list;
 571         return new_list;
 572 }
 573
 574 int commit_list_contains(struct commit *item, struct commit_list *list)
 575 {
 576         while (list) {
 577                 if (list->item == item)
 578                         return 1;
 579                 list = list->next;
 580         }
 581
 582         return 0;
 583 }
 584
 585 unsigned commit_list_count(const struct commit_list *l)
 586 {
 587         unsigned c = 0;
 588         for (; l; l = l->next )
 589                 c++;
 590         return c;
 591 }
 592
 593 struct commit_list *copy_commit_list(struct commit_list *list)
 594 {
 595         struct commit_list *head = NULL;
 596         struct commit_list **pp = &head;
 597         while (list) {
 598                 pp = commit_list_append(list->item, pp);
 599                 list = list->next;
 600         }
 601         return head;
 602 }
 603
 604 struct commit_list *reverse_commit_list(struct commit_list *list)
 605 {
 606         struct commit_list *next = NULL, *current, *backup;
 607         for (current = list; current; current = backup) {
 608                 backup = current->next;
 609                 current->next = next;
 610                 next = current;
 611         }
 612         return next;
 613 }
 614
 615 void free_commit_list(struct commit_list *list)
 616 {
 617         while (list)
 618                 pop_commit(&list);
 619 }
 620
 621 struct commit_list * commit_list_insert_by_date(struct commit *item, struct commit_list **list)
 622 {
 623         struct commit_list **pp = list;
 624         struct commit_list *p;
 625         while ((p = *pp) != NULL) {
 626                 if (p->item->date < item->date) {
 627                         break;
 628                 }
 629                 pp = &p->next;
 630         }
 631         return commit_list_insert(item, pp);
 632 }
 633
 634 static int commit_list_compare_by_date(const void *a, const void *b)
 635 {
 636         timestamp_t a_date = ((const struct commit_list *)a)->item->date;
 637         timestamp_t b_date = ((const struct commit_list *)b)->item->date;
 638         if (a_date < b_date)
 639                 return 1;
 640         if (a_date > b_date)
 641                 return -1;
 642         return 0;
 643 }
 644
 645 static void *commit_list_get_next(const void *a)
 646 {
 647         return ((const struct commit_list *)a)->next;
 648 }
 649
 650 static void commit_list_set_next(void *a, void *next)
 651 {
 652         ((struct commit_list *)a)->next = next;
 653 }
 654
 655 void commit_list_sort_by_date(struct commit_list **list)
 656 {
 657         *list = llist_mergesort(*list, commit_list_get_next, commit_list_set_next,
 658                                 commit_list_compare_by_date);
 659 }
 660
 661 struct commit *pop_most_recent_commit(struct commit_list **list,
 662                                       unsigned int mark)
 663 {
 664         struct commit *ret = pop_commit(list);
 665         struct commit_list *parents = ret->parents;
 666
 667         while (parents) {
 668                 struct commit *commit = parents->item;
 669                 if (!parse_commit(commit) && !(commit->object.flags & mark)) {
 670                         commit->object.flags |= mark;
 671                         commit_list_insert_by_date(commit, list);
 672                 }
 673                 parents = parents->next;
 674         }
 675         return ret;
 676 }
 677
 678 static void clear_commit_marks_1(struct commit_list **plist,
 679                                  struct commit *commit, unsigned int mark)
 680 {
 681         while (commit) {
 682                 struct commit_list *parents;
 683
 684                 if (!(mark & commit->object.flags))
 685                         return;
 686
 687                 commit->object.flags &= ~mark;
 688
 689                 parents = commit->parents;
 690                 if (!parents)
 691                         return;
 692
 693                 while ((parents = parents->next))
 694                         commit_list_insert(parents->item, plist);
 695
 696                 commit = commit->parents->item;
 697         }
 698 }
 699
 700 void clear_commit_marks_many(int nr, struct commit **commit, unsigned int mark)
 701 {
 702         struct commit_list *list = NULL;
 703
 704         while (nr--) {
 705                 clear_commit_marks_1(&list, *commit, mark);
 706                 commit++;
 707         }
 708         while (list)
 709                 clear_commit_marks_1(&list, pop_commit(&list), mark);
 710 }
 711
 712 void clear_commit_marks(struct commit *commit, unsigned int mark)
 713 {
 714         clear_commit_marks_many(1, &commit, mark);
 715 }
 716
 717 struct commit *pop_commit(struct commit_list **stack)
 718 {
 719         struct commit_list *top = *stack;
 720         struct commit *item = top ? top->item : NULL;
 721
 722         if (top) {
 723                 *stack = top->next;
 724                 free(top);
 725         }
 726         return item;
 727 }
 728
 729 /*
 730  * Topological sort support
 731  */
 732
 733 /* count number of children that have not been emitted */
 734 define_commit_slab(indegree_slab, int);
 735
 736 define_commit_slab(author_date_slab, timestamp_t);
 737
 738 void record_author_date(struct author_date_slab *author_date,
 739                         struct commit *commit)
 740 {
 741         const char *buffer = get_commit_buffer(commit, NULL);
 742         struct ident_split ident;
 743         const char *ident_line;
 744         size_t ident_len;
 745         char *date_end;
 746         timestamp_t date;
 747
 748         ident_line = find_commit_header(buffer, "author", &ident_len);
 749         if (!ident_line)
 750                 goto fail_exit; /* no author line */
 751         if (split_ident_line(&ident, ident_line, ident_len) ||
 752             !ident.date_begin || !ident.date_end)
 753                 goto fail_exit; /* malformed "author" line */
 754
 755         date = parse_timestamp(ident.date_begin, &date_end, 10);
 756         if (date_end != ident.date_end)
 757                 goto fail_exit; /* malformed date */
 758         *(author_date_slab_at(author_date, commit)) = date;
 759
 760 fail_exit:
 761         unuse_commit_buffer(commit, buffer);
 762 }
 763
 764 int compare_commits_by_author_date(const void *a_, const void *b_,
 765                                    void *cb_data)
 766 {
 767         const struct commit *a = a_, *b = b_;
 768         struct author_date_slab *author_date = cb_data;
 769         timestamp_t a_date = *(author_date_slab_at(author_date, a));
 770         timestamp_t b_date = *(author_date_slab_at(author_date, b));
 771
 772         /* newer commits with larger date first */
 773         if (a_date < b_date)
 774                 return 1;
 775         else if (a_date > b_date)
 776                 return -1;
 777         return 0;
 778 }
 779
 780 int compare_commits_by_gen_then_commit_date(const void *a_, const void *b_, void *unused)
 781 {
 782         const struct commit *a = a_, *b = b_;
 783         const timestamp_t generation_a = commit_graph_generation(a),
 784                           generation_b = commit_graph_generation(b);
 785
 786         /* newer commits first */
 787         if (generation_a < generation_b)
 788                 return 1;
 789         else if (generation_a > generation_b)
 790                 return -1;
 791
 792         /* use date as a heuristic when generations are equal */
 793         if (a->date < b->date)
 794                 return 1;
 795         else if (a->date > b->date)
 796                 return -1;
 797         return 0;
 798 }
 799
 800 int compare_commits_by_commit_date(const void *a_, const void *b_, void *unused)
 801 {
 802         const struct commit *a = a_, *b = b_;
 803         /* newer commits with larger date first */
 804         if (a->date < b->date)
 805                 return 1;
 806         else if (a->date > b->date)
 807                 return -1;
 808         return 0;
 809 }
 810
 811 /*
 812  * Performs an in-place topological sort on the list supplied.
 813  */
 814 void sort_in_topological_order(struct commit_list **list, enum rev_sort_order sort_order)
 815 {
 816         struct commit_list *next, *orig = *list;
 817         struct commit_list **pptr;
 818         struct indegree_slab indegree;
 819         struct prio_queue queue;
 820         struct commit *commit;
 821         struct author_date_slab author_date;
 822
 823         if (!orig)
 824                 return;
 825         *list = NULL;
 826
 827         init_indegree_slab(&indegree);
 828         memset(&queue, '\0', sizeof(queue));
 829
 830         switch (sort_order) {
 831         default: /* REV_SORT_IN_GRAPH_ORDER */
 832                 queue.compare = NULL;
 833                 break;
 834         case REV_SORT_BY_COMMIT_DATE:
 835                 queue.compare = compare_commits_by_commit_date;
 836                 break;
 837         case REV_SORT_BY_AUTHOR_DATE:
 838                 init_author_date_slab(&author_date);
 839                 queue.compare = compare_commits_by_author_date;
 840                 queue.cb_data = &author_date;
 841                 break;
 842         }
 843
 844         /* Mark them and clear the indegree */
 845         for (next = orig; next; next = next->next) {
 846                 struct commit *commit = next->item;
 847                 *(indegree_slab_at(&indegree, commit)) = 1;
 848                 /* also record the author dates, if needed */
 849                 if (sort_order == REV_SORT_BY_AUTHOR_DATE)
 850                         record_author_date(&author_date, commit);
 851         }
 852
 853         /* update the indegree */
 854         for (next = orig; next; next = next->next) {
 855                 struct commit_list *parents = next->item->parents;
 856                 while (parents) {
 857                         struct commit *parent = parents->item;
 858                         int *pi = indegree_slab_at(&indegree, parent);
 859
 860                         if (*pi)
 861                                 (*pi)++;
 862                         parents = parents->next;
 863                 }
 864         }
 865
 866         /*
 867          * find the tips
 868          *
 869          * tips are nodes not reachable from any other node in the list
 870          *
 871          * the tips serve as a starting set for the work queue.
 872          */
 873         for (next = orig; next; next = next->next) {
 874                 struct commit *commit = next->item;
 875
 876                 if (*(indegree_slab_at(&indegree, commit)) == 1)
 877                         prio_queue_put(&queue, commit);
 878         }
 879
 880         /*
 881          * This is unfortunate; the initial tips need to be shown
 882          * in the order given from the revision traversal machinery.
 883          */
 884         if (sort_order == REV_SORT_IN_GRAPH_ORDER)
 885                 prio_queue_reverse(&queue);
 886
 887         /* We no longer need the commit list */
 888         free_commit_list(orig);
 889
 890         pptr = list;
 891         *list = NULL;
 892         while ((commit = prio_queue_get(&queue)) != NULL) {
 893                 struct commit_list *parents;
 894
 895                 for (parents = commit->parents; parents ; parents = parents->next) {
 896                         struct commit *parent = parents->item;
 897                         int *pi = indegree_slab_at(&indegree, parent);
 898
 899                         if (!*pi)
 900                                 continue;
 901
 902                         /*
 903                          * parents are only enqueued for emission
 904                          * when all their children have been emitted thereby
 905                          * guaranteeing topological order.
 906                          */
 907                         if (--(*pi) == 1)
 908                                 prio_queue_put(&queue, parent);
 909                 }
 910                 /*
 911                  * all children of commit have already been
 912                  * emitted. we can emit it now.
 913                  */
 914                 *(indegree_slab_at(&indegree, commit)) = 0;
 915
 916                 pptr = &commit_list_insert(commit, pptr)->next;
 917         }
 918
 919         clear_indegree_slab(&indegree);
 920         clear_prio_queue(&queue);
 921         if (sort_order == REV_SORT_BY_AUTHOR_DATE)
 922                 clear_author_date_slab(&author_date);
 923 }
 924
 925 struct rev_collect {
 926         struct commit **commit;
 927         int nr;
 928         int alloc;
 929         unsigned int initial : 1;
 930 };
 931
 932 static void add_one_commit(struct object_id *oid, struct rev_collect *revs)
 933 {
 934         struct commit *commit;
 935
 936         if (is_null_oid(oid))
 937                 return;
 938
 939         commit = lookup_commit(the_repository, oid);
 940         if (!commit ||
 941             (commit->object.flags & TMP_MARK) ||
 942             parse_commit(commit))
 943                 return;
 944
 945         ALLOC_GROW(revs->commit, revs->nr + 1, revs->alloc);
 946         revs->commit[revs->nr++] = commit;
 947         commit->object.flags |= TMP_MARK;
 948 }
 949
 950 static int collect_one_reflog_ent(struct object_id *ooid, struct object_id *noid,
 951                                   const char *ident, timestamp_t timestamp,
 952                                   int tz, const char *message, void *cbdata)
 953 {
 954         struct rev_collect *revs = cbdata;
 955
 956         if (revs->initial) {
 957                 revs->initial = 0;
 958                 add_one_commit(ooid, revs);
 959         }
 960         add_one_commit(noid, revs);
 961         return 0;
 962 }
 963
 964 struct commit *get_fork_point(const char *refname, struct commit *commit)
 965 {
 966         struct object_id oid;
 967         struct rev_collect revs;
 968         struct commit_list *bases;
 969         int i;
 970         struct commit *ret = NULL;
 971         char *full_refname;
 972
 973         switch (dwim_ref(refname, strlen(refname), &oid, &full_refname, 0)) {
 974         case 0:
 975                 die("No such ref: '%s'", refname);
 976         case 1:
 977                 break; /* good */
 978         default:
 979                 die("Ambiguous refname: '%s'", refname);
 980         }
 981
 982         memset(&revs, 0, sizeof(revs));
 983         revs.initial = 1;
 984         for_each_reflog_ent(full_refname, collect_one_reflog_ent, &revs);
 985
 986         if (!revs.nr)
 987                 add_one_commit(&oid, &revs);
 988
 989         for (i = 0; i < revs.nr; i++)
 990                 revs.commit[i]->object.flags &= ~TMP_MARK;
 991
 992         bases = get_merge_bases_many(commit, revs.nr, revs.commit);
 993
 994         /*
 995          * There should be one and only one merge base, when we found
 996          * a common ancestor among reflog entries.
 997          */
 998         if (!bases || bases->next)
 999                 goto cleanup_return;
1000
1001         /* And the found one must be one of the reflog entries */
1002         for (i = 0; i < revs.nr; i++)
1003                 if (&bases->item->object == &revs.commit[i]->object)
1004                         break; /* found */
1005         if (revs.nr <= i)
1006                 goto cleanup_return;
1007
1008         ret = bases->item;
1009
1010 cleanup_return:
1011         free_commit_list(bases);
1012         free(full_refname);
1013         return ret;
1014 }
1015
1016 /*
1017  * Indexed by hash algorithm identifier.
1018  */
1019 static const char *gpg_sig_headers[] = {
1020         NULL,
1021         "gpgsig",
1022         "gpgsig-sha256",
1023 };
1024
1025 int sign_with_header(struct strbuf *buf, const char *keyid)
1026 {
1027         struct strbuf sig = STRBUF_INIT;
1028         int inspos, copypos;
1029         const char *eoh;
1030         const char *gpg_sig_header = gpg_sig_headers[hash_algo_by_ptr(the_hash_algo)];
1031         int gpg_sig_header_len = strlen(gpg_sig_header);
1032
1033         /* find the end of the header */
1034         eoh = strstr(buf->buf, "\n\n");
1035         if (!eoh)
1036                 inspos = buf->len;
1037         else
1038                 inspos = eoh - buf->buf + 1;
1039
1040         if (!keyid || !*keyid)
1041                 keyid = get_signing_key();
1042         if (sign_buffer(buf, &sig, keyid)) {
1043                 strbuf_release(&sig);
1044                 return -1;
1045         }
1046
1047         for (copypos = 0; sig.buf[copypos]; ) {
1048                 const char *bol = sig.buf + copypos;
1049                 const char *eol = strchrnul(bol, '\n');
1050                 int len = (eol - bol) + !!*eol;
1051
1052                 if (!copypos) {
1053                         strbuf_insert(buf, inspos, gpg_sig_header, gpg_sig_header_len);
1054                         inspos += gpg_sig_header_len;
1055                 }
1056                 strbuf_insertstr(buf, inspos++, " ");
1057                 strbuf_insert(buf, inspos, bol, len);
1058                 inspos += len;
1059                 copypos += len;
1060         }
1061         strbuf_release(&sig);
1062         return 0;
1063 }
1064
1065
1066
1067 int parse_signed_commit(const struct commit *commit,
1068                         struct strbuf *payload, struct strbuf *signature,
1069                         const struct git_hash_algo *algop)
1070 {
1071         unsigned long size;
1072         const char *buffer = get_commit_buffer(commit, &size);
1073         int ret = parse_buffer_signed_by_header(buffer, size, payload, signature, algop);
1074
1075         unuse_commit_buffer(commit, buffer);
1076         return ret;
1077 }
1078
1079 int parse_buffer_signed_by_header(const char *buffer,
1080                                   unsigned long size,
1081                                   struct strbuf *payload,
1082                                   struct strbuf *signature,
1083                                   const struct git_hash_algo *algop)
1084 {
1085         int in_signature = 0, saw_signature = 0, other_signature = 0;
1086         const char *line, *tail, *p;
1087         const char *gpg_sig_header = gpg_sig_headers[hash_algo_by_ptr(algop)];
1088
1089         line = buffer;
1090         tail = buffer + size;
1091         while (line < tail) {
1092                 const char *sig = NULL;
1093                 const char *next = memchr(line, '\n', tail - line);
1094
1095                 next = next ? next + 1 : tail;
1096                 if (in_signature && line[0] == ' ')
1097                         sig = line + 1;
1098                 else if (skip_prefix(line, gpg_sig_header, &p) &&
1099                          *p == ' ') {
1100                         sig = line + strlen(gpg_sig_header) + 1;
1101                         other_signature = 0;
1102                 }
1103                 else if (starts_with(line, "gpgsig"))
1104                         other_signature = 1;
1105                 else if (other_signature && line[0] != ' ')
1106                         other_signature = 0;
1107                 if (sig) {
1108                         strbuf_add(signature, sig, next - sig);
1109                         saw_signature = 1;
1110                         in_signature = 1;
1111                 } else {
1112                         if (*line == '\n')
1113                                 /* dump the whole remainder of the buffer */
1114                                 next = tail;
1115                         if (!other_signature)
1116                                 strbuf_add(payload, line, next - line);
1117                         in_signature = 0;
1118                 }
1119                 line = next;
1120         }
1121         return saw_signature;
1122 }
1123
1124 int remove_signature(struct strbuf *buf)
1125 {
1126         const char *line = buf->buf;
1127         const char *tail = buf->buf + buf->len;
1128         int in_signature = 0;
1129         struct sigbuf {
1130                 const char *start;
1131                 const char *end;
1132         } sigs[2], *sigp = &sigs[0];
1133         int i;
1134         const char *orig_buf = buf->buf;
1135
1136         memset(sigs, 0, sizeof(sigs));
1137
1138         while (line < tail) {
1139                 const char *next = memchr(line, '\n', tail - line);
1140                 next = next ? next + 1 : tail;
1141
1142                 if (in_signature && line[0] == ' ')
1143                         sigp->end = next;
1144                 else if (starts_with(line, "gpgsig")) {
1145                         int i;
1146                         for (i = 1; i < GIT_HASH_NALGOS; i++) {
1147                                 const char *p;
1148                                 if (skip_prefix(line, gpg_sig_headers[i], &p) &&
1149                                     *p == ' ') {
1150                                         sigp->start = line;
1151                                         sigp->end = next;
1152                                         in_signature = 1;
1153                                 }
1154                         }
1155                 } else {
1156                         if (*line == '\n')
1157                                 /* dump the whole remainder of the buffer */
1158                                 next = tail;
1159                         if (in_signature && sigp - sigs != ARRAY_SIZE(sigs))
1160                                 sigp++;
1161                         in_signature = 0;
1162                 }
1163                 line = next;
1164         }
1165
1166         for (i = ARRAY_SIZE(sigs) - 1; i >= 0; i--)
1167                 if (sigs[i].start)
1168                         strbuf_remove(buf, sigs[i].start - orig_buf, sigs[i].end - sigs[i].start);
1169
1170         return sigs[0].start != NULL;
1171 }
1172
1173 static void handle_signed_tag(struct commit *parent, struct commit_extra_header ***tail)
1174 {
1175         struct merge_remote_desc *desc;
1176         struct commit_extra_header *mergetag;
1177         char *buf;
1178         unsigned long size;
1179         enum object_type type;
1180         struct strbuf payload = STRBUF_INIT;
1181         struct strbuf signature = STRBUF_INIT;
1182
1183         desc = merge_remote_util(parent);
1184         if (!desc || !desc->obj)
1185                 return;
1186         buf = read_object_file(&desc->obj->oid, &type, &size);
1187         if (!buf || type != OBJ_TAG)
1188                 goto free_return;
1189         if (!parse_signature(buf, size, &payload, &signature))
1190                 goto free_return;
1191         /*
1192          * We could verify this signature and either omit the tag when
1193          * it does not validate, but the integrator may not have the
1194          * public key of the signer of the tag being merged, while a
1195          * later auditor may have it while auditing, so let's not run
1196          * verify-signed-buffer here for now...
1197          *
1198          * if (verify_signed_buffer(buf, len, buf + len, size - len, ...))
1199          *      warn("warning: signed tag unverified.");
1200          */
1201         CALLOC_ARRAY(mergetag, 1);
1202         mergetag->key = xstrdup("mergetag");
1203         mergetag->value = buf;
1204         mergetag->len = size;
1205
1206         **tail = mergetag;
1207         *tail = &mergetag->next;
1208         strbuf_release(&payload);
1209         strbuf_release(&signature);
1210         return;
1211
1212 free_return:
1213         free(buf);
1214 }
1215
1216 int check_commit_signature(const struct commit *commit, struct signature_check *sigc)
1217 {
1218         struct strbuf payload = STRBUF_INIT;
1219         struct strbuf signature = STRBUF_INIT;
1220         int ret = 1;
1221
1222         sigc->result = 'N';
1223
1224         if (parse_signed_commit(commit, &payload, &signature, the_hash_algo) <= 0)
1225                 goto out;
1226
1227         sigc->payload_type = SIGNATURE_PAYLOAD_COMMIT;
1228         sigc->payload = strbuf_detach(&payload, &sigc->payload_len);
1229         ret = check_signature(sigc, signature.buf, signature.len);
1230
1231  out:
1232         strbuf_release(&payload);
1233         strbuf_release(&signature);
1234
1235         return ret;
1236 }
1237
1238 void verify_merge_signature(struct commit *commit, int verbosity,
1239                             int check_trust)
1240 {
1241         char hex[GIT_MAX_HEXSZ + 1];
1242         struct signature_check signature_check;
1243         int ret;
1244         memset(&signature_check, 0, sizeof(signature_check));
1245
1246         ret = check_commit_signature(commit, &signature_check);
1247
1248         find_unique_abbrev_r(hex, &commit->object.oid, DEFAULT_ABBREV);
1249         switch (signature_check.result) {
1250         case 'G':
1251                 if (ret || (check_trust && signature_check.trust_level < TRUST_MARGINAL))
1252                         die(_("Commit %s has an untrusted GPG signature, "
1253                               "allegedly by %s."), hex, signature_check.signer);
1254                 break;
1255         case 'B':
1256                 die(_("Commit %s has a bad GPG signature "
1257                       "allegedly by %s."), hex, signature_check.signer);
1258         default: /* 'N' */
1259                 die(_("Commit %s does not have a GPG signature."), hex);
1260         }
1261         if (verbosity >= 0 && signature_check.result == 'G')
1262                 printf(_("Commit %s has a good GPG signature by %s\n"),
1263                        hex, signature_check.signer);
1264
1265         signature_check_clear(&signature_check);
1266 }
1267
1268 void append_merge_tag_headers(struct commit_list *parents,
1269                               struct commit_extra_header ***tail)
1270 {
1271         while (parents) {
1272                 struct commit *parent = parents->item;
1273                 handle_signed_tag(parent, tail);
1274                 parents = parents->next;
1275         }
1276 }
1277
1278 static void add_extra_header(struct strbuf *buffer,
1279                              struct commit_extra_header *extra)
1280 {
1281         strbuf_addstr(buffer, extra->key);
1282         if (extra->len)
1283                 strbuf_add_lines(buffer, " ", extra->value, extra->len);
1284         else
1285                 strbuf_addch(buffer, '\n');
1286 }
1287
1288 struct commit_extra_header *read_commit_extra_headers(struct commit *commit,
1289                                                       const char **exclude)
1290 {
1291         struct commit_extra_header *extra = NULL;
1292         unsigned long size;
1293         const char *buffer = get_commit_buffer(commit, &size);
1294         extra = read_commit_extra_header_lines(buffer, size, exclude);
1295         unuse_commit_buffer(commit, buffer);
1296         return extra;
1297 }
1298
1299 int for_each_mergetag(each_mergetag_fn fn, struct commit *commit, void *data)
1300 {
1301         struct commit_extra_header *extra, *to_free;
1302         int res = 0;
1303
1304         to_free = read_commit_extra_headers(commit, NULL);
1305         for (extra = to_free; !res && extra; extra = extra->next) {
1306                 if (strcmp(extra->key, "mergetag"))
1307                         continue; /* not a merge tag */
1308                 res = fn(commit, extra, data);
1309         }
1310         free_commit_extra_headers(to_free);
1311         return res;
1312 }
1313
1314 static inline int standard_header_field(const char *field, size_t len)
1315 {
1316         return ((len == 4 && !memcmp(field, "tree", 4)) ||
1317                 (len == 6 && !memcmp(field, "parent", 6)) ||
1318                 (len == 6 && !memcmp(field, "author", 6)) ||
1319                 (len == 9 && !memcmp(field, "committer", 9)) ||
1320                 (len == 8 && !memcmp(field, "encoding", 8)));
1321 }
1322
1323 static int excluded_header_field(const char *field, size_t len, const char **exclude)
1324 {
1325         if (!exclude)
1326                 return 0;
1327
1328         while (*exclude) {
1329                 size_t xlen = strlen(*exclude);
1330                 if (len == xlen && !memcmp(field, *exclude, xlen))
1331                         return 1;
1332                 exclude++;
1333         }
1334         return 0;
1335 }
1336
1337 static struct commit_extra_header *read_commit_extra_header_lines(
1338         const char *buffer, size_t size,
1339         const char **exclude)
1340 {
1341         struct commit_extra_header *extra = NULL, **tail = &extra, *it = NULL;
1342         const char *line, *next, *eof, *eob;
1343         struct strbuf buf = STRBUF_INIT;
1344
1345         for (line = buffer, eob = line + size;
1346              line < eob && *line != '\n';
1347              line = next) {
1348                 next = memchr(line, '\n', eob - line);
1349                 next = next ? next + 1 : eob;
1350                 if (*line == ' ') {
1351                         /* continuation */
1352                         if (it)
1353                                 strbuf_add(&buf, line + 1, next - (line + 1));
1354                         continue;
1355                 }
1356                 if (it)
1357                         it->value = strbuf_detach(&buf, &it->len);
1358                 strbuf_reset(&buf);
1359                 it = NULL;
1360
1361                 eof = memchr(line, ' ', next - line);
1362                 if (!eof)
1363                         eof = next;
1364                 else if (standard_header_field(line, eof - line) ||
1365                          excluded_header_field(line, eof - line, exclude))
1366                         continue;
1367
1368                 CALLOC_ARRAY(it, 1);
1369                 it->key = xmemdupz(line, eof-line);
1370                 *tail = it;
1371                 tail = &it->next;
1372                 if (eof + 1 < next)
1373                         strbuf_add(&buf, eof + 1, next - (eof + 1));
1374         }
1375         if (it)
1376                 it->value = strbuf_detach(&buf, &it->len);
1377         return extra;
1378 }
1379
1380 void free_commit_extra_headers(struct commit_extra_header *extra)
1381 {
1382         while (extra) {
1383                 struct commit_extra_header *next = extra->next;
1384                 free(extra->key);
1385                 free(extra->value);
1386                 free(extra);
1387                 extra = next;
1388         }
1389 }
1390
1391 int commit_tree(const char *msg, size_t msg_len, const struct object_id *tree,
1392                 struct commit_list *parents, struct object_id *ret,
1393                 const char *author, const char *sign_commit)
1394 {
1395         struct commit_extra_header *extra = NULL, **tail = &extra;
1396         int result;
1397
1398         append_merge_tag_headers(parents, &tail);
1399         result = commit_tree_extended(msg, msg_len, tree, parents, ret, author,
1400                                       NULL, sign_commit, extra);
1401         free_commit_extra_headers(extra);
1402         return result;
1403 }
1404
1405 static int find_invalid_utf8(const char *buf, int len)
1406 {
1407         int offset = 0;
1408         static const unsigned int max_codepoint[] = {
1409                 0x7f, 0x7ff, 0xffff, 0x10ffff
1410         };
1411
1412         while (len) {
1413                 unsigned char c = *buf++;
1414                 int bytes, bad_offset;
1415                 unsigned int codepoint;
1416                 unsigned int min_val, max_val;
1417
1418                 len--;
1419                 offset++;
1420
1421                 /* Simple US-ASCII? No worries. */
1422                 if (c < 0x80)
1423                         continue;
1424
1425                 bad_offset = offset-1;
1426
1427                 /*
1428                  * Count how many more high bits set: that's how
1429                  * many more bytes this sequence should have.
1430                  */
1431                 bytes = 0;
1432                 while (c & 0x40) {
1433                         c <<= 1;
1434                         bytes++;
1435                 }
1436
1437                 /*
1438                  * Must be between 1 and 3 more bytes.  Longer sequences result in
1439                  * codepoints beyond U+10FFFF, which are guaranteed never to exist.
1440                  */
1441                 if (bytes < 1 || 3 < bytes)
1442                         return bad_offset;
1443
1444                 /* Do we *have* that many bytes? */
1445                 if (len < bytes)
1446                         return bad_offset;
1447
1448                 /*
1449                  * Place the encoded bits at the bottom of the value and compute the
1450                  * valid range.
1451                  */
1452                 codepoint = (c & 0x7f) >> bytes;
1453                 min_val = max_codepoint[bytes-1] + 1;
1454                 max_val = max_codepoint[bytes];
1455
1456                 offset += bytes;
1457                 len -= bytes;
1458
1459                 /* And verify that they are good continuation bytes */
1460                 do {
1461                         codepoint <<= 6;
1462                         codepoint |= *buf & 0x3f;
1463                         if ((*buf++ & 0xc0) != 0x80)
1464                                 return bad_offset;
1465                 } while (--bytes);
1466
1467                 /* Reject codepoints that are out of range for the sequence length. */
1468                 if (codepoint < min_val || codepoint > max_val)
1469                         return bad_offset;
1470                 /* Surrogates are only for UTF-16 and cannot be encoded in UTF-8. */
1471                 if ((codepoint & 0x1ff800) == 0xd800)
1472                         return bad_offset;
1473                 /* U+xxFFFE and U+xxFFFF are guaranteed non-characters. */
1474                 if ((codepoint & 0xfffe) == 0xfffe)
1475                         return bad_offset;
1476                 /* So are anything in the range U+FDD0..U+FDEF. */
1477                 if (codepoint >= 0xfdd0 && codepoint <= 0xfdef)
1478                         return bad_offset;
1479         }
1480         return -1;
1481 }
1482
1483 /*
1484  * This verifies that the buffer is in proper utf8 format.
1485  *
1486  * If it isn't, it assumes any non-utf8 characters are Latin1,
1487  * and does the conversion.
1488  */
1489 static int verify_utf8(struct strbuf *buf)
1490 {
1491         int ok = 1;
1492         long pos = 0;
1493
1494         for (;;) {
1495                 int bad;
1496                 unsigned char c;
1497                 unsigned char replace[2];
1498
1499                 bad = find_invalid_utf8(buf->buf + pos, buf->len - pos);
1500                 if (bad < 0)
1501                         return ok;
1502                 pos += bad;
1503                 ok = 0;
1504                 c = buf->buf[pos];
1505                 strbuf_remove(buf, pos, 1);
1506
1507                 /* We know 'c' must be in the range 128-255 */
1508                 replace[0] = 0xc0 + (c >> 6);
1509                 replace[1] = 0x80 + (c & 0x3f);
1510                 strbuf_insert(buf, pos, replace, 2);
1511                 pos += 2;
1512         }
1513 }
1514
1515 static const char commit_utf8_warn[] =
1516 N_("Warning: commit message did not conform to UTF-8.\n"
1517    "You may want to amend it after fixing the message, or set the config\n"
1518    "variable i18n.commitencoding to the encoding your project uses.\n");
1519
1520 int commit_tree_extended(const char *msg, size_t msg_len,
1521                          const struct object_id *tree,
1522                          struct commit_list *parents, struct object_id *ret,
1523                          const char *author, const char *committer,
1524                          const char *sign_commit,
1525                          struct commit_extra_header *extra)
1526 {
1527         int result;
1528         int encoding_is_utf8;
1529         struct strbuf buffer;
1530
1531         assert_oid_type(tree, OBJ_TREE);
1532
1533         if (memchr(msg, '\0', msg_len))
1534                 return error("a NUL byte in commit log message not allowed.");
1535
1536         /* Not having i18n.commitencoding is the same as having utf-8 */
1537         encoding_is_utf8 = is_encoding_utf8(git_commit_encoding);
1538
1539         strbuf_init(&buffer, 8192); /* should avoid reallocs for the headers */
1540         strbuf_addf(&buffer, "tree %s\n", oid_to_hex(tree));
1541
1542         /*
1543          * NOTE! This ordering means that the same exact tree merged with a
1544          * different order of parents will be a _different_ changeset even
1545          * if everything else stays the same.
1546          */
1547         while (parents) {
1548                 struct commit *parent = pop_commit(&parents);
1549                 strbuf_addf(&buffer, "parent %s\n",
1550                             oid_to_hex(&parent->object.oid));
1551         }
1552
1553         /* Person/date information */
1554         if (!author)
1555                 author = git_author_info(IDENT_STRICT);
1556         strbuf_addf(&buffer, "author %s\n", author);
1557         if (!committer)
1558                 committer = git_committer_info(IDENT_STRICT);
1559         strbuf_addf(&buffer, "committer %s\n", committer);
1560         if (!encoding_is_utf8)
1561                 strbuf_addf(&buffer, "encoding %s\n", git_commit_encoding);
1562
1563         while (extra) {
1564                 add_extra_header(&buffer, extra);
1565                 extra = extra->next;
1566         }
1567         strbuf_addch(&buffer, '\n');
1568
1569         /* And add the comment */
1570         strbuf_add(&buffer, msg, msg_len);
1571
1572         /* And check the encoding */
1573         if (encoding_is_utf8 && !verify_utf8(&buffer))
1574                 fprintf(stderr, _(commit_utf8_warn));
1575
1576         if (sign_commit && sign_with_header(&buffer, sign_commit)) {
1577                 result = -1;
1578                 goto out;
1579         }
1580
1581         result = write_object_file(buffer.buf, buffer.len, OBJ_COMMIT, ret);
1582 out:
1583         strbuf_release(&buffer);
1584         return result;
1585 }
1586
1587 define_commit_slab(merge_desc_slab, struct merge_remote_desc *);
1588 static struct merge_desc_slab merge_desc_slab = COMMIT_SLAB_INIT(1, merge_desc_slab);
1589
1590 struct merge_remote_desc *merge_remote_util(struct commit *commit)
1591 {
1592         return *merge_desc_slab_at(&merge_desc_slab, commit);
1593 }
1594
1595 void set_merge_remote_desc(struct commit *commit,
1596                            const char *name, struct object *obj)
1597 {
1598         struct merge_remote_desc *desc;
1599         FLEX_ALLOC_STR(desc, name, name);
1600         desc->obj = obj;
1601         *merge_desc_slab_at(&merge_desc_slab, commit) = desc;
1602 }
1603
1604 struct commit *get_merge_parent(const char *name)
1605 {
1606         struct object *obj;
1607         struct commit *commit;
1608         struct object_id oid;
1609         if (get_oid(name, &oid))
1610                 return NULL;
1611         obj = parse_object(the_repository, &oid);
1612         commit = (struct commit *)peel_to_type(name, 0, obj, OBJ_COMMIT);
1613         if (commit && !merge_remote_util(commit))
1614                 set_merge_remote_desc(commit, name, obj);
1615         return commit;
1616 }
1617
1618 /*
1619  * Append a commit to the end of the commit_list.
1620  *
1621  * next starts by pointing to the variable that holds the head of an
1622  * empty commit_list, and is updated to point to the "next" field of
1623  * the last item on the list as new commits are appended.
1624  *
1625  * Usage example:
1626  *
1627  *     struct commit_list *list;
1628  *     struct commit_list **next = &list;
1629  *
1630  *     next = commit_list_append(c1, next);
1631  *     next = commit_list_append(c2, next);
1632  *     assert(commit_list_count(list) == 2);
1633  *     return list;
1634  */
1635 struct commit_list **commit_list_append(struct commit *commit,
1636                                         struct commit_list **next)
1637 {
1638         struct commit_list *new_commit = xmalloc(sizeof(struct commit_list));
1639         new_commit->item = commit;
1640         *next = new_commit;
1641         new_commit->next = NULL;
1642         return &new_commit->next;
1643 }
1644
1645 const char *find_header_mem(const char *msg, size_t len,
1646                         const char *key, size_t *out_len)
1647 {
1648         int key_len = strlen(key);
1649         const char *line = msg;
1650
1651         /*
1652          * NEEDSWORK: It's possible for strchrnul() to scan beyond the range
1653          * given by len. However, current callers are safe because they compute
1654          * len by scanning a NUL-terminated block of memory starting at msg.
1655          * Nonetheless, it would be better to ensure the function does not look
1656          * at msg beyond the len provided by the caller.
1657          */
1658         while (line && line < msg + len) {
1659                 const char *eol = strchrnul(line, '\n');
1660
1661                 if (line == eol)
1662                         return NULL;
1663
1664                 if (eol - line > key_len &&
1665                     !strncmp(line, key, key_len) &&
1666                     line[key_len] == ' ') {
1667                         *out_len = eol - line - key_len - 1;
1668                         return line + key_len + 1;
1669                 }
1670                 line = *eol ? eol + 1 : NULL;
1671         }
1672         return NULL;
1673 }
1674
1675 const char *find_commit_header(const char *msg, const char *key, size_t *out_len)
1676 {
1677         return find_header_mem(msg, strlen(msg), key, out_len);
1678 }
1679 /*
1680  * Inspect the given string and determine the true "end" of the log message, in
1681  * order to find where to put a new Signed-off-by trailer.  Ignored are
1682  * trailing comment lines and blank lines.  To support "git commit -s
1683  * --amend" on an existing commit, we also ignore "Conflicts:".  To
1684  * support "git commit -v", we truncate at cut lines.
1685  *
1686  * Returns the number of bytes from the tail to ignore, to be fed as
1687  * the second parameter to append_signoff().
1688  */
1689 size_t ignore_non_trailer(const char *buf, size_t len)
1690 {
1691         size_t boc = 0;
1692         size_t bol = 0;
1693         int in_old_conflicts_block = 0;
1694         size_t cutoff = wt_status_locate_end(buf, len);
1695
1696         while (bol < cutoff) {
1697                 const char *next_line = memchr(buf + bol, '\n', len - bol);
1698
1699                 if (!next_line)
1700                         next_line = buf + len;
1701                 else
1702                         next_line++;
1703
1704                 if (buf[bol] == comment_line_char || buf[bol] == '\n') {
1705                         /* is this the first of the run of comments? */
1706                         if (!boc)
1707                                 boc = bol;
1708                         /* otherwise, it is just continuing */
1709                 } else if (starts_with(buf + bol, "Conflicts:\n")) {
1710                         in_old_conflicts_block = 1;
1711                         if (!boc)
1712                                 boc = bol;
1713                 } else if (in_old_conflicts_block && buf[bol] == '\t') {
1714                         ; /* a pathname in the conflicts block */
1715                 } else if (boc) {
1716                         /* the previous was not trailing comment */
1717                         boc = 0;
1718                         in_old_conflicts_block = 0;
1719                 }
1720                 bol = next_line - buf;
1721         }
1722         return boc ? len - boc : len - cutoff;
1723 }
1724
1725 int run_commit_hook(int editor_is_used, const char *index_file,
1726                     int *invoked_hook, const char *name, ...)
1727 {
1728         struct run_hooks_opt opt = RUN_HOOKS_OPT_INIT;
1729         va_list args;
1730         const char *arg;
1731
1732         strvec_pushf(&opt.env, "GIT_INDEX_FILE=%s", index_file);
1733
1734         /*
1735          * Let the hook know that no editor will be launched.
1736          */
1737         if (!editor_is_used)
1738                 strvec_push(&opt.env, "GIT_EDITOR=:");
1739
1740         va_start(args, name);
1741         while ((arg = va_arg(args, const char *)))
1742                 strvec_push(&opt.args, arg);
1743         va_end(args);
1744
1745         opt.invoked_hook = invoked_hook;
1746         return run_hooks_opt(name, &opt);
1747 }