bisect.c

   1 #include "cache.h"
   2 #include "config.h"
   3 #include "commit.h"
   4 #include "diff.h"
   5 #include "revision.h"
   6 #include "refs.h"
   7 #include "list-objects.h"
   8 #include "quote.h"
   9 #include "sha1-lookup.h"
  10 #include "run-command.h"
  11 #include "log-tree.h"
  12 #include "bisect.h"
  13 #include "sha1-array.h"
  14 #include "argv-array.h"
  15
  16 static struct oid_array good_revs;
  17 static struct oid_array skipped_revs;
  18
  19 static struct object_id *current_bad_oid;
  20
  21 static const char *argv_checkout[] = {"checkout", "-q", NULL, "--", NULL};
  22 static const char *argv_show_branch[] = {"show-branch", NULL, NULL};
  23
  24 static const char *term_bad;
  25 static const char *term_good;
  26
  27 /* Remember to update object flag allocation in object.h */
  28 #define COUNTED         (1u<<16)
  29
  30 /*
  31  * This is a truly stupid algorithm, but it's only
  32  * used for bisection, and we just don't care enough.
  33  *
  34  * We care just barely enough to avoid recursing for
  35  * non-merge entries.
  36  */
  37 static int count_distance(struct commit_list *entry)
  38 {
  39         int nr = 0;
  40
  41         while (entry) {
  42                 struct commit *commit = entry->item;
  43                 struct commit_list *p;
  44
  45                 if (commit->object.flags & (UNINTERESTING | COUNTED))
  46                         break;
  47                 if (!(commit->object.flags & TREESAME))
  48                         nr++;
  49                 commit->object.flags |= COUNTED;
  50                 p = commit->parents;
  51                 entry = p;
  52                 if (p) {
  53                         p = p->next;
  54                         while (p) {
  55                                 nr += count_distance(p);
  56                                 p = p->next;
  57                         }
  58                 }
  59         }
  60
  61         return nr;
  62 }
  63
  64 static void clear_distance(struct commit_list *list)
  65 {
  66         while (list) {
  67                 struct commit *commit = list->item;
  68                 commit->object.flags &= ~COUNTED;
  69                 list = list->next;
  70         }
  71 }
  72
  73 #define DEBUG_BISECT 0
  74
  75 static inline int weight(struct commit_list *elem)
  76 {
  77         return *((int*)(elem->item->util));
  78 }
  79
  80 static inline void weight_set(struct commit_list *elem, int weight)
  81 {
  82         *((int*)(elem->item->util)) = weight;
  83 }
  84
  85 static int count_interesting_parents(struct commit *commit)
  86 {
  87         struct commit_list *p;
  88         int count;
  89
  90         for (count = 0, p = commit->parents; p; p = p->next) {
  91                 if (p->item->object.flags & UNINTERESTING)
  92                         continue;
  93                 count++;
  94         }
  95         return count;
  96 }
  97
  98 static inline int halfway(struct commit_list *p, int nr)
  99 {
 100         /*
 101          * Don't short-cut something we are not going to return!
 102          */
 103         if (p->item->object.flags & TREESAME)
 104                 return 0;
 105         if (DEBUG_BISECT)
 106                 return 0;
 107         /*
 108          * 2 and 3 are halfway of 5.
 109          * 3 is halfway of 6 but 2 and 4 are not.
 110          */
 111         switch (2 * weight(p) - nr) {
 112         case -1: case 0: case 1:
 113                 return 1;
 114         default:
 115                 return 0;
 116         }
 117 }
 118
 119 #if !DEBUG_BISECT
 120 #define show_list(a,b,c,d) do { ; } while (0)
 121 #else
 122 static void show_list(const char *debug, int counted, int nr,
 123                       struct commit_list *list)
 124 {
 125         struct commit_list *p;
 126
 127         fprintf(stderr, "%s (%d/%d)\n", debug, counted, nr);
 128
 129         for (p = list; p; p = p->next) {
 130                 struct commit_list *pp;
 131                 struct commit *commit = p->item;
 132                 unsigned flags = commit->object.flags;
 133                 enum object_type type;
 134                 unsigned long size;
 135                 char *buf = read_sha1_file(commit->object.oid.hash, &type, &size);
 136                 const char *subject_start;
 137                 int subject_len;
 138
 139                 fprintf(stderr, "%c%c%c ",
 140                         (flags & TREESAME) ? ' ' : 'T',
 141                         (flags & UNINTERESTING) ? 'U' : ' ',
 142                         (flags & COUNTED) ? 'C' : ' ');
 143                 if (commit->util)
 144                         fprintf(stderr, "%3d", weight(p));
 145                 else
 146                         fprintf(stderr, "---");
 147                 fprintf(stderr, " %.*s", 8, sha1_to_hex(commit->object.oid.hash));
 148                 for (pp = commit->parents; pp; pp = pp->next)
 149                         fprintf(stderr, " %.*s", 8,
 150                                 sha1_to_hex(pp->item->object.oid.hash));
 151
 152                 subject_len = find_commit_subject(buf, &subject_start);
 153                 if (subject_len)
 154                         fprintf(stderr, " %.*s", subject_len, subject_start);
 155                 fprintf(stderr, "\n");
 156         }
 157 }
 158 #endif /* DEBUG_BISECT */
 159
 160 static struct commit_list *best_bisection(struct commit_list *list, int nr)
 161 {
 162         struct commit_list *p, *best;
 163         int best_distance = -1;
 164
 165         best = list;
 166         for (p = list; p; p = p->next) {
 167                 int distance;
 168                 unsigned flags = p->item->object.flags;
 169
 170                 if (flags & TREESAME)
 171                         continue;
 172                 distance = weight(p);
 173                 if (nr - distance < distance)
 174                         distance = nr - distance;
 175                 if (distance > best_distance) {
 176                         best = p;
 177                         best_distance = distance;
 178                 }
 179         }
 180
 181         return best;
 182 }
 183
 184 struct commit_dist {
 185         struct commit *commit;
 186         int distance;
 187 };
 188
 189 static int compare_commit_dist(const void *a_, const void *b_)
 190 {
 191         struct commit_dist *a, *b;
 192
 193         a = (struct commit_dist *)a_;
 194         b = (struct commit_dist *)b_;
 195         if (a->distance != b->distance)
 196                 return b->distance - a->distance; /* desc sort */
 197         return oidcmp(&a->commit->object.oid, &b->commit->object.oid);
 198 }
 199
 200 static struct commit_list *best_bisection_sorted(struct commit_list *list, int nr)
 201 {
 202         struct commit_list *p;
 203         struct commit_dist *array = xcalloc(nr, sizeof(*array));
 204         struct strbuf buf = STRBUF_INIT;
 205         int cnt, i;
 206
 207         for (p = list, cnt = 0; p; p = p->next) {
 208                 int distance;
 209                 unsigned flags = p->item->object.flags;
 210
 211                 if (flags & TREESAME)
 212                         continue;
 213                 distance = weight(p);
 214                 if (nr - distance < distance)
 215                         distance = nr - distance;
 216                 array[cnt].commit = p->item;
 217                 array[cnt].distance = distance;
 218                 cnt++;
 219         }
 220         QSORT(array, cnt, compare_commit_dist);
 221         for (p = list, i = 0; i < cnt; i++) {
 222                 struct object *obj = &(array[i].commit->object);
 223
 224                 strbuf_reset(&buf);
 225                 strbuf_addf(&buf, "dist=%d", array[i].distance);
 226                 add_name_decoration(DECORATION_NONE, buf.buf, obj);
 227
 228                 p->item = array[i].commit;
 229                 if (i < cnt - 1)
 230                         p = p->next;
 231         }
 232         if (p) {
 233                 free_commit_list(p->next);
 234                 p->next = NULL;
 235         }
 236         strbuf_release(&buf);
 237         free(array);
 238         return list;
 239 }
 240
 241 /*
 242  * zero or positive weight is the number of interesting commits it can
 243  * reach, including itself.  Especially, weight = 0 means it does not
 244  * reach any tree-changing commits (e.g. just above uninteresting one
 245  * but traversal is with pathspec).
 246  *
 247  * weight = -1 means it has one parent and its distance is yet to
 248  * be computed.
 249  *
 250  * weight = -2 means it has more than one parent and its distance is
 251  * unknown.  After running count_distance() first, they will get zero
 252  * or positive distance.
 253  */
 254 static struct commit_list *do_find_bisection(struct commit_list *list,
 255                                              int nr, int *weights,
 256                                              int find_all)
 257 {
 258         int n, counted;
 259         struct commit_list *p;
 260
 261         counted = 0;
 262
 263         for (n = 0, p = list; p; p = p->next) {
 264                 struct commit *commit = p->item;
 265                 unsigned flags = commit->object.flags;
 266
 267                 p->item->util = &weights[n++];
 268                 switch (count_interesting_parents(commit)) {
 269                 case 0:
 270                         if (!(flags & TREESAME)) {
 271                                 weight_set(p, 1);
 272                                 counted++;
 273                                 show_list("bisection 2 count one",
 274                                           counted, nr, list);
 275                         }
 276                         /*
 277                          * otherwise, it is known not to reach any
 278                          * tree-changing commit and gets weight 0.
 279                          */
 280                         break;
 281                 case 1:
 282                         weight_set(p, -1);
 283                         break;
 284                 default:
 285                         weight_set(p, -2);
 286                         break;
 287                 }
 288         }
 289
 290         show_list("bisection 2 initialize", counted, nr, list);
 291
 292         /*
 293          * If you have only one parent in the resulting set
 294          * then you can reach one commit more than that parent
 295          * can reach.  So we do not have to run the expensive
 296          * count_distance() for single strand of pearls.
 297          *
 298          * However, if you have more than one parents, you cannot
 299          * just add their distance and one for yourself, since
 300          * they usually reach the same ancestor and you would
 301          * end up counting them twice that way.
 302          *
 303          * So we will first count distance of merges the usual
 304          * way, and then fill the blanks using cheaper algorithm.
 305          */
 306         for (p = list; p; p = p->next) {
 307                 if (p->item->object.flags & UNINTERESTING)
 308                         continue;
 309                 if (weight(p) != -2)
 310                         continue;
 311                 weight_set(p, count_distance(p));
 312                 clear_distance(list);
 313
 314                 /* Does it happen to be at exactly half-way? */
 315                 if (!find_all && halfway(p, nr))
 316                         return p;
 317                 counted++;
 318         }
 319
 320         show_list("bisection 2 count_distance", counted, nr, list);
 321
 322         while (counted < nr) {
 323                 for (p = list; p; p = p->next) {
 324                         struct commit_list *q;
 325                         unsigned flags = p->item->object.flags;
 326
 327                         if (0 <= weight(p))
 328                                 continue;
 329                         for (q = p->item->parents; q; q = q->next) {
 330                                 if (q->item->object.flags & UNINTERESTING)
 331                                         continue;
 332                                 if (0 <= weight(q))
 333                                         break;
 334                         }
 335                         if (!q)
 336                                 continue;
 337
 338                         /*
 339                          * weight for p is unknown but q is known.
 340                          * add one for p itself if p is to be counted,
 341                          * otherwise inherit it from q directly.
 342                          */
 343                         if (!(flags & TREESAME)) {
 344                                 weight_set(p, weight(q)+1);
 345                                 counted++;
 346                                 show_list("bisection 2 count one",
 347                                           counted, nr, list);
 348                         }
 349                         else
 350                                 weight_set(p, weight(q));
 351
 352                         /* Does it happen to be at exactly half-way? */
 353                         if (!find_all && halfway(p, nr))
 354                                 return p;
 355                 }
 356         }
 357
 358         show_list("bisection 2 counted all", counted, nr, list);
 359
 360         if (!find_all)
 361                 return best_bisection(list, nr);
 362         else
 363                 return best_bisection_sorted(list, nr);
 364 }
 365
 366 void find_bisection(struct commit_list **commit_list, int *reaches,
 367                     int *all, int find_all)
 368 {
 369         int nr, on_list;
 370         struct commit_list *list, *p, *best, *next, *last;
 371         int *weights;
 372
 373         show_list("bisection 2 entry", 0, 0, *commit_list);
 374
 375         /*
 376          * Count the number of total and tree-changing items on the
 377          * list, while reversing the list.
 378          */
 379         for (nr = on_list = 0, last = NULL, p = *commit_list;
 380              p;
 381              p = next) {
 382                 unsigned flags = p->item->object.flags;
 383
 384                 next = p->next;
 385                 if (flags & UNINTERESTING) {
 386                         free(p);
 387                         continue;
 388                 }
 389                 p->next = last;
 390                 last = p;
 391                 if (!(flags & TREESAME))
 392                         nr++;
 393                 on_list++;
 394         }
 395         list = last;
 396         show_list("bisection 2 sorted", 0, nr, list);
 397
 398         *all = nr;
 399         weights = xcalloc(on_list, sizeof(*weights));
 400
 401         /* Do the real work of finding bisection commit. */
 402         best = do_find_bisection(list, nr, weights, find_all);
 403         if (best) {
 404                 if (!find_all) {
 405                         list->item = best->item;
 406                         free_commit_list(list->next);
 407                         best = list;
 408                         best->next = NULL;
 409                 }
 410                 *reaches = weight(best);
 411         }
 412         free(weights);
 413         *commit_list = best;
 414 }
 415
 416 static int register_ref(const char *refname, const struct object_id *oid,
 417                         int flags, void *cb_data)
 418 {
 419         struct strbuf good_prefix = STRBUF_INIT;
 420         strbuf_addstr(&good_prefix, term_good);
 421         strbuf_addstr(&good_prefix, "-");
 422
 423         if (!strcmp(refname, term_bad)) {
 424                 current_bad_oid = xmalloc(sizeof(*current_bad_oid));
 425                 oidcpy(current_bad_oid, oid);
 426         } else if (starts_with(refname, good_prefix.buf)) {
 427                 oid_array_append(&good_revs, oid);
 428         } else if (starts_with(refname, "skip-")) {
 429                 oid_array_append(&skipped_revs, oid);
 430         }
 431
 432         strbuf_release(&good_prefix);
 433
 434         return 0;
 435 }
 436
 437 static int read_bisect_refs(void)
 438 {
 439         return for_each_ref_in("refs/bisect/", register_ref, NULL);
 440 }
 441
 442 static GIT_PATH_FUNC(git_path_bisect_names, "BISECT_NAMES")
 443 static GIT_PATH_FUNC(git_path_bisect_expected_rev, "BISECT_EXPECTED_REV")
 444 static GIT_PATH_FUNC(git_path_bisect_ancestors_ok, "BISECT_ANCESTORS_OK")
 445 static GIT_PATH_FUNC(git_path_bisect_run, "BISECT_RUN")
 446 static GIT_PATH_FUNC(git_path_bisect_start, "BISECT_START")
 447 static GIT_PATH_FUNC(git_path_bisect_log, "BISECT_LOG")
 448 static GIT_PATH_FUNC(git_path_bisect_terms, "BISECT_TERMS")
 449 static GIT_PATH_FUNC(git_path_head_name, "head-name")
 450
 451 static void read_bisect_paths(struct argv_array *array)
 452 {
 453         struct strbuf str = STRBUF_INIT;
 454         const char *filename = git_path_bisect_names();
 455         FILE *fp = xfopen(filename, "r");
 456
 457         while (strbuf_getline_lf(&str, fp) != EOF) {
 458                 strbuf_trim(&str);
 459                 if (sq_dequote_to_argv_array(str.buf, array))
 460                         die(_("Badly quoted content in file '%s': %s"),
 461                             filename, str.buf);
 462         }
 463
 464         strbuf_release(&str);
 465         fclose(fp);
 466 }
 467
 468 static char *join_sha1_array_hex(struct oid_array *array, char delim)
 469 {
 470         struct strbuf joined_hexs = STRBUF_INIT;
 471         int i;
 472
 473         for (i = 0; i < array->nr; i++) {
 474                 strbuf_addstr(&joined_hexs, oid_to_hex(array->oid + i));
 475                 if (i + 1 < array->nr)
 476                         strbuf_addch(&joined_hexs, delim);
 477         }
 478
 479         return strbuf_detach(&joined_hexs, NULL);
 480 }
 481
 482 /*
 483  * In this function, passing a not NULL skipped_first is very special.
 484  * It means that we want to know if the first commit in the list is
 485  * skipped because we will want to test a commit away from it if it is
 486  * indeed skipped.
 487  * So if the first commit is skipped, we cannot take the shortcut to
 488  * just "return list" when we find the first non skipped commit, we
 489  * have to return a fully filtered list.
 490  *
 491  * We use (*skipped_first == -1) to mean "it has been found that the
 492  * first commit is not skipped". In this case *skipped_first is set back
 493  * to 0 just before the function returns.
 494  */
 495 struct commit_list *filter_skipped(struct commit_list *list,
 496                                    struct commit_list **tried,
 497                                    int show_all,
 498                                    int *count,
 499                                    int *skipped_first)
 500 {
 501         struct commit_list *filtered = NULL, **f = &filtered;
 502
 503         *tried = NULL;
 504
 505         if (skipped_first)
 506                 *skipped_first = 0;
 507         if (count)
 508                 *count = 0;
 509
 510         if (!skipped_revs.nr)
 511                 return list;
 512
 513         while (list) {
 514                 struct commit_list *next = list->next;
 515                 list->next = NULL;
 516                 if (0 <= oid_array_lookup(&skipped_revs, &list->item->object.oid)) {
 517                         if (skipped_first && !*skipped_first)
 518                                 *skipped_first = 1;
 519                         /* Move current to tried list */
 520                         *tried = list;
 521                         tried = &list->next;
 522                 } else {
 523                         if (!show_all) {
 524                                 if (!skipped_first || !*skipped_first)
 525                                         return list;
 526                         } else if (skipped_first && !*skipped_first) {
 527                                 /* This means we know it's not skipped */
 528                                 *skipped_first = -1;
 529                         }
 530                         /* Move current to filtered list */
 531                         *f = list;
 532                         f = &list->next;
 533                         if (count)
 534                                 (*count)++;
 535                 }
 536                 list = next;
 537         }
 538
 539         if (skipped_first && *skipped_first == -1)
 540                 *skipped_first = 0;
 541
 542         return filtered;
 543 }
 544
 545 #define PRN_MODULO 32768
 546
 547 /*
 548  * This is a pseudo random number generator based on "man 3 rand".
 549  * It is not used properly because the seed is the argument and it
 550  * is increased by one between each call, but that should not matter
 551  * for this application.
 552  */
 553 static unsigned get_prn(unsigned count) {
 554         count = count * 1103515245 + 12345;
 555         return (count/65536) % PRN_MODULO;
 556 }
 557
 558 /*
 559  * Custom integer square root from
 560  * https://en.wikipedia.org/wiki/Integer_square_root
 561  */
 562 static int sqrti(int val)
 563 {
 564         float d, x = val;
 565
 566         if (val == 0)
 567                 return 0;
 568
 569         do {
 570                 float y = (x + (float)val / x) / 2;
 571                 d = (y > x) ? y - x : x - y;
 572                 x = y;
 573         } while (d >= 0.5);
 574
 575         return (int)x;
 576 }
 577
 578 static struct commit_list *skip_away(struct commit_list *list, int count)
 579 {
 580         struct commit_list *cur, *previous;
 581         int prn, index, i;
 582
 583         prn = get_prn(count);
 584         index = (count * prn / PRN_MODULO) * sqrti(prn) / sqrti(PRN_MODULO);
 585
 586         cur = list;
 587         previous = NULL;
 588
 589         for (i = 0; cur; cur = cur->next, i++) {
 590                 if (i == index) {
 591                         if (oidcmp(&cur->item->object.oid, current_bad_oid))
 592                                 return cur;
 593                         if (previous)
 594                                 return previous;
 595                         return list;
 596                 }
 597                 previous = cur;
 598         }
 599
 600         return list;
 601 }
 602
 603 static struct commit_list *managed_skipped(struct commit_list *list,
 604                                            struct commit_list **tried)
 605 {
 606         int count, skipped_first;
 607
 608         *tried = NULL;
 609
 610         if (!skipped_revs.nr)
 611                 return list;
 612
 613         list = filter_skipped(list, tried, 0, &count, &skipped_first);
 614
 615         if (!skipped_first)
 616                 return list;
 617
 618         return skip_away(list, count);
 619 }
 620
 621 static void bisect_rev_setup(struct rev_info *revs, const char *prefix,
 622                              const char *bad_format, const char *good_format,
 623                              int read_paths)
 624 {
 625         struct argv_array rev_argv = ARGV_ARRAY_INIT;
 626         int i;
 627
 628         init_revisions(revs, prefix);
 629         revs->abbrev = 0;
 630         revs->commit_format = CMIT_FMT_UNSPECIFIED;
 631
 632         /* rev_argv.argv[0] will be ignored by setup_revisions */
 633         argv_array_push(&rev_argv, "bisect_rev_setup");
 634         argv_array_pushf(&rev_argv, bad_format, oid_to_hex(current_bad_oid));
 635         for (i = 0; i < good_revs.nr; i++)
 636                 argv_array_pushf(&rev_argv, good_format,
 637                                  oid_to_hex(good_revs.oid + i));
 638         argv_array_push(&rev_argv, "--");
 639         if (read_paths)
 640                 read_bisect_paths(&rev_argv);
 641
 642         setup_revisions(rev_argv.argc, rev_argv.argv, revs, NULL);
 643         /* XXX leak rev_argv, as "revs" may still be pointing to it */
 644 }
 645
 646 static void bisect_common(struct rev_info *revs)
 647 {
 648         if (prepare_revision_walk(revs))
 649                 die("revision walk setup failed");
 650         if (revs->tree_objects)
 651                 mark_edges_uninteresting(revs, NULL);
 652 }
 653
 654 static void exit_if_skipped_commits(struct commit_list *tried,
 655                                     const struct object_id *bad)
 656 {
 657         if (!tried)
 658                 return;
 659
 660         printf("There are only 'skip'ped commits left to test.\n"
 661                "The first %s commit could be any of:\n", term_bad);
 662
 663         for ( ; tried; tried = tried->next)
 664                 printf("%s\n", oid_to_hex(&tried->item->object.oid));
 665
 666         if (bad)
 667                 printf("%s\n", oid_to_hex(bad));
 668         printf(_("We cannot bisect more!\n"));
 669         exit(2);
 670 }
 671
 672 static int is_expected_rev(const struct object_id *oid)
 673 {
 674         const char *filename = git_path_bisect_expected_rev();
 675         struct stat st;
 676         struct strbuf str = STRBUF_INIT;
 677         FILE *fp;
 678         int res = 0;
 679
 680         if (stat(filename, &st) || !S_ISREG(st.st_mode))
 681                 return 0;
 682
 683         fp = fopen_or_warn(filename, "r");
 684         if (!fp)
 685                 return 0;
 686
 687         if (strbuf_getline_lf(&str, fp) != EOF)
 688                 res = !strcmp(str.buf, oid_to_hex(oid));
 689
 690         strbuf_release(&str);
 691         fclose(fp);
 692
 693         return res;
 694 }
 695
 696 static int bisect_checkout(const struct object_id *bisect_rev, int no_checkout)
 697 {
 698         char bisect_rev_hex[GIT_MAX_HEXSZ + 1];
 699
 700         memcpy(bisect_rev_hex, oid_to_hex(bisect_rev), GIT_SHA1_HEXSZ + 1);
 701         update_ref(NULL, "BISECT_EXPECTED_REV", bisect_rev, NULL, 0, UPDATE_REFS_DIE_ON_ERR);
 702
 703         argv_checkout[2] = bisect_rev_hex;
 704         if (no_checkout) {
 705                 update_ref(NULL, "BISECT_HEAD", bisect_rev, NULL, 0,
 706                            UPDATE_REFS_DIE_ON_ERR);
 707         } else {
 708                 int res;
 709                 res = run_command_v_opt(argv_checkout, RUN_GIT_CMD);
 710                 if (res)
 711                         exit(res);
 712         }
 713
 714         argv_show_branch[1] = bisect_rev_hex;
 715         return run_command_v_opt(argv_show_branch, RUN_GIT_CMD);
 716 }
 717
 718 static struct commit *get_commit_reference(const struct object_id *oid)
 719 {
 720         struct commit *r = lookup_commit_reference(oid);
 721         if (!r)
 722                 die(_("Not a valid commit name %s"), oid_to_hex(oid));
 723         return r;
 724 }
 725
 726 static struct commit **get_bad_and_good_commits(int *rev_nr)
 727 {
 728         struct commit **rev;
 729         int i, n = 0;
 730
 731         ALLOC_ARRAY(rev, 1 + good_revs.nr);
 732         rev[n++] = get_commit_reference(current_bad_oid);
 733         for (i = 0; i < good_revs.nr; i++)
 734                 rev[n++] = get_commit_reference(good_revs.oid + i);
 735         *rev_nr = n;
 736
 737         return rev;
 738 }
 739
 740 static void handle_bad_merge_base(void)
 741 {
 742         if (is_expected_rev(current_bad_oid)) {
 743                 char *bad_hex = oid_to_hex(current_bad_oid);
 744                 char *good_hex = join_sha1_array_hex(&good_revs, ' ');
 745                 if (!strcmp(term_bad, "bad") && !strcmp(term_good, "good")) {
 746                         fprintf(stderr, _("The merge base %s is bad.\n"
 747                                 "This means the bug has been fixed "
 748                                 "between %s and [%s].\n"),
 749                                 bad_hex, bad_hex, good_hex);
 750                 } else if (!strcmp(term_bad, "new") && !strcmp(term_good, "old")) {
 751                         fprintf(stderr, _("The merge base %s is new.\n"
 752                                 "The property has changed "
 753                                 "between %s and [%s].\n"),
 754                                 bad_hex, bad_hex, good_hex);
 755                 } else {
 756                         fprintf(stderr, _("The merge base %s is %s.\n"
 757                                 "This means the first '%s' commit is "
 758                                 "between %s and [%s].\n"),
 759                                 bad_hex, term_bad, term_good, bad_hex, good_hex);
 760                 }
 761                 exit(3);
 762         }
 763
 764         fprintf(stderr, _("Some %s revs are not ancestors of the %s rev.\n"
 765                 "git bisect cannot work properly in this case.\n"
 766                 "Maybe you mistook %s and %s revs?\n"),
 767                 term_good, term_bad, term_good, term_bad);
 768         exit(1);
 769 }
 770
 771 static void handle_skipped_merge_base(const struct object_id *mb)
 772 {
 773         char *mb_hex = oid_to_hex(mb);
 774         char *bad_hex = oid_to_hex(current_bad_oid);
 775         char *good_hex = join_sha1_array_hex(&good_revs, ' ');
 776
 777         warning(_("the merge base between %s and [%s] "
 778                 "must be skipped.\n"
 779                 "So we cannot be sure the first %s commit is "
 780                 "between %s and %s.\n"
 781                 "We continue anyway."),
 782                 bad_hex, good_hex, term_bad, mb_hex, bad_hex);
 783         free(good_hex);
 784 }
 785
 786 /*
 787  * "check_merge_bases" checks that merge bases are not "bad" (or "new").
 788  *
 789  * - If one is "bad" (or "new"), it means the user assumed something wrong
 790  * and we must exit with a non 0 error code.
 791  * - If one is "good" (or "old"), that's good, we have nothing to do.
 792  * - If one is "skipped", we can't know but we should warn.
 793  * - If we don't know, we should check it out and ask the user to test.
 794  */
 795 static void check_merge_bases(int rev_nr, struct commit **rev, int no_checkout)
 796 {
 797         struct commit_list *result;
 798
 799         result = get_merge_bases_many(rev[0], rev_nr - 1, rev + 1);
 800
 801         for (; result; result = result->next) {
 802                 const struct object_id *mb = &result->item->object.oid;
 803                 if (!oidcmp(mb, current_bad_oid)) {
 804                         handle_bad_merge_base();
 805                 } else if (0 <= oid_array_lookup(&good_revs, mb)) {
 806                         continue;
 807                 } else if (0 <= oid_array_lookup(&skipped_revs, mb)) {
 808                         handle_skipped_merge_base(mb);
 809                 } else {
 810                         printf(_("Bisecting: a merge base must be tested\n"));
 811                         exit(bisect_checkout(mb, no_checkout));
 812                 }
 813         }
 814
 815         free_commit_list(result);
 816 }
 817
 818 static int check_ancestors(int rev_nr, struct commit **rev, const char *prefix)
 819 {
 820         struct rev_info revs;
 821         int res;
 822
 823         bisect_rev_setup(&revs, prefix, "^%s", "%s", 0);
 824
 825         bisect_common(&revs);
 826         res = (revs.commits != NULL);
 827
 828         /* Clean up objects used, as they will be reused. */
 829         clear_commit_marks_many(rev_nr, rev, ALL_REV_FLAGS);
 830
 831         return res;
 832 }
 833
 834 /*
 835  * "check_good_are_ancestors_of_bad" checks that all "good" revs are
 836  * ancestor of the "bad" rev.
 837  *
 838  * If that's not the case, we need to check the merge bases.
 839  * If a merge base must be tested by the user, its source code will be
 840  * checked out to be tested by the user and we will exit.
 841  */
 842 static void check_good_are_ancestors_of_bad(const char *prefix, int no_checkout)
 843 {
 844         char *filename = git_pathdup("BISECT_ANCESTORS_OK");
 845         struct stat st;
 846         int fd, rev_nr;
 847         struct commit **rev;
 848
 849         if (!current_bad_oid)
 850                 die(_("a %s revision is needed"), term_bad);
 851
 852         /* Check if file BISECT_ANCESTORS_OK exists. */
 853         if (!stat(filename, &st) && S_ISREG(st.st_mode))
 854                 goto done;
 855
 856         /* Bisecting with no good rev is ok. */
 857         if (good_revs.nr == 0)
 858                 goto done;
 859
 860         /* Check if all good revs are ancestor of the bad rev. */
 861         rev = get_bad_and_good_commits(&rev_nr);
 862         if (check_ancestors(rev_nr, rev, prefix))
 863                 check_merge_bases(rev_nr, rev, no_checkout);
 864         free(rev);
 865
 866         /* Create file BISECT_ANCESTORS_OK. */
 867         fd = open(filename, O_CREAT | O_TRUNC | O_WRONLY, 0600);
 868         if (fd < 0)
 869                 warning_errno(_("could not create file '%s'"),
 870                               filename);
 871         else
 872                 close(fd);
 873  done:
 874         free(filename);
 875 }
 876
 877 /*
 878  * This does "git diff-tree --pretty COMMIT" without one fork+exec.
 879  */
 880 static void show_diff_tree(const char *prefix, struct commit *commit)
 881 {
 882         struct rev_info opt;
 883
 884         /* diff-tree init */
 885         init_revisions(&opt, prefix);
 886         git_config(git_diff_basic_config, NULL); /* no "diff" UI options */
 887         opt.abbrev = 0;
 888         opt.diff = 1;
 889
 890         /* This is what "--pretty" does */
 891         opt.verbose_header = 1;
 892         opt.use_terminator = 0;
 893         opt.commit_format = CMIT_FMT_DEFAULT;
 894
 895         /* diff-tree init */
 896         if (!opt.diffopt.output_format)
 897                 opt.diffopt.output_format = DIFF_FORMAT_RAW;
 898
 899         setup_revisions(0, NULL, &opt, NULL);
 900         log_tree_commit(&opt, commit);
 901 }
 902
 903 /*
 904  * The terms used for this bisect session are stored in BISECT_TERMS.
 905  * We read them and store them to adapt the messages accordingly.
 906  * Default is bad/good.
 907  */
 908 void read_bisect_terms(const char **read_bad, const char **read_good)
 909 {
 910         struct strbuf str = STRBUF_INIT;
 911         const char *filename = git_path_bisect_terms();
 912         FILE *fp = fopen(filename, "r");
 913
 914         if (!fp) {
 915                 if (errno == ENOENT) {
 916                         *read_bad = "bad";
 917                         *read_good = "good";
 918                         return;
 919                 } else {
 920                         die_errno(_("could not read file '%s'"), filename);
 921                 }
 922         } else {
 923                 strbuf_getline_lf(&str, fp);
 924                 *read_bad = strbuf_detach(&str, NULL);
 925                 strbuf_getline_lf(&str, fp);
 926                 *read_good = strbuf_detach(&str, NULL);
 927         }
 928         strbuf_release(&str);
 929         fclose(fp);
 930 }
 931
 932 /*
 933  * We use the convention that exiting with an exit code 10 means that
 934  * the bisection process finished successfully.
 935  * In this case the calling shell script should exit 0.
 936  *
 937  * If no_checkout is non-zero, the bisection process does not
 938  * checkout the trial commit but instead simply updates BISECT_HEAD.
 939  */
 940 int bisect_next_all(const char *prefix, int no_checkout)
 941 {
 942         struct rev_info revs;
 943         struct commit_list *tried;
 944         int reaches = 0, all = 0, nr, steps;
 945         struct object_id *bisect_rev;
 946         char *steps_msg;
 947
 948         read_bisect_terms(&term_bad, &term_good);
 949         if (read_bisect_refs())
 950                 die(_("reading bisect refs failed"));
 951
 952         check_good_are_ancestors_of_bad(prefix, no_checkout);
 953
 954         bisect_rev_setup(&revs, prefix, "%s", "^%s", 1);
 955         revs.limited = 1;
 956
 957         bisect_common(&revs);
 958
 959         find_bisection(&revs.commits, &reaches, &all, !!skipped_revs.nr);
 960         revs.commits = managed_skipped(revs.commits, &tried);
 961
 962         if (!revs.commits) {
 963                 /*
 964                  * We should exit here only if the "bad"
 965                  * commit is also a "skip" commit.
 966                  */
 967                 exit_if_skipped_commits(tried, NULL);
 968
 969                 printf(_("%s was both %s and %s\n"),
 970                        oid_to_hex(current_bad_oid),
 971                        term_good,
 972                        term_bad);
 973                 exit(1);
 974         }
 975
 976         if (!all) {
 977                 fprintf(stderr, _("No testable commit found.\n"
 978                         "Maybe you started with bad path parameters?\n"));
 979                 exit(4);
 980         }
 981
 982         bisect_rev = &revs.commits->item->object.oid;
 983
 984         if (!oidcmp(bisect_rev, current_bad_oid)) {
 985                 exit_if_skipped_commits(tried, current_bad_oid);
 986                 printf("%s is the first %s commit\n", oid_to_hex(bisect_rev),
 987                         term_bad);
 988                 show_diff_tree(prefix, revs.commits->item);
 989                 /* This means the bisection process succeeded. */
 990                 exit(10);
 991         }
 992
 993         nr = all - reaches - 1;
 994         steps = estimate_bisect_steps(all);
 995
 996         steps_msg = xstrfmt(Q_("(roughly %d step)", "(roughly %d steps)",
 997                   steps), steps);
 998         /*
 999          * TRANSLATORS: the last %s will be replaced with "(roughly %d
1000          * steps)" translation.
1001          */
1002         printf(Q_("Bisecting: %d revision left to test after this %s\n",
1003                   "Bisecting: %d revisions left to test after this %s\n",
1004                   nr), nr, steps_msg);
1005         free(steps_msg);
1006
1007         return bisect_checkout(bisect_rev, no_checkout);
1008 }
1009
1010 static inline int log2i(int n)
1011 {
1012         int log2 = 0;
1013
1014         for (; n > 1; n >>= 1)
1015                 log2++;
1016
1017         return log2;
1018 }
1019
1020 static inline int exp2i(int n)
1021 {
1022         return 1 << n;
1023 }
1024
1025 /*
1026  * Estimate the number of bisect steps left (after the current step)
1027  *
1028  * For any x between 0 included and 2^n excluded, the probability for
1029  * n - 1 steps left looks like:
1030  *
1031  * P(2^n + x) == (2^n - x) / (2^n + x)
1032  *
1033  * and P(2^n + x) < 0.5 means 2^n < 3x
1034  */
1035 int estimate_bisect_steps(int all)
1036 {
1037         int n, x, e;
1038
1039         if (all < 3)
1040                 return 0;
1041
1042         n = log2i(all);
1043         e = exp2i(n);
1044         x = all - e;
1045
1046         return (e < 3 * x) ? n : n - 1;
1047 }
1048
1049 static int mark_for_removal(const char *refname, const struct object_id *oid,
1050                             int flag, void *cb_data)
1051 {
1052         struct string_list *refs = cb_data;
1053         char *ref = xstrfmt("refs/bisect%s", refname);
1054         string_list_append(refs, ref);
1055         return 0;
1056 }
1057
1058 int bisect_clean_state(void)
1059 {
1060         int result = 0;
1061
1062         /* There may be some refs packed during bisection */
1063         struct string_list refs_for_removal = STRING_LIST_INIT_NODUP;
1064         for_each_ref_in("refs/bisect", mark_for_removal, (void *) &refs_for_removal);
1065         string_list_append(&refs_for_removal, xstrdup("BISECT_HEAD"));
1066         result = delete_refs("bisect: remove", &refs_for_removal, REF_NO_DEREF);
1067         refs_for_removal.strdup_strings = 1;
1068         string_list_clear(&refs_for_removal, 0);
1069         unlink_or_warn(git_path_bisect_expected_rev());
1070         unlink_or_warn(git_path_bisect_ancestors_ok());
1071         unlink_or_warn(git_path_bisect_log());
1072         unlink_or_warn(git_path_bisect_names());
1073         unlink_or_warn(git_path_bisect_run());
1074         unlink_or_warn(git_path_bisect_terms());
1075         /* Cleanup head-name if it got left by an old version of git-bisect */
1076         unlink_or_warn(git_path_head_name());
1077         /*
1078          * Cleanup BISECT_START last to support the --no-checkout option
1079          * introduced in the commit 4796e823a.
1080          */
1081         unlink_or_warn(git_path_bisect_start());
1082
1083         return result;
1084 }