Git 2.19.5
[git] / 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
17 static struct oid_array good_revs;
18 static struct oid_array skipped_revs;
19
20 static struct object_id *current_bad_oid;
21
22 static const char *argv_checkout[] = {"checkout", "-q", NULL, "--", NULL};
23 static const char *argv_show_branch[] = {"show-branch", NULL, NULL};
24
25 static const char *term_bad;
26 static const char *term_good;
27
28 /* Remember to update object flag allocation in object.h */
29 #define COUNTED         (1u<<16)
30
31 /*
32  * This is a truly stupid algorithm, but it's only
33  * used for bisection, and we just don't care enough.
34  *
35  * We care just barely enough to avoid recursing for
36  * non-merge entries.
37  */
38 static int count_distance(struct commit_list *entry)
39 {
40         int nr = 0;
41
42         while (entry) {
43                 struct commit *commit = entry->item;
44                 struct commit_list *p;
45
46                 if (commit->object.flags & (UNINTERESTING | COUNTED))
47                         break;
48                 if (!(commit->object.flags & TREESAME))
49                         nr++;
50                 commit->object.flags |= COUNTED;
51                 p = commit->parents;
52                 entry = p;
53                 if (p) {
54                         p = p->next;
55                         while (p) {
56                                 nr += count_distance(p);
57                                 p = p->next;
58                         }
59                 }
60         }
61
62         return nr;
63 }
64
65 static void clear_distance(struct commit_list *list)
66 {
67         while (list) {
68                 struct commit *commit = list->item;
69                 commit->object.flags &= ~COUNTED;
70                 list = list->next;
71         }
72 }
73
74 define_commit_slab(commit_weight, int *);
75 static struct commit_weight commit_weight;
76
77 #define DEBUG_BISECT 0
78
79 static inline int weight(struct commit_list *elem)
80 {
81         return **commit_weight_at(&commit_weight, elem->item);
82 }
83
84 static inline void weight_set(struct commit_list *elem, int weight)
85 {
86         **commit_weight_at(&commit_weight, elem->item) = weight;
87 }
88
89 static int count_interesting_parents(struct commit *commit)
90 {
91         struct commit_list *p;
92         int count;
93
94         for (count = 0, p = commit->parents; p; p = p->next) {
95                 if (p->item->object.flags & UNINTERESTING)
96                         continue;
97                 count++;
98         }
99         return count;
100 }
101
102 static inline int halfway(struct commit_list *p, int nr)
103 {
104         /*
105          * Don't short-cut something we are not going to return!
106          */
107         if (p->item->object.flags & TREESAME)
108                 return 0;
109         if (DEBUG_BISECT)
110                 return 0;
111         /*
112          * 2 and 3 are halfway of 5.
113          * 3 is halfway of 6 but 2 and 4 are not.
114          */
115         switch (2 * weight(p) - nr) {
116         case -1: case 0: case 1:
117                 return 1;
118         default:
119                 return 0;
120         }
121 }
122
123 #if !DEBUG_BISECT
124 #define show_list(a,b,c,d) do { ; } while (0)
125 #else
126 static void show_list(const char *debug, int counted, int nr,
127                       struct commit_list *list)
128 {
129         struct commit_list *p;
130
131         fprintf(stderr, "%s (%d/%d)\n", debug, counted, nr);
132
133         for (p = list; p; p = p->next) {
134                 struct commit_list *pp;
135                 struct commit *commit = p->item;
136                 unsigned flags = commit->object.flags;
137                 enum object_type type;
138                 unsigned long size;
139                 char *buf = read_object_file(&commit->object.oid, &type,
140                                              &size);
141                 const char *subject_start;
142                 int subject_len;
143
144                 fprintf(stderr, "%c%c%c ",
145                         (flags & TREESAME) ? ' ' : 'T',
146                         (flags & UNINTERESTING) ? 'U' : ' ',
147                         (flags & COUNTED) ? 'C' : ' ');
148                 if (commit->util)
149                         fprintf(stderr, "%3d", weight(p));
150                 else
151                         fprintf(stderr, "---");
152                 fprintf(stderr, " %.*s", 8, oid_to_hex(&commit->object.oid));
153                 for (pp = commit->parents; pp; pp = pp->next)
154                         fprintf(stderr, " %.*s", 8,
155                                 oid_to_hex(&pp->item->object.oid));
156
157                 subject_len = find_commit_subject(buf, &subject_start);
158                 if (subject_len)
159                         fprintf(stderr, " %.*s", subject_len, subject_start);
160                 fprintf(stderr, "\n");
161         }
162 }
163 #endif /* DEBUG_BISECT */
164
165 static struct commit_list *best_bisection(struct commit_list *list, int nr)
166 {
167         struct commit_list *p, *best;
168         int best_distance = -1;
169
170         best = list;
171         for (p = list; p; p = p->next) {
172                 int distance;
173                 unsigned flags = p->item->object.flags;
174
175                 if (flags & TREESAME)
176                         continue;
177                 distance = weight(p);
178                 if (nr - distance < distance)
179                         distance = nr - distance;
180                 if (distance > best_distance) {
181                         best = p;
182                         best_distance = distance;
183                 }
184         }
185
186         return best;
187 }
188
189 struct commit_dist {
190         struct commit *commit;
191         int distance;
192 };
193
194 static int compare_commit_dist(const void *a_, const void *b_)
195 {
196         struct commit_dist *a, *b;
197
198         a = (struct commit_dist *)a_;
199         b = (struct commit_dist *)b_;
200         if (a->distance != b->distance)
201                 return b->distance - a->distance; /* desc sort */
202         return oidcmp(&a->commit->object.oid, &b->commit->object.oid);
203 }
204
205 static struct commit_list *best_bisection_sorted(struct commit_list *list, int nr)
206 {
207         struct commit_list *p;
208         struct commit_dist *array = xcalloc(nr, sizeof(*array));
209         struct strbuf buf = STRBUF_INIT;
210         int cnt, i;
211
212         for (p = list, cnt = 0; p; p = p->next) {
213                 int distance;
214                 unsigned flags = p->item->object.flags;
215
216                 if (flags & TREESAME)
217                         continue;
218                 distance = weight(p);
219                 if (nr - distance < distance)
220                         distance = nr - distance;
221                 array[cnt].commit = p->item;
222                 array[cnt].distance = distance;
223                 cnt++;
224         }
225         QSORT(array, cnt, compare_commit_dist);
226         for (p = list, i = 0; i < cnt; i++) {
227                 struct object *obj = &(array[i].commit->object);
228
229                 strbuf_reset(&buf);
230                 strbuf_addf(&buf, "dist=%d", array[i].distance);
231                 add_name_decoration(DECORATION_NONE, buf.buf, obj);
232
233                 p->item = array[i].commit;
234                 if (i < cnt - 1)
235                         p = p->next;
236         }
237         if (p) {
238                 free_commit_list(p->next);
239                 p->next = NULL;
240         }
241         strbuf_release(&buf);
242         free(array);
243         return list;
244 }
245
246 /*
247  * zero or positive weight is the number of interesting commits it can
248  * reach, including itself.  Especially, weight = 0 means it does not
249  * reach any tree-changing commits (e.g. just above uninteresting one
250  * but traversal is with pathspec).
251  *
252  * weight = -1 means it has one parent and its distance is yet to
253  * be computed.
254  *
255  * weight = -2 means it has more than one parent and its distance is
256  * unknown.  After running count_distance() first, they will get zero
257  * or positive distance.
258  */
259 static struct commit_list *do_find_bisection(struct commit_list *list,
260                                              int nr, int *weights,
261                                              int find_all)
262 {
263         int n, counted;
264         struct commit_list *p;
265
266         counted = 0;
267
268         for (n = 0, p = list; p; p = p->next) {
269                 struct commit *commit = p->item;
270                 unsigned flags = commit->object.flags;
271
272                 *commit_weight_at(&commit_weight, p->item) = &weights[n++];
273                 switch (count_interesting_parents(commit)) {
274                 case 0:
275                         if (!(flags & TREESAME)) {
276                                 weight_set(p, 1);
277                                 counted++;
278                                 show_list("bisection 2 count one",
279                                           counted, nr, list);
280                         }
281                         /*
282                          * otherwise, it is known not to reach any
283                          * tree-changing commit and gets weight 0.
284                          */
285                         break;
286                 case 1:
287                         weight_set(p, -1);
288                         break;
289                 default:
290                         weight_set(p, -2);
291                         break;
292                 }
293         }
294
295         show_list("bisection 2 initialize", counted, nr, list);
296
297         /*
298          * If you have only one parent in the resulting set
299          * then you can reach one commit more than that parent
300          * can reach.  So we do not have to run the expensive
301          * count_distance() for single strand of pearls.
302          *
303          * However, if you have more than one parents, you cannot
304          * just add their distance and one for yourself, since
305          * they usually reach the same ancestor and you would
306          * end up counting them twice that way.
307          *
308          * So we will first count distance of merges the usual
309          * way, and then fill the blanks using cheaper algorithm.
310          */
311         for (p = list; p; p = p->next) {
312                 if (p->item->object.flags & UNINTERESTING)
313                         continue;
314                 if (weight(p) != -2)
315                         continue;
316                 weight_set(p, count_distance(p));
317                 clear_distance(list);
318
319                 /* Does it happen to be at exactly half-way? */
320                 if (!find_all && halfway(p, nr))
321                         return p;
322                 counted++;
323         }
324
325         show_list("bisection 2 count_distance", counted, nr, list);
326
327         while (counted < nr) {
328                 for (p = list; p; p = p->next) {
329                         struct commit_list *q;
330                         unsigned flags = p->item->object.flags;
331
332                         if (0 <= weight(p))
333                                 continue;
334                         for (q = p->item->parents; q; q = q->next) {
335                                 if (q->item->object.flags & UNINTERESTING)
336                                         continue;
337                                 if (0 <= weight(q))
338                                         break;
339                         }
340                         if (!q)
341                                 continue;
342
343                         /*
344                          * weight for p is unknown but q is known.
345                          * add one for p itself if p is to be counted,
346                          * otherwise inherit it from q directly.
347                          */
348                         if (!(flags & TREESAME)) {
349                                 weight_set(p, weight(q)+1);
350                                 counted++;
351                                 show_list("bisection 2 count one",
352                                           counted, nr, list);
353                         }
354                         else
355                                 weight_set(p, weight(q));
356
357                         /* Does it happen to be at exactly half-way? */
358                         if (!find_all && halfway(p, nr))
359                                 return p;
360                 }
361         }
362
363         show_list("bisection 2 counted all", counted, nr, list);
364
365         if (!find_all)
366                 return best_bisection(list, nr);
367         else
368                 return best_bisection_sorted(list, nr);
369 }
370
371 void find_bisection(struct commit_list **commit_list, int *reaches,
372                     int *all, int find_all)
373 {
374         int nr, on_list;
375         struct commit_list *list, *p, *best, *next, *last;
376         int *weights;
377
378         show_list("bisection 2 entry", 0, 0, *commit_list);
379         init_commit_weight(&commit_weight);
380
381         /*
382          * Count the number of total and tree-changing items on the
383          * list, while reversing the list.
384          */
385         for (nr = on_list = 0, last = NULL, p = *commit_list;
386              p;
387              p = next) {
388                 unsigned flags = p->item->object.flags;
389
390                 next = p->next;
391                 if (flags & UNINTERESTING) {
392                         free(p);
393                         continue;
394                 }
395                 p->next = last;
396                 last = p;
397                 if (!(flags & TREESAME))
398                         nr++;
399                 on_list++;
400         }
401         list = last;
402         show_list("bisection 2 sorted", 0, nr, list);
403
404         *all = nr;
405         weights = xcalloc(on_list, sizeof(*weights));
406
407         /* Do the real work of finding bisection commit. */
408         best = do_find_bisection(list, nr, weights, find_all);
409         if (best) {
410                 if (!find_all) {
411                         list->item = best->item;
412                         free_commit_list(list->next);
413                         best = list;
414                         best->next = NULL;
415                 }
416                 *reaches = weight(best);
417         }
418         free(weights);
419         *commit_list = best;
420         clear_commit_weight(&commit_weight);
421 }
422
423 static int register_ref(const char *refname, const struct object_id *oid,
424                         int flags, void *cb_data)
425 {
426         struct strbuf good_prefix = STRBUF_INIT;
427         strbuf_addstr(&good_prefix, term_good);
428         strbuf_addstr(&good_prefix, "-");
429
430         if (!strcmp(refname, term_bad)) {
431                 current_bad_oid = xmalloc(sizeof(*current_bad_oid));
432                 oidcpy(current_bad_oid, oid);
433         } else if (starts_with(refname, good_prefix.buf)) {
434                 oid_array_append(&good_revs, oid);
435         } else if (starts_with(refname, "skip-")) {
436                 oid_array_append(&skipped_revs, oid);
437         }
438
439         strbuf_release(&good_prefix);
440
441         return 0;
442 }
443
444 static int read_bisect_refs(void)
445 {
446         return for_each_ref_in("refs/bisect/", register_ref, NULL);
447 }
448
449 static GIT_PATH_FUNC(git_path_bisect_names, "BISECT_NAMES")
450 static GIT_PATH_FUNC(git_path_bisect_expected_rev, "BISECT_EXPECTED_REV")
451 static GIT_PATH_FUNC(git_path_bisect_ancestors_ok, "BISECT_ANCESTORS_OK")
452 static GIT_PATH_FUNC(git_path_bisect_run, "BISECT_RUN")
453 static GIT_PATH_FUNC(git_path_bisect_start, "BISECT_START")
454 static GIT_PATH_FUNC(git_path_bisect_log, "BISECT_LOG")
455 static GIT_PATH_FUNC(git_path_bisect_terms, "BISECT_TERMS")
456 static GIT_PATH_FUNC(git_path_head_name, "head-name")
457
458 static void read_bisect_paths(struct argv_array *array)
459 {
460         struct strbuf str = STRBUF_INIT;
461         const char *filename = git_path_bisect_names();
462         FILE *fp = xfopen(filename, "r");
463
464         while (strbuf_getline_lf(&str, fp) != EOF) {
465                 strbuf_trim(&str);
466                 if (sq_dequote_to_argv_array(str.buf, array))
467                         die(_("Badly quoted content in file '%s': %s"),
468                             filename, str.buf);
469         }
470
471         strbuf_release(&str);
472         fclose(fp);
473 }
474
475 static char *join_sha1_array_hex(struct oid_array *array, char delim)
476 {
477         struct strbuf joined_hexs = STRBUF_INIT;
478         int i;
479
480         for (i = 0; i < array->nr; i++) {
481                 strbuf_addstr(&joined_hexs, oid_to_hex(array->oid + i));
482                 if (i + 1 < array->nr)
483                         strbuf_addch(&joined_hexs, delim);
484         }
485
486         return strbuf_detach(&joined_hexs, NULL);
487 }
488
489 /*
490  * In this function, passing a not NULL skipped_first is very special.
491  * It means that we want to know if the first commit in the list is
492  * skipped because we will want to test a commit away from it if it is
493  * indeed skipped.
494  * So if the first commit is skipped, we cannot take the shortcut to
495  * just "return list" when we find the first non skipped commit, we
496  * have to return a fully filtered list.
497  *
498  * We use (*skipped_first == -1) to mean "it has been found that the
499  * first commit is not skipped". In this case *skipped_first is set back
500  * to 0 just before the function returns.
501  */
502 struct commit_list *filter_skipped(struct commit_list *list,
503                                    struct commit_list **tried,
504                                    int show_all,
505                                    int *count,
506                                    int *skipped_first)
507 {
508         struct commit_list *filtered = NULL, **f = &filtered;
509
510         *tried = NULL;
511
512         if (skipped_first)
513                 *skipped_first = 0;
514         if (count)
515                 *count = 0;
516
517         if (!skipped_revs.nr)
518                 return list;
519
520         while (list) {
521                 struct commit_list *next = list->next;
522                 list->next = NULL;
523                 if (0 <= oid_array_lookup(&skipped_revs, &list->item->object.oid)) {
524                         if (skipped_first && !*skipped_first)
525                                 *skipped_first = 1;
526                         /* Move current to tried list */
527                         *tried = list;
528                         tried = &list->next;
529                 } else {
530                         if (!show_all) {
531                                 if (!skipped_first || !*skipped_first)
532                                         return list;
533                         } else if (skipped_first && !*skipped_first) {
534                                 /* This means we know it's not skipped */
535                                 *skipped_first = -1;
536                         }
537                         /* Move current to filtered list */
538                         *f = list;
539                         f = &list->next;
540                         if (count)
541                                 (*count)++;
542                 }
543                 list = next;
544         }
545
546         if (skipped_first && *skipped_first == -1)
547                 *skipped_first = 0;
548
549         return filtered;
550 }
551
552 #define PRN_MODULO 32768
553
554 /*
555  * This is a pseudo random number generator based on "man 3 rand".
556  * It is not used properly because the seed is the argument and it
557  * is increased by one between each call, but that should not matter
558  * for this application.
559  */
560 static unsigned get_prn(unsigned count) {
561         count = count * 1103515245 + 12345;
562         return (count/65536) % PRN_MODULO;
563 }
564
565 /*
566  * Custom integer square root from
567  * https://en.wikipedia.org/wiki/Integer_square_root
568  */
569 static int sqrti(int val)
570 {
571         float d, x = val;
572
573         if (val == 0)
574                 return 0;
575
576         do {
577                 float y = (x + (float)val / x) / 2;
578                 d = (y > x) ? y - x : x - y;
579                 x = y;
580         } while (d >= 0.5);
581
582         return (int)x;
583 }
584
585 static struct commit_list *skip_away(struct commit_list *list, int count)
586 {
587         struct commit_list *cur, *previous;
588         int prn, index, i;
589
590         prn = get_prn(count);
591         index = (count * prn / PRN_MODULO) * sqrti(prn) / sqrti(PRN_MODULO);
592
593         cur = list;
594         previous = NULL;
595
596         for (i = 0; cur; cur = cur->next, i++) {
597                 if (i == index) {
598                         if (oidcmp(&cur->item->object.oid, current_bad_oid))
599                                 return cur;
600                         if (previous)
601                                 return previous;
602                         return list;
603                 }
604                 previous = cur;
605         }
606
607         return list;
608 }
609
610 static struct commit_list *managed_skipped(struct commit_list *list,
611                                            struct commit_list **tried)
612 {
613         int count, skipped_first;
614
615         *tried = NULL;
616
617         if (!skipped_revs.nr)
618                 return list;
619
620         list = filter_skipped(list, tried, 0, &count, &skipped_first);
621
622         if (!skipped_first)
623                 return list;
624
625         return skip_away(list, count);
626 }
627
628 static void bisect_rev_setup(struct rev_info *revs, const char *prefix,
629                              const char *bad_format, const char *good_format,
630                              int read_paths)
631 {
632         struct argv_array rev_argv = ARGV_ARRAY_INIT;
633         int i;
634
635         init_revisions(revs, prefix);
636         revs->abbrev = 0;
637         revs->commit_format = CMIT_FMT_UNSPECIFIED;
638
639         /* rev_argv.argv[0] will be ignored by setup_revisions */
640         argv_array_push(&rev_argv, "bisect_rev_setup");
641         argv_array_pushf(&rev_argv, bad_format, oid_to_hex(current_bad_oid));
642         for (i = 0; i < good_revs.nr; i++)
643                 argv_array_pushf(&rev_argv, good_format,
644                                  oid_to_hex(good_revs.oid + i));
645         argv_array_push(&rev_argv, "--");
646         if (read_paths)
647                 read_bisect_paths(&rev_argv);
648
649         setup_revisions(rev_argv.argc, rev_argv.argv, revs, NULL);
650         /* XXX leak rev_argv, as "revs" may still be pointing to it */
651 }
652
653 static void bisect_common(struct rev_info *revs)
654 {
655         if (prepare_revision_walk(revs))
656                 die("revision walk setup failed");
657         if (revs->tree_objects)
658                 mark_edges_uninteresting(revs, NULL);
659 }
660
661 static void exit_if_skipped_commits(struct commit_list *tried,
662                                     const struct object_id *bad)
663 {
664         if (!tried)
665                 return;
666
667         printf("There are only 'skip'ped commits left to test.\n"
668                "The first %s commit could be any of:\n", term_bad);
669
670         for ( ; tried; tried = tried->next)
671                 printf("%s\n", oid_to_hex(&tried->item->object.oid));
672
673         if (bad)
674                 printf("%s\n", oid_to_hex(bad));
675         printf(_("We cannot bisect more!\n"));
676         exit(2);
677 }
678
679 static int is_expected_rev(const struct object_id *oid)
680 {
681         const char *filename = git_path_bisect_expected_rev();
682         struct stat st;
683         struct strbuf str = STRBUF_INIT;
684         FILE *fp;
685         int res = 0;
686
687         if (stat(filename, &st) || !S_ISREG(st.st_mode))
688                 return 0;
689
690         fp = fopen_or_warn(filename, "r");
691         if (!fp)
692                 return 0;
693
694         if (strbuf_getline_lf(&str, fp) != EOF)
695                 res = !strcmp(str.buf, oid_to_hex(oid));
696
697         strbuf_release(&str);
698         fclose(fp);
699
700         return res;
701 }
702
703 static int bisect_checkout(const struct object_id *bisect_rev, int no_checkout)
704 {
705         char bisect_rev_hex[GIT_MAX_HEXSZ + 1];
706
707         memcpy(bisect_rev_hex, oid_to_hex(bisect_rev), GIT_SHA1_HEXSZ + 1);
708         update_ref(NULL, "BISECT_EXPECTED_REV", bisect_rev, NULL, 0, UPDATE_REFS_DIE_ON_ERR);
709
710         argv_checkout[2] = bisect_rev_hex;
711         if (no_checkout) {
712                 update_ref(NULL, "BISECT_HEAD", bisect_rev, NULL, 0,
713                            UPDATE_REFS_DIE_ON_ERR);
714         } else {
715                 int res;
716                 res = run_command_v_opt(argv_checkout, RUN_GIT_CMD);
717                 if (res)
718                         exit(res);
719         }
720
721         argv_show_branch[1] = bisect_rev_hex;
722         return run_command_v_opt(argv_show_branch, RUN_GIT_CMD);
723 }
724
725 static struct commit *get_commit_reference(const struct object_id *oid)
726 {
727         struct commit *r = lookup_commit_reference(the_repository, oid);
728         if (!r)
729                 die(_("Not a valid commit name %s"), oid_to_hex(oid));
730         return r;
731 }
732
733 static struct commit **get_bad_and_good_commits(int *rev_nr)
734 {
735         struct commit **rev;
736         int i, n = 0;
737
738         ALLOC_ARRAY(rev, 1 + good_revs.nr);
739         rev[n++] = get_commit_reference(current_bad_oid);
740         for (i = 0; i < good_revs.nr; i++)
741                 rev[n++] = get_commit_reference(good_revs.oid + i);
742         *rev_nr = n;
743
744         return rev;
745 }
746
747 static void handle_bad_merge_base(void)
748 {
749         if (is_expected_rev(current_bad_oid)) {
750                 char *bad_hex = oid_to_hex(current_bad_oid);
751                 char *good_hex = join_sha1_array_hex(&good_revs, ' ');
752                 if (!strcmp(term_bad, "bad") && !strcmp(term_good, "good")) {
753                         fprintf(stderr, _("The merge base %s is bad.\n"
754                                 "This means the bug has been fixed "
755                                 "between %s and [%s].\n"),
756                                 bad_hex, bad_hex, good_hex);
757                 } else if (!strcmp(term_bad, "new") && !strcmp(term_good, "old")) {
758                         fprintf(stderr, _("The merge base %s is new.\n"
759                                 "The property has changed "
760                                 "between %s and [%s].\n"),
761                                 bad_hex, bad_hex, good_hex);
762                 } else {
763                         fprintf(stderr, _("The merge base %s is %s.\n"
764                                 "This means the first '%s' commit is "
765                                 "between %s and [%s].\n"),
766                                 bad_hex, term_bad, term_good, bad_hex, good_hex);
767                 }
768                 exit(3);
769         }
770
771         fprintf(stderr, _("Some %s revs are not ancestors of the %s rev.\n"
772                 "git bisect cannot work properly in this case.\n"
773                 "Maybe you mistook %s and %s revs?\n"),
774                 term_good, term_bad, term_good, term_bad);
775         exit(1);
776 }
777
778 static void handle_skipped_merge_base(const struct object_id *mb)
779 {
780         char *mb_hex = oid_to_hex(mb);
781         char *bad_hex = oid_to_hex(current_bad_oid);
782         char *good_hex = join_sha1_array_hex(&good_revs, ' ');
783
784         warning(_("the merge base between %s and [%s] "
785                 "must be skipped.\n"
786                 "So we cannot be sure the first %s commit is "
787                 "between %s and %s.\n"
788                 "We continue anyway."),
789                 bad_hex, good_hex, term_bad, mb_hex, bad_hex);
790         free(good_hex);
791 }
792
793 /*
794  * "check_merge_bases" checks that merge bases are not "bad" (or "new").
795  *
796  * - If one is "bad" (or "new"), it means the user assumed something wrong
797  * and we must exit with a non 0 error code.
798  * - If one is "good" (or "old"), that's good, we have nothing to do.
799  * - If one is "skipped", we can't know but we should warn.
800  * - If we don't know, we should check it out and ask the user to test.
801  */
802 static void check_merge_bases(int rev_nr, struct commit **rev, int no_checkout)
803 {
804         struct commit_list *result;
805
806         result = get_merge_bases_many(rev[0], rev_nr - 1, rev + 1);
807
808         for (; result; result = result->next) {
809                 const struct object_id *mb = &result->item->object.oid;
810                 if (!oidcmp(mb, current_bad_oid)) {
811                         handle_bad_merge_base();
812                 } else if (0 <= oid_array_lookup(&good_revs, mb)) {
813                         continue;
814                 } else if (0 <= oid_array_lookup(&skipped_revs, mb)) {
815                         handle_skipped_merge_base(mb);
816                 } else {
817                         printf(_("Bisecting: a merge base must be tested\n"));
818                         exit(bisect_checkout(mb, no_checkout));
819                 }
820         }
821
822         free_commit_list(result);
823 }
824
825 static int check_ancestors(int rev_nr, struct commit **rev, const char *prefix)
826 {
827         struct rev_info revs;
828         int res;
829
830         bisect_rev_setup(&revs, prefix, "^%s", "%s", 0);
831
832         bisect_common(&revs);
833         res = (revs.commits != NULL);
834
835         /* Clean up objects used, as they will be reused. */
836         clear_commit_marks_many(rev_nr, rev, ALL_REV_FLAGS);
837
838         return res;
839 }
840
841 /*
842  * "check_good_are_ancestors_of_bad" checks that all "good" revs are
843  * ancestor of the "bad" rev.
844  *
845  * If that's not the case, we need to check the merge bases.
846  * If a merge base must be tested by the user, its source code will be
847  * checked out to be tested by the user and we will exit.
848  */
849 static void check_good_are_ancestors_of_bad(const char *prefix, int no_checkout)
850 {
851         char *filename = git_pathdup("BISECT_ANCESTORS_OK");
852         struct stat st;
853         int fd, rev_nr;
854         struct commit **rev;
855
856         if (!current_bad_oid)
857                 die(_("a %s revision is needed"), term_bad);
858
859         /* Check if file BISECT_ANCESTORS_OK exists. */
860         if (!stat(filename, &st) && S_ISREG(st.st_mode))
861                 goto done;
862
863         /* Bisecting with no good rev is ok. */
864         if (good_revs.nr == 0)
865                 goto done;
866
867         /* Check if all good revs are ancestor of the bad rev. */
868         rev = get_bad_and_good_commits(&rev_nr);
869         if (check_ancestors(rev_nr, rev, prefix))
870                 check_merge_bases(rev_nr, rev, no_checkout);
871         free(rev);
872
873         /* Create file BISECT_ANCESTORS_OK. */
874         fd = open(filename, O_CREAT | O_TRUNC | O_WRONLY, 0600);
875         if (fd < 0)
876                 warning_errno(_("could not create file '%s'"),
877                               filename);
878         else
879                 close(fd);
880  done:
881         free(filename);
882 }
883
884 /*
885  * This does "git diff-tree --pretty COMMIT" without one fork+exec.
886  */
887 static void show_diff_tree(const char *prefix, struct commit *commit)
888 {
889         struct rev_info opt;
890
891         /* diff-tree init */
892         init_revisions(&opt, prefix);
893         git_config(git_diff_basic_config, NULL); /* no "diff" UI options */
894         opt.abbrev = 0;
895         opt.diff = 1;
896
897         /* This is what "--pretty" does */
898         opt.verbose_header = 1;
899         opt.use_terminator = 0;
900         opt.commit_format = CMIT_FMT_DEFAULT;
901
902         /* diff-tree init */
903         if (!opt.diffopt.output_format)
904                 opt.diffopt.output_format = DIFF_FORMAT_RAW;
905
906         setup_revisions(0, NULL, &opt, NULL);
907         log_tree_commit(&opt, commit);
908 }
909
910 /*
911  * The terms used for this bisect session are stored in BISECT_TERMS.
912  * We read them and store them to adapt the messages accordingly.
913  * Default is bad/good.
914  */
915 void read_bisect_terms(const char **read_bad, const char **read_good)
916 {
917         struct strbuf str = STRBUF_INIT;
918         const char *filename = git_path_bisect_terms();
919         FILE *fp = fopen(filename, "r");
920
921         if (!fp) {
922                 if (errno == ENOENT) {
923                         *read_bad = "bad";
924                         *read_good = "good";
925                         return;
926                 } else {
927                         die_errno(_("could not read file '%s'"), filename);
928                 }
929         } else {
930                 strbuf_getline_lf(&str, fp);
931                 *read_bad = strbuf_detach(&str, NULL);
932                 strbuf_getline_lf(&str, fp);
933                 *read_good = strbuf_detach(&str, NULL);
934         }
935         strbuf_release(&str);
936         fclose(fp);
937 }
938
939 /*
940  * We use the convention that exiting with an exit code 10 means that
941  * the bisection process finished successfully.
942  * In this case the calling shell script should exit 0.
943  *
944  * If no_checkout is non-zero, the bisection process does not
945  * checkout the trial commit but instead simply updates BISECT_HEAD.
946  */
947 int bisect_next_all(const char *prefix, int no_checkout)
948 {
949         struct rev_info revs;
950         struct commit_list *tried;
951         int reaches = 0, all = 0, nr, steps;
952         struct object_id *bisect_rev;
953         char *steps_msg;
954
955         read_bisect_terms(&term_bad, &term_good);
956         if (read_bisect_refs())
957                 die(_("reading bisect refs failed"));
958
959         check_good_are_ancestors_of_bad(prefix, no_checkout);
960
961         bisect_rev_setup(&revs, prefix, "%s", "^%s", 1);
962         revs.limited = 1;
963
964         bisect_common(&revs);
965
966         find_bisection(&revs.commits, &reaches, &all, !!skipped_revs.nr);
967         revs.commits = managed_skipped(revs.commits, &tried);
968
969         if (!revs.commits) {
970                 /*
971                  * We should exit here only if the "bad"
972                  * commit is also a "skip" commit.
973                  */
974                 exit_if_skipped_commits(tried, NULL);
975
976                 printf(_("%s was both %s and %s\n"),
977                        oid_to_hex(current_bad_oid),
978                        term_good,
979                        term_bad);
980                 exit(1);
981         }
982
983         if (!all) {
984                 fprintf(stderr, _("No testable commit found.\n"
985                         "Maybe you started with bad path parameters?\n"));
986                 exit(4);
987         }
988
989         bisect_rev = &revs.commits->item->object.oid;
990
991         if (!oidcmp(bisect_rev, current_bad_oid)) {
992                 exit_if_skipped_commits(tried, current_bad_oid);
993                 printf("%s is the first %s commit\n", oid_to_hex(bisect_rev),
994                         term_bad);
995                 show_diff_tree(prefix, revs.commits->item);
996                 /* This means the bisection process succeeded. */
997                 exit(10);
998         }
999
1000         nr = all - reaches - 1;
1001         steps = estimate_bisect_steps(all);
1002
1003         steps_msg = xstrfmt(Q_("(roughly %d step)", "(roughly %d steps)",
1004                   steps), steps);
1005         /*
1006          * TRANSLATORS: the last %s will be replaced with "(roughly %d
1007          * steps)" translation.
1008          */
1009         printf(Q_("Bisecting: %d revision left to test after this %s\n",
1010                   "Bisecting: %d revisions left to test after this %s\n",
1011                   nr), nr, steps_msg);
1012         free(steps_msg);
1013
1014         return bisect_checkout(bisect_rev, no_checkout);
1015 }
1016
1017 static inline int log2i(int n)
1018 {
1019         int log2 = 0;
1020
1021         for (; n > 1; n >>= 1)
1022                 log2++;
1023
1024         return log2;
1025 }
1026
1027 static inline int exp2i(int n)
1028 {
1029         return 1 << n;
1030 }
1031
1032 /*
1033  * Estimate the number of bisect steps left (after the current step)
1034  *
1035  * For any x between 0 included and 2^n excluded, the probability for
1036  * n - 1 steps left looks like:
1037  *
1038  * P(2^n + x) == (2^n - x) / (2^n + x)
1039  *
1040  * and P(2^n + x) < 0.5 means 2^n < 3x
1041  */
1042 int estimate_bisect_steps(int all)
1043 {
1044         int n, x, e;
1045
1046         if (all < 3)
1047                 return 0;
1048
1049         n = log2i(all);
1050         e = exp2i(n);
1051         x = all - e;
1052
1053         return (e < 3 * x) ? n : n - 1;
1054 }
1055
1056 static int mark_for_removal(const char *refname, const struct object_id *oid,
1057                             int flag, void *cb_data)
1058 {
1059         struct string_list *refs = cb_data;
1060         char *ref = xstrfmt("refs/bisect%s", refname);
1061         string_list_append(refs, ref);
1062         return 0;
1063 }
1064
1065 int bisect_clean_state(void)
1066 {
1067         int result = 0;
1068
1069         /* There may be some refs packed during bisection */
1070         struct string_list refs_for_removal = STRING_LIST_INIT_NODUP;
1071         for_each_ref_in("refs/bisect", mark_for_removal, (void *) &refs_for_removal);
1072         string_list_append(&refs_for_removal, xstrdup("BISECT_HEAD"));
1073         result = delete_refs("bisect: remove", &refs_for_removal, REF_NO_DEREF);
1074         refs_for_removal.strdup_strings = 1;
1075         string_list_clear(&refs_for_removal, 0);
1076         unlink_or_warn(git_path_bisect_expected_rev());
1077         unlink_or_warn(git_path_bisect_ancestors_ok());
1078         unlink_or_warn(git_path_bisect_log());
1079         unlink_or_warn(git_path_bisect_names());
1080         unlink_or_warn(git_path_bisect_run());
1081         unlink_or_warn(git_path_bisect_terms());
1082         /* Cleanup head-name if it got left by an old version of git-bisect */
1083         unlink_or_warn(git_path_head_name());
1084         /*
1085          * Cleanup BISECT_START last to support the --no-checkout option
1086          * introduced in the commit 4796e823a.
1087          */
1088         unlink_or_warn(git_path_bisect_start());
1089
1090         return result;
1091 }