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 #include "commit-slab.h"
  16 #include "commit-reach.h"
  17 #include "object-store.h"
  18
  19 static struct oid_array good_revs;
  20 static struct oid_array skipped_revs;
  21
  22 static struct object_id *current_bad_oid;
  23
  24 static const char *argv_checkout[] = {"checkout", "-q", NULL, "--", NULL};
  25 static const char *argv_show_branch[] = {"show-branch", NULL, NULL};
  26
  27 static const char *term_bad;
  28 static const char *term_good;
  29
  30 /* Remember to update object flag allocation in object.h */
  31 #define COUNTED         (1u<<16)
  32
  33 /*
  34  * This is a truly stupid algorithm, but it's only
  35  * used for bisection, and we just don't care enough.
  36  *
  37  * We care just barely enough to avoid recursing for
  38  * non-merge entries.
  39  */
  40 static int count_distance(struct commit_list *entry)
  41 {
  42         int nr = 0;
  43
  44         while (entry) {
  45                 struct commit *commit = entry->item;
  46                 struct commit_list *p;
  47
  48                 if (commit->object.flags & (UNINTERESTING | COUNTED))
  49                         break;
  50                 if (!(commit->object.flags & TREESAME))
  51                         nr++;
  52                 commit->object.flags |= COUNTED;
  53                 p = commit->parents;
  54                 entry = p;
  55                 if (p) {
  56                         p = p->next;
  57                         while (p) {
  58                                 nr += count_distance(p);
  59                                 p = p->next;
  60                         }
  61                 }
  62         }
  63
  64         return nr;
  65 }
  66
  67 static void clear_distance(struct commit_list *list)
  68 {
  69         while (list) {
  70                 struct commit *commit = list->item;
  71                 commit->object.flags &= ~COUNTED;
  72                 list = list->next;
  73         }
  74 }
  75
  76 define_commit_slab(commit_weight, int *);
  77 static struct commit_weight commit_weight;
  78
  79 #define DEBUG_BISECT 0
  80
  81 static inline int weight(struct commit_list *elem)
  82 {
  83         return **commit_weight_at(&commit_weight, elem->item);
  84 }
  85
  86 static inline void weight_set(struct commit_list *elem, int weight)
  87 {
  88         **commit_weight_at(&commit_weight, elem->item) = weight;
  89 }
  90
  91 static int count_interesting_parents(struct commit *commit)
  92 {
  93         struct commit_list *p;
  94         int count;
  95
  96         for (count = 0, p = commit->parents; p; p = p->next) {
  97                 if (p->item->object.flags & UNINTERESTING)
  98                         continue;
  99                 count++;
 100         }
 101         return count;
 102 }
 103
 104 static inline int halfway(struct commit_list *p, int nr)
 105 {
 106         /*
 107          * Don't short-cut something we are not going to return!
 108          */
 109         if (p->item->object.flags & TREESAME)
 110                 return 0;
 111         if (DEBUG_BISECT)
 112                 return 0;
 113         /*
 114          * 2 and 3 are halfway of 5.
 115          * 3 is halfway of 6 but 2 and 4 are not.
 116          */
 117         switch (2 * weight(p) - nr) {
 118         case -1: case 0: case 1:
 119                 return 1;
 120         default:
 121                 return 0;
 122         }
 123 }
 124
 125 static void show_list(const char *debug, int counted, int nr,
 126                       struct commit_list *list)
 127 {
 128         struct commit_list *p;
 129
 130         if (!DEBUG_BISECT)
 131                 return;
 132
 133         fprintf(stderr, "%s (%d/%d)\n", debug, counted, nr);
 134
 135         for (p = list; p; p = p->next) {
 136                 struct commit_list *pp;
 137                 struct commit *commit = p->item;
 138                 unsigned flags = commit->object.flags;
 139                 enum object_type type;
 140                 unsigned long size;
 141                 char *buf = read_object_file(&commit->object.oid, &type,
 142                                              &size);
 143                 const char *subject_start;
 144                 int subject_len;
 145
 146                 fprintf(stderr, "%c%c%c ",
 147                         (flags & TREESAME) ? ' ' : 'T',
 148                         (flags & UNINTERESTING) ? 'U' : ' ',
 149                         (flags & COUNTED) ? 'C' : ' ');
 150                 if (*commit_weight_at(&commit_weight, p->item))
 151                         fprintf(stderr, "%3d", weight(p));
 152                 else
 153                         fprintf(stderr, "---");
 154                 fprintf(stderr, " %.*s", 8, oid_to_hex(&commit->object.oid));
 155                 for (pp = commit->parents; pp; pp = pp->next)
 156                         fprintf(stderr, " %.*s", 8,
 157                                 oid_to_hex(&pp->item->object.oid));
 158
 159                 subject_len = find_commit_subject(buf, &subject_start);
 160                 if (subject_len)
 161                         fprintf(stderr, " %.*s", subject_len, subject_start);
 162                 fprintf(stderr, "\n");
 163         }
 164 }
 165
 166 static struct commit_list *best_bisection(struct commit_list *list, int nr)
 167 {
 168         struct commit_list *p, *best;
 169         int best_distance = -1;
 170
 171         best = list;
 172         for (p = list; p; p = p->next) {
 173                 int distance;
 174                 unsigned flags = p->item->object.flags;
 175
 176                 if (flags & TREESAME)
 177                         continue;
 178                 distance = weight(p);
 179                 if (nr - distance < distance)
 180                         distance = nr - distance;
 181                 if (distance > best_distance) {
 182                         best = p;
 183                         best_distance = distance;
 184                 }
 185         }
 186
 187         return best;
 188 }
 189
 190 struct commit_dist {
 191         struct commit *commit;
 192         int distance;
 193 };
 194
 195 static int compare_commit_dist(const void *a_, const void *b_)
 196 {
 197         struct commit_dist *a, *b;
 198
 199         a = (struct commit_dist *)a_;
 200         b = (struct commit_dist *)b_;
 201         if (a->distance != b->distance)
 202                 return b->distance - a->distance; /* desc sort */
 203         return oidcmp(&a->commit->object.oid, &b->commit->object.oid);
 204 }
 205
 206 static struct commit_list *best_bisection_sorted(struct commit_list *list, int nr)
 207 {
 208         struct commit_list *p;
 209         struct commit_dist *array = xcalloc(nr, sizeof(*array));
 210         struct strbuf buf = STRBUF_INIT;
 211         int cnt, i;
 212
 213         for (p = list, cnt = 0; p; p = p->next) {
 214                 int distance;
 215                 unsigned flags = p->item->object.flags;
 216
 217                 if (flags & TREESAME)
 218                         continue;
 219                 distance = weight(p);
 220                 if (nr - distance < distance)
 221                         distance = nr - distance;
 222                 array[cnt].commit = p->item;
 223                 array[cnt].distance = distance;
 224                 cnt++;
 225         }
 226         QSORT(array, cnt, compare_commit_dist);
 227         for (p = list, i = 0; i < cnt; i++) {
 228                 struct object *obj = &(array[i].commit->object);
 229
 230                 strbuf_reset(&buf);
 231                 strbuf_addf(&buf, "dist=%d", array[i].distance);
 232                 add_name_decoration(DECORATION_NONE, buf.buf, obj);
 233
 234                 p->item = array[i].commit;
 235                 if (i < cnt - 1)
 236                         p = p->next;
 237         }
 238         if (p) {
 239                 free_commit_list(p->next);
 240                 p->next = NULL;
 241         }
 242         strbuf_release(&buf);
 243         free(array);
 244         return list;
 245 }
 246
 247 /*
 248  * zero or positive weight is the number of interesting commits it can
 249  * reach, including itself.  Especially, weight = 0 means it does not
 250  * reach any tree-changing commits (e.g. just above uninteresting one
 251  * but traversal is with pathspec).
 252  *
 253  * weight = -1 means it has one parent and its distance is yet to
 254  * be computed.
 255  *
 256  * weight = -2 means it has more than one parent and its distance is
 257  * unknown.  After running count_distance() first, they will get zero
 258  * or positive distance.
 259  */
 260 static struct commit_list *do_find_bisection(struct commit_list *list,
 261                                              int nr, int *weights,
 262                                              int find_all)
 263 {
 264         int n, counted;
 265         struct commit_list *p;
 266
 267         counted = 0;
 268
 269         for (n = 0, p = list; p; p = p->next) {
 270                 struct commit *commit = p->item;
 271                 unsigned flags = commit->object.flags;
 272
 273                 *commit_weight_at(&commit_weight, p->item) = &weights[n++];
 274                 switch (count_interesting_parents(commit)) {
 275                 case 0:
 276                         if (!(flags & TREESAME)) {
 277                                 weight_set(p, 1);
 278                                 counted++;
 279                                 show_list("bisection 2 count one",
 280                                           counted, nr, list);
 281                         }
 282                         /*
 283                          * otherwise, it is known not to reach any
 284                          * tree-changing commit and gets weight 0.
 285                          */
 286                         break;
 287                 case 1:
 288                         weight_set(p, -1);
 289                         break;
 290                 default:
 291                         weight_set(p, -2);
 292                         break;
 293                 }
 294         }
 295
 296         show_list("bisection 2 initialize", counted, nr, list);
 297
 298         /*
 299          * If you have only one parent in the resulting set
 300          * then you can reach one commit more than that parent
 301          * can reach.  So we do not have to run the expensive
 302          * count_distance() for single strand of pearls.
 303          *
 304          * However, if you have more than one parents, you cannot
 305          * just add their distance and one for yourself, since
 306          * they usually reach the same ancestor and you would
 307          * end up counting them twice that way.
 308          *
 309          * So we will first count distance of merges the usual
 310          * way, and then fill the blanks using cheaper algorithm.
 311          */
 312         for (p = list; p; p = p->next) {
 313                 if (p->item->object.flags & UNINTERESTING)
 314                         continue;
 315                 if (weight(p) != -2)
 316                         continue;
 317                 weight_set(p, count_distance(p));
 318                 clear_distance(list);
 319
 320                 /* Does it happen to be at exactly half-way? */
 321                 if (!find_all && halfway(p, nr))
 322                         return p;
 323                 counted++;
 324         }
 325
 326         show_list("bisection 2 count_distance", counted, nr, list);
 327
 328         while (counted < nr) {
 329                 for (p = list; p; p = p->next) {
 330                         struct commit_list *q;
 331                         unsigned flags = p->item->object.flags;
 332
 333                         if (0 <= weight(p))
 334                                 continue;
 335                         for (q = p->item->parents; q; q = q->next) {
 336                                 if (q->item->object.flags & UNINTERESTING)
 337                                         continue;
 338                                 if (0 <= weight(q))
 339                                         break;
 340                         }
 341                         if (!q)
 342                                 continue;
 343
 344                         /*
 345                          * weight for p is unknown but q is known.
 346                          * add one for p itself if p is to be counted,
 347                          * otherwise inherit it from q directly.
 348                          */
 349                         if (!(flags & TREESAME)) {
 350                                 weight_set(p, weight(q)+1);
 351                                 counted++;
 352                                 show_list("bisection 2 count one",
 353                                           counted, nr, list);
 354                         }
 355                         else
 356                                 weight_set(p, weight(q));
 357
 358                         /* Does it happen to be at exactly half-way? */
 359                         if (!find_all && halfway(p, nr))
 360                                 return p;
 361                 }
 362         }
 363
 364         show_list("bisection 2 counted all", counted, nr, list);
 365
 366         if (!find_all)
 367                 return best_bisection(list, nr);
 368         else
 369                 return best_bisection_sorted(list, nr);
 370 }
 371
 372 void find_bisection(struct commit_list **commit_list, int *reaches,
 373                     int *all, int find_all)
 374 {
 375         int nr, on_list;
 376         struct commit_list *list, *p, *best, *next, *last;
 377         int *weights;
 378
 379         show_list("bisection 2 entry", 0, 0, *commit_list);
 380         init_commit_weight(&commit_weight);
 381
 382         /*
 383          * Count the number of total and tree-changing items on the
 384          * list, while reversing the list.
 385          */
 386         for (nr = on_list = 0, last = NULL, p = *commit_list;
 387              p;
 388              p = next) {
 389                 unsigned flags = p->item->object.flags;
 390
 391                 next = p->next;
 392                 if (flags & UNINTERESTING) {
 393                         free(p);
 394                         continue;
 395                 }
 396                 p->next = last;
 397                 last = p;
 398                 if (!(flags & TREESAME))
 399                         nr++;
 400                 on_list++;
 401         }
 402         list = last;
 403         show_list("bisection 2 sorted", 0, nr, list);
 404
 405         *all = nr;
 406         weights = xcalloc(on_list, sizeof(*weights));
 407
 408         /* Do the real work of finding bisection commit. */
 409         best = do_find_bisection(list, nr, weights, find_all);
 410         if (best) {
 411                 if (!find_all) {
 412                         list->item = best->item;
 413                         free_commit_list(list->next);
 414                         best = list;
 415                         best->next = NULL;
 416                 }
 417                 *reaches = weight(best);
 418         }
 419         free(weights);
 420         *commit_list = best;
 421         clear_commit_weight(&commit_weight);
 422 }
 423
 424 static int register_ref(const char *refname, const struct object_id *oid,
 425                         int flags, void *cb_data)
 426 {
 427         struct strbuf good_prefix = STRBUF_INIT;
 428         strbuf_addstr(&good_prefix, term_good);
 429         strbuf_addstr(&good_prefix, "-");
 430
 431         if (!strcmp(refname, term_bad)) {
 432                 current_bad_oid = xmalloc(sizeof(*current_bad_oid));
 433                 oidcpy(current_bad_oid, oid);
 434         } else if (starts_with(refname, good_prefix.buf)) {
 435                 oid_array_append(&good_revs, oid);
 436         } else if (starts_with(refname, "skip-")) {
 437                 oid_array_append(&skipped_revs, oid);
 438         }
 439
 440         strbuf_release(&good_prefix);
 441
 442         return 0;
 443 }
 444
 445 static int read_bisect_refs(void)
 446 {
 447         return for_each_ref_in("refs/bisect/", register_ref, NULL);
 448 }
 449
 450 static GIT_PATH_FUNC(git_path_bisect_names, "BISECT_NAMES")
 451 static GIT_PATH_FUNC(git_path_bisect_expected_rev, "BISECT_EXPECTED_REV")
 452 static GIT_PATH_FUNC(git_path_bisect_ancestors_ok, "BISECT_ANCESTORS_OK")
 453 static GIT_PATH_FUNC(git_path_bisect_run, "BISECT_RUN")
 454 static GIT_PATH_FUNC(git_path_bisect_start, "BISECT_START")
 455 static GIT_PATH_FUNC(git_path_bisect_log, "BISECT_LOG")
 456 static GIT_PATH_FUNC(git_path_bisect_terms, "BISECT_TERMS")
 457 static GIT_PATH_FUNC(git_path_head_name, "head-name")
 458
 459 static void read_bisect_paths(struct argv_array *array)
 460 {
 461         struct strbuf str = STRBUF_INIT;
 462         const char *filename = git_path_bisect_names();
 463         FILE *fp = xfopen(filename, "r");
 464
 465         while (strbuf_getline_lf(&str, fp) != EOF) {
 466                 strbuf_trim(&str);
 467                 if (sq_dequote_to_argv_array(str.buf, array))
 468                         die(_("Badly quoted content in file '%s': %s"),
 469                             filename, str.buf);
 470         }
 471
 472         strbuf_release(&str);
 473         fclose(fp);
 474 }
 475
 476 static char *join_sha1_array_hex(struct oid_array *array, char delim)
 477 {
 478         struct strbuf joined_hexs = STRBUF_INIT;
 479         int i;
 480
 481         for (i = 0; i < array->nr; i++) {
 482                 strbuf_addstr(&joined_hexs, oid_to_hex(array->oid + i));
 483                 if (i + 1 < array->nr)
 484                         strbuf_addch(&joined_hexs, delim);
 485         }
 486
 487         return strbuf_detach(&joined_hexs, NULL);
 488 }
 489
 490 /*
 491  * In this function, passing a not NULL skipped_first is very special.
 492  * It means that we want to know if the first commit in the list is
 493  * skipped because we will want to test a commit away from it if it is
 494  * indeed skipped.
 495  * So if the first commit is skipped, we cannot take the shortcut to
 496  * just "return list" when we find the first non skipped commit, we
 497  * have to return a fully filtered list.
 498  *
 499  * We use (*skipped_first == -1) to mean "it has been found that the
 500  * first commit is not skipped". In this case *skipped_first is set back
 501  * to 0 just before the function returns.
 502  */
 503 struct commit_list *filter_skipped(struct commit_list *list,
 504                                    struct commit_list **tried,
 505                                    int show_all,
 506                                    int *count,
 507                                    int *skipped_first)
 508 {
 509         struct commit_list *filtered = NULL, **f = &filtered;
 510
 511         *tried = NULL;
 512
 513         if (skipped_first)
 514                 *skipped_first = 0;
 515         if (count)
 516                 *count = 0;
 517
 518         if (!skipped_revs.nr)
 519                 return list;
 520
 521         while (list) {
 522                 struct commit_list *next = list->next;
 523                 list->next = NULL;
 524                 if (0 <= oid_array_lookup(&skipped_revs, &list->item->object.oid)) {
 525                         if (skipped_first && !*skipped_first)
 526                                 *skipped_first = 1;
 527                         /* Move current to tried list */
 528                         *tried = list;
 529                         tried = &list->next;
 530                 } else {
 531                         if (!show_all) {
 532                                 if (!skipped_first || !*skipped_first)
 533                                         return list;
 534                         } else if (skipped_first && !*skipped_first) {
 535                                 /* This means we know it's not skipped */
 536                                 *skipped_first = -1;
 537                         }
 538                         /* Move current to filtered list */
 539                         *f = list;
 540                         f = &list->next;
 541                         if (count)
 542                                 (*count)++;
 543                 }
 544                 list = next;
 545         }
 546
 547         if (skipped_first && *skipped_first == -1)
 548                 *skipped_first = 0;
 549
 550         return filtered;
 551 }
 552
 553 #define PRN_MODULO 32768
 554
 555 /*
 556  * This is a pseudo random number generator based on "man 3 rand".
 557  * It is not used properly because the seed is the argument and it
 558  * is increased by one between each call, but that should not matter
 559  * for this application.
 560  */
 561 static unsigned get_prn(unsigned count)
 562 {
 563         count = count * 1103515245 + 12345;
 564         return (count/65536) % PRN_MODULO;
 565 }
 566
 567 /*
 568  * Custom integer square root from
 569  * https://en.wikipedia.org/wiki/Integer_square_root
 570  */
 571 static int sqrti(int val)
 572 {
 573         float d, x = val;
 574
 575         if (val == 0)
 576                 return 0;
 577
 578         do {
 579                 float y = (x + (float)val / x) / 2;
 580                 d = (y > x) ? y - x : x - y;
 581                 x = y;
 582         } while (d >= 0.5);
 583
 584         return (int)x;
 585 }
 586
 587 static struct commit_list *skip_away(struct commit_list *list, int count)
 588 {
 589         struct commit_list *cur, *previous;
 590         int prn, index, i;
 591
 592         prn = get_prn(count);
 593         index = (count * prn / PRN_MODULO) * sqrti(prn) / sqrti(PRN_MODULO);
 594
 595         cur = list;
 596         previous = NULL;
 597
 598         for (i = 0; cur; cur = cur->next, i++) {
 599                 if (i == index) {
 600                         if (!oideq(&cur->item->object.oid, current_bad_oid))
 601                                 return cur;
 602                         if (previous)
 603                                 return previous;
 604                         return list;
 605                 }
 606                 previous = cur;
 607         }
 608
 609         return list;
 610 }
 611
 612 static struct commit_list *managed_skipped(struct commit_list *list,
 613                                            struct commit_list **tried)
 614 {
 615         int count, skipped_first;
 616
 617         *tried = NULL;
 618
 619         if (!skipped_revs.nr)
 620                 return list;
 621
 622         list = filter_skipped(list, tried, 0, &count, &skipped_first);
 623
 624         if (!skipped_first)
 625                 return list;
 626
 627         return skip_away(list, count);
 628 }
 629
 630 static void bisect_rev_setup(struct repository *r, struct rev_info *revs,
 631                              const char *prefix,
 632                              const char *bad_format, const char *good_format,
 633                              int read_paths)
 634 {
 635         struct argv_array rev_argv = ARGV_ARRAY_INIT;
 636         int i;
 637
 638         repo_init_revisions(r, revs, prefix);
 639         revs->abbrev = 0;
 640         revs->commit_format = CMIT_FMT_UNSPECIFIED;
 641
 642         /* rev_argv.argv[0] will be ignored by setup_revisions */
 643         argv_array_push(&rev_argv, "bisect_rev_setup");
 644         argv_array_pushf(&rev_argv, bad_format, oid_to_hex(current_bad_oid));
 645         for (i = 0; i < good_revs.nr; i++)
 646                 argv_array_pushf(&rev_argv, good_format,
 647                                  oid_to_hex(good_revs.oid + i));
 648         argv_array_push(&rev_argv, "--");
 649         if (read_paths)
 650                 read_bisect_paths(&rev_argv);
 651
 652         setup_revisions(rev_argv.argc, rev_argv.argv, revs, NULL);
 653         /* XXX leak rev_argv, as "revs" may still be pointing to it */
 654 }
 655
 656 static void bisect_common(struct rev_info *revs)
 657 {
 658         if (prepare_revision_walk(revs))
 659                 die("revision walk setup failed");
 660         if (revs->tree_objects)
 661                 mark_edges_uninteresting(revs, NULL, 0);
 662 }
 663
 664 static void exit_if_skipped_commits(struct commit_list *tried,
 665                                     const struct object_id *bad)
 666 {
 667         if (!tried)
 668                 return;
 669
 670         printf("There are only 'skip'ped commits left to test.\n"
 671                "The first %s commit could be any of:\n", term_bad);
 672
 673         for ( ; tried; tried = tried->next)
 674                 printf("%s\n", oid_to_hex(&tried->item->object.oid));
 675
 676         if (bad)
 677                 printf("%s\n", oid_to_hex(bad));
 678         printf(_("We cannot bisect more!\n"));
 679         exit(2);
 680 }
 681
 682 static int is_expected_rev(const struct object_id *oid)
 683 {
 684         const char *filename = git_path_bisect_expected_rev();
 685         struct stat st;
 686         struct strbuf str = STRBUF_INIT;
 687         FILE *fp;
 688         int res = 0;
 689
 690         if (stat(filename, &st) || !S_ISREG(st.st_mode))
 691                 return 0;
 692
 693         fp = fopen_or_warn(filename, "r");
 694         if (!fp)
 695                 return 0;
 696
 697         if (strbuf_getline_lf(&str, fp) != EOF)
 698                 res = !strcmp(str.buf, oid_to_hex(oid));
 699
 700         strbuf_release(&str);
 701         fclose(fp);
 702
 703         return res;
 704 }
 705
 706 static int bisect_checkout(const struct object_id *bisect_rev, int no_checkout)
 707 {
 708         char bisect_rev_hex[GIT_MAX_HEXSZ + 1];
 709
 710         memcpy(bisect_rev_hex, oid_to_hex(bisect_rev), the_hash_algo->hexsz + 1);
 711         update_ref(NULL, "BISECT_EXPECTED_REV", bisect_rev, NULL, 0, UPDATE_REFS_DIE_ON_ERR);
 712
 713         argv_checkout[2] = bisect_rev_hex;
 714         if (no_checkout) {
 715                 update_ref(NULL, "BISECT_HEAD", bisect_rev, NULL, 0,
 716                            UPDATE_REFS_DIE_ON_ERR);
 717         } else {
 718                 int res;
 719                 res = run_command_v_opt(argv_checkout, RUN_GIT_CMD);
 720                 if (res)
 721                         exit(res);
 722         }
 723
 724         argv_show_branch[1] = bisect_rev_hex;
 725         return run_command_v_opt(argv_show_branch, RUN_GIT_CMD);
 726 }
 727
 728 static struct commit *get_commit_reference(struct repository *r,
 729                                            const struct object_id *oid)
 730 {
 731         struct commit *c = lookup_commit_reference(r, oid);
 732         if (!c)
 733                 die(_("Not a valid commit name %s"), oid_to_hex(oid));
 734         return c;
 735 }
 736
 737 static struct commit **get_bad_and_good_commits(struct repository *r,
 738                                                 int *rev_nr)
 739 {
 740         struct commit **rev;
 741         int i, n = 0;
 742
 743         ALLOC_ARRAY(rev, 1 + good_revs.nr);
 744         rev[n++] = get_commit_reference(r, current_bad_oid);
 745         for (i = 0; i < good_revs.nr; i++)
 746                 rev[n++] = get_commit_reference(r, good_revs.oid + i);
 747         *rev_nr = n;
 748
 749         return rev;
 750 }
 751
 752 static void handle_bad_merge_base(void)
 753 {
 754         if (is_expected_rev(current_bad_oid)) {
 755                 char *bad_hex = oid_to_hex(current_bad_oid);
 756                 char *good_hex = join_sha1_array_hex(&good_revs, ' ');
 757                 if (!strcmp(term_bad, "bad") && !strcmp(term_good, "good")) {
 758                         fprintf(stderr, _("The merge base %s is bad.\n"
 759                                 "This means the bug has been fixed "
 760                                 "between %s and [%s].\n"),
 761                                 bad_hex, bad_hex, good_hex);
 762                 } else if (!strcmp(term_bad, "new") && !strcmp(term_good, "old")) {
 763                         fprintf(stderr, _("The merge base %s is new.\n"
 764                                 "The property has changed "
 765                                 "between %s and [%s].\n"),
 766                                 bad_hex, bad_hex, good_hex);
 767                 } else {
 768                         fprintf(stderr, _("The merge base %s is %s.\n"
 769                                 "This means the first '%s' commit is "
 770                                 "between %s and [%s].\n"),
 771                                 bad_hex, term_bad, term_good, bad_hex, good_hex);
 772                 }
 773                 exit(3);
 774         }
 775
 776         fprintf(stderr, _("Some %s revs are not ancestors of the %s rev.\n"
 777                 "git bisect cannot work properly in this case.\n"
 778                 "Maybe you mistook %s and %s revs?\n"),
 779                 term_good, term_bad, term_good, term_bad);
 780         exit(1);
 781 }
 782
 783 static void handle_skipped_merge_base(const struct object_id *mb)
 784 {
 785         char *mb_hex = oid_to_hex(mb);
 786         char *bad_hex = oid_to_hex(current_bad_oid);
 787         char *good_hex = join_sha1_array_hex(&good_revs, ' ');
 788
 789         warning(_("the merge base between %s and [%s] "
 790                 "must be skipped.\n"
 791                 "So we cannot be sure the first %s commit is "
 792                 "between %s and %s.\n"
 793                 "We continue anyway."),
 794                 bad_hex, good_hex, term_bad, mb_hex, bad_hex);
 795         free(good_hex);
 796 }
 797
 798 /*
 799  * "check_merge_bases" checks that merge bases are not "bad" (or "new").
 800  *
 801  * - If one is "bad" (or "new"), it means the user assumed something wrong
 802  * and we must exit with a non 0 error code.
 803  * - If one is "good" (or "old"), that's good, we have nothing to do.
 804  * - If one is "skipped", we can't know but we should warn.
 805  * - If we don't know, we should check it out and ask the user to test.
 806  */
 807 static void check_merge_bases(int rev_nr, struct commit **rev, int no_checkout)
 808 {
 809         struct commit_list *result;
 810
 811         result = get_merge_bases_many(rev[0], rev_nr - 1, rev + 1);
 812
 813         for (; result; result = result->next) {
 814                 const struct object_id *mb = &result->item->object.oid;
 815                 if (oideq(mb, current_bad_oid)) {
 816                         handle_bad_merge_base();
 817                 } else if (0 <= oid_array_lookup(&good_revs, mb)) {
 818                         continue;
 819                 } else if (0 <= oid_array_lookup(&skipped_revs, mb)) {
 820                         handle_skipped_merge_base(mb);
 821                 } else {
 822                         printf(_("Bisecting: a merge base must be tested\n"));
 823                         exit(bisect_checkout(mb, no_checkout));
 824                 }
 825         }
 826
 827         free_commit_list(result);
 828 }
 829
 830 static int check_ancestors(struct repository *r, int rev_nr,
 831                            struct commit **rev, const char *prefix)
 832 {
 833         struct rev_info revs;
 834         int res;
 835
 836         bisect_rev_setup(r, &revs, prefix, "^%s", "%s", 0);
 837
 838         bisect_common(&revs);
 839         res = (revs.commits != NULL);
 840
 841         /* Clean up objects used, as they will be reused. */
 842         clear_commit_marks_many(rev_nr, rev, ALL_REV_FLAGS);
 843
 844         return res;
 845 }
 846
 847 /*
 848  * "check_good_are_ancestors_of_bad" checks that all "good" revs are
 849  * ancestor of the "bad" rev.
 850  *
 851  * If that's not the case, we need to check the merge bases.
 852  * If a merge base must be tested by the user, its source code will be
 853  * checked out to be tested by the user and we will exit.
 854  */
 855 static void check_good_are_ancestors_of_bad(struct repository *r,
 856                                             const char *prefix,
 857                                             int no_checkout)
 858 {
 859         char *filename = git_pathdup("BISECT_ANCESTORS_OK");
 860         struct stat st;
 861         int fd, rev_nr;
 862         struct commit **rev;
 863
 864         if (!current_bad_oid)
 865                 die(_("a %s revision is needed"), term_bad);
 866
 867         /* Check if file BISECT_ANCESTORS_OK exists. */
 868         if (!stat(filename, &st) && S_ISREG(st.st_mode))
 869                 goto done;
 870
 871         /* Bisecting with no good rev is ok. */
 872         if (good_revs.nr == 0)
 873                 goto done;
 874
 875         /* Check if all good revs are ancestor of the bad rev. */
 876         rev = get_bad_and_good_commits(r, &rev_nr);
 877         if (check_ancestors(r, rev_nr, rev, prefix))
 878                 check_merge_bases(rev_nr, rev, no_checkout);
 879         free(rev);
 880
 881         /* Create file BISECT_ANCESTORS_OK. */
 882         fd = open(filename, O_CREAT | O_TRUNC | O_WRONLY, 0600);
 883         if (fd < 0)
 884                 warning_errno(_("could not create file '%s'"),
 885                               filename);
 886         else
 887                 close(fd);
 888  done:
 889         free(filename);
 890 }
 891
 892 /*
 893  * This does "git diff-tree --pretty COMMIT" without one fork+exec.
 894  */
 895 static void show_diff_tree(struct repository *r,
 896                            const char *prefix,
 897                            struct commit *commit)
 898 {
 899         const char *argv[] = {
 900                 "diff-tree", "--pretty", "--stat", "--summary", "--cc", NULL
 901         };
 902         struct rev_info opt;
 903
 904         git_config(git_diff_ui_config, NULL);
 905         repo_init_revisions(r, &opt, prefix);
 906
 907         setup_revisions(ARRAY_SIZE(argv) - 1, argv, &opt, NULL);
 908         log_tree_commit(&opt, commit);
 909 }
 910
 911 /*
 912  * The terms used for this bisect session are stored in BISECT_TERMS.
 913  * We read them and store them to adapt the messages accordingly.
 914  * Default is bad/good.
 915  */
 916 void read_bisect_terms(const char **read_bad, const char **read_good)
 917 {
 918         struct strbuf str = STRBUF_INIT;
 919         const char *filename = git_path_bisect_terms();
 920         FILE *fp = fopen(filename, "r");
 921
 922         if (!fp) {
 923                 if (errno == ENOENT) {
 924                         *read_bad = "bad";
 925                         *read_good = "good";
 926                         return;
 927                 } else {
 928                         die_errno(_("could not read file '%s'"), filename);
 929                 }
 930         } else {
 931                 strbuf_getline_lf(&str, fp);
 932                 *read_bad = strbuf_detach(&str, NULL);
 933                 strbuf_getline_lf(&str, fp);
 934                 *read_good = strbuf_detach(&str, NULL);
 935         }
 936         strbuf_release(&str);
 937         fclose(fp);
 938 }
 939
 940 /*
 941  * We use the convention that exiting with an exit code 10 means that
 942  * the bisection process finished successfully.
 943  * In this case the calling shell script should exit 0.
 944  *
 945  * If no_checkout is non-zero, the bisection process does not
 946  * checkout the trial commit but instead simply updates BISECT_HEAD.
 947  */
 948 int bisect_next_all(struct repository *r, const char *prefix, int no_checkout)
 949 {
 950         struct rev_info revs;
 951         struct commit_list *tried;
 952         int reaches = 0, all = 0, nr, steps;
 953         struct object_id *bisect_rev;
 954         char *steps_msg;
 955
 956         read_bisect_terms(&term_bad, &term_good);
 957         if (read_bisect_refs())
 958                 die(_("reading bisect refs failed"));
 959
 960         check_good_are_ancestors_of_bad(r, prefix, no_checkout);
 961
 962         bisect_rev_setup(r, &revs, prefix, "%s", "^%s", 1);
 963         revs.limited = 1;
 964
 965         bisect_common(&revs);
 966
 967         find_bisection(&revs.commits, &reaches, &all, !!skipped_revs.nr);
 968         revs.commits = managed_skipped(revs.commits, &tried);
 969
 970         if (!revs.commits) {
 971                 /*
 972                  * We should exit here only if the "bad"
 973                  * commit is also a "skip" commit.
 974                  */
 975                 exit_if_skipped_commits(tried, NULL);
 976
 977                 printf(_("%s was both %s and %s\n"),
 978                        oid_to_hex(current_bad_oid),
 979                        term_good,
 980                        term_bad);
 981                 exit(1);
 982         }
 983
 984         if (!all) {
 985                 fprintf(stderr, _("No testable commit found.\n"
 986                         "Maybe you started with bad path parameters?\n"));
 987                 exit(4);
 988         }
 989
 990         bisect_rev = &revs.commits->item->object.oid;
 991
 992         if (oideq(bisect_rev, current_bad_oid)) {
 993                 exit_if_skipped_commits(tried, current_bad_oid);
 994                 printf("%s is the first %s commit\n", oid_to_hex(bisect_rev),
 995                         term_bad);
 996                 show_diff_tree(r, prefix, revs.commits->item);
 997                 /* This means the bisection process succeeded. */
 998                 exit(10);
 999         }
1000
1001         nr = all - reaches - 1;
1002         steps = estimate_bisect_steps(all);
1003
1004         steps_msg = xstrfmt(Q_("(roughly %d step)", "(roughly %d steps)",
1005                   steps), steps);
1006         /*
1007          * TRANSLATORS: the last %s will be replaced with "(roughly %d
1008          * steps)" translation.
1009          */
1010         printf(Q_("Bisecting: %d revision left to test after this %s\n",
1011                   "Bisecting: %d revisions left to test after this %s\n",
1012                   nr), nr, steps_msg);
1013         free(steps_msg);
1014
1015         return bisect_checkout(bisect_rev, no_checkout);
1016 }
1017
1018 static inline int log2i(int n)
1019 {
1020         int log2 = 0;
1021
1022         for (; n > 1; n >>= 1)
1023                 log2++;
1024
1025         return log2;
1026 }
1027
1028 static inline int exp2i(int n)
1029 {
1030         return 1 << n;
1031 }
1032
1033 /*
1034  * Estimate the number of bisect steps left (after the current step)
1035  *
1036  * For any x between 0 included and 2^n excluded, the probability for
1037  * n - 1 steps left looks like:
1038  *
1039  * P(2^n + x) == (2^n - x) / (2^n + x)
1040  *
1041  * and P(2^n + x) < 0.5 means 2^n < 3x
1042  */
1043 int estimate_bisect_steps(int all)
1044 {
1045         int n, x, e;
1046
1047         if (all < 3)
1048                 return 0;
1049
1050         n = log2i(all);
1051         e = exp2i(n);
1052         x = all - e;
1053
1054         return (e < 3 * x) ? n : n - 1;
1055 }
1056
1057 static int mark_for_removal(const char *refname, const struct object_id *oid,
1058                             int flag, void *cb_data)
1059 {
1060         struct string_list *refs = cb_data;
1061         char *ref = xstrfmt("refs/bisect%s", refname);
1062         string_list_append(refs, ref);
1063         return 0;
1064 }
1065
1066 int bisect_clean_state(void)
1067 {
1068         int result = 0;
1069
1070         /* There may be some refs packed during bisection */
1071         struct string_list refs_for_removal = STRING_LIST_INIT_NODUP;
1072         for_each_ref_in("refs/bisect", mark_for_removal, (void *) &refs_for_removal);
1073         string_list_append(&refs_for_removal, xstrdup("BISECT_HEAD"));
1074         result = delete_refs("bisect: remove", &refs_for_removal, REF_NO_DEREF);
1075         refs_for_removal.strdup_strings = 1;
1076         string_list_clear(&refs_for_removal, 0);
1077         unlink_or_warn(git_path_bisect_expected_rev());
1078         unlink_or_warn(git_path_bisect_ancestors_ok());
1079         unlink_or_warn(git_path_bisect_log());
1080         unlink_or_warn(git_path_bisect_names());
1081         unlink_or_warn(git_path_bisect_run());
1082         unlink_or_warn(git_path_bisect_terms());
1083         /* Cleanup head-name if it got left by an old version of git-bisect */
1084         unlink_or_warn(git_path_head_name());
1085         /*
1086          * Cleanup BISECT_START last to support the --no-checkout option
1087          * introduced in the commit 4796e823a.
1088          */
1089         unlink_or_warn(git_path_bisect_start());
1090
1091         return result;
1092 }