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_object(struct repository *r,
  63                                     const struct object_id *oid)
  64 {
  65         struct object *obj = parse_object(r, oid);
  66         return obj ? object_as_type(obj, OBJ_COMMIT, 0) : NULL;
  67
  68 }
  69
  70 struct commit *lookup_commit(struct repository *r, const struct object_id *oid)
  71 {
  72         struct object *obj = lookup_object(r, oid);
  73         if (!obj)
  74                 return create_object(r, oid, alloc_commit_node(r));
  75         return object_as_type(obj, OBJ_COMMIT, 0);
  76 }
  77
  78 struct commit *lookup_commit_reference_by_name(const char *name)
  79 {
  80         struct object_id oid;
  81         struct commit *commit;
  82
  83         if (get_oid_committish(name, &oid))
  84                 return NULL;
  85         commit = lookup_commit_reference(the_repository, &oid);
  86         if (parse_commit(commit))
  87                 return NULL;
  88         return commit;
  89 }
  90
  91 static timestamp_t parse_commit_date(const char *buf, const char *tail)
  92 {
  93         const char *dateptr;
  94
  95         if (buf + 6 >= tail)
  96                 return 0;
  97         if (memcmp(buf, "author", 6))
  98                 return 0;
  99         while (buf < tail && *buf++ != '\n')
 100                 /* nada */;
 101         if (buf + 9 >= tail)
 102                 return 0;
 103         if (memcmp(buf, "committer", 9))
 104                 return 0;
 105         while (buf < tail && *buf++ != '>')
 106                 /* nada */;
 107         if (buf >= tail)
 108                 return 0;
 109         dateptr = buf;
 110         while (buf < tail && *buf++ != '\n')
 111                 /* nada */;
 112         if (buf >= tail)
 113                 return 0;
 114         /* dateptr < buf && buf[-1] == '\n', so parsing will stop at buf-1 */
 115         return parse_timestamp(dateptr, NULL, 10);
 116 }
 117
 118 static const struct object_id *commit_graft_oid_access(size_t index, const void *table)
 119 {
 120         const struct commit_graft * const *commit_graft_table = table;
 121         return &commit_graft_table[index]->oid;
 122 }
 123
 124 int commit_graft_pos(struct repository *r, const struct object_id *oid)
 125 {
 126         return oid_pos(oid, r->parsed_objects->grafts,
 127                        r->parsed_objects->grafts_nr,
 128                        commit_graft_oid_access);
 129 }
 130
 131 static void unparse_commit(struct repository *r, const struct object_id *oid)
 132 {
 133         struct commit *c = lookup_commit(r, oid);
 134
 135         if (!c->object.parsed)
 136                 return;
 137         free_commit_list(c->parents);
 138         c->parents = NULL;
 139         c->object.parsed = 0;
 140 }
 141
 142 int register_commit_graft(struct repository *r, struct commit_graft *graft,
 143                           int ignore_dups)
 144 {
 145         int pos = commit_graft_pos(r, &graft->oid);
 146
 147         if (0 <= pos) {
 148                 if (ignore_dups)
 149                         free(graft);
 150                 else {
 151                         free(r->parsed_objects->grafts[pos]);
 152                         r->parsed_objects->grafts[pos] = graft;
 153                 }
 154                 return 1;
 155         }
 156         pos = -pos - 1;
 157         ALLOC_GROW(r->parsed_objects->grafts,
 158                    r->parsed_objects->grafts_nr + 1,
 159                    r->parsed_objects->grafts_alloc);
 160         r->parsed_objects->grafts_nr++;
 161         if (pos < r->parsed_objects->grafts_nr)
 162                 memmove(r->parsed_objects->grafts + pos + 1,
 163                         r->parsed_objects->grafts + pos,
 164                         (r->parsed_objects->grafts_nr - pos - 1) *
 165                         sizeof(*r->parsed_objects->grafts));
 166         r->parsed_objects->grafts[pos] = graft;
 167         unparse_commit(r, &graft->oid);
 168         return 0;
 169 }
 170
 171 struct commit_graft *read_graft_line(struct strbuf *line)
 172 {
 173         /* The format is just "Commit Parent1 Parent2 ...\n" */
 174         int i, phase;
 175         const char *tail = NULL;
 176         struct commit_graft *graft = NULL;
 177         struct object_id dummy_oid, *oid;
 178
 179         strbuf_rtrim(line);
 180         if (!line->len || line->buf[0] == '#')
 181                 return NULL;
 182         /*
 183          * phase 0 verifies line, counts hashes in line and allocates graft
 184          * phase 1 fills graft
 185          */
 186         for (phase = 0; phase < 2; phase++) {
 187                 oid = graft ? &graft->oid : &dummy_oid;
 188                 if (parse_oid_hex(line->buf, oid, &tail))
 189                         goto bad_graft_data;
 190                 for (i = 0; *tail != '\0'; i++) {
 191                         oid = graft ? &graft->parent[i] : &dummy_oid;
 192                         if (!isspace(*tail++) || parse_oid_hex(tail, oid, &tail))
 193                                 goto bad_graft_data;
 194                 }
 195                 if (!graft) {
 196                         graft = xmalloc(st_add(sizeof(*graft),
 197                                                st_mult(sizeof(struct object_id), i)));
 198                         graft->nr_parent = i;
 199                 }
 200         }
 201         return graft;
 202
 203 bad_graft_data:
 204         error("bad graft data: %s", line->buf);
 205         assert(!graft);
 206         return NULL;
 207 }
 208
 209 static int read_graft_file(struct repository *r, const char *graft_file)
 210 {
 211         FILE *fp = fopen_or_warn(graft_file, "r");
 212         struct strbuf buf = STRBUF_INIT;
 213         if (!fp)
 214                 return -1;
 215         if (!no_graft_file_deprecated_advice &&
 216             advice_enabled(ADVICE_GRAFT_FILE_DEPRECATED))
 217                 advise(_("Support for <GIT_DIR>/info/grafts is deprecated\n"
 218                          "and will be removed in a future Git version.\n"
 219                          "\n"
 220                          "Please use \"git replace --convert-graft-file\"\n"
 221                          "to convert the grafts into replace refs.\n"
 222                          "\n"
 223                          "Turn this message off by running\n"
 224                          "\"git config advice.graftFileDeprecated false\""));
 225         while (!strbuf_getwholeline(&buf, fp, '\n')) {
 226                 /* The format is just "Commit Parent1 Parent2 ...\n" */
 227                 struct commit_graft *graft = read_graft_line(&buf);
 228                 if (!graft)
 229                         continue;
 230                 if (register_commit_graft(r, graft, 1))
 231                         error("duplicate graft data: %s", buf.buf);
 232         }
 233         fclose(fp);
 234         strbuf_release(&buf);
 235         return 0;
 236 }
 237
 238 void prepare_commit_graft(struct repository *r)
 239 {
 240         char *graft_file;
 241
 242         if (r->parsed_objects->commit_graft_prepared)
 243                 return;
 244         if (!startup_info->have_repository)
 245                 return;
 246
 247         graft_file = get_graft_file(r);
 248         read_graft_file(r, graft_file);
 249         /* make sure shallows are read */
 250         is_repository_shallow(r);
 251         r->parsed_objects->commit_graft_prepared = 1;
 252 }
 253
 254 struct commit_graft *lookup_commit_graft(struct repository *r, const struct object_id *oid)
 255 {
 256         int pos;
 257         prepare_commit_graft(r);
 258         pos = commit_graft_pos(r, oid);
 259         if (pos < 0)
 260                 return NULL;
 261         return r->parsed_objects->grafts[pos];
 262 }
 263
 264 int for_each_commit_graft(each_commit_graft_fn fn, void *cb_data)
 265 {
 266         int i, ret;
 267         for (i = ret = 0; i < the_repository->parsed_objects->grafts_nr && !ret; i++)
 268                 ret = fn(the_repository->parsed_objects->grafts[i], cb_data);
 269         return ret;
 270 }
 271
 272 void reset_commit_grafts(struct repository *r)
 273 {
 274         int i;
 275
 276         for (i = 0; i < r->parsed_objects->grafts_nr; i++) {
 277                 unparse_commit(r, &r->parsed_objects->grafts[i]->oid);
 278                 free(r->parsed_objects->grafts[i]);
 279         }
 280         r->parsed_objects->grafts_nr = 0;
 281         r->parsed_objects->commit_graft_prepared = 0;
 282 }
 283
 284 struct commit_buffer {
 285         void *buffer;
 286         unsigned long size;
 287 };
 288 define_commit_slab(buffer_slab, struct commit_buffer);
 289
 290 struct buffer_slab *allocate_commit_buffer_slab(void)
 291 {
 292         struct buffer_slab *bs = xmalloc(sizeof(*bs));
 293         init_buffer_slab(bs);
 294         return bs;
 295 }
 296
 297 void free_commit_buffer_slab(struct buffer_slab *bs)
 298 {
 299         clear_buffer_slab(bs);
 300         free(bs);
 301 }
 302
 303 void set_commit_buffer(struct repository *r, struct commit *commit, void *buffer, unsigned long size)
 304 {
 305         struct commit_buffer *v = buffer_slab_at(
 306                 r->parsed_objects->buffer_slab, commit);
 307         v->buffer = buffer;
 308         v->size = size;
 309 }
 310
 311 const void *get_cached_commit_buffer(struct repository *r, const struct commit *commit, unsigned long *sizep)
 312 {
 313         struct commit_buffer *v = buffer_slab_peek(
 314                 r->parsed_objects->buffer_slab, commit);
 315         if (!v) {
 316                 if (sizep)
 317                         *sizep = 0;
 318                 return NULL;
 319         }
 320         if (sizep)
 321                 *sizep = v->size;
 322         return v->buffer;
 323 }
 324
 325 const void *repo_get_commit_buffer(struct repository *r,
 326                                    const struct commit *commit,
 327                                    unsigned long *sizep)
 328 {
 329         const void *ret = get_cached_commit_buffer(r, commit, sizep);
 330         if (!ret) {
 331                 enum object_type type;
 332                 unsigned long size;
 333                 ret = repo_read_object_file(r, &commit->object.oid, &type, &size);
 334                 if (!ret)
 335                         die("cannot read commit object %s",
 336                             oid_to_hex(&commit->object.oid));
 337                 if (type != OBJ_COMMIT)
 338                         die("expected commit for %s, got %s",
 339                             oid_to_hex(&commit->object.oid), type_name(type));
 340                 if (sizep)
 341                         *sizep = size;
 342         }
 343         return ret;
 344 }
 345
 346 void repo_unuse_commit_buffer(struct repository *r,
 347                               const struct commit *commit,
 348                               const void *buffer)
 349 {
 350         struct commit_buffer *v = buffer_slab_peek(
 351                 r->parsed_objects->buffer_slab, commit);
 352         if (!(v && v->buffer == buffer))
 353                 free((void *)buffer);
 354 }
 355
 356 void free_commit_buffer(struct parsed_object_pool *pool, struct commit *commit)
 357 {
 358         struct commit_buffer *v = buffer_slab_peek(
 359                 pool->buffer_slab, commit);
 360         if (v) {
 361                 FREE_AND_NULL(v->buffer);
 362                 v->size = 0;
 363         }
 364 }
 365
 366 static inline void set_commit_tree(struct commit *c, struct tree *t)
 367 {
 368         c->maybe_tree = t;
 369 }
 370
 371 struct tree *repo_get_commit_tree(struct repository *r,
 372                                   const struct commit *commit)
 373 {
 374         if (commit->maybe_tree || !commit->object.parsed)
 375                 return commit->maybe_tree;
 376
 377         if (commit_graph_position(commit) != COMMIT_NOT_FROM_GRAPH)
 378                 return get_commit_tree_in_graph(r, commit);
 379
 380         return NULL;
 381 }
 382
 383 struct object_id *get_commit_tree_oid(const struct commit *commit)
 384 {
 385         struct tree *tree = get_commit_tree(commit);
 386         return tree ? &tree->object.oid : NULL;
 387 }
 388
 389 void release_commit_memory(struct parsed_object_pool *pool, struct commit *c)
 390 {
 391         set_commit_tree(c, NULL);
 392         free_commit_buffer(pool, c);
 393         c->index = 0;
 394         free_commit_list(c->parents);
 395
 396         c->object.parsed = 0;
 397 }
 398
 399 const void *detach_commit_buffer(struct commit *commit, unsigned long *sizep)
 400 {
 401         struct commit_buffer *v = buffer_slab_peek(
 402                 the_repository->parsed_objects->buffer_slab, commit);
 403         void *ret;
 404
 405         if (!v) {
 406                 if (sizep)
 407                         *sizep = 0;
 408                 return NULL;
 409         }
 410         ret = v->buffer;
 411         if (sizep)
 412                 *sizep = v->size;
 413
 414         v->buffer = NULL;
 415         v->size = 0;
 416         return ret;
 417 }
 418
 419 int parse_commit_buffer(struct repository *r, struct commit *item, const void *buffer, unsigned long size, int check_graph)
 420 {
 421         const char *tail = buffer;
 422         const char *bufptr = buffer;
 423         struct object_id parent;
 424         struct commit_list **pptr;
 425         struct commit_graft *graft;
 426         const int tree_entry_len = the_hash_algo->hexsz + 5;
 427         const int parent_entry_len = the_hash_algo->hexsz + 7;
 428         struct tree *tree;
 429
 430         if (item->object.parsed)
 431                 return 0;
 432         /*
 433          * Presumably this is leftover from an earlier failed parse;
 434          * clear it out in preparation for us re-parsing (we'll hit the
 435          * same error, but that's good, since it lets our caller know
 436          * the result cannot be trusted.
 437          */
 438         free_commit_list(item->parents);
 439         item->parents = NULL;
 440
 441         tail += size;
 442         if (tail <= bufptr + tree_entry_len + 1 || memcmp(bufptr, "tree ", 5) ||
 443                         bufptr[tree_entry_len] != '\n')
 444                 return error("bogus commit object %s", oid_to_hex(&item->object.oid));
 445         if (get_oid_hex(bufptr + 5, &parent) < 0)
 446                 return error("bad tree pointer in commit %s",
 447                              oid_to_hex(&item->object.oid));
 448         tree = lookup_tree(r, &parent);
 449         if (!tree)
 450                 return error("bad tree pointer %s in commit %s",
 451                              oid_to_hex(&parent),
 452                              oid_to_hex(&item->object.oid));
 453         set_commit_tree(item, tree);
 454         bufptr += tree_entry_len + 1; /* "tree " + "hex sha1" + "\n" */
 455         pptr = &item->parents;
 456
 457         graft = lookup_commit_graft(r, &item->object.oid);
 458         if (graft)
 459                 r->parsed_objects->substituted_parent = 1;
 460         while (bufptr + parent_entry_len < tail && !memcmp(bufptr, "parent ", 7)) {
 461                 struct commit *new_parent;
 462
 463                 if (tail <= bufptr + parent_entry_len + 1 ||
 464                     get_oid_hex(bufptr + 7, &parent) ||
 465                     bufptr[parent_entry_len] != '\n')
 466                         return error("bad parents in commit %s", oid_to_hex(&item->object.oid));
 467                 bufptr += parent_entry_len + 1;
 468                 /*
 469                  * The clone is shallow if nr_parent < 0, and we must
 470                  * not traverse its real parents even when we unhide them.
 471                  */
 472                 if (graft && (graft->nr_parent < 0 || grafts_replace_parents))
 473                         continue;
 474                 new_parent = lookup_commit(r, &parent);
 475                 if (!new_parent)
 476                         return error("bad parent %s in commit %s",
 477                                      oid_to_hex(&parent),
 478                                      oid_to_hex(&item->object.oid));
 479                 pptr = &commit_list_insert(new_parent, pptr)->next;
 480         }
 481         if (graft) {
 482                 int i;
 483                 struct commit *new_parent;
 484                 for (i = 0; i < graft->nr_parent; i++) {
 485                         new_parent = lookup_commit(r,
 486                                                    &graft->parent[i]);
 487                         if (!new_parent)
 488                                 return error("bad graft parent %s in commit %s",
 489                                              oid_to_hex(&graft->parent[i]),
 490                                              oid_to_hex(&item->object.oid));
 491                         pptr = &commit_list_insert(new_parent, pptr)->next;
 492                 }
 493         }
 494         item->date = parse_commit_date(bufptr, tail);
 495
 496         if (check_graph)
 497                 load_commit_graph_info(r, item);
 498
 499         item->object.parsed = 1;
 500         return 0;
 501 }
 502
 503 int repo_parse_commit_internal(struct repository *r,
 504                                struct commit *item,
 505                                int quiet_on_missing,
 506                                int use_commit_graph)
 507 {
 508         enum object_type type;
 509         void *buffer;
 510         unsigned long size;
 511         int ret;
 512
 513         if (!item)
 514                 return -1;
 515         if (item->object.parsed)
 516                 return 0;
 517         if (use_commit_graph && parse_commit_in_graph(r, item))
 518                 return 0;
 519         buffer = repo_read_object_file(r, &item->object.oid, &type, &size);
 520         if (!buffer)
 521                 return quiet_on_missing ? -1 :
 522                         error("Could not read %s",
 523                              oid_to_hex(&item->object.oid));
 524         if (type != OBJ_COMMIT) {
 525                 free(buffer);
 526                 return error("Object %s not a commit",
 527                              oid_to_hex(&item->object.oid));
 528         }
 529
 530         ret = parse_commit_buffer(r, item, buffer, size, 0);
 531         if (save_commit_buffer && !ret) {
 532                 set_commit_buffer(r, item, buffer, size);
 533                 return 0;
 534         }
 535         free(buffer);
 536         return ret;
 537 }
 538
 539 int repo_parse_commit_gently(struct repository *r,
 540                              struct commit *item, int quiet_on_missing)
 541 {
 542         return repo_parse_commit_internal(r, item, quiet_on_missing, 1);
 543 }
 544
 545 void parse_commit_or_die(struct commit *item)
 546 {
 547         if (parse_commit(item))
 548                 die("unable to parse commit %s",
 549                     item ? oid_to_hex(&item->object.oid) : "(null)");
 550 }
 551
 552 int find_commit_subject(const char *commit_buffer, const char **subject)
 553 {
 554         const char *eol;
 555         const char *p = commit_buffer;
 556
 557         while (*p && (*p != '\n' || p[1] != '\n'))
 558                 p++;
 559         if (*p) {
 560                 p = skip_blank_lines(p + 2);
 561                 eol = strchrnul(p, '\n');
 562         } else
 563                 eol = p;
 564
 565         *subject = p;
 566
 567         return eol - p;
 568 }
 569
 570 size_t commit_subject_length(const char *body)
 571 {
 572         const char *p = body;
 573         while (*p) {
 574                 const char *next = skip_blank_lines(p);
 575                 if (next != p)
 576                         break;
 577                 p = strchrnul(p, '\n');
 578                 if (*p)
 579                         p++;
 580         }
 581         return p - body;
 582 }
 583
 584 struct commit_list *commit_list_insert(struct commit *item, struct commit_list **list_p)
 585 {
 586         struct commit_list *new_list = xmalloc(sizeof(struct commit_list));
 587         new_list->item = item;
 588         new_list->next = *list_p;
 589         *list_p = new_list;
 590         return new_list;
 591 }
 592
 593 int commit_list_contains(struct commit *item, struct commit_list *list)
 594 {
 595         while (list) {
 596                 if (list->item == item)
 597                         return 1;
 598                 list = list->next;
 599         }
 600
 601         return 0;
 602 }
 603
 604 unsigned commit_list_count(const struct commit_list *l)
 605 {
 606         unsigned c = 0;
 607         for (; l; l = l->next )
 608                 c++;
 609         return c;
 610 }
 611
 612 struct commit_list *copy_commit_list(struct commit_list *list)
 613 {
 614         struct commit_list *head = NULL;
 615         struct commit_list **pp = &head;
 616         while (list) {
 617                 pp = commit_list_append(list->item, pp);
 618                 list = list->next;
 619         }
 620         return head;
 621 }
 622
 623 struct commit_list *reverse_commit_list(struct commit_list *list)
 624 {
 625         struct commit_list *next = NULL, *current, *backup;
 626         for (current = list; current; current = backup) {
 627                 backup = current->next;
 628                 current->next = next;
 629                 next = current;
 630         }
 631         return next;
 632 }
 633
 634 void free_commit_list(struct commit_list *list)
 635 {
 636         while (list)
 637                 pop_commit(&list);
 638 }
 639
 640 struct commit_list * commit_list_insert_by_date(struct commit *item, struct commit_list **list)
 641 {
 642         struct commit_list **pp = list;
 643         struct commit_list *p;
 644         while ((p = *pp) != NULL) {
 645                 if (p->item->date < item->date) {
 646                         break;
 647                 }
 648                 pp = &p->next;
 649         }
 650         return commit_list_insert(item, pp);
 651 }
 652
 653 static int commit_list_compare_by_date(const void *a, const void *b)
 654 {
 655         timestamp_t a_date = ((const struct commit_list *)a)->item->date;
 656         timestamp_t b_date = ((const struct commit_list *)b)->item->date;
 657         if (a_date < b_date)
 658                 return 1;
 659         if (a_date > b_date)
 660                 return -1;
 661         return 0;
 662 }
 663
 664 static void *commit_list_get_next(const void *a)
 665 {
 666         return ((const struct commit_list *)a)->next;
 667 }
 668
 669 static void commit_list_set_next(void *a, void *next)
 670 {
 671         ((struct commit_list *)a)->next = next;
 672 }
 673
 674 void commit_list_sort_by_date(struct commit_list **list)
 675 {
 676         *list = llist_mergesort(*list, commit_list_get_next, commit_list_set_next,
 677                                 commit_list_compare_by_date);
 678 }
 679
 680 struct commit *pop_most_recent_commit(struct commit_list **list,
 681                                       unsigned int mark)
 682 {
 683         struct commit *ret = pop_commit(list);
 684         struct commit_list *parents = ret->parents;
 685
 686         while (parents) {
 687                 struct commit *commit = parents->item;
 688                 if (!parse_commit(commit) && !(commit->object.flags & mark)) {
 689                         commit->object.flags |= mark;
 690                         commit_list_insert_by_date(commit, list);
 691                 }
 692                 parents = parents->next;
 693         }
 694         return ret;
 695 }
 696
 697 static void clear_commit_marks_1(struct commit_list **plist,
 698                                  struct commit *commit, unsigned int mark)
 699 {
 700         while (commit) {
 701                 struct commit_list *parents;
 702
 703                 if (!(mark & commit->object.flags))
 704                         return;
 705
 706                 commit->object.flags &= ~mark;
 707
 708                 parents = commit->parents;
 709                 if (!parents)
 710                         return;
 711
 712                 while ((parents = parents->next))
 713                         commit_list_insert(parents->item, plist);
 714
 715                 commit = commit->parents->item;
 716         }
 717 }
 718
 719 void clear_commit_marks_many(int nr, struct commit **commit, unsigned int mark)
 720 {
 721         struct commit_list *list = NULL;
 722
 723         while (nr--) {
 724                 clear_commit_marks_1(&list, *commit, mark);
 725                 commit++;
 726         }
 727         while (list)
 728                 clear_commit_marks_1(&list, pop_commit(&list), mark);
 729 }
 730
 731 void clear_commit_marks(struct commit *commit, unsigned int mark)
 732 {
 733         clear_commit_marks_many(1, &commit, mark);
 734 }
 735
 736 struct commit *pop_commit(struct commit_list **stack)
 737 {
 738         struct commit_list *top = *stack;
 739         struct commit *item = top ? top->item : NULL;
 740
 741         if (top) {
 742                 *stack = top->next;
 743                 free(top);
 744         }
 745         return item;
 746 }
 747
 748 /*
 749  * Topological sort support
 750  */
 751
 752 /* count number of children that have not been emitted */
 753 define_commit_slab(indegree_slab, int);
 754
 755 define_commit_slab(author_date_slab, timestamp_t);
 756
 757 void record_author_date(struct author_date_slab *author_date,
 758                         struct commit *commit)
 759 {
 760         const char *buffer = get_commit_buffer(commit, NULL);
 761         struct ident_split ident;
 762         const char *ident_line;
 763         size_t ident_len;
 764         char *date_end;
 765         timestamp_t date;
 766
 767         ident_line = find_commit_header(buffer, "author", &ident_len);
 768         if (!ident_line)
 769                 goto fail_exit; /* no author line */
 770         if (split_ident_line(&ident, ident_line, ident_len) ||
 771             !ident.date_begin || !ident.date_end)
 772                 goto fail_exit; /* malformed "author" line */
 773
 774         date = parse_timestamp(ident.date_begin, &date_end, 10);
 775         if (date_end != ident.date_end)
 776                 goto fail_exit; /* malformed date */
 777         *(author_date_slab_at(author_date, commit)) = date;
 778
 779 fail_exit:
 780         unuse_commit_buffer(commit, buffer);
 781 }
 782
 783 int compare_commits_by_author_date(const void *a_, const void *b_,
 784                                    void *cb_data)
 785 {
 786         const struct commit *a = a_, *b = b_;
 787         struct author_date_slab *author_date = cb_data;
 788         timestamp_t a_date = *(author_date_slab_at(author_date, a));
 789         timestamp_t b_date = *(author_date_slab_at(author_date, b));
 790
 791         /* newer commits with larger date first */
 792         if (a_date < b_date)
 793                 return 1;
 794         else if (a_date > b_date)
 795                 return -1;
 796         return 0;
 797 }
 798
 799 int compare_commits_by_gen_then_commit_date(const void *a_, const void *b_, void *unused)
 800 {
 801         const struct commit *a = a_, *b = b_;
 802         const timestamp_t generation_a = commit_graph_generation(a),
 803                           generation_b = commit_graph_generation(b);
 804
 805         /* newer commits first */
 806         if (generation_a < generation_b)
 807                 return 1;
 808         else if (generation_a > generation_b)
 809                 return -1;
 810
 811         /* use date as a heuristic when generations are equal */
 812         if (a->date < b->date)
 813                 return 1;
 814         else if (a->date > b->date)
 815                 return -1;
 816         return 0;
 817 }
 818
 819 int compare_commits_by_commit_date(const void *a_, const void *b_, void *unused)
 820 {
 821         const struct commit *a = a_, *b = b_;
 822         /* newer commits with larger date first */
 823         if (a->date < b->date)
 824                 return 1;
 825         else if (a->date > b->date)
 826                 return -1;
 827         return 0;
 828 }
 829
 830 /*
 831  * Performs an in-place topological sort on the list supplied.
 832  */
 833 void sort_in_topological_order(struct commit_list **list, enum rev_sort_order sort_order)
 834 {
 835         struct commit_list *next, *orig = *list;
 836         struct commit_list **pptr;
 837         struct indegree_slab indegree;
 838         struct prio_queue queue;
 839         struct commit *commit;
 840         struct author_date_slab author_date;
 841
 842         if (!orig)
 843                 return;
 844         *list = NULL;
 845
 846         init_indegree_slab(&indegree);
 847         memset(&queue, '\0', sizeof(queue));
 848
 849         switch (sort_order) {
 850         default: /* REV_SORT_IN_GRAPH_ORDER */
 851                 queue.compare = NULL;
 852                 break;
 853         case REV_SORT_BY_COMMIT_DATE:
 854                 queue.compare = compare_commits_by_commit_date;
 855                 break;
 856         case REV_SORT_BY_AUTHOR_DATE:
 857                 init_author_date_slab(&author_date);
 858                 queue.compare = compare_commits_by_author_date;
 859                 queue.cb_data = &author_date;
 860                 break;
 861         }
 862
 863         /* Mark them and clear the indegree */
 864         for (next = orig; next; next = next->next) {
 865                 struct commit *commit = next->item;
 866                 *(indegree_slab_at(&indegree, commit)) = 1;
 867                 /* also record the author dates, if needed */
 868                 if (sort_order == REV_SORT_BY_AUTHOR_DATE)
 869                         record_author_date(&author_date, commit);
 870         }
 871
 872         /* update the indegree */
 873         for (next = orig; next; next = next->next) {
 874                 struct commit_list *parents = next->item->parents;
 875                 while (parents) {
 876                         struct commit *parent = parents->item;
 877                         int *pi = indegree_slab_at(&indegree, parent);
 878
 879                         if (*pi)
 880                                 (*pi)++;
 881                         parents = parents->next;
 882                 }
 883         }
 884
 885         /*
 886          * find the tips
 887          *
 888          * tips are nodes not reachable from any other node in the list
 889          *
 890          * the tips serve as a starting set for the work queue.
 891          */
 892         for (next = orig; next; next = next->next) {
 893                 struct commit *commit = next->item;
 894
 895                 if (*(indegree_slab_at(&indegree, commit)) == 1)
 896                         prio_queue_put(&queue, commit);
 897         }
 898
 899         /*
 900          * This is unfortunate; the initial tips need to be shown
 901          * in the order given from the revision traversal machinery.
 902          */
 903         if (sort_order == REV_SORT_IN_GRAPH_ORDER)
 904                 prio_queue_reverse(&queue);
 905
 906         /* We no longer need the commit list */
 907         free_commit_list(orig);
 908
 909         pptr = list;
 910         *list = NULL;
 911         while ((commit = prio_queue_get(&queue)) != NULL) {
 912                 struct commit_list *parents;
 913
 914                 for (parents = commit->parents; parents ; parents = parents->next) {
 915                         struct commit *parent = parents->item;
 916                         int *pi = indegree_slab_at(&indegree, parent);
 917
 918                         if (!*pi)
 919                                 continue;
 920
 921                         /*
 922                          * parents are only enqueued for emission
 923                          * when all their children have been emitted thereby
 924                          * guaranteeing topological order.
 925                          */
 926                         if (--(*pi) == 1)
 927                                 prio_queue_put(&queue, parent);
 928                 }
 929                 /*
 930                  * all children of commit have already been
 931                  * emitted. we can emit it now.
 932                  */
 933                 *(indegree_slab_at(&indegree, commit)) = 0;
 934
 935                 pptr = &commit_list_insert(commit, pptr)->next;
 936         }
 937
 938         clear_indegree_slab(&indegree);
 939         clear_prio_queue(&queue);
 940         if (sort_order == REV_SORT_BY_AUTHOR_DATE)
 941                 clear_author_date_slab(&author_date);
 942 }
 943
 944 struct rev_collect {
 945         struct commit **commit;
 946         int nr;
 947         int alloc;
 948         unsigned int initial : 1;
 949 };
 950
 951 static void add_one_commit(struct object_id *oid, struct rev_collect *revs)
 952 {
 953         struct commit *commit;
 954
 955         if (is_null_oid(oid))
 956                 return;
 957
 958         commit = lookup_commit(the_repository, oid);
 959         if (!commit ||
 960             (commit->object.flags & TMP_MARK) ||
 961             parse_commit(commit))
 962                 return;
 963
 964         ALLOC_GROW(revs->commit, revs->nr + 1, revs->alloc);
 965         revs->commit[revs->nr++] = commit;
 966         commit->object.flags |= TMP_MARK;
 967 }
 968
 969 static int collect_one_reflog_ent(struct object_id *ooid, struct object_id *noid,
 970                                   const char *ident, timestamp_t timestamp,
 971                                   int tz, const char *message, void *cbdata)
 972 {
 973         struct rev_collect *revs = cbdata;
 974
 975         if (revs->initial) {
 976                 revs->initial = 0;
 977                 add_one_commit(ooid, revs);
 978         }
 979         add_one_commit(noid, revs);
 980         return 0;
 981 }
 982
 983 struct commit *get_fork_point(const char *refname, struct commit *commit)
 984 {
 985         struct object_id oid;
 986         struct rev_collect revs;
 987         struct commit_list *bases;
 988         int i;
 989         struct commit *ret = NULL;
 990         char *full_refname;
 991
 992         switch (dwim_ref(refname, strlen(refname), &oid, &full_refname, 0)) {
 993         case 0:
 994                 die("No such ref: '%s'", refname);
 995         case 1:
 996                 break; /* good */
 997         default:
 998                 die("Ambiguous refname: '%s'", refname);
 999         }
1000
1001         memset(&revs, 0, sizeof(revs));
1002         revs.initial = 1;
1003         for_each_reflog_ent(full_refname, collect_one_reflog_ent, &revs);
1004
1005         if (!revs.nr)
1006                 add_one_commit(&oid, &revs);
1007
1008         for (i = 0; i < revs.nr; i++)
1009                 revs.commit[i]->object.flags &= ~TMP_MARK;
1010
1011         bases = get_merge_bases_many(commit, revs.nr, revs.commit);
1012
1013         /*
1014          * There should be one and only one merge base, when we found
1015          * a common ancestor among reflog entries.
1016          */
1017         if (!bases || bases->next)
1018                 goto cleanup_return;
1019
1020         /* And the found one must be one of the reflog entries */
1021         for (i = 0; i < revs.nr; i++)
1022                 if (&bases->item->object == &revs.commit[i]->object)
1023                         break; /* found */
1024         if (revs.nr <= i)
1025                 goto cleanup_return;
1026
1027         ret = bases->item;
1028
1029 cleanup_return:
1030         free_commit_list(bases);
1031         free(full_refname);
1032         return ret;
1033 }
1034
1035 /*
1036  * Indexed by hash algorithm identifier.
1037  */
1038 static const char *gpg_sig_headers[] = {
1039         NULL,
1040         "gpgsig",
1041         "gpgsig-sha256",
1042 };
1043
1044 int sign_with_header(struct strbuf *buf, const char *keyid)
1045 {
1046         struct strbuf sig = STRBUF_INIT;
1047         int inspos, copypos;
1048         const char *eoh;
1049         const char *gpg_sig_header = gpg_sig_headers[hash_algo_by_ptr(the_hash_algo)];
1050         int gpg_sig_header_len = strlen(gpg_sig_header);
1051
1052         /* find the end of the header */
1053         eoh = strstr(buf->buf, "\n\n");
1054         if (!eoh)
1055                 inspos = buf->len;
1056         else
1057                 inspos = eoh - buf->buf + 1;
1058
1059         if (!keyid || !*keyid)
1060                 keyid = get_signing_key();
1061         if (sign_buffer(buf, &sig, keyid)) {
1062                 strbuf_release(&sig);
1063                 return -1;
1064         }
1065
1066         for (copypos = 0; sig.buf[copypos]; ) {
1067                 const char *bol = sig.buf + copypos;
1068                 const char *eol = strchrnul(bol, '\n');
1069                 int len = (eol - bol) + !!*eol;
1070
1071                 if (!copypos) {
1072                         strbuf_insert(buf, inspos, gpg_sig_header, gpg_sig_header_len);
1073                         inspos += gpg_sig_header_len;
1074                 }
1075                 strbuf_insertstr(buf, inspos++, " ");
1076                 strbuf_insert(buf, inspos, bol, len);
1077                 inspos += len;
1078                 copypos += len;
1079         }
1080         strbuf_release(&sig);
1081         return 0;
1082 }
1083
1084
1085
1086 int parse_signed_commit(const struct commit *commit,
1087                         struct strbuf *payload, struct strbuf *signature,
1088                         const struct git_hash_algo *algop)
1089 {
1090         unsigned long size;
1091         const char *buffer = get_commit_buffer(commit, &size);
1092         int ret = parse_buffer_signed_by_header(buffer, size, payload, signature, algop);
1093
1094         unuse_commit_buffer(commit, buffer);
1095         return ret;
1096 }
1097
1098 int parse_buffer_signed_by_header(const char *buffer,
1099                                   unsigned long size,
1100                                   struct strbuf *payload,
1101                                   struct strbuf *signature,
1102                                   const struct git_hash_algo *algop)
1103 {
1104         int in_signature = 0, saw_signature = 0, other_signature = 0;
1105         const char *line, *tail, *p;
1106         const char *gpg_sig_header = gpg_sig_headers[hash_algo_by_ptr(algop)];
1107
1108         line = buffer;
1109         tail = buffer + size;
1110         while (line < tail) {
1111                 const char *sig = NULL;
1112                 const char *next = memchr(line, '\n', tail - line);
1113
1114                 next = next ? next + 1 : tail;
1115                 if (in_signature && line[0] == ' ')
1116                         sig = line + 1;
1117                 else if (skip_prefix(line, gpg_sig_header, &p) &&
1118                          *p == ' ') {
1119                         sig = line + strlen(gpg_sig_header) + 1;
1120                         other_signature = 0;
1121                 }
1122                 else if (starts_with(line, "gpgsig"))
1123                         other_signature = 1;
1124                 else if (other_signature && line[0] != ' ')
1125                         other_signature = 0;
1126                 if (sig) {
1127                         strbuf_add(signature, sig, next - sig);
1128                         saw_signature = 1;
1129                         in_signature = 1;
1130                 } else {
1131                         if (*line == '\n')
1132                                 /* dump the whole remainder of the buffer */
1133                                 next = tail;
1134                         if (!other_signature)
1135                                 strbuf_add(payload, line, next - line);
1136                         in_signature = 0;
1137                 }
1138                 line = next;
1139         }
1140         return saw_signature;
1141 }
1142
1143 int remove_signature(struct strbuf *buf)
1144 {
1145         const char *line = buf->buf;
1146         const char *tail = buf->buf + buf->len;
1147         int in_signature = 0;
1148         struct sigbuf {
1149                 const char *start;
1150                 const char *end;
1151         } sigs[2], *sigp = &sigs[0];
1152         int i;
1153         const char *orig_buf = buf->buf;
1154
1155         memset(sigs, 0, sizeof(sigs));
1156
1157         while (line < tail) {
1158                 const char *next = memchr(line, '\n', tail - line);
1159                 next = next ? next + 1 : tail;
1160
1161                 if (in_signature && line[0] == ' ')
1162                         sigp->end = next;
1163                 else if (starts_with(line, "gpgsig")) {
1164                         int i;
1165                         for (i = 1; i < GIT_HASH_NALGOS; i++) {
1166                                 const char *p;
1167                                 if (skip_prefix(line, gpg_sig_headers[i], &p) &&
1168                                     *p == ' ') {
1169                                         sigp->start = line;
1170                                         sigp->end = next;
1171                                         in_signature = 1;
1172                                 }
1173                         }
1174                 } else {
1175                         if (*line == '\n')
1176                                 /* dump the whole remainder of the buffer */
1177                                 next = tail;
1178                         if (in_signature && sigp - sigs != ARRAY_SIZE(sigs))
1179                                 sigp++;
1180                         in_signature = 0;
1181                 }
1182                 line = next;
1183         }
1184
1185         for (i = ARRAY_SIZE(sigs) - 1; i >= 0; i--)
1186                 if (sigs[i].start)
1187                         strbuf_remove(buf, sigs[i].start - orig_buf, sigs[i].end - sigs[i].start);
1188
1189         return sigs[0].start != NULL;
1190 }
1191
1192 static void handle_signed_tag(struct commit *parent, struct commit_extra_header ***tail)
1193 {
1194         struct merge_remote_desc *desc;
1195         struct commit_extra_header *mergetag;
1196         char *buf;
1197         unsigned long size;
1198         enum object_type type;
1199         struct strbuf payload = STRBUF_INIT;
1200         struct strbuf signature = STRBUF_INIT;
1201
1202         desc = merge_remote_util(parent);
1203         if (!desc || !desc->obj)
1204                 return;
1205         buf = read_object_file(&desc->obj->oid, &type, &size);
1206         if (!buf || type != OBJ_TAG)
1207                 goto free_return;
1208         if (!parse_signature(buf, size, &payload, &signature))
1209                 goto free_return;
1210         /*
1211          * We could verify this signature and either omit the tag when
1212          * it does not validate, but the integrator may not have the
1213          * public key of the signer of the tag being merged, while a
1214          * later auditor may have it while auditing, so let's not run
1215          * verify-signed-buffer here for now...
1216          *
1217          * if (verify_signed_buffer(buf, len, buf + len, size - len, ...))
1218          *      warn("warning: signed tag unverified.");
1219          */
1220         CALLOC_ARRAY(mergetag, 1);
1221         mergetag->key = xstrdup("mergetag");
1222         mergetag->value = buf;
1223         mergetag->len = size;
1224
1225         **tail = mergetag;
1226         *tail = &mergetag->next;
1227         strbuf_release(&payload);
1228         strbuf_release(&signature);
1229         return;
1230
1231 free_return:
1232         free(buf);
1233 }
1234
1235 int check_commit_signature(const struct commit *commit, struct signature_check *sigc)
1236 {
1237         struct strbuf payload = STRBUF_INIT;
1238         struct strbuf signature = STRBUF_INIT;
1239         int ret = 1;
1240
1241         sigc->result = 'N';
1242
1243         if (parse_signed_commit(commit, &payload, &signature, the_hash_algo) <= 0)
1244                 goto out;
1245
1246         sigc->payload_type = SIGNATURE_PAYLOAD_COMMIT;
1247         sigc->payload = strbuf_detach(&payload, &sigc->payload_len);
1248         ret = check_signature(sigc, signature.buf, signature.len);
1249
1250  out:
1251         strbuf_release(&payload);
1252         strbuf_release(&signature);
1253
1254         return ret;
1255 }
1256
1257 void verify_merge_signature(struct commit *commit, int verbosity,
1258                             int check_trust)
1259 {
1260         char hex[GIT_MAX_HEXSZ + 1];
1261         struct signature_check signature_check;
1262         int ret;
1263         memset(&signature_check, 0, sizeof(signature_check));
1264
1265         ret = check_commit_signature(commit, &signature_check);
1266
1267         find_unique_abbrev_r(hex, &commit->object.oid, DEFAULT_ABBREV);
1268         switch (signature_check.result) {
1269         case 'G':
1270                 if (ret || (check_trust && signature_check.trust_level < TRUST_MARGINAL))
1271                         die(_("Commit %s has an untrusted GPG signature, "
1272                               "allegedly by %s."), hex, signature_check.signer);
1273                 break;
1274         case 'B':
1275                 die(_("Commit %s has a bad GPG signature "
1276                       "allegedly by %s."), hex, signature_check.signer);
1277         default: /* 'N' */
1278                 die(_("Commit %s does not have a GPG signature."), hex);
1279         }
1280         if (verbosity >= 0 && signature_check.result == 'G')
1281                 printf(_("Commit %s has a good GPG signature by %s\n"),
1282                        hex, signature_check.signer);
1283
1284         signature_check_clear(&signature_check);
1285 }
1286
1287 void append_merge_tag_headers(struct commit_list *parents,
1288                               struct commit_extra_header ***tail)
1289 {
1290         while (parents) {
1291                 struct commit *parent = parents->item;
1292                 handle_signed_tag(parent, tail);
1293                 parents = parents->next;
1294         }
1295 }
1296
1297 static void add_extra_header(struct strbuf *buffer,
1298                              struct commit_extra_header *extra)
1299 {
1300         strbuf_addstr(buffer, extra->key);
1301         if (extra->len)
1302                 strbuf_add_lines(buffer, " ", extra->value, extra->len);
1303         else
1304                 strbuf_addch(buffer, '\n');
1305 }
1306
1307 struct commit_extra_header *read_commit_extra_headers(struct commit *commit,
1308                                                       const char **exclude)
1309 {
1310         struct commit_extra_header *extra = NULL;
1311         unsigned long size;
1312         const char *buffer = get_commit_buffer(commit, &size);
1313         extra = read_commit_extra_header_lines(buffer, size, exclude);
1314         unuse_commit_buffer(commit, buffer);
1315         return extra;
1316 }
1317
1318 int for_each_mergetag(each_mergetag_fn fn, struct commit *commit, void *data)
1319 {
1320         struct commit_extra_header *extra, *to_free;
1321         int res = 0;
1322
1323         to_free = read_commit_extra_headers(commit, NULL);
1324         for (extra = to_free; !res && extra; extra = extra->next) {
1325                 if (strcmp(extra->key, "mergetag"))
1326                         continue; /* not a merge tag */
1327                 res = fn(commit, extra, data);
1328         }
1329         free_commit_extra_headers(to_free);
1330         return res;
1331 }
1332
1333 static inline int standard_header_field(const char *field, size_t len)
1334 {
1335         return ((len == 4 && !memcmp(field, "tree", 4)) ||
1336                 (len == 6 && !memcmp(field, "parent", 6)) ||
1337                 (len == 6 && !memcmp(field, "author", 6)) ||
1338                 (len == 9 && !memcmp(field, "committer", 9)) ||
1339                 (len == 8 && !memcmp(field, "encoding", 8)));
1340 }
1341
1342 static int excluded_header_field(const char *field, size_t len, const char **exclude)
1343 {
1344         if (!exclude)
1345                 return 0;
1346
1347         while (*exclude) {
1348                 size_t xlen = strlen(*exclude);
1349                 if (len == xlen && !memcmp(field, *exclude, xlen))
1350                         return 1;
1351                 exclude++;
1352         }
1353         return 0;
1354 }
1355
1356 static struct commit_extra_header *read_commit_extra_header_lines(
1357         const char *buffer, size_t size,
1358         const char **exclude)
1359 {
1360         struct commit_extra_header *extra = NULL, **tail = &extra, *it = NULL;
1361         const char *line, *next, *eof, *eob;
1362         struct strbuf buf = STRBUF_INIT;
1363
1364         for (line = buffer, eob = line + size;
1365              line < eob && *line != '\n';
1366              line = next) {
1367                 next = memchr(line, '\n', eob - line);
1368                 next = next ? next + 1 : eob;
1369                 if (*line == ' ') {
1370                         /* continuation */
1371                         if (it)
1372                                 strbuf_add(&buf, line + 1, next - (line + 1));
1373                         continue;
1374                 }
1375                 if (it)
1376                         it->value = strbuf_detach(&buf, &it->len);
1377                 strbuf_reset(&buf);
1378                 it = NULL;
1379
1380                 eof = memchr(line, ' ', next - line);
1381                 if (!eof)
1382                         eof = next;
1383                 else if (standard_header_field(line, eof - line) ||
1384                          excluded_header_field(line, eof - line, exclude))
1385                         continue;
1386
1387                 CALLOC_ARRAY(it, 1);
1388                 it->key = xmemdupz(line, eof-line);
1389                 *tail = it;
1390                 tail = &it->next;
1391                 if (eof + 1 < next)
1392                         strbuf_add(&buf, eof + 1, next - (eof + 1));
1393         }
1394         if (it)
1395                 it->value = strbuf_detach(&buf, &it->len);
1396         return extra;
1397 }
1398
1399 void free_commit_extra_headers(struct commit_extra_header *extra)
1400 {
1401         while (extra) {
1402                 struct commit_extra_header *next = extra->next;
1403                 free(extra->key);
1404                 free(extra->value);
1405                 free(extra);
1406                 extra = next;
1407         }
1408 }
1409
1410 int commit_tree(const char *msg, size_t msg_len, const struct object_id *tree,
1411                 struct commit_list *parents, struct object_id *ret,
1412                 const char *author, const char *sign_commit)
1413 {
1414         struct commit_extra_header *extra = NULL, **tail = &extra;
1415         int result;
1416
1417         append_merge_tag_headers(parents, &tail);
1418         result = commit_tree_extended(msg, msg_len, tree, parents, ret, author,
1419                                       NULL, sign_commit, extra);
1420         free_commit_extra_headers(extra);
1421         return result;
1422 }
1423
1424 static int find_invalid_utf8(const char *buf, int len)
1425 {
1426         int offset = 0;
1427         static const unsigned int max_codepoint[] = {
1428                 0x7f, 0x7ff, 0xffff, 0x10ffff
1429         };
1430
1431         while (len) {
1432                 unsigned char c = *buf++;
1433                 int bytes, bad_offset;
1434                 unsigned int codepoint;
1435                 unsigned int min_val, max_val;
1436
1437                 len--;
1438                 offset++;
1439
1440                 /* Simple US-ASCII? No worries. */
1441                 if (c < 0x80)
1442                         continue;
1443
1444                 bad_offset = offset-1;
1445
1446                 /*
1447                  * Count how many more high bits set: that's how
1448                  * many more bytes this sequence should have.
1449                  */
1450                 bytes = 0;
1451                 while (c & 0x40) {
1452                         c <<= 1;
1453                         bytes++;
1454                 }
1455
1456                 /*
1457                  * Must be between 1 and 3 more bytes.  Longer sequences result in
1458                  * codepoints beyond U+10FFFF, which are guaranteed never to exist.
1459                  */
1460                 if (bytes < 1 || 3 < bytes)
1461                         return bad_offset;
1462
1463                 /* Do we *have* that many bytes? */
1464                 if (len < bytes)
1465                         return bad_offset;
1466
1467                 /*
1468                  * Place the encoded bits at the bottom of the value and compute the
1469                  * valid range.
1470                  */
1471                 codepoint = (c & 0x7f) >> bytes;
1472                 min_val = max_codepoint[bytes-1] + 1;
1473                 max_val = max_codepoint[bytes];
1474
1475                 offset += bytes;
1476                 len -= bytes;
1477
1478                 /* And verify that they are good continuation bytes */
1479                 do {
1480                         codepoint <<= 6;
1481                         codepoint |= *buf & 0x3f;
1482                         if ((*buf++ & 0xc0) != 0x80)
1483                                 return bad_offset;
1484                 } while (--bytes);
1485
1486                 /* Reject codepoints that are out of range for the sequence length. */
1487                 if (codepoint < min_val || codepoint > max_val)
1488                         return bad_offset;
1489                 /* Surrogates are only for UTF-16 and cannot be encoded in UTF-8. */
1490                 if ((codepoint & 0x1ff800) == 0xd800)
1491                         return bad_offset;
1492                 /* U+xxFFFE and U+xxFFFF are guaranteed non-characters. */
1493                 if ((codepoint & 0xfffe) == 0xfffe)
1494                         return bad_offset;
1495                 /* So are anything in the range U+FDD0..U+FDEF. */
1496                 if (codepoint >= 0xfdd0 && codepoint <= 0xfdef)
1497                         return bad_offset;
1498         }
1499         return -1;
1500 }
1501
1502 /*
1503  * This verifies that the buffer is in proper utf8 format.
1504  *
1505  * If it isn't, it assumes any non-utf8 characters are Latin1,
1506  * and does the conversion.
1507  */
1508 static int verify_utf8(struct strbuf *buf)
1509 {
1510         int ok = 1;
1511         long pos = 0;
1512
1513         for (;;) {
1514                 int bad;
1515                 unsigned char c;
1516                 unsigned char replace[2];
1517
1518                 bad = find_invalid_utf8(buf->buf + pos, buf->len - pos);
1519                 if (bad < 0)
1520                         return ok;
1521                 pos += bad;
1522                 ok = 0;
1523                 c = buf->buf[pos];
1524                 strbuf_remove(buf, pos, 1);
1525
1526                 /* We know 'c' must be in the range 128-255 */
1527                 replace[0] = 0xc0 + (c >> 6);
1528                 replace[1] = 0x80 + (c & 0x3f);
1529                 strbuf_insert(buf, pos, replace, 2);
1530                 pos += 2;
1531         }
1532 }
1533
1534 static const char commit_utf8_warn[] =
1535 N_("Warning: commit message did not conform to UTF-8.\n"
1536    "You may want to amend it after fixing the message, or set the config\n"
1537    "variable i18n.commitEncoding to the encoding your project uses.\n");
1538
1539 int commit_tree_extended(const char *msg, size_t msg_len,
1540                          const struct object_id *tree,
1541                          struct commit_list *parents, struct object_id *ret,
1542                          const char *author, const char *committer,
1543                          const char *sign_commit,
1544                          struct commit_extra_header *extra)
1545 {
1546         int result;
1547         int encoding_is_utf8;
1548         struct strbuf buffer;
1549
1550         assert_oid_type(tree, OBJ_TREE);
1551
1552         if (memchr(msg, '\0', msg_len))
1553                 return error("a NUL byte in commit log message not allowed.");
1554
1555         /* Not having i18n.commitencoding is the same as having utf-8 */
1556         encoding_is_utf8 = is_encoding_utf8(git_commit_encoding);
1557
1558         strbuf_init(&buffer, 8192); /* should avoid reallocs for the headers */
1559         strbuf_addf(&buffer, "tree %s\n", oid_to_hex(tree));
1560
1561         /*
1562          * NOTE! This ordering means that the same exact tree merged with a
1563          * different order of parents will be a _different_ changeset even
1564          * if everything else stays the same.
1565          */
1566         while (parents) {
1567                 struct commit *parent = pop_commit(&parents);
1568                 strbuf_addf(&buffer, "parent %s\n",
1569                             oid_to_hex(&parent->object.oid));
1570         }
1571
1572         /* Person/date information */
1573         if (!author)
1574                 author = git_author_info(IDENT_STRICT);
1575         strbuf_addf(&buffer, "author %s\n", author);
1576         if (!committer)
1577                 committer = git_committer_info(IDENT_STRICT);
1578         strbuf_addf(&buffer, "committer %s\n", committer);
1579         if (!encoding_is_utf8)
1580                 strbuf_addf(&buffer, "encoding %s\n", git_commit_encoding);
1581
1582         while (extra) {
1583                 add_extra_header(&buffer, extra);
1584                 extra = extra->next;
1585         }
1586         strbuf_addch(&buffer, '\n');
1587
1588         /* And add the comment */
1589         strbuf_add(&buffer, msg, msg_len);
1590
1591         /* And check the encoding */
1592         if (encoding_is_utf8 && !verify_utf8(&buffer))
1593                 fprintf(stderr, _(commit_utf8_warn));
1594
1595         if (sign_commit && sign_with_header(&buffer, sign_commit)) {
1596                 result = -1;
1597                 goto out;
1598         }
1599
1600         result = write_object_file(buffer.buf, buffer.len, OBJ_COMMIT, ret);
1601 out:
1602         strbuf_release(&buffer);
1603         return result;
1604 }
1605
1606 define_commit_slab(merge_desc_slab, struct merge_remote_desc *);
1607 static struct merge_desc_slab merge_desc_slab = COMMIT_SLAB_INIT(1, merge_desc_slab);
1608
1609 struct merge_remote_desc *merge_remote_util(struct commit *commit)
1610 {
1611         return *merge_desc_slab_at(&merge_desc_slab, commit);
1612 }
1613
1614 void set_merge_remote_desc(struct commit *commit,
1615                            const char *name, struct object *obj)
1616 {
1617         struct merge_remote_desc *desc;
1618         FLEX_ALLOC_STR(desc, name, name);
1619         desc->obj = obj;
1620         *merge_desc_slab_at(&merge_desc_slab, commit) = desc;
1621 }
1622
1623 struct commit *get_merge_parent(const char *name)
1624 {
1625         struct object *obj;
1626         struct commit *commit;
1627         struct object_id oid;
1628         if (get_oid(name, &oid))
1629                 return NULL;
1630         obj = parse_object(the_repository, &oid);
1631         commit = (struct commit *)peel_to_type(name, 0, obj, OBJ_COMMIT);
1632         if (commit && !merge_remote_util(commit))
1633                 set_merge_remote_desc(commit, name, obj);
1634         return commit;
1635 }
1636
1637 /*
1638  * Append a commit to the end of the commit_list.
1639  *
1640  * next starts by pointing to the variable that holds the head of an
1641  * empty commit_list, and is updated to point to the "next" field of
1642  * the last item on the list as new commits are appended.
1643  *
1644  * Usage example:
1645  *
1646  *     struct commit_list *list;
1647  *     struct commit_list **next = &list;
1648  *
1649  *     next = commit_list_append(c1, next);
1650  *     next = commit_list_append(c2, next);
1651  *     assert(commit_list_count(list) == 2);
1652  *     return list;
1653  */
1654 struct commit_list **commit_list_append(struct commit *commit,
1655                                         struct commit_list **next)
1656 {
1657         struct commit_list *new_commit = xmalloc(sizeof(struct commit_list));
1658         new_commit->item = commit;
1659         *next = new_commit;
1660         new_commit->next = NULL;
1661         return &new_commit->next;
1662 }
1663
1664 const char *find_header_mem(const char *msg, size_t len,
1665                         const char *key, size_t *out_len)
1666 {
1667         int key_len = strlen(key);
1668         const char *line = msg;
1669
1670         /*
1671          * NEEDSWORK: It's possible for strchrnul() to scan beyond the range
1672          * given by len. However, current callers are safe because they compute
1673          * len by scanning a NUL-terminated block of memory starting at msg.
1674          * Nonetheless, it would be better to ensure the function does not look
1675          * at msg beyond the len provided by the caller.
1676          */
1677         while (line && line < msg + len) {
1678                 const char *eol = strchrnul(line, '\n');
1679
1680                 if (line == eol)
1681                         return NULL;
1682
1683                 if (eol - line > key_len &&
1684                     !strncmp(line, key, key_len) &&
1685                     line[key_len] == ' ') {
1686                         *out_len = eol - line - key_len - 1;
1687                         return line + key_len + 1;
1688                 }
1689                 line = *eol ? eol + 1 : NULL;
1690         }
1691         return NULL;
1692 }
1693
1694 const char *find_commit_header(const char *msg, const char *key, size_t *out_len)
1695 {
1696         return find_header_mem(msg, strlen(msg), key, out_len);
1697 }
1698 /*
1699  * Inspect the given string and determine the true "end" of the log message, in
1700  * order to find where to put a new Signed-off-by trailer.  Ignored are
1701  * trailing comment lines and blank lines.  To support "git commit -s
1702  * --amend" on an existing commit, we also ignore "Conflicts:".  To
1703  * support "git commit -v", we truncate at cut lines.
1704  *
1705  * Returns the number of bytes from the tail to ignore, to be fed as
1706  * the second parameter to append_signoff().
1707  */
1708 size_t ignore_non_trailer(const char *buf, size_t len)
1709 {
1710         size_t boc = 0;
1711         size_t bol = 0;
1712         int in_old_conflicts_block = 0;
1713         size_t cutoff = wt_status_locate_end(buf, len);
1714
1715         while (bol < cutoff) {
1716                 const char *next_line = memchr(buf + bol, '\n', len - bol);
1717
1718                 if (!next_line)
1719                         next_line = buf + len;
1720                 else
1721                         next_line++;
1722
1723                 if (buf[bol] == comment_line_char || buf[bol] == '\n') {
1724                         /* is this the first of the run of comments? */
1725                         if (!boc)
1726                                 boc = bol;
1727                         /* otherwise, it is just continuing */
1728                 } else if (starts_with(buf + bol, "Conflicts:\n")) {
1729                         in_old_conflicts_block = 1;
1730                         if (!boc)
1731                                 boc = bol;
1732                 } else if (in_old_conflicts_block && buf[bol] == '\t') {
1733                         ; /* a pathname in the conflicts block */
1734                 } else if (boc) {
1735                         /* the previous was not trailing comment */
1736                         boc = 0;
1737                         in_old_conflicts_block = 0;
1738                 }
1739                 bol = next_line - buf;
1740         }
1741         return boc ? len - boc : len - cutoff;
1742 }
1743
1744 int run_commit_hook(int editor_is_used, const char *index_file,
1745                     int *invoked_hook, const char *name, ...)
1746 {
1747         struct run_hooks_opt opt = RUN_HOOKS_OPT_INIT;
1748         va_list args;
1749         const char *arg;
1750
1751         strvec_pushf(&opt.env, "GIT_INDEX_FILE=%s", index_file);
1752
1753         /*
1754          * Let the hook know that no editor will be launched.
1755          */
1756         if (!editor_is_used)
1757                 strvec_push(&opt.env, "GIT_EDITOR=:");
1758
1759         va_start(args, name);
1760         while ((arg = va_arg(args, const char *)))
1761                 strvec_push(&opt.args, arg);
1762         va_end(args);
1763
1764         opt.invoked_hook = invoked_hook;
1765         return run_hooks_opt(name, &opt);
1766 }