bisect.c

   1 #include "cache.h"
   2 #include "commit.h"
   3 #include "diff.h"
   4 #include "revision.h"
   5 #include "refs.h"
   6 #include "list-objects.h"
   7 #include "quote.h"
   8 #include "sha1-lookup.h"
   9 #include "run-command.h"
  10 #include "log-tree.h"
  11 #include "bisect.h"
  12 #include "sha1-array.h"
  13 #include "argv-array.h"
  14
  15 static struct oid_array good_revs;
  16 static struct oid_array skipped_revs;
  17
  18 static struct object_id *current_bad_oid;
  19
  20 static const char *argv_checkout[] = {"checkout", "-q", NULL, "--", NULL};
  21 static const char *argv_show_branch[] = {"show-branch", NULL, NULL};
  22
  23 static const char *term_bad;
  24 static const char *term_good;
  25
  26 /* Remember to update object flag allocation in object.h */
  27 #define COUNTED         (1u<<16)
  28
  29 /*
  30  * This is a truly stupid algorithm, but it's only
  31  * used for bisection, and we just don't care enough.
  32  *
  33  * We care just barely enough to avoid recursing for
  34  * non-merge entries.
  35  */
  36 static int count_distance(struct commit_list *entry)
  37 {
  38         int nr = 0;
  39
  40         while (entry) {
  41                 struct commit *commit = entry->item;
  42                 struct commit_list *p;
  43
  44                 if (commit->object.flags & (UNINTERESTING | COUNTED))
  45                         break;
  46                 if (!(commit->object.flags & TREESAME))
  47                         nr++;
  48                 commit->object.flags |= COUNTED;
  49                 p = commit->parents;
  50                 entry = p;
  51                 if (p) {
  52                         p = p->next;
  53                         while (p) {
  54                                 nr += count_distance(p);
  55                                 p = p->next;
  56                         }
  57                 }
  58         }
  59
  60         return nr;
  61 }
  62
  63 static void clear_distance(struct commit_list *list)
  64 {
  65         while (list) {
  66                 struct commit *commit = list->item;
  67                 commit->object.flags &= ~COUNTED;
  68                 list = list->next;
  69         }
  70 }
  71
  72 #define DEBUG_BISECT 0
  73
  74 static inline int weight(struct commit_list *elem)
  75 {
  76         return *((int*)(elem->item->util));
  77 }
  78
  79 static inline void weight_set(struct commit_list *elem, int weight)
  80 {
  81         *((int*)(elem->item->util)) = weight;
  82 }
  83
  84 static int count_interesting_parents(struct commit *commit)
  85 {
  86         struct commit_list *p;
  87         int count;
  88
  89         for (count = 0, p = commit->parents; p; p = p->next) {
  90                 if (p->item->object.flags & UNINTERESTING)
  91                         continue;
  92                 count++;
  93         }
  94         return count;
  95 }
  96
  97 static inline int halfway(struct commit_list *p, int nr)
  98 {
  99         /*
 100          * Don't short-cut something we are not going to return!
 101          */
 102         if (p->item->object.flags & TREESAME)
 103                 return 0;
 104         if (DEBUG_BISECT)
 105                 return 0;
 106         /*
 107          * 2 and 3 are halfway of 5.
 108          * 3 is halfway of 6 but 2 and 4 are not.
 109          */
 110         switch (2 * weight(p) - nr) {
 111         case -1: case 0: case 1:
 112                 return 1;
 113         default:
 114                 return 0;
 115         }
 116 }
 117
 118 #if !DEBUG_BISECT
 119 #define show_list(a,b,c,d) do { ; } while (0)
 120 #else
 121 static void show_list(const char *debug, int counted, int nr,
 122                       struct commit_list *list)
 123 {
 124         struct commit_list *p;
 125
 126         fprintf(stderr, "%s (%d/%d)\n", debug, counted, nr);
 127
 128         for (p = list; p; p = p->next) {
 129                 struct commit_list *pp;
 130                 struct commit *commit = p->item;
 131                 unsigned flags = commit->object.flags;
 132                 enum object_type type;
 133                 unsigned long size;
 134                 char *buf = read_sha1_file(commit->object.sha1, &type, &size);
 135                 const char *subject_start;
 136                 int subject_len;
 137
 138                 fprintf(stderr, "%c%c%c ",
 139                         (flags & TREESAME) ? ' ' : 'T',
 140                         (flags & UNINTERESTING) ? 'U' : ' ',
 141                         (flags & COUNTED) ? 'C' : ' ');
 142                 if (commit->util)
 143                         fprintf(stderr, "%3d", weight(p));
 144                 else
 145                         fprintf(stderr, "---");
 146                 fprintf(stderr, " %.*s", 8, sha1_to_hex(commit->object.sha1));
 147                 for (pp = commit->parents; pp; pp = pp->next)
 148                         fprintf(stderr, " %.*s", 8,
 149                                 sha1_to_hex(pp->item->object.sha1));
 150
 151                 subject_len = find_commit_subject(buf, &subject_start);
 152                 if (subject_len)
 153                         fprintf(stderr, " %.*s", subject_len, subject_start);
 154                 fprintf(stderr, "\n");
 155         }
 156 }
 157 #endif /* DEBUG_BISECT */
 158
 159 static struct commit_list *best_bisection(struct commit_list *list, int nr)
 160 {
 161         struct commit_list *p, *best;
 162         int best_distance = -1;
 163
 164         best = list;
 165         for (p = list; p; p = p->next) {
 166                 int distance;
 167                 unsigned flags = p->item->object.flags;
 168
 169                 if (flags & TREESAME)
 170                         continue;
 171                 distance = weight(p);
 172                 if (nr - distance < distance)
 173                         distance = nr - distance;
 174                 if (distance > best_distance) {
 175                         best = p;
 176                         best_distance = distance;
 177                 }
 178         }
 179
 180         return best;
 181 }
 182
 183 struct commit_dist {
 184         struct commit *commit;
 185         int distance;
 186 };
 187
 188 static int compare_commit_dist(const void *a_, const void *b_)
 189 {
 190         struct commit_dist *a, *b;
 191
 192         a = (struct commit_dist *)a_;
 193         b = (struct commit_dist *)b_;
 194         if (a->distance != b->distance)
 195                 return b->distance - a->distance; /* desc sort */
 196         return oidcmp(&a->commit->object.oid, &b->commit->object.oid);
 197 }
 198
 199 static struct commit_list *best_bisection_sorted(struct commit_list *list, int nr)
 200 {
 201         struct commit_list *p;
 202         struct commit_dist *array = xcalloc(nr, sizeof(*array));
 203         struct strbuf buf = STRBUF_INIT;
 204         int cnt, i;
 205
 206         for (p = list, cnt = 0; p; p = p->next) {
 207                 int distance;
 208                 unsigned flags = p->item->object.flags;
 209
 210                 if (flags & TREESAME)
 211                         continue;
 212                 distance = weight(p);
 213                 if (nr - distance < distance)
 214                         distance = nr - distance;
 215                 array[cnt].commit = p->item;
 216                 array[cnt].distance = distance;
 217                 cnt++;
 218         }
 219         QSORT(array, cnt, compare_commit_dist);
 220         for (p = list, i = 0; i < cnt; i++) {
 221                 struct object *obj = &(array[i].commit->object);
 222
 223                 strbuf_reset(&buf);
 224                 strbuf_addf(&buf, "dist=%d", array[i].distance);
 225                 add_name_decoration(DECORATION_NONE, buf.buf, obj);
 226
 227                 p->item = array[i].commit;
 228                 p = p->next;
 229         }
 230         if (p)
 231                 p->next = NULL;
 232         strbuf_release(&buf);
 233         free(array);
 234         return list;
 235 }
 236
 237 /*
 238  * zero or positive weight is the number of interesting commits it can
 239  * reach, including itself.  Especially, weight = 0 means it does not
 240  * reach any tree-changing commits (e.g. just above uninteresting one
 241  * but traversal is with pathspec).
 242  *
 243  * weight = -1 means it has one parent and its distance is yet to
 244  * be computed.
 245  *
 246  * weight = -2 means it has more than one parent and its distance is
 247  * unknown.  After running count_distance() first, they will get zero
 248  * or positive distance.
 249  */
 250 static struct commit_list *do_find_bisection(struct commit_list *list,
 251                                              int nr, int *weights,
 252                                              int find_all)
 253 {
 254         int n, counted;
 255         struct commit_list *p;
 256
 257         counted = 0;
 258
 259         for (n = 0, p = list; p; p = p->next) {
 260                 struct commit *commit = p->item;
 261                 unsigned flags = commit->object.flags;
 262
 263                 p->item->util = &weights[n++];
 264                 switch (count_interesting_parents(commit)) {
 265                 case 0:
 266                         if (!(flags & TREESAME)) {
 267                                 weight_set(p, 1);
 268                                 counted++;
 269                                 show_list("bisection 2 count one",
 270                                           counted, nr, list);
 271                         }
 272                         /*
 273                          * otherwise, it is known not to reach any
 274                          * tree-changing commit and gets weight 0.
 275                          */
 276                         break;
 277                 case 1:
 278                         weight_set(p, -1);
 279                         break;
 280                 default:
 281                         weight_set(p, -2);
 282                         break;
 283                 }
 284         }
 285
 286         show_list("bisection 2 initialize", counted, nr, list);
 287
 288         /*
 289          * If you have only one parent in the resulting set
 290          * then you can reach one commit more than that parent
 291          * can reach.  So we do not have to run the expensive
 292          * count_distance() for single strand of pearls.
 293          *
 294          * However, if you have more than one parents, you cannot
 295          * just add their distance and one for yourself, since
 296          * they usually reach the same ancestor and you would
 297          * end up counting them twice that way.
 298          *
 299          * So we will first count distance of merges the usual
 300          * way, and then fill the blanks using cheaper algorithm.
 301          */
 302         for (p = list; p; p = p->next) {
 303                 if (p->item->object.flags & UNINTERESTING)
 304                         continue;
 305                 if (weight(p) != -2)
 306                         continue;
 307                 weight_set(p, count_distance(p));
 308                 clear_distance(list);
 309
 310                 /* Does it happen to be at exactly half-way? */
 311                 if (!find_all && halfway(p, nr))
 312                         return p;
 313                 counted++;
 314         }
 315
 316         show_list("bisection 2 count_distance", counted, nr, list);
 317
 318         while (counted < nr) {
 319                 for (p = list; p; p = p->next) {
 320                         struct commit_list *q;
 321                         unsigned flags = p->item->object.flags;
 322
 323                         if (0 <= weight(p))
 324                                 continue;
 325                         for (q = p->item->parents; q; q = q->next) {
 326                                 if (q->item->object.flags & UNINTERESTING)
 327                                         continue;
 328                                 if (0 <= weight(q))
 329                                         break;
 330                         }
 331                         if (!q)
 332                                 continue;
 333
 334                         /*
 335                          * weight for p is unknown but q is known.
 336                          * add one for p itself if p is to be counted,
 337                          * otherwise inherit it from q directly.
 338                          */
 339                         if (!(flags & TREESAME)) {
 340                                 weight_set(p, weight(q)+1);
 341                                 counted++;
 342                                 show_list("bisection 2 count one",
 343                                           counted, nr, list);
 344                         }
 345                         else
 346                                 weight_set(p, weight(q));
 347
 348                         /* Does it happen to be at exactly half-way? */
 349                         if (!find_all && halfway(p, nr))
 350                                 return p;
 351                 }
 352         }
 353
 354         show_list("bisection 2 counted all", counted, nr, list);
 355
 356         if (!find_all)
 357                 return best_bisection(list, nr);
 358         else
 359                 return best_bisection_sorted(list, nr);
 360 }
 361
 362 struct commit_list *find_bisection(struct commit_list *list,
 363                                           int *reaches, int *all,
 364                                           int find_all)
 365 {
 366         int nr, on_list;
 367         struct commit_list *p, *best, *next, *last;
 368         int *weights;
 369
 370         show_list("bisection 2 entry", 0, 0, list);
 371
 372         /*
 373          * Count the number of total and tree-changing items on the
 374          * list, while reversing the list.
 375          */
 376         for (nr = on_list = 0, last = NULL, p = list;
 377              p;
 378              p = next) {
 379                 unsigned flags = p->item->object.flags;
 380
 381                 next = p->next;
 382                 if (flags & UNINTERESTING)
 383                         continue;
 384                 p->next = last;
 385                 last = p;
 386                 if (!(flags & TREESAME))
 387                         nr++;
 388                 on_list++;
 389         }
 390         list = last;
 391         show_list("bisection 2 sorted", 0, nr, list);
 392
 393         *all = nr;
 394         weights = xcalloc(on_list, sizeof(*weights));
 395
 396         /* Do the real work of finding bisection commit. */
 397         best = do_find_bisection(list, nr, weights, find_all);
 398         if (best) {
 399                 if (!find_all)
 400                         best->next = NULL;
 401                 *reaches = weight(best);
 402         }
 403         free(weights);
 404         return best;
 405 }
 406
 407 static int register_ref(const char *refname, const struct object_id *oid,
 408                         int flags, void *cb_data)
 409 {
 410         struct strbuf good_prefix = STRBUF_INIT;
 411         strbuf_addstr(&good_prefix, term_good);
 412         strbuf_addstr(&good_prefix, "-");
 413
 414         if (!strcmp(refname, term_bad)) {
 415                 current_bad_oid = xmalloc(sizeof(*current_bad_oid));
 416                 oidcpy(current_bad_oid, oid);
 417         } else if (starts_with(refname, good_prefix.buf)) {
 418                 oid_array_append(&good_revs, oid);
 419         } else if (starts_with(refname, "skip-")) {
 420                 oid_array_append(&skipped_revs, oid);
 421         }
 422
 423         strbuf_release(&good_prefix);
 424
 425         return 0;
 426 }
 427
 428 static int read_bisect_refs(void)
 429 {
 430         return for_each_ref_in("refs/bisect/", register_ref, NULL);
 431 }
 432
 433 static GIT_PATH_FUNC(git_path_bisect_names, "BISECT_NAMES")
 434 static GIT_PATH_FUNC(git_path_bisect_expected_rev, "BISECT_EXPECTED_REV")
 435 static GIT_PATH_FUNC(git_path_bisect_terms, "BISECT_TERMS")
 436
 437 static void read_bisect_paths(struct argv_array *array)
 438 {
 439         struct strbuf str = STRBUF_INIT;
 440         const char *filename = git_path_bisect_names();
 441         FILE *fp = xfopen(filename, "r");
 442
 443         while (strbuf_getline_lf(&str, fp) != EOF) {
 444                 strbuf_trim(&str);
 445                 if (sq_dequote_to_argv_array(str.buf, array))
 446                         die(_("Badly quoted content in file '%s': %s"),
 447                             filename, str.buf);
 448         }
 449
 450         strbuf_release(&str);
 451         fclose(fp);
 452 }
 453
 454 static char *join_sha1_array_hex(struct oid_array *array, char delim)
 455 {
 456         struct strbuf joined_hexs = STRBUF_INIT;
 457         int i;
 458
 459         for (i = 0; i < array->nr; i++) {
 460                 strbuf_addstr(&joined_hexs, oid_to_hex(array->oid + i));
 461                 if (i + 1 < array->nr)
 462                         strbuf_addch(&joined_hexs, delim);
 463         }
 464
 465         return strbuf_detach(&joined_hexs, NULL);
 466 }
 467
 468 /*
 469  * In this function, passing a not NULL skipped_first is very special.
 470  * It means that we want to know if the first commit in the list is
 471  * skipped because we will want to test a commit away from it if it is
 472  * indeed skipped.
 473  * So if the first commit is skipped, we cannot take the shortcut to
 474  * just "return list" when we find the first non skipped commit, we
 475  * have to return a fully filtered list.
 476  *
 477  * We use (*skipped_first == -1) to mean "it has been found that the
 478  * first commit is not skipped". In this case *skipped_first is set back
 479  * to 0 just before the function returns.
 480  */
 481 struct commit_list *filter_skipped(struct commit_list *list,
 482                                    struct commit_list **tried,
 483                                    int show_all,
 484                                    int *count,
 485                                    int *skipped_first)
 486 {
 487         struct commit_list *filtered = NULL, **f = &filtered;
 488
 489         *tried = NULL;
 490
 491         if (skipped_first)
 492                 *skipped_first = 0;
 493         if (count)
 494                 *count = 0;
 495
 496         if (!skipped_revs.nr)
 497                 return list;
 498
 499         while (list) {
 500                 struct commit_list *next = list->next;
 501                 list->next = NULL;
 502                 if (0 <= oid_array_lookup(&skipped_revs, &list->item->object.oid)) {
 503                         if (skipped_first && !*skipped_first)
 504                                 *skipped_first = 1;
 505                         /* Move current to tried list */
 506                         *tried = list;
 507                         tried = &list->next;
 508                 } else {
 509                         if (!show_all) {
 510                                 if (!skipped_first || !*skipped_first)
 511                                         return list;
 512                         } else if (skipped_first && !*skipped_first) {
 513                                 /* This means we know it's not skipped */
 514                                 *skipped_first = -1;
 515                         }
 516                         /* Move current to filtered list */
 517                         *f = list;
 518                         f = &list->next;
 519                         if (count)
 520                                 (*count)++;
 521                 }
 522                 list = next;
 523         }
 524
 525         if (skipped_first && *skipped_first == -1)
 526                 *skipped_first = 0;
 527
 528         return filtered;
 529 }
 530
 531 #define PRN_MODULO 32768
 532
 533 /*
 534  * This is a pseudo random number generator based on "man 3 rand".
 535  * It is not used properly because the seed is the argument and it
 536  * is increased by one between each call, but that should not matter
 537  * for this application.
 538  */
 539 static unsigned get_prn(unsigned count) {
 540         count = count * 1103515245 + 12345;
 541         return (count/65536) % PRN_MODULO;
 542 }
 543
 544 /*
 545  * Custom integer square root from
 546  * https://en.wikipedia.org/wiki/Integer_square_root
 547  */
 548 static int sqrti(int val)
 549 {
 550         float d, x = val;
 551
 552         if (val == 0)
 553                 return 0;
 554
 555         do {
 556                 float y = (x + (float)val / x) / 2;
 557                 d = (y > x) ? y - x : x - y;
 558                 x = y;
 559         } while (d >= 0.5);
 560
 561         return (int)x;
 562 }
 563
 564 static struct commit_list *skip_away(struct commit_list *list, int count)
 565 {
 566         struct commit_list *cur, *previous;
 567         int prn, index, i;
 568
 569         prn = get_prn(count);
 570         index = (count * prn / PRN_MODULO) * sqrti(prn) / sqrti(PRN_MODULO);
 571
 572         cur = list;
 573         previous = NULL;
 574
 575         for (i = 0; cur; cur = cur->next, i++) {
 576                 if (i == index) {
 577                         if (oidcmp(&cur->item->object.oid, current_bad_oid))
 578                                 return cur;
 579                         if (previous)
 580                                 return previous;
 581                         return list;
 582                 }
 583                 previous = cur;
 584         }
 585
 586         return list;
 587 }
 588
 589 static struct commit_list *managed_skipped(struct commit_list *list,
 590                                            struct commit_list **tried)
 591 {
 592         int count, skipped_first;
 593
 594         *tried = NULL;
 595
 596         if (!skipped_revs.nr)
 597                 return list;
 598
 599         list = filter_skipped(list, tried, 0, &count, &skipped_first);
 600
 601         if (!skipped_first)
 602                 return list;
 603
 604         return skip_away(list, count);
 605 }
 606
 607 static void bisect_rev_setup(struct rev_info *revs, const char *prefix,
 608                              const char *bad_format, const char *good_format,
 609                              int read_paths)
 610 {
 611         struct argv_array rev_argv = ARGV_ARRAY_INIT;
 612         int i;
 613
 614         init_revisions(revs, prefix);
 615         revs->abbrev = 0;
 616         revs->commit_format = CMIT_FMT_UNSPECIFIED;
 617
 618         /* rev_argv.argv[0] will be ignored by setup_revisions */
 619         argv_array_push(&rev_argv, "bisect_rev_setup");
 620         argv_array_pushf(&rev_argv, bad_format, oid_to_hex(current_bad_oid));
 621         for (i = 0; i < good_revs.nr; i++)
 622                 argv_array_pushf(&rev_argv, good_format,
 623                                  oid_to_hex(good_revs.oid + i));
 624         argv_array_push(&rev_argv, "--");
 625         if (read_paths)
 626                 read_bisect_paths(&rev_argv);
 627
 628         setup_revisions(rev_argv.argc, rev_argv.argv, revs, NULL);
 629         /* XXX leak rev_argv, as "revs" may still be pointing to it */
 630 }
 631
 632 static void bisect_common(struct rev_info *revs)
 633 {
 634         if (prepare_revision_walk(revs))
 635                 die("revision walk setup failed");
 636         if (revs->tree_objects)
 637                 mark_edges_uninteresting(revs, NULL);
 638 }
 639
 640 static void exit_if_skipped_commits(struct commit_list *tried,
 641                                     const struct object_id *bad)
 642 {
 643         if (!tried)
 644                 return;
 645
 646         printf("There are only 'skip'ped commits left to test.\n"
 647                "The first %s commit could be any of:\n", term_bad);
 648
 649         for ( ; tried; tried = tried->next)
 650                 printf("%s\n", oid_to_hex(&tried->item->object.oid));
 651
 652         if (bad)
 653                 printf("%s\n", oid_to_hex(bad));
 654         printf(_("We cannot bisect more!\n"));
 655         exit(2);
 656 }
 657
 658 static int is_expected_rev(const struct object_id *oid)
 659 {
 660         const char *filename = git_path_bisect_expected_rev();
 661         struct stat st;
 662         struct strbuf str = STRBUF_INIT;
 663         FILE *fp;
 664         int res = 0;
 665
 666         if (stat(filename, &st) || !S_ISREG(st.st_mode))
 667                 return 0;
 668
 669         fp = fopen_or_warn(filename, "r");
 670         if (!fp)
 671                 return 0;
 672
 673         if (strbuf_getline_lf(&str, fp) != EOF)
 674                 res = !strcmp(str.buf, oid_to_hex(oid));
 675
 676         strbuf_release(&str);
 677         fclose(fp);
 678
 679         return res;
 680 }
 681
 682 static int bisect_checkout(const unsigned char *bisect_rev, int no_checkout)
 683 {
 684         char bisect_rev_hex[GIT_MAX_HEXSZ + 1];
 685
 686         memcpy(bisect_rev_hex, sha1_to_hex(bisect_rev), GIT_SHA1_HEXSZ + 1);
 687         update_ref(NULL, "BISECT_EXPECTED_REV", bisect_rev, NULL, 0, UPDATE_REFS_DIE_ON_ERR);
 688
 689         argv_checkout[2] = bisect_rev_hex;
 690         if (no_checkout) {
 691                 update_ref(NULL, "BISECT_HEAD", bisect_rev, NULL, 0, UPDATE_REFS_DIE_ON_ERR);
 692         } else {
 693                 int res;
 694                 res = run_command_v_opt(argv_checkout, RUN_GIT_CMD);
 695                 if (res)
 696                         exit(res);
 697         }
 698
 699         argv_show_branch[1] = bisect_rev_hex;
 700         return run_command_v_opt(argv_show_branch, RUN_GIT_CMD);
 701 }
 702
 703 static struct commit *get_commit_reference(const struct object_id *oid)
 704 {
 705         struct commit *r = lookup_commit_reference(oid);
 706         if (!r)
 707                 die(_("Not a valid commit name %s"), oid_to_hex(oid));
 708         return r;
 709 }
 710
 711 static struct commit **get_bad_and_good_commits(int *rev_nr)
 712 {
 713         struct commit **rev;
 714         int i, n = 0;
 715
 716         ALLOC_ARRAY(rev, 1 + good_revs.nr);
 717         rev[n++] = get_commit_reference(current_bad_oid);
 718         for (i = 0; i < good_revs.nr; i++)
 719                 rev[n++] = get_commit_reference(good_revs.oid + i);
 720         *rev_nr = n;
 721
 722         return rev;
 723 }
 724
 725 static void handle_bad_merge_base(void)
 726 {
 727         if (is_expected_rev(current_bad_oid)) {
 728                 char *bad_hex = oid_to_hex(current_bad_oid);
 729                 char *good_hex = join_sha1_array_hex(&good_revs, ' ');
 730                 if (!strcmp(term_bad, "bad") && !strcmp(term_good, "good")) {
 731                         fprintf(stderr, _("The merge base %s is bad.\n"
 732                                 "This means the bug has been fixed "
 733                                 "between %s and [%s].\n"),
 734                                 bad_hex, bad_hex, good_hex);
 735                 } else if (!strcmp(term_bad, "new") && !strcmp(term_good, "old")) {
 736                         fprintf(stderr, _("The merge base %s is new.\n"
 737                                 "The property has changed "
 738                                 "between %s and [%s].\n"),
 739                                 bad_hex, bad_hex, good_hex);
 740                 } else {
 741                         fprintf(stderr, _("The merge base %s is %s.\n"
 742                                 "This means the first '%s' commit is "
 743                                 "between %s and [%s].\n"),
 744                                 bad_hex, term_bad, term_good, bad_hex, good_hex);
 745                 }
 746                 exit(3);
 747         }
 748
 749         fprintf(stderr, _("Some %s revs are not ancestors of the %s rev.\n"
 750                 "git bisect cannot work properly in this case.\n"
 751                 "Maybe you mistook %s and %s revs?\n"),
 752                 term_good, term_bad, term_good, term_bad);
 753         exit(1);
 754 }
 755
 756 static void handle_skipped_merge_base(const struct object_id *mb)
 757 {
 758         char *mb_hex = oid_to_hex(mb);
 759         char *bad_hex = oid_to_hex(current_bad_oid);
 760         char *good_hex = join_sha1_array_hex(&good_revs, ' ');
 761
 762         warning(_("the merge base between %s and [%s] "
 763                 "must be skipped.\n"
 764                 "So we cannot be sure the first %s commit is "
 765                 "between %s and %s.\n"
 766                 "We continue anyway."),
 767                 bad_hex, good_hex, term_bad, mb_hex, bad_hex);
 768         free(good_hex);
 769 }
 770
 771 /*
 772  * "check_merge_bases" checks that merge bases are not "bad" (or "new").
 773  *
 774  * - If one is "bad" (or "new"), it means the user assumed something wrong
 775  * and we must exit with a non 0 error code.
 776  * - If one is "good" (or "old"), that's good, we have nothing to do.
 777  * - If one is "skipped", we can't know but we should warn.
 778  * - If we don't know, we should check it out and ask the user to test.
 779  */
 780 static void check_merge_bases(int no_checkout)
 781 {
 782         struct commit_list *result;
 783         int rev_nr;
 784         struct commit **rev = get_bad_and_good_commits(&rev_nr);
 785
 786         result = get_merge_bases_many(rev[0], rev_nr - 1, rev + 1);
 787
 788         for (; result; result = result->next) {
 789                 const struct object_id *mb = &result->item->object.oid;
 790                 if (!oidcmp(mb, current_bad_oid)) {
 791                         handle_bad_merge_base();
 792                 } else if (0 <= oid_array_lookup(&good_revs, mb)) {
 793                         continue;
 794                 } else if (0 <= oid_array_lookup(&skipped_revs, mb)) {
 795                         handle_skipped_merge_base(mb);
 796                 } else {
 797                         printf(_("Bisecting: a merge base must be tested\n"));
 798                         exit(bisect_checkout(mb->hash, no_checkout));
 799                 }
 800         }
 801
 802         free(rev);
 803         free_commit_list(result);
 804 }
 805
 806 static int check_ancestors(const char *prefix)
 807 {
 808         struct rev_info revs;
 809         struct object_array pending_copy;
 810         int res;
 811
 812         bisect_rev_setup(&revs, prefix, "^%s", "%s", 0);
 813
 814         /* Save pending objects, so they can be cleaned up later. */
 815         pending_copy = revs.pending;
 816         revs.leak_pending = 1;
 817
 818         /*
 819          * bisect_common calls prepare_revision_walk right away, which
 820          * (together with .leak_pending = 1) makes us the sole owner of
 821          * the list of pending objects.
 822          */
 823         bisect_common(&revs);
 824         res = (revs.commits != NULL);
 825
 826         /* Clean up objects used, as they will be reused. */
 827         clear_commit_marks_for_object_array(&pending_copy, ALL_REV_FLAGS);
 828         free(pending_copy.objects);
 829
 830         return res;
 831 }
 832
 833 /*
 834  * "check_good_are_ancestors_of_bad" checks that all "good" revs are
 835  * ancestor of the "bad" rev.
 836  *
 837  * If that's not the case, we need to check the merge bases.
 838  * If a merge base must be tested by the user, its source code will be
 839  * checked out to be tested by the user and we will exit.
 840  */
 841 static void check_good_are_ancestors_of_bad(const char *prefix, int no_checkout)
 842 {
 843         char *filename = git_pathdup("BISECT_ANCESTORS_OK");
 844         struct stat st;
 845         int fd;
 846
 847         if (!current_bad_oid)
 848                 die(_("a %s revision is needed"), term_bad);
 849
 850         /* Check if file BISECT_ANCESTORS_OK exists. */
 851         if (!stat(filename, &st) && S_ISREG(st.st_mode))
 852                 goto done;
 853
 854         /* Bisecting with no good rev is ok. */
 855         if (good_revs.nr == 0)
 856                 goto done;
 857
 858         /* Check if all good revs are ancestor of the bad rev. */
 859         if (check_ancestors(prefix))
 860                 check_merge_bases(no_checkout);
 861
 862         /* Create file BISECT_ANCESTORS_OK. */
 863         fd = open(filename, O_CREAT | O_TRUNC | O_WRONLY, 0600);
 864         if (fd < 0)
 865                 warning_errno(_("could not create file '%s'"),
 866                               filename);
 867         else
 868                 close(fd);
 869  done:
 870         free(filename);
 871 }
 872
 873 /*
 874  * This does "git diff-tree --pretty COMMIT" without one fork+exec.
 875  */
 876 static void show_diff_tree(const char *prefix, struct commit *commit)
 877 {
 878         struct rev_info opt;
 879
 880         /* diff-tree init */
 881         init_revisions(&opt, prefix);
 882         git_config(git_diff_basic_config, NULL); /* no "diff" UI options */
 883         opt.abbrev = 0;
 884         opt.diff = 1;
 885
 886         /* This is what "--pretty" does */
 887         opt.verbose_header = 1;
 888         opt.use_terminator = 0;
 889         opt.commit_format = CMIT_FMT_DEFAULT;
 890
 891         /* diff-tree init */
 892         if (!opt.diffopt.output_format)
 893                 opt.diffopt.output_format = DIFF_FORMAT_RAW;
 894
 895         setup_revisions(0, NULL, &opt, NULL);
 896         log_tree_commit(&opt, commit);
 897 }
 898
 899 /*
 900  * The terms used for this bisect session are stored in BISECT_TERMS.
 901  * We read them and store them to adapt the messages accordingly.
 902  * Default is bad/good.
 903  */
 904 void read_bisect_terms(const char **read_bad, const char **read_good)
 905 {
 906         struct strbuf str = STRBUF_INIT;
 907         const char *filename = git_path_bisect_terms();
 908         FILE *fp = fopen(filename, "r");
 909
 910         if (!fp) {
 911                 if (errno == ENOENT) {
 912                         *read_bad = "bad";
 913                         *read_good = "good";
 914                         return;
 915                 } else {
 916                         die_errno(_("could not read file '%s'"), filename);
 917                 }
 918         } else {
 919                 strbuf_getline_lf(&str, fp);
 920                 *read_bad = strbuf_detach(&str, NULL);
 921                 strbuf_getline_lf(&str, fp);
 922                 *read_good = strbuf_detach(&str, NULL);
 923         }
 924         strbuf_release(&str);
 925         fclose(fp);
 926 }
 927
 928 /*
 929  * We use the convention that exiting with an exit code 10 means that
 930  * the bisection process finished successfully.
 931  * In this case the calling shell script should exit 0.
 932  *
 933  * If no_checkout is non-zero, the bisection process does not
 934  * checkout the trial commit but instead simply updates BISECT_HEAD.
 935  */
 936 int bisect_next_all(const char *prefix, int no_checkout)
 937 {
 938         struct rev_info revs;
 939         struct commit_list *tried;
 940         int reaches = 0, all = 0, nr, steps;
 941         const unsigned char *bisect_rev;
 942         char *steps_msg;
 943
 944         read_bisect_terms(&term_bad, &term_good);
 945         if (read_bisect_refs())
 946                 die(_("reading bisect refs failed"));
 947
 948         check_good_are_ancestors_of_bad(prefix, no_checkout);
 949
 950         bisect_rev_setup(&revs, prefix, "%s", "^%s", 1);
 951         revs.limited = 1;
 952
 953         bisect_common(&revs);
 954
 955         revs.commits = find_bisection(revs.commits, &reaches, &all,
 956                                        !!skipped_revs.nr);
 957         revs.commits = managed_skipped(revs.commits, &tried);
 958
 959         if (!revs.commits) {
 960                 /*
 961                  * We should exit here only if the "bad"
 962                  * commit is also a "skip" commit.
 963                  */
 964                 exit_if_skipped_commits(tried, NULL);
 965
 966                 printf(_("%s was both %s and %s\n"),
 967                        oid_to_hex(current_bad_oid),
 968                        term_good,
 969                        term_bad);
 970                 exit(1);
 971         }
 972
 973         if (!all) {
 974                 fprintf(stderr, _("No testable commit found.\n"
 975                         "Maybe you started with bad path parameters?\n"));
 976                 exit(4);
 977         }
 978
 979         bisect_rev = revs.commits->item->object.oid.hash;
 980
 981         if (!hashcmp(bisect_rev, current_bad_oid->hash)) {
 982                 exit_if_skipped_commits(tried, current_bad_oid);
 983                 printf("%s is the first %s commit\n", sha1_to_hex(bisect_rev),
 984                         term_bad);
 985                 show_diff_tree(prefix, revs.commits->item);
 986                 /* This means the bisection process succeeded. */
 987                 exit(10);
 988         }
 989
 990         nr = all - reaches - 1;
 991         steps = estimate_bisect_steps(all);
 992
 993         steps_msg = xstrfmt(Q_("(roughly %d step)", "(roughly %d steps)",
 994                   steps), steps);
 995         /*
 996          * TRANSLATORS: the last %s will be replaced with "(roughly %d
 997          * steps)" translation.
 998          */
 999         printf(Q_("Bisecting: %d revision left to test after this %s\n",
1000                   "Bisecting: %d revisions left to test after this %s\n",
1001                   nr), nr, steps_msg);
1002         free(steps_msg);
1003
1004         return bisect_checkout(bisect_rev, no_checkout);
1005 }
1006
1007 static inline int log2i(int n)
1008 {
1009         int log2 = 0;
1010
1011         for (; n > 1; n >>= 1)
1012                 log2++;
1013
1014         return log2;
1015 }
1016
1017 static inline int exp2i(int n)
1018 {
1019         return 1 << n;
1020 }
1021
1022 /*
1023  * Estimate the number of bisect steps left (after the current step)
1024  *
1025  * For any x between 0 included and 2^n excluded, the probability for
1026  * n - 1 steps left looks like:
1027  *
1028  * P(2^n + x) == (2^n - x) / (2^n + x)
1029  *
1030  * and P(2^n + x) < 0.5 means 2^n < 3x
1031  */
1032 int estimate_bisect_steps(int all)
1033 {
1034         int n, x, e;
1035
1036         if (all < 3)
1037                 return 0;
1038
1039         n = log2i(all);
1040         e = exp2i(n);
1041         x = all - e;
1042
1043         return (e < 3 * x) ? n : n - 1;
1044 }