Merge branch 'ph/pack-objects-mutex-fix'
[git] / revision.c
1 #include "cache.h"
2 #include "object-store.h"
3 #include "tag.h"
4 #include "blob.h"
5 #include "tree.h"
6 #include "commit.h"
7 #include "diff.h"
8 #include "refs.h"
9 #include "revision.h"
10 #include "repository.h"
11 #include "graph.h"
12 #include "grep.h"
13 #include "reflog-walk.h"
14 #include "patch-ids.h"
15 #include "decorate.h"
16 #include "log-tree.h"
17 #include "string-list.h"
18 #include "line-log.h"
19 #include "mailmap.h"
20 #include "commit-slab.h"
21 #include "dir.h"
22 #include "cache-tree.h"
23 #include "bisect.h"
24 #include "packfile.h"
25 #include "worktree.h"
26 #include "argv-array.h"
27 #include "commit-reach.h"
28 #include "commit-graph.h"
29 #include "prio-queue.h"
30
31 volatile show_early_output_fn_t show_early_output;
32
33 static const char *term_bad;
34 static const char *term_good;
35
36 implement_shared_commit_slab(revision_sources, char *);
37
38 void show_object_with_name(FILE *out, struct object *obj, const char *name)
39 {
40         const char *p;
41
42         fprintf(out, "%s ", oid_to_hex(&obj->oid));
43         for (p = name; *p && *p != '\n'; p++)
44                 fputc(*p, out);
45         fputc('\n', out);
46 }
47
48 static void mark_blob_uninteresting(struct blob *blob)
49 {
50         if (!blob)
51                 return;
52         if (blob->object.flags & UNINTERESTING)
53                 return;
54         blob->object.flags |= UNINTERESTING;
55 }
56
57 static void mark_tree_contents_uninteresting(struct repository *r,
58                                              struct tree *tree)
59 {
60         struct tree_desc desc;
61         struct name_entry entry;
62
63         if (parse_tree_gently(tree, 1) < 0)
64                 return;
65
66         init_tree_desc(&desc, tree->buffer, tree->size);
67         while (tree_entry(&desc, &entry)) {
68                 switch (object_type(entry.mode)) {
69                 case OBJ_TREE:
70                         mark_tree_uninteresting(r, lookup_tree(r, &entry.oid));
71                         break;
72                 case OBJ_BLOB:
73                         mark_blob_uninteresting(lookup_blob(r, &entry.oid));
74                         break;
75                 default:
76                         /* Subproject commit - not in this repository */
77                         break;
78                 }
79         }
80
81         /*
82          * We don't care about the tree any more
83          * after it has been marked uninteresting.
84          */
85         free_tree_buffer(tree);
86 }
87
88 void mark_tree_uninteresting(struct repository *r, struct tree *tree)
89 {
90         struct object *obj;
91
92         if (!tree)
93                 return;
94
95         obj = &tree->object;
96         if (obj->flags & UNINTERESTING)
97                 return;
98         obj->flags |= UNINTERESTING;
99         mark_tree_contents_uninteresting(r, tree);
100 }
101
102 struct commit_stack {
103         struct commit **items;
104         size_t nr, alloc;
105 };
106 #define COMMIT_STACK_INIT { NULL, 0, 0 }
107
108 static void commit_stack_push(struct commit_stack *stack, struct commit *commit)
109 {
110         ALLOC_GROW(stack->items, stack->nr + 1, stack->alloc);
111         stack->items[stack->nr++] = commit;
112 }
113
114 static struct commit *commit_stack_pop(struct commit_stack *stack)
115 {
116         return stack->nr ? stack->items[--stack->nr] : NULL;
117 }
118
119 static void commit_stack_clear(struct commit_stack *stack)
120 {
121         FREE_AND_NULL(stack->items);
122         stack->nr = stack->alloc = 0;
123 }
124
125 static void mark_one_parent_uninteresting(struct commit *commit,
126                                           struct commit_stack *pending)
127 {
128         struct commit_list *l;
129
130         if (commit->object.flags & UNINTERESTING)
131                 return;
132         commit->object.flags |= UNINTERESTING;
133
134         /*
135          * Normally we haven't parsed the parent
136          * yet, so we won't have a parent of a parent
137          * here. However, it may turn out that we've
138          * reached this commit some other way (where it
139          * wasn't uninteresting), in which case we need
140          * to mark its parents recursively too..
141          */
142         for (l = commit->parents; l; l = l->next)
143                 commit_stack_push(pending, l->item);
144 }
145
146 void mark_parents_uninteresting(struct commit *commit)
147 {
148         struct commit_stack pending = COMMIT_STACK_INIT;
149         struct commit_list *l;
150
151         for (l = commit->parents; l; l = l->next)
152                 mark_one_parent_uninteresting(l->item, &pending);
153
154         while (pending.nr > 0)
155                 mark_one_parent_uninteresting(commit_stack_pop(&pending),
156                                               &pending);
157
158         commit_stack_clear(&pending);
159 }
160
161 static void add_pending_object_with_path(struct rev_info *revs,
162                                          struct object *obj,
163                                          const char *name, unsigned mode,
164                                          const char *path)
165 {
166         if (!obj)
167                 return;
168         if (revs->no_walk && (obj->flags & UNINTERESTING))
169                 revs->no_walk = 0;
170         if (revs->reflog_info && obj->type == OBJ_COMMIT) {
171                 struct strbuf buf = STRBUF_INIT;
172                 int len = interpret_branch_name(name, 0, &buf, 0);
173
174                 if (0 < len && name[len] && buf.len)
175                         strbuf_addstr(&buf, name + len);
176                 add_reflog_for_walk(revs->reflog_info,
177                                     (struct commit *)obj,
178                                     buf.buf[0] ? buf.buf: name);
179                 strbuf_release(&buf);
180                 return; /* do not add the commit itself */
181         }
182         add_object_array_with_path(obj, name, &revs->pending, mode, path);
183 }
184
185 static void add_pending_object_with_mode(struct rev_info *revs,
186                                          struct object *obj,
187                                          const char *name, unsigned mode)
188 {
189         add_pending_object_with_path(revs, obj, name, mode, NULL);
190 }
191
192 void add_pending_object(struct rev_info *revs,
193                         struct object *obj, const char *name)
194 {
195         add_pending_object_with_mode(revs, obj, name, S_IFINVALID);
196 }
197
198 void add_head_to_pending(struct rev_info *revs)
199 {
200         struct object_id oid;
201         struct object *obj;
202         if (get_oid("HEAD", &oid))
203                 return;
204         obj = parse_object(revs->repo, &oid);
205         if (!obj)
206                 return;
207         add_pending_object(revs, obj, "HEAD");
208 }
209
210 static struct object *get_reference(struct rev_info *revs, const char *name,
211                                     const struct object_id *oid,
212                                     unsigned int flags)
213 {
214         struct object *object;
215
216         /*
217          * If the repository has commit graphs, repo_parse_commit() avoids
218          * reading the object buffer, so use it whenever possible.
219          */
220         if (oid_object_info(revs->repo, oid, NULL) == OBJ_COMMIT) {
221                 struct commit *c = lookup_commit(revs->repo, oid);
222                 if (!repo_parse_commit(revs->repo, c))
223                         object = (struct object *) c;
224                 else
225                         object = NULL;
226         } else {
227                 object = parse_object(revs->repo, oid);
228         }
229
230         if (!object) {
231                 if (revs->ignore_missing)
232                         return object;
233                 if (revs->exclude_promisor_objects && is_promisor_object(oid))
234                         return NULL;
235                 die("bad object %s", name);
236         }
237         object->flags |= flags;
238         return object;
239 }
240
241 void add_pending_oid(struct rev_info *revs, const char *name,
242                       const struct object_id *oid, unsigned int flags)
243 {
244         struct object *object = get_reference(revs, name, oid, flags);
245         add_pending_object(revs, object, name);
246 }
247
248 static struct commit *handle_commit(struct rev_info *revs,
249                                     struct object_array_entry *entry)
250 {
251         struct object *object = entry->item;
252         const char *name = entry->name;
253         const char *path = entry->path;
254         unsigned int mode = entry->mode;
255         unsigned long flags = object->flags;
256
257         /*
258          * Tag object? Look what it points to..
259          */
260         while (object->type == OBJ_TAG) {
261                 struct tag *tag = (struct tag *) object;
262                 if (revs->tag_objects && !(flags & UNINTERESTING))
263                         add_pending_object(revs, object, tag->tag);
264                 if (!tag->tagged)
265                         die("bad tag");
266                 object = parse_object(revs->repo, &tag->tagged->oid);
267                 if (!object) {
268                         if (revs->ignore_missing_links || (flags & UNINTERESTING))
269                                 return NULL;
270                         if (revs->exclude_promisor_objects &&
271                             is_promisor_object(&tag->tagged->oid))
272                                 return NULL;
273                         die("bad object %s", oid_to_hex(&tag->tagged->oid));
274                 }
275                 object->flags |= flags;
276                 /*
277                  * We'll handle the tagged object by looping or dropping
278                  * through to the non-tag handlers below. Do not
279                  * propagate path data from the tag's pending entry.
280                  */
281                 path = NULL;
282                 mode = 0;
283         }
284
285         /*
286          * Commit object? Just return it, we'll do all the complex
287          * reachability crud.
288          */
289         if (object->type == OBJ_COMMIT) {
290                 struct commit *commit = (struct commit *)object;
291
292                 if (parse_commit(commit) < 0)
293                         die("unable to parse commit %s", name);
294                 if (flags & UNINTERESTING) {
295                         mark_parents_uninteresting(commit);
296                         revs->limited = 1;
297                 }
298                 if (revs->sources) {
299                         char **slot = revision_sources_at(revs->sources, commit);
300
301                         if (!*slot)
302                                 *slot = xstrdup(name);
303                 }
304                 return commit;
305         }
306
307         /*
308          * Tree object? Either mark it uninteresting, or add it
309          * to the list of objects to look at later..
310          */
311         if (object->type == OBJ_TREE) {
312                 struct tree *tree = (struct tree *)object;
313                 if (!revs->tree_objects)
314                         return NULL;
315                 if (flags & UNINTERESTING) {
316                         mark_tree_contents_uninteresting(revs->repo, tree);
317                         return NULL;
318                 }
319                 add_pending_object_with_path(revs, object, name, mode, path);
320                 return NULL;
321         }
322
323         /*
324          * Blob object? You know the drill by now..
325          */
326         if (object->type == OBJ_BLOB) {
327                 if (!revs->blob_objects)
328                         return NULL;
329                 if (flags & UNINTERESTING)
330                         return NULL;
331                 add_pending_object_with_path(revs, object, name, mode, path);
332                 return NULL;
333         }
334         die("%s is unknown object", name);
335 }
336
337 static int everybody_uninteresting(struct commit_list *orig,
338                                    struct commit **interesting_cache)
339 {
340         struct commit_list *list = orig;
341
342         if (*interesting_cache) {
343                 struct commit *commit = *interesting_cache;
344                 if (!(commit->object.flags & UNINTERESTING))
345                         return 0;
346         }
347
348         while (list) {
349                 struct commit *commit = list->item;
350                 list = list->next;
351                 if (commit->object.flags & UNINTERESTING)
352                         continue;
353
354                 *interesting_cache = commit;
355                 return 0;
356         }
357         return 1;
358 }
359
360 /*
361  * A definition of "relevant" commit that we can use to simplify limited graphs
362  * by eliminating side branches.
363  *
364  * A "relevant" commit is one that is !UNINTERESTING (ie we are including it
365  * in our list), or that is a specified BOTTOM commit. Then after computing
366  * a limited list, during processing we can generally ignore boundary merges
367  * coming from outside the graph, (ie from irrelevant parents), and treat
368  * those merges as if they were single-parent. TREESAME is defined to consider
369  * only relevant parents, if any. If we are TREESAME to our on-graph parents,
370  * we don't care if we were !TREESAME to non-graph parents.
371  *
372  * Treating bottom commits as relevant ensures that a limited graph's
373  * connection to the actual bottom commit is not viewed as a side branch, but
374  * treated as part of the graph. For example:
375  *
376  *   ....Z...A---X---o---o---B
377  *        .     /
378  *         W---Y
379  *
380  * When computing "A..B", the A-X connection is at least as important as
381  * Y-X, despite A being flagged UNINTERESTING.
382  *
383  * And when computing --ancestry-path "A..B", the A-X connection is more
384  * important than Y-X, despite both A and Y being flagged UNINTERESTING.
385  */
386 static inline int relevant_commit(struct commit *commit)
387 {
388         return (commit->object.flags & (UNINTERESTING | BOTTOM)) != UNINTERESTING;
389 }
390
391 /*
392  * Return a single relevant commit from a parent list. If we are a TREESAME
393  * commit, and this selects one of our parents, then we can safely simplify to
394  * that parent.
395  */
396 static struct commit *one_relevant_parent(const struct rev_info *revs,
397                                           struct commit_list *orig)
398 {
399         struct commit_list *list = orig;
400         struct commit *relevant = NULL;
401
402         if (!orig)
403                 return NULL;
404
405         /*
406          * For 1-parent commits, or if first-parent-only, then return that
407          * first parent (even if not "relevant" by the above definition).
408          * TREESAME will have been set purely on that parent.
409          */
410         if (revs->first_parent_only || !orig->next)
411                 return orig->item;
412
413         /*
414          * For multi-parent commits, identify a sole relevant parent, if any.
415          * If we have only one relevant parent, then TREESAME will be set purely
416          * with regard to that parent, and we can simplify accordingly.
417          *
418          * If we have more than one relevant parent, or no relevant parents
419          * (and multiple irrelevant ones), then we can't select a parent here
420          * and return NULL.
421          */
422         while (list) {
423                 struct commit *commit = list->item;
424                 list = list->next;
425                 if (relevant_commit(commit)) {
426                         if (relevant)
427                                 return NULL;
428                         relevant = commit;
429                 }
430         }
431         return relevant;
432 }
433
434 /*
435  * The goal is to get REV_TREE_NEW as the result only if the
436  * diff consists of all '+' (and no other changes), REV_TREE_OLD
437  * if the whole diff is removal of old data, and otherwise
438  * REV_TREE_DIFFERENT (of course if the trees are the same we
439  * want REV_TREE_SAME).
440  *
441  * The only time we care about the distinction is when
442  * remove_empty_trees is in effect, in which case we care only about
443  * whether the whole change is REV_TREE_NEW, or if there's another type
444  * of change. Which means we can stop the diff early in either of these
445  * cases:
446  *
447  *   1. We're not using remove_empty_trees at all.
448  *
449  *   2. We saw anything except REV_TREE_NEW.
450  */
451 static int tree_difference = REV_TREE_SAME;
452
453 static void file_add_remove(struct diff_options *options,
454                     int addremove, unsigned mode,
455                     const struct object_id *oid,
456                     int oid_valid,
457                     const char *fullpath, unsigned dirty_submodule)
458 {
459         int diff = addremove == '+' ? REV_TREE_NEW : REV_TREE_OLD;
460         struct rev_info *revs = options->change_fn_data;
461
462         tree_difference |= diff;
463         if (!revs->remove_empty_trees || tree_difference != REV_TREE_NEW)
464                 options->flags.has_changes = 1;
465 }
466
467 static void file_change(struct diff_options *options,
468                  unsigned old_mode, unsigned new_mode,
469                  const struct object_id *old_oid,
470                  const struct object_id *new_oid,
471                  int old_oid_valid, int new_oid_valid,
472                  const char *fullpath,
473                  unsigned old_dirty_submodule, unsigned new_dirty_submodule)
474 {
475         tree_difference = REV_TREE_DIFFERENT;
476         options->flags.has_changes = 1;
477 }
478
479 static int rev_compare_tree(struct rev_info *revs,
480                             struct commit *parent, struct commit *commit)
481 {
482         struct tree *t1 = get_commit_tree(parent);
483         struct tree *t2 = get_commit_tree(commit);
484
485         if (!t1)
486                 return REV_TREE_NEW;
487         if (!t2)
488                 return REV_TREE_OLD;
489
490         if (revs->simplify_by_decoration) {
491                 /*
492                  * If we are simplifying by decoration, then the commit
493                  * is worth showing if it has a tag pointing at it.
494                  */
495                 if (get_name_decoration(&commit->object))
496                         return REV_TREE_DIFFERENT;
497                 /*
498                  * A commit that is not pointed by a tag is uninteresting
499                  * if we are not limited by path.  This means that you will
500                  * see the usual "commits that touch the paths" plus any
501                  * tagged commit by specifying both --simplify-by-decoration
502                  * and pathspec.
503                  */
504                 if (!revs->prune_data.nr)
505                         return REV_TREE_SAME;
506         }
507
508         tree_difference = REV_TREE_SAME;
509         revs->pruning.flags.has_changes = 0;
510         if (diff_tree_oid(&t1->object.oid, &t2->object.oid, "",
511                            &revs->pruning) < 0)
512                 return REV_TREE_DIFFERENT;
513         return tree_difference;
514 }
515
516 static int rev_same_tree_as_empty(struct rev_info *revs, struct commit *commit)
517 {
518         int retval;
519         struct tree *t1 = get_commit_tree(commit);
520
521         if (!t1)
522                 return 0;
523
524         tree_difference = REV_TREE_SAME;
525         revs->pruning.flags.has_changes = 0;
526         retval = diff_tree_oid(NULL, &t1->object.oid, "", &revs->pruning);
527
528         return retval >= 0 && (tree_difference == REV_TREE_SAME);
529 }
530
531 struct treesame_state {
532         unsigned int nparents;
533         unsigned char treesame[FLEX_ARRAY];
534 };
535
536 static struct treesame_state *initialise_treesame(struct rev_info *revs, struct commit *commit)
537 {
538         unsigned n = commit_list_count(commit->parents);
539         struct treesame_state *st = xcalloc(1, st_add(sizeof(*st), n));
540         st->nparents = n;
541         add_decoration(&revs->treesame, &commit->object, st);
542         return st;
543 }
544
545 /*
546  * Must be called immediately after removing the nth_parent from a commit's
547  * parent list, if we are maintaining the per-parent treesame[] decoration.
548  * This does not recalculate the master TREESAME flag - update_treesame()
549  * should be called to update it after a sequence of treesame[] modifications
550  * that may have affected it.
551  */
552 static int compact_treesame(struct rev_info *revs, struct commit *commit, unsigned nth_parent)
553 {
554         struct treesame_state *st;
555         int old_same;
556
557         if (!commit->parents) {
558                 /*
559                  * Have just removed the only parent from a non-merge.
560                  * Different handling, as we lack decoration.
561                  */
562                 if (nth_parent != 0)
563                         die("compact_treesame %u", nth_parent);
564                 old_same = !!(commit->object.flags & TREESAME);
565                 if (rev_same_tree_as_empty(revs, commit))
566                         commit->object.flags |= TREESAME;
567                 else
568                         commit->object.flags &= ~TREESAME;
569                 return old_same;
570         }
571
572         st = lookup_decoration(&revs->treesame, &commit->object);
573         if (!st || nth_parent >= st->nparents)
574                 die("compact_treesame %u", nth_parent);
575
576         old_same = st->treesame[nth_parent];
577         memmove(st->treesame + nth_parent,
578                 st->treesame + nth_parent + 1,
579                 st->nparents - nth_parent - 1);
580
581         /*
582          * If we've just become a non-merge commit, update TREESAME
583          * immediately, and remove the no-longer-needed decoration.
584          * If still a merge, defer update until update_treesame().
585          */
586         if (--st->nparents == 1) {
587                 if (commit->parents->next)
588                         die("compact_treesame parents mismatch");
589                 if (st->treesame[0] && revs->dense)
590                         commit->object.flags |= TREESAME;
591                 else
592                         commit->object.flags &= ~TREESAME;
593                 free(add_decoration(&revs->treesame, &commit->object, NULL));
594         }
595
596         return old_same;
597 }
598
599 static unsigned update_treesame(struct rev_info *revs, struct commit *commit)
600 {
601         if (commit->parents && commit->parents->next) {
602                 unsigned n;
603                 struct treesame_state *st;
604                 struct commit_list *p;
605                 unsigned relevant_parents;
606                 unsigned relevant_change, irrelevant_change;
607
608                 st = lookup_decoration(&revs->treesame, &commit->object);
609                 if (!st)
610                         die("update_treesame %s", oid_to_hex(&commit->object.oid));
611                 relevant_parents = 0;
612                 relevant_change = irrelevant_change = 0;
613                 for (p = commit->parents, n = 0; p; n++, p = p->next) {
614                         if (relevant_commit(p->item)) {
615                                 relevant_change |= !st->treesame[n];
616                                 relevant_parents++;
617                         } else
618                                 irrelevant_change |= !st->treesame[n];
619                 }
620                 if (relevant_parents ? relevant_change : irrelevant_change)
621                         commit->object.flags &= ~TREESAME;
622                 else
623                         commit->object.flags |= TREESAME;
624         }
625
626         return commit->object.flags & TREESAME;
627 }
628
629 static inline int limiting_can_increase_treesame(const struct rev_info *revs)
630 {
631         /*
632          * TREESAME is irrelevant unless prune && dense;
633          * if simplify_history is set, we can't have a mixture of TREESAME and
634          *    !TREESAME INTERESTING parents (and we don't have treesame[]
635          *    decoration anyway);
636          * if first_parent_only is set, then the TREESAME flag is locked
637          *    against the first parent (and again we lack treesame[] decoration).
638          */
639         return revs->prune && revs->dense &&
640                !revs->simplify_history &&
641                !revs->first_parent_only;
642 }
643
644 static void try_to_simplify_commit(struct rev_info *revs, struct commit *commit)
645 {
646         struct commit_list **pp, *parent;
647         struct treesame_state *ts = NULL;
648         int relevant_change = 0, irrelevant_change = 0;
649         int relevant_parents, nth_parent;
650
651         /*
652          * If we don't do pruning, everything is interesting
653          */
654         if (!revs->prune)
655                 return;
656
657         if (!get_commit_tree(commit))
658                 return;
659
660         if (!commit->parents) {
661                 if (rev_same_tree_as_empty(revs, commit))
662                         commit->object.flags |= TREESAME;
663                 return;
664         }
665
666         /*
667          * Normal non-merge commit? If we don't want to make the
668          * history dense, we consider it always to be a change..
669          */
670         if (!revs->dense && !commit->parents->next)
671                 return;
672
673         for (pp = &commit->parents, nth_parent = 0, relevant_parents = 0;
674              (parent = *pp) != NULL;
675              pp = &parent->next, nth_parent++) {
676                 struct commit *p = parent->item;
677                 if (relevant_commit(p))
678                         relevant_parents++;
679
680                 if (nth_parent == 1) {
681                         /*
682                          * This our second loop iteration - so we now know
683                          * we're dealing with a merge.
684                          *
685                          * Do not compare with later parents when we care only about
686                          * the first parent chain, in order to avoid derailing the
687                          * traversal to follow a side branch that brought everything
688                          * in the path we are limited to by the pathspec.
689                          */
690                         if (revs->first_parent_only)
691                                 break;
692                         /*
693                          * If this will remain a potentially-simplifiable
694                          * merge, remember per-parent treesame if needed.
695                          * Initialise the array with the comparison from our
696                          * first iteration.
697                          */
698                         if (revs->treesame.name &&
699                             !revs->simplify_history &&
700                             !(commit->object.flags & UNINTERESTING)) {
701                                 ts = initialise_treesame(revs, commit);
702                                 if (!(irrelevant_change || relevant_change))
703                                         ts->treesame[0] = 1;
704                         }
705                 }
706                 if (parse_commit(p) < 0)
707                         die("cannot simplify commit %s (because of %s)",
708                             oid_to_hex(&commit->object.oid),
709                             oid_to_hex(&p->object.oid));
710                 switch (rev_compare_tree(revs, p, commit)) {
711                 case REV_TREE_SAME:
712                         if (!revs->simplify_history || !relevant_commit(p)) {
713                                 /* Even if a merge with an uninteresting
714                                  * side branch brought the entire change
715                                  * we are interested in, we do not want
716                                  * to lose the other branches of this
717                                  * merge, so we just keep going.
718                                  */
719                                 if (ts)
720                                         ts->treesame[nth_parent] = 1;
721                                 continue;
722                         }
723                         parent->next = NULL;
724                         commit->parents = parent;
725                         commit->object.flags |= TREESAME;
726                         return;
727
728                 case REV_TREE_NEW:
729                         if (revs->remove_empty_trees &&
730                             rev_same_tree_as_empty(revs, p)) {
731                                 /* We are adding all the specified
732                                  * paths from this parent, so the
733                                  * history beyond this parent is not
734                                  * interesting.  Remove its parents
735                                  * (they are grandparents for us).
736                                  * IOW, we pretend this parent is a
737                                  * "root" commit.
738                                  */
739                                 if (parse_commit(p) < 0)
740                                         die("cannot simplify commit %s (invalid %s)",
741                                             oid_to_hex(&commit->object.oid),
742                                             oid_to_hex(&p->object.oid));
743                                 p->parents = NULL;
744                         }
745                 /* fallthrough */
746                 case REV_TREE_OLD:
747                 case REV_TREE_DIFFERENT:
748                         if (relevant_commit(p))
749                                 relevant_change = 1;
750                         else
751                                 irrelevant_change = 1;
752                         continue;
753                 }
754                 die("bad tree compare for commit %s", oid_to_hex(&commit->object.oid));
755         }
756
757         /*
758          * TREESAME is straightforward for single-parent commits. For merge
759          * commits, it is most useful to define it so that "irrelevant"
760          * parents cannot make us !TREESAME - if we have any relevant
761          * parents, then we only consider TREESAMEness with respect to them,
762          * allowing irrelevant merges from uninteresting branches to be
763          * simplified away. Only if we have only irrelevant parents do we
764          * base TREESAME on them. Note that this logic is replicated in
765          * update_treesame, which should be kept in sync.
766          */
767         if (relevant_parents ? !relevant_change : !irrelevant_change)
768                 commit->object.flags |= TREESAME;
769 }
770
771 static void commit_list_insert_by_date_cached(struct commit *p, struct commit_list **head,
772                     struct commit_list *cached_base, struct commit_list **cache)
773 {
774         struct commit_list *new_entry;
775
776         if (cached_base && p->date < cached_base->item->date)
777                 new_entry = commit_list_insert_by_date(p, &cached_base->next);
778         else
779                 new_entry = commit_list_insert_by_date(p, head);
780
781         if (cache && (!*cache || p->date < (*cache)->item->date))
782                 *cache = new_entry;
783 }
784
785 static int process_parents(struct rev_info *revs, struct commit *commit,
786                            struct commit_list **list, struct commit_list **cache_ptr)
787 {
788         struct commit_list *parent = commit->parents;
789         unsigned left_flag;
790         struct commit_list *cached_base = cache_ptr ? *cache_ptr : NULL;
791
792         if (commit->object.flags & ADDED)
793                 return 0;
794         commit->object.flags |= ADDED;
795
796         if (revs->include_check &&
797             !revs->include_check(commit, revs->include_check_data))
798                 return 0;
799
800         /*
801          * If the commit is uninteresting, don't try to
802          * prune parents - we want the maximal uninteresting
803          * set.
804          *
805          * Normally we haven't parsed the parent
806          * yet, so we won't have a parent of a parent
807          * here. However, it may turn out that we've
808          * reached this commit some other way (where it
809          * wasn't uninteresting), in which case we need
810          * to mark its parents recursively too..
811          */
812         if (commit->object.flags & UNINTERESTING) {
813                 while (parent) {
814                         struct commit *p = parent->item;
815                         parent = parent->next;
816                         if (p)
817                                 p->object.flags |= UNINTERESTING;
818                         if (parse_commit_gently(p, 1) < 0)
819                                 continue;
820                         if (p->parents)
821                                 mark_parents_uninteresting(p);
822                         if (p->object.flags & SEEN)
823                                 continue;
824                         p->object.flags |= SEEN;
825                         if (list)
826                                 commit_list_insert_by_date_cached(p, list, cached_base, cache_ptr);
827                 }
828                 return 0;
829         }
830
831         /*
832          * Ok, the commit wasn't uninteresting. Try to
833          * simplify the commit history and find the parent
834          * that has no differences in the path set if one exists.
835          */
836         try_to_simplify_commit(revs, commit);
837
838         if (revs->no_walk)
839                 return 0;
840
841         left_flag = (commit->object.flags & SYMMETRIC_LEFT);
842
843         for (parent = commit->parents; parent; parent = parent->next) {
844                 struct commit *p = parent->item;
845                 int gently = revs->ignore_missing_links ||
846                              revs->exclude_promisor_objects;
847                 if (parse_commit_gently(p, gently) < 0) {
848                         if (revs->exclude_promisor_objects &&
849                             is_promisor_object(&p->object.oid)) {
850                                 if (revs->first_parent_only)
851                                         break;
852                                 continue;
853                         }
854                         return -1;
855                 }
856                 if (revs->sources) {
857                         char **slot = revision_sources_at(revs->sources, p);
858
859                         if (!*slot)
860                                 *slot = *revision_sources_at(revs->sources, commit);
861                 }
862                 p->object.flags |= left_flag;
863                 if (!(p->object.flags & SEEN)) {
864                         p->object.flags |= SEEN;
865                         if (list)
866                                 commit_list_insert_by_date_cached(p, list, cached_base, cache_ptr);
867                 }
868                 if (revs->first_parent_only)
869                         break;
870         }
871         return 0;
872 }
873
874 static void cherry_pick_list(struct commit_list *list, struct rev_info *revs)
875 {
876         struct commit_list *p;
877         int left_count = 0, right_count = 0;
878         int left_first;
879         struct patch_ids ids;
880         unsigned cherry_flag;
881
882         /* First count the commits on the left and on the right */
883         for (p = list; p; p = p->next) {
884                 struct commit *commit = p->item;
885                 unsigned flags = commit->object.flags;
886                 if (flags & BOUNDARY)
887                         ;
888                 else if (flags & SYMMETRIC_LEFT)
889                         left_count++;
890                 else
891                         right_count++;
892         }
893
894         if (!left_count || !right_count)
895                 return;
896
897         left_first = left_count < right_count;
898         init_patch_ids(revs->repo, &ids);
899         ids.diffopts.pathspec = revs->diffopt.pathspec;
900
901         /* Compute patch-ids for one side */
902         for (p = list; p; p = p->next) {
903                 struct commit *commit = p->item;
904                 unsigned flags = commit->object.flags;
905
906                 if (flags & BOUNDARY)
907                         continue;
908                 /*
909                  * If we have fewer left, left_first is set and we omit
910                  * commits on the right branch in this loop.  If we have
911                  * fewer right, we skip the left ones.
912                  */
913                 if (left_first != !!(flags & SYMMETRIC_LEFT))
914                         continue;
915                 add_commit_patch_id(commit, &ids);
916         }
917
918         /* either cherry_mark or cherry_pick are true */
919         cherry_flag = revs->cherry_mark ? PATCHSAME : SHOWN;
920
921         /* Check the other side */
922         for (p = list; p; p = p->next) {
923                 struct commit *commit = p->item;
924                 struct patch_id *id;
925                 unsigned flags = commit->object.flags;
926
927                 if (flags & BOUNDARY)
928                         continue;
929                 /*
930                  * If we have fewer left, left_first is set and we omit
931                  * commits on the left branch in this loop.
932                  */
933                 if (left_first == !!(flags & SYMMETRIC_LEFT))
934                         continue;
935
936                 /*
937                  * Have we seen the same patch id?
938                  */
939                 id = has_commit_patch_id(commit, &ids);
940                 if (!id)
941                         continue;
942
943                 commit->object.flags |= cherry_flag;
944                 id->commit->object.flags |= cherry_flag;
945         }
946
947         free_patch_ids(&ids);
948 }
949
950 /* How many extra uninteresting commits we want to see.. */
951 #define SLOP 5
952
953 static int still_interesting(struct commit_list *src, timestamp_t date, int slop,
954                              struct commit **interesting_cache)
955 {
956         /*
957          * No source list at all? We're definitely done..
958          */
959         if (!src)
960                 return 0;
961
962         /*
963          * Does the destination list contain entries with a date
964          * before the source list? Definitely _not_ done.
965          */
966         if (date <= src->item->date)
967                 return SLOP;
968
969         /*
970          * Does the source list still have interesting commits in
971          * it? Definitely not done..
972          */
973         if (!everybody_uninteresting(src, interesting_cache))
974                 return SLOP;
975
976         /* Ok, we're closing in.. */
977         return slop-1;
978 }
979
980 /*
981  * "rev-list --ancestry-path A..B" computes commits that are ancestors
982  * of B but not ancestors of A but further limits the result to those
983  * that are descendants of A.  This takes the list of bottom commits and
984  * the result of "A..B" without --ancestry-path, and limits the latter
985  * further to the ones that can reach one of the commits in "bottom".
986  */
987 static void limit_to_ancestry(struct commit_list *bottom, struct commit_list *list)
988 {
989         struct commit_list *p;
990         struct commit_list *rlist = NULL;
991         int made_progress;
992
993         /*
994          * Reverse the list so that it will be likely that we would
995          * process parents before children.
996          */
997         for (p = list; p; p = p->next)
998                 commit_list_insert(p->item, &rlist);
999
1000         for (p = bottom; p; p = p->next)
1001                 p->item->object.flags |= TMP_MARK;
1002
1003         /*
1004          * Mark the ones that can reach bottom commits in "list",
1005          * in a bottom-up fashion.
1006          */
1007         do {
1008                 made_progress = 0;
1009                 for (p = rlist; p; p = p->next) {
1010                         struct commit *c = p->item;
1011                         struct commit_list *parents;
1012                         if (c->object.flags & (TMP_MARK | UNINTERESTING))
1013                                 continue;
1014                         for (parents = c->parents;
1015                              parents;
1016                              parents = parents->next) {
1017                                 if (!(parents->item->object.flags & TMP_MARK))
1018                                         continue;
1019                                 c->object.flags |= TMP_MARK;
1020                                 made_progress = 1;
1021                                 break;
1022                         }
1023                 }
1024         } while (made_progress);
1025
1026         /*
1027          * NEEDSWORK: decide if we want to remove parents that are
1028          * not marked with TMP_MARK from commit->parents for commits
1029          * in the resulting list.  We may not want to do that, though.
1030          */
1031
1032         /*
1033          * The ones that are not marked with TMP_MARK are uninteresting
1034          */
1035         for (p = list; p; p = p->next) {
1036                 struct commit *c = p->item;
1037                 if (c->object.flags & TMP_MARK)
1038                         continue;
1039                 c->object.flags |= UNINTERESTING;
1040         }
1041
1042         /* We are done with the TMP_MARK */
1043         for (p = list; p; p = p->next)
1044                 p->item->object.flags &= ~TMP_MARK;
1045         for (p = bottom; p; p = p->next)
1046                 p->item->object.flags &= ~TMP_MARK;
1047         free_commit_list(rlist);
1048 }
1049
1050 /*
1051  * Before walking the history, keep the set of "negative" refs the
1052  * caller has asked to exclude.
1053  *
1054  * This is used to compute "rev-list --ancestry-path A..B", as we need
1055  * to filter the result of "A..B" further to the ones that can actually
1056  * reach A.
1057  */
1058 static struct commit_list *collect_bottom_commits(struct commit_list *list)
1059 {
1060         struct commit_list *elem, *bottom = NULL;
1061         for (elem = list; elem; elem = elem->next)
1062                 if (elem->item->object.flags & BOTTOM)
1063                         commit_list_insert(elem->item, &bottom);
1064         return bottom;
1065 }
1066
1067 /* Assumes either left_only or right_only is set */
1068 static void limit_left_right(struct commit_list *list, struct rev_info *revs)
1069 {
1070         struct commit_list *p;
1071
1072         for (p = list; p; p = p->next) {
1073                 struct commit *commit = p->item;
1074
1075                 if (revs->right_only) {
1076                         if (commit->object.flags & SYMMETRIC_LEFT)
1077                                 commit->object.flags |= SHOWN;
1078                 } else  /* revs->left_only is set */
1079                         if (!(commit->object.flags & SYMMETRIC_LEFT))
1080                                 commit->object.flags |= SHOWN;
1081         }
1082 }
1083
1084 static int limit_list(struct rev_info *revs)
1085 {
1086         int slop = SLOP;
1087         timestamp_t date = TIME_MAX;
1088         struct commit_list *list = revs->commits;
1089         struct commit_list *newlist = NULL;
1090         struct commit_list **p = &newlist;
1091         struct commit_list *bottom = NULL;
1092         struct commit *interesting_cache = NULL;
1093
1094         if (revs->ancestry_path) {
1095                 bottom = collect_bottom_commits(list);
1096                 if (!bottom)
1097                         die("--ancestry-path given but there are no bottom commits");
1098         }
1099
1100         while (list) {
1101                 struct commit *commit = pop_commit(&list);
1102                 struct object *obj = &commit->object;
1103                 show_early_output_fn_t show;
1104
1105                 if (commit == interesting_cache)
1106                         interesting_cache = NULL;
1107
1108                 if (revs->max_age != -1 && (commit->date < revs->max_age))
1109                         obj->flags |= UNINTERESTING;
1110                 if (process_parents(revs, commit, &list, NULL) < 0)
1111                         return -1;
1112                 if (obj->flags & UNINTERESTING) {
1113                         mark_parents_uninteresting(commit);
1114                         slop = still_interesting(list, date, slop, &interesting_cache);
1115                         if (slop)
1116                                 continue;
1117                         break;
1118                 }
1119                 if (revs->min_age != -1 && (commit->date > revs->min_age))
1120                         continue;
1121                 date = commit->date;
1122                 p = &commit_list_insert(commit, p)->next;
1123
1124                 show = show_early_output;
1125                 if (!show)
1126                         continue;
1127
1128                 show(revs, newlist);
1129                 show_early_output = NULL;
1130         }
1131         if (revs->cherry_pick || revs->cherry_mark)
1132                 cherry_pick_list(newlist, revs);
1133
1134         if (revs->left_only || revs->right_only)
1135                 limit_left_right(newlist, revs);
1136
1137         if (bottom) {
1138                 limit_to_ancestry(bottom, newlist);
1139                 free_commit_list(bottom);
1140         }
1141
1142         /*
1143          * Check if any commits have become TREESAME by some of their parents
1144          * becoming UNINTERESTING.
1145          */
1146         if (limiting_can_increase_treesame(revs))
1147                 for (list = newlist; list; list = list->next) {
1148                         struct commit *c = list->item;
1149                         if (c->object.flags & (UNINTERESTING | TREESAME))
1150                                 continue;
1151                         update_treesame(revs, c);
1152                 }
1153
1154         revs->commits = newlist;
1155         return 0;
1156 }
1157
1158 /*
1159  * Add an entry to refs->cmdline with the specified information.
1160  * *name is copied.
1161  */
1162 static void add_rev_cmdline(struct rev_info *revs,
1163                             struct object *item,
1164                             const char *name,
1165                             int whence,
1166                             unsigned flags)
1167 {
1168         struct rev_cmdline_info *info = &revs->cmdline;
1169         unsigned int nr = info->nr;
1170
1171         ALLOC_GROW(info->rev, nr + 1, info->alloc);
1172         info->rev[nr].item = item;
1173         info->rev[nr].name = xstrdup(name);
1174         info->rev[nr].whence = whence;
1175         info->rev[nr].flags = flags;
1176         info->nr++;
1177 }
1178
1179 static void add_rev_cmdline_list(struct rev_info *revs,
1180                                  struct commit_list *commit_list,
1181                                  int whence,
1182                                  unsigned flags)
1183 {
1184         while (commit_list) {
1185                 struct object *object = &commit_list->item->object;
1186                 add_rev_cmdline(revs, object, oid_to_hex(&object->oid),
1187                                 whence, flags);
1188                 commit_list = commit_list->next;
1189         }
1190 }
1191
1192 struct all_refs_cb {
1193         int all_flags;
1194         int warned_bad_reflog;
1195         struct rev_info *all_revs;
1196         const char *name_for_errormsg;
1197         struct worktree *wt;
1198 };
1199
1200 int ref_excluded(struct string_list *ref_excludes, const char *path)
1201 {
1202         struct string_list_item *item;
1203
1204         if (!ref_excludes)
1205                 return 0;
1206         for_each_string_list_item(item, ref_excludes) {
1207                 if (!wildmatch(item->string, path, 0))
1208                         return 1;
1209         }
1210         return 0;
1211 }
1212
1213 static int handle_one_ref(const char *path, const struct object_id *oid,
1214                           int flag, void *cb_data)
1215 {
1216         struct all_refs_cb *cb = cb_data;
1217         struct object *object;
1218
1219         if (ref_excluded(cb->all_revs->ref_excludes, path))
1220             return 0;
1221
1222         object = get_reference(cb->all_revs, path, oid, cb->all_flags);
1223         add_rev_cmdline(cb->all_revs, object, path, REV_CMD_REF, cb->all_flags);
1224         add_pending_oid(cb->all_revs, path, oid, cb->all_flags);
1225         return 0;
1226 }
1227
1228 static void init_all_refs_cb(struct all_refs_cb *cb, struct rev_info *revs,
1229         unsigned flags)
1230 {
1231         cb->all_revs = revs;
1232         cb->all_flags = flags;
1233         revs->rev_input_given = 1;
1234         cb->wt = NULL;
1235 }
1236
1237 void clear_ref_exclusion(struct string_list **ref_excludes_p)
1238 {
1239         if (*ref_excludes_p) {
1240                 string_list_clear(*ref_excludes_p, 0);
1241                 free(*ref_excludes_p);
1242         }
1243         *ref_excludes_p = NULL;
1244 }
1245
1246 void add_ref_exclusion(struct string_list **ref_excludes_p, const char *exclude)
1247 {
1248         if (!*ref_excludes_p) {
1249                 *ref_excludes_p = xcalloc(1, sizeof(**ref_excludes_p));
1250                 (*ref_excludes_p)->strdup_strings = 1;
1251         }
1252         string_list_append(*ref_excludes_p, exclude);
1253 }
1254
1255 static void handle_refs(struct ref_store *refs,
1256                         struct rev_info *revs, unsigned flags,
1257                         int (*for_each)(struct ref_store *, each_ref_fn, void *))
1258 {
1259         struct all_refs_cb cb;
1260
1261         if (!refs) {
1262                 /* this could happen with uninitialized submodules */
1263                 return;
1264         }
1265
1266         init_all_refs_cb(&cb, revs, flags);
1267         for_each(refs, handle_one_ref, &cb);
1268 }
1269
1270 static void handle_one_reflog_commit(struct object_id *oid, void *cb_data)
1271 {
1272         struct all_refs_cb *cb = cb_data;
1273         if (!is_null_oid(oid)) {
1274                 struct object *o = parse_object(cb->all_revs->repo, oid);
1275                 if (o) {
1276                         o->flags |= cb->all_flags;
1277                         /* ??? CMDLINEFLAGS ??? */
1278                         add_pending_object(cb->all_revs, o, "");
1279                 }
1280                 else if (!cb->warned_bad_reflog) {
1281                         warning("reflog of '%s' references pruned commits",
1282                                 cb->name_for_errormsg);
1283                         cb->warned_bad_reflog = 1;
1284                 }
1285         }
1286 }
1287
1288 static int handle_one_reflog_ent(struct object_id *ooid, struct object_id *noid,
1289                 const char *email, timestamp_t timestamp, int tz,
1290                 const char *message, void *cb_data)
1291 {
1292         handle_one_reflog_commit(ooid, cb_data);
1293         handle_one_reflog_commit(noid, cb_data);
1294         return 0;
1295 }
1296
1297 static int handle_one_reflog(const char *refname_in_wt,
1298                              const struct object_id *oid,
1299                              int flag, void *cb_data)
1300 {
1301         struct all_refs_cb *cb = cb_data;
1302         struct strbuf refname = STRBUF_INIT;
1303
1304         cb->warned_bad_reflog = 0;
1305         strbuf_worktree_ref(cb->wt, &refname, refname_in_wt);
1306         cb->name_for_errormsg = refname.buf;
1307         refs_for_each_reflog_ent(get_main_ref_store(the_repository),
1308                                  refname.buf,
1309                                  handle_one_reflog_ent, cb_data);
1310         strbuf_release(&refname);
1311         return 0;
1312 }
1313
1314 static void add_other_reflogs_to_pending(struct all_refs_cb *cb)
1315 {
1316         struct worktree **worktrees, **p;
1317
1318         worktrees = get_worktrees(0);
1319         for (p = worktrees; *p; p++) {
1320                 struct worktree *wt = *p;
1321
1322                 if (wt->is_current)
1323                         continue;
1324
1325                 cb->wt = wt;
1326                 refs_for_each_reflog(get_worktree_ref_store(wt),
1327                                      handle_one_reflog,
1328                                      cb);
1329         }
1330         free_worktrees(worktrees);
1331 }
1332
1333 void add_reflogs_to_pending(struct rev_info *revs, unsigned flags)
1334 {
1335         struct all_refs_cb cb;
1336
1337         cb.all_revs = revs;
1338         cb.all_flags = flags;
1339         cb.wt = NULL;
1340         for_each_reflog(handle_one_reflog, &cb);
1341
1342         if (!revs->single_worktree)
1343                 add_other_reflogs_to_pending(&cb);
1344 }
1345
1346 static void add_cache_tree(struct cache_tree *it, struct rev_info *revs,
1347                            struct strbuf *path, unsigned int flags)
1348 {
1349         size_t baselen = path->len;
1350         int i;
1351
1352         if (it->entry_count >= 0) {
1353                 struct tree *tree = lookup_tree(revs->repo, &it->oid);
1354                 tree->object.flags |= flags;
1355                 add_pending_object_with_path(revs, &tree->object, "",
1356                                              040000, path->buf);
1357         }
1358
1359         for (i = 0; i < it->subtree_nr; i++) {
1360                 struct cache_tree_sub *sub = it->down[i];
1361                 strbuf_addf(path, "%s%s", baselen ? "/" : "", sub->name);
1362                 add_cache_tree(sub->cache_tree, revs, path, flags);
1363                 strbuf_setlen(path, baselen);
1364         }
1365
1366 }
1367
1368 static void do_add_index_objects_to_pending(struct rev_info *revs,
1369                                             struct index_state *istate,
1370                                             unsigned int flags)
1371 {
1372         int i;
1373
1374         for (i = 0; i < istate->cache_nr; i++) {
1375                 struct cache_entry *ce = istate->cache[i];
1376                 struct blob *blob;
1377
1378                 if (S_ISGITLINK(ce->ce_mode))
1379                         continue;
1380
1381                 blob = lookup_blob(revs->repo, &ce->oid);
1382                 if (!blob)
1383                         die("unable to add index blob to traversal");
1384                 blob->object.flags |= flags;
1385                 add_pending_object_with_path(revs, &blob->object, "",
1386                                              ce->ce_mode, ce->name);
1387         }
1388
1389         if (istate->cache_tree) {
1390                 struct strbuf path = STRBUF_INIT;
1391                 add_cache_tree(istate->cache_tree, revs, &path, flags);
1392                 strbuf_release(&path);
1393         }
1394 }
1395
1396 void add_index_objects_to_pending(struct rev_info *revs, unsigned int flags)
1397 {
1398         struct worktree **worktrees, **p;
1399
1400         read_index(revs->repo->index);
1401         do_add_index_objects_to_pending(revs, revs->repo->index, flags);
1402
1403         if (revs->single_worktree)
1404                 return;
1405
1406         worktrees = get_worktrees(0);
1407         for (p = worktrees; *p; p++) {
1408                 struct worktree *wt = *p;
1409                 struct index_state istate = { NULL };
1410
1411                 if (wt->is_current)
1412                         continue; /* current index already taken care of */
1413
1414                 if (read_index_from(&istate,
1415                                     worktree_git_path(wt, "index"),
1416                                     get_worktree_git_dir(wt)) > 0)
1417                         do_add_index_objects_to_pending(revs, &istate, flags);
1418                 discard_index(&istate);
1419         }
1420         free_worktrees(worktrees);
1421 }
1422
1423 static int add_parents_only(struct rev_info *revs, const char *arg_, int flags,
1424                             int exclude_parent)
1425 {
1426         struct object_id oid;
1427         struct object *it;
1428         struct commit *commit;
1429         struct commit_list *parents;
1430         int parent_number;
1431         const char *arg = arg_;
1432
1433         if (*arg == '^') {
1434                 flags ^= UNINTERESTING | BOTTOM;
1435                 arg++;
1436         }
1437         if (get_oid_committish(arg, &oid))
1438                 return 0;
1439         while (1) {
1440                 it = get_reference(revs, arg, &oid, 0);
1441                 if (!it && revs->ignore_missing)
1442                         return 0;
1443                 if (it->type != OBJ_TAG)
1444                         break;
1445                 if (!((struct tag*)it)->tagged)
1446                         return 0;
1447                 oidcpy(&oid, &((struct tag*)it)->tagged->oid);
1448         }
1449         if (it->type != OBJ_COMMIT)
1450                 return 0;
1451         commit = (struct commit *)it;
1452         if (exclude_parent &&
1453             exclude_parent > commit_list_count(commit->parents))
1454                 return 0;
1455         for (parents = commit->parents, parent_number = 1;
1456              parents;
1457              parents = parents->next, parent_number++) {
1458                 if (exclude_parent && parent_number != exclude_parent)
1459                         continue;
1460
1461                 it = &parents->item->object;
1462                 it->flags |= flags;
1463                 add_rev_cmdline(revs, it, arg_, REV_CMD_PARENTS_ONLY, flags);
1464                 add_pending_object(revs, it, arg);
1465         }
1466         return 1;
1467 }
1468
1469 void repo_init_revisions(struct repository *r,
1470                          struct rev_info *revs,
1471                          const char *prefix)
1472 {
1473         memset(revs, 0, sizeof(*revs));
1474
1475         revs->repo = r;
1476         revs->abbrev = DEFAULT_ABBREV;
1477         revs->ignore_merges = 1;
1478         revs->simplify_history = 1;
1479         revs->pruning.repo = r;
1480         revs->pruning.flags.recursive = 1;
1481         revs->pruning.flags.quick = 1;
1482         revs->pruning.add_remove = file_add_remove;
1483         revs->pruning.change = file_change;
1484         revs->pruning.change_fn_data = revs;
1485         revs->sort_order = REV_SORT_IN_GRAPH_ORDER;
1486         revs->dense = 1;
1487         revs->prefix = prefix;
1488         revs->max_age = -1;
1489         revs->min_age = -1;
1490         revs->skip_count = -1;
1491         revs->max_count = -1;
1492         revs->max_parents = -1;
1493         revs->expand_tabs_in_log = -1;
1494
1495         revs->commit_format = CMIT_FMT_DEFAULT;
1496         revs->expand_tabs_in_log_default = 8;
1497
1498         init_grep_defaults(revs->repo);
1499         grep_init(&revs->grep_filter, revs->repo, prefix);
1500         revs->grep_filter.status_only = 1;
1501
1502         repo_diff_setup(revs->repo, &revs->diffopt);
1503         if (prefix && !revs->diffopt.prefix) {
1504                 revs->diffopt.prefix = prefix;
1505                 revs->diffopt.prefix_length = strlen(prefix);
1506         }
1507
1508         revs->notes_opt.use_default_notes = -1;
1509 }
1510
1511 static void add_pending_commit_list(struct rev_info *revs,
1512                                     struct commit_list *commit_list,
1513                                     unsigned int flags)
1514 {
1515         while (commit_list) {
1516                 struct object *object = &commit_list->item->object;
1517                 object->flags |= flags;
1518                 add_pending_object(revs, object, oid_to_hex(&object->oid));
1519                 commit_list = commit_list->next;
1520         }
1521 }
1522
1523 static void prepare_show_merge(struct rev_info *revs)
1524 {
1525         struct commit_list *bases;
1526         struct commit *head, *other;
1527         struct object_id oid;
1528         const char **prune = NULL;
1529         int i, prune_num = 1; /* counting terminating NULL */
1530         struct index_state *istate = revs->repo->index;
1531
1532         if (get_oid("HEAD", &oid))
1533                 die("--merge without HEAD?");
1534         head = lookup_commit_or_die(&oid, "HEAD");
1535         if (get_oid("MERGE_HEAD", &oid))
1536                 die("--merge without MERGE_HEAD?");
1537         other = lookup_commit_or_die(&oid, "MERGE_HEAD");
1538         add_pending_object(revs, &head->object, "HEAD");
1539         add_pending_object(revs, &other->object, "MERGE_HEAD");
1540         bases = get_merge_bases(head, other);
1541         add_rev_cmdline_list(revs, bases, REV_CMD_MERGE_BASE, UNINTERESTING | BOTTOM);
1542         add_pending_commit_list(revs, bases, UNINTERESTING | BOTTOM);
1543         free_commit_list(bases);
1544         head->object.flags |= SYMMETRIC_LEFT;
1545
1546         if (!istate->cache_nr)
1547                 read_index(istate);
1548         for (i = 0; i < istate->cache_nr; i++) {
1549                 const struct cache_entry *ce = istate->cache[i];
1550                 if (!ce_stage(ce))
1551                         continue;
1552                 if (ce_path_match(istate, ce, &revs->prune_data, NULL)) {
1553                         prune_num++;
1554                         REALLOC_ARRAY(prune, prune_num);
1555                         prune[prune_num-2] = ce->name;
1556                         prune[prune_num-1] = NULL;
1557                 }
1558                 while ((i+1 < istate->cache_nr) &&
1559                        ce_same_name(ce, istate->cache[i+1]))
1560                         i++;
1561         }
1562         clear_pathspec(&revs->prune_data);
1563         parse_pathspec(&revs->prune_data, PATHSPEC_ALL_MAGIC & ~PATHSPEC_LITERAL,
1564                        PATHSPEC_PREFER_FULL | PATHSPEC_LITERAL_PATH, "", prune);
1565         revs->limited = 1;
1566 }
1567
1568 static int dotdot_missing(const char *arg, char *dotdot,
1569                           struct rev_info *revs, int symmetric)
1570 {
1571         if (revs->ignore_missing)
1572                 return 0;
1573         /* de-munge so we report the full argument */
1574         *dotdot = '.';
1575         die(symmetric
1576             ? "Invalid symmetric difference expression %s"
1577             : "Invalid revision range %s", arg);
1578 }
1579
1580 static int handle_dotdot_1(const char *arg, char *dotdot,
1581                            struct rev_info *revs, int flags,
1582                            int cant_be_filename,
1583                            struct object_context *a_oc,
1584                            struct object_context *b_oc)
1585 {
1586         const char *a_name, *b_name;
1587         struct object_id a_oid, b_oid;
1588         struct object *a_obj, *b_obj;
1589         unsigned int a_flags, b_flags;
1590         int symmetric = 0;
1591         unsigned int flags_exclude = flags ^ (UNINTERESTING | BOTTOM);
1592         unsigned int oc_flags = GET_OID_COMMITTISH | GET_OID_RECORD_PATH;
1593
1594         a_name = arg;
1595         if (!*a_name)
1596                 a_name = "HEAD";
1597
1598         b_name = dotdot + 2;
1599         if (*b_name == '.') {
1600                 symmetric = 1;
1601                 b_name++;
1602         }
1603         if (!*b_name)
1604                 b_name = "HEAD";
1605
1606         if (get_oid_with_context(a_name, oc_flags, &a_oid, a_oc) ||
1607             get_oid_with_context(b_name, oc_flags, &b_oid, b_oc))
1608                 return -1;
1609
1610         if (!cant_be_filename) {
1611                 *dotdot = '.';
1612                 verify_non_filename(revs->prefix, arg);
1613                 *dotdot = '\0';
1614         }
1615
1616         a_obj = parse_object(revs->repo, &a_oid);
1617         b_obj = parse_object(revs->repo, &b_oid);
1618         if (!a_obj || !b_obj)
1619                 return dotdot_missing(arg, dotdot, revs, symmetric);
1620
1621         if (!symmetric) {
1622                 /* just A..B */
1623                 b_flags = flags;
1624                 a_flags = flags_exclude;
1625         } else {
1626                 /* A...B -- find merge bases between the two */
1627                 struct commit *a, *b;
1628                 struct commit_list *exclude;
1629
1630                 a = lookup_commit_reference(revs->repo, &a_obj->oid);
1631                 b = lookup_commit_reference(revs->repo, &b_obj->oid);
1632                 if (!a || !b)
1633                         return dotdot_missing(arg, dotdot, revs, symmetric);
1634
1635                 exclude = get_merge_bases(a, b);
1636                 add_rev_cmdline_list(revs, exclude, REV_CMD_MERGE_BASE,
1637                                      flags_exclude);
1638                 add_pending_commit_list(revs, exclude, flags_exclude);
1639                 free_commit_list(exclude);
1640
1641                 b_flags = flags;
1642                 a_flags = flags | SYMMETRIC_LEFT;
1643         }
1644
1645         a_obj->flags |= a_flags;
1646         b_obj->flags |= b_flags;
1647         add_rev_cmdline(revs, a_obj, a_name, REV_CMD_LEFT, a_flags);
1648         add_rev_cmdline(revs, b_obj, b_name, REV_CMD_RIGHT, b_flags);
1649         add_pending_object_with_path(revs, a_obj, a_name, a_oc->mode, a_oc->path);
1650         add_pending_object_with_path(revs, b_obj, b_name, b_oc->mode, b_oc->path);
1651         return 0;
1652 }
1653
1654 static int handle_dotdot(const char *arg,
1655                          struct rev_info *revs, int flags,
1656                          int cant_be_filename)
1657 {
1658         struct object_context a_oc, b_oc;
1659         char *dotdot = strstr(arg, "..");
1660         int ret;
1661
1662         if (!dotdot)
1663                 return -1;
1664
1665         memset(&a_oc, 0, sizeof(a_oc));
1666         memset(&b_oc, 0, sizeof(b_oc));
1667
1668         *dotdot = '\0';
1669         ret = handle_dotdot_1(arg, dotdot, revs, flags, cant_be_filename,
1670                               &a_oc, &b_oc);
1671         *dotdot = '.';
1672
1673         free(a_oc.path);
1674         free(b_oc.path);
1675
1676         return ret;
1677 }
1678
1679 int handle_revision_arg(const char *arg_, struct rev_info *revs, int flags, unsigned revarg_opt)
1680 {
1681         struct object_context oc;
1682         char *mark;
1683         struct object *object;
1684         struct object_id oid;
1685         int local_flags;
1686         const char *arg = arg_;
1687         int cant_be_filename = revarg_opt & REVARG_CANNOT_BE_FILENAME;
1688         unsigned get_sha1_flags = GET_OID_RECORD_PATH;
1689
1690         flags = flags & UNINTERESTING ? flags | BOTTOM : flags & ~BOTTOM;
1691
1692         if (!cant_be_filename && !strcmp(arg, "..")) {
1693                 /*
1694                  * Just ".."?  That is not a range but the
1695                  * pathspec for the parent directory.
1696                  */
1697                 return -1;
1698         }
1699
1700         if (!handle_dotdot(arg, revs, flags, revarg_opt))
1701                 return 0;
1702
1703         mark = strstr(arg, "^@");
1704         if (mark && !mark[2]) {
1705                 *mark = 0;
1706                 if (add_parents_only(revs, arg, flags, 0))
1707                         return 0;
1708                 *mark = '^';
1709         }
1710         mark = strstr(arg, "^!");
1711         if (mark && !mark[2]) {
1712                 *mark = 0;
1713                 if (!add_parents_only(revs, arg, flags ^ (UNINTERESTING | BOTTOM), 0))
1714                         *mark = '^';
1715         }
1716         mark = strstr(arg, "^-");
1717         if (mark) {
1718                 int exclude_parent = 1;
1719
1720                 if (mark[2]) {
1721                         char *end;
1722                         exclude_parent = strtoul(mark + 2, &end, 10);
1723                         if (*end != '\0' || !exclude_parent)
1724                                 return -1;
1725                 }
1726
1727                 *mark = 0;
1728                 if (!add_parents_only(revs, arg, flags ^ (UNINTERESTING | BOTTOM), exclude_parent))
1729                         *mark = '^';
1730         }
1731
1732         local_flags = 0;
1733         if (*arg == '^') {
1734                 local_flags = UNINTERESTING | BOTTOM;
1735                 arg++;
1736         }
1737
1738         if (revarg_opt & REVARG_COMMITTISH)
1739                 get_sha1_flags |= GET_OID_COMMITTISH;
1740
1741         if (get_oid_with_context(arg, get_sha1_flags, &oid, &oc))
1742                 return revs->ignore_missing ? 0 : -1;
1743         if (!cant_be_filename)
1744                 verify_non_filename(revs->prefix, arg);
1745         object = get_reference(revs, arg, &oid, flags ^ local_flags);
1746         if (!object)
1747                 return revs->ignore_missing ? 0 : -1;
1748         add_rev_cmdline(revs, object, arg_, REV_CMD_REV, flags ^ local_flags);
1749         add_pending_object_with_path(revs, object, arg, oc.mode, oc.path);
1750         free(oc.path);
1751         return 0;
1752 }
1753
1754 static void read_pathspec_from_stdin(struct rev_info *revs, struct strbuf *sb,
1755                                      struct argv_array *prune)
1756 {
1757         while (strbuf_getline(sb, stdin) != EOF)
1758                 argv_array_push(prune, sb->buf);
1759 }
1760
1761 static void read_revisions_from_stdin(struct rev_info *revs,
1762                                       struct argv_array *prune)
1763 {
1764         struct strbuf sb;
1765         int seen_dashdash = 0;
1766         int save_warning;
1767
1768         save_warning = warn_on_object_refname_ambiguity;
1769         warn_on_object_refname_ambiguity = 0;
1770
1771         strbuf_init(&sb, 1000);
1772         while (strbuf_getline(&sb, stdin) != EOF) {
1773                 int len = sb.len;
1774                 if (!len)
1775                         break;
1776                 if (sb.buf[0] == '-') {
1777                         if (len == 2 && sb.buf[1] == '-') {
1778                                 seen_dashdash = 1;
1779                                 break;
1780                         }
1781                         die("options not supported in --stdin mode");
1782                 }
1783                 if (handle_revision_arg(sb.buf, revs, 0,
1784                                         REVARG_CANNOT_BE_FILENAME))
1785                         die("bad revision '%s'", sb.buf);
1786         }
1787         if (seen_dashdash)
1788                 read_pathspec_from_stdin(revs, &sb, prune);
1789
1790         strbuf_release(&sb);
1791         warn_on_object_refname_ambiguity = save_warning;
1792 }
1793
1794 static void add_grep(struct rev_info *revs, const char *ptn, enum grep_pat_token what)
1795 {
1796         append_grep_pattern(&revs->grep_filter, ptn, "command line", 0, what);
1797 }
1798
1799 static void add_header_grep(struct rev_info *revs, enum grep_header_field field, const char *pattern)
1800 {
1801         append_header_grep_pattern(&revs->grep_filter, field, pattern);
1802 }
1803
1804 static void add_message_grep(struct rev_info *revs, const char *pattern)
1805 {
1806         add_grep(revs, pattern, GREP_PATTERN_BODY);
1807 }
1808
1809 static int handle_revision_opt(struct rev_info *revs, int argc, const char **argv,
1810                                int *unkc, const char **unkv,
1811                                const struct setup_revision_opt* opt)
1812 {
1813         const char *arg = argv[0];
1814         const char *optarg;
1815         int argcount;
1816         const unsigned hexsz = the_hash_algo->hexsz;
1817
1818         /* pseudo revision arguments */
1819         if (!strcmp(arg, "--all") || !strcmp(arg, "--branches") ||
1820             !strcmp(arg, "--tags") || !strcmp(arg, "--remotes") ||
1821             !strcmp(arg, "--reflog") || !strcmp(arg, "--not") ||
1822             !strcmp(arg, "--no-walk") || !strcmp(arg, "--do-walk") ||
1823             !strcmp(arg, "--bisect") || starts_with(arg, "--glob=") ||
1824             !strcmp(arg, "--indexed-objects") ||
1825             starts_with(arg, "--exclude=") ||
1826             starts_with(arg, "--branches=") || starts_with(arg, "--tags=") ||
1827             starts_with(arg, "--remotes=") || starts_with(arg, "--no-walk="))
1828         {
1829                 unkv[(*unkc)++] = arg;
1830                 return 1;
1831         }
1832
1833         if ((argcount = parse_long_opt("max-count", argv, &optarg))) {
1834                 revs->max_count = atoi(optarg);
1835                 revs->no_walk = 0;
1836                 return argcount;
1837         } else if ((argcount = parse_long_opt("skip", argv, &optarg))) {
1838                 revs->skip_count = atoi(optarg);
1839                 return argcount;
1840         } else if ((*arg == '-') && isdigit(arg[1])) {
1841                 /* accept -<digit>, like traditional "head" */
1842                 if (strtol_i(arg + 1, 10, &revs->max_count) < 0 ||
1843                     revs->max_count < 0)
1844                         die("'%s': not a non-negative integer", arg + 1);
1845                 revs->no_walk = 0;
1846         } else if (!strcmp(arg, "-n")) {
1847                 if (argc <= 1)
1848                         return error("-n requires an argument");
1849                 revs->max_count = atoi(argv[1]);
1850                 revs->no_walk = 0;
1851                 return 2;
1852         } else if (skip_prefix(arg, "-n", &optarg)) {
1853                 revs->max_count = atoi(optarg);
1854                 revs->no_walk = 0;
1855         } else if ((argcount = parse_long_opt("max-age", argv, &optarg))) {
1856                 revs->max_age = atoi(optarg);
1857                 return argcount;
1858         } else if ((argcount = parse_long_opt("since", argv, &optarg))) {
1859                 revs->max_age = approxidate(optarg);
1860                 return argcount;
1861         } else if ((argcount = parse_long_opt("after", argv, &optarg))) {
1862                 revs->max_age = approxidate(optarg);
1863                 return argcount;
1864         } else if ((argcount = parse_long_opt("min-age", argv, &optarg))) {
1865                 revs->min_age = atoi(optarg);
1866                 return argcount;
1867         } else if ((argcount = parse_long_opt("before", argv, &optarg))) {
1868                 revs->min_age = approxidate(optarg);
1869                 return argcount;
1870         } else if ((argcount = parse_long_opt("until", argv, &optarg))) {
1871                 revs->min_age = approxidate(optarg);
1872                 return argcount;
1873         } else if (!strcmp(arg, "--first-parent")) {
1874                 revs->first_parent_only = 1;
1875         } else if (!strcmp(arg, "--ancestry-path")) {
1876                 revs->ancestry_path = 1;
1877                 revs->simplify_history = 0;
1878                 revs->limited = 1;
1879         } else if (!strcmp(arg, "-g") || !strcmp(arg, "--walk-reflogs")) {
1880                 init_reflog_walk(&revs->reflog_info);
1881         } else if (!strcmp(arg, "--default")) {
1882                 if (argc <= 1)
1883                         return error("bad --default argument");
1884                 revs->def = argv[1];
1885                 return 2;
1886         } else if (!strcmp(arg, "--merge")) {
1887                 revs->show_merge = 1;
1888         } else if (!strcmp(arg, "--topo-order")) {
1889                 revs->sort_order = REV_SORT_IN_GRAPH_ORDER;
1890                 revs->topo_order = 1;
1891         } else if (!strcmp(arg, "--simplify-merges")) {
1892                 revs->simplify_merges = 1;
1893                 revs->topo_order = 1;
1894                 revs->rewrite_parents = 1;
1895                 revs->simplify_history = 0;
1896                 revs->limited = 1;
1897         } else if (!strcmp(arg, "--simplify-by-decoration")) {
1898                 revs->simplify_merges = 1;
1899                 revs->topo_order = 1;
1900                 revs->rewrite_parents = 1;
1901                 revs->simplify_history = 0;
1902                 revs->simplify_by_decoration = 1;
1903                 revs->limited = 1;
1904                 revs->prune = 1;
1905                 load_ref_decorations(NULL, DECORATE_SHORT_REFS);
1906         } else if (!strcmp(arg, "--date-order")) {
1907                 revs->sort_order = REV_SORT_BY_COMMIT_DATE;
1908                 revs->topo_order = 1;
1909         } else if (!strcmp(arg, "--author-date-order")) {
1910                 revs->sort_order = REV_SORT_BY_AUTHOR_DATE;
1911                 revs->topo_order = 1;
1912         } else if (!strcmp(arg, "--early-output")) {
1913                 revs->early_output = 100;
1914                 revs->topo_order = 1;
1915         } else if (skip_prefix(arg, "--early-output=", &optarg)) {
1916                 if (strtoul_ui(optarg, 10, &revs->early_output) < 0)
1917                         die("'%s': not a non-negative integer", optarg);
1918                 revs->topo_order = 1;
1919         } else if (!strcmp(arg, "--parents")) {
1920                 revs->rewrite_parents = 1;
1921                 revs->print_parents = 1;
1922         } else if (!strcmp(arg, "--dense")) {
1923                 revs->dense = 1;
1924         } else if (!strcmp(arg, "--sparse")) {
1925                 revs->dense = 0;
1926         } else if (!strcmp(arg, "--in-commit-order")) {
1927                 revs->tree_blobs_in_commit_order = 1;
1928         } else if (!strcmp(arg, "--remove-empty")) {
1929                 revs->remove_empty_trees = 1;
1930         } else if (!strcmp(arg, "--merges")) {
1931                 revs->min_parents = 2;
1932         } else if (!strcmp(arg, "--no-merges")) {
1933                 revs->max_parents = 1;
1934         } else if (skip_prefix(arg, "--min-parents=", &optarg)) {
1935                 revs->min_parents = atoi(optarg);
1936         } else if (!strcmp(arg, "--no-min-parents")) {
1937                 revs->min_parents = 0;
1938         } else if (skip_prefix(arg, "--max-parents=", &optarg)) {
1939                 revs->max_parents = atoi(optarg);
1940         } else if (!strcmp(arg, "--no-max-parents")) {
1941                 revs->max_parents = -1;
1942         } else if (!strcmp(arg, "--boundary")) {
1943                 revs->boundary = 1;
1944         } else if (!strcmp(arg, "--left-right")) {
1945                 revs->left_right = 1;
1946         } else if (!strcmp(arg, "--left-only")) {
1947                 if (revs->right_only)
1948                         die("--left-only is incompatible with --right-only"
1949                             " or --cherry");
1950                 revs->left_only = 1;
1951         } else if (!strcmp(arg, "--right-only")) {
1952                 if (revs->left_only)
1953                         die("--right-only is incompatible with --left-only");
1954                 revs->right_only = 1;
1955         } else if (!strcmp(arg, "--cherry")) {
1956                 if (revs->left_only)
1957                         die("--cherry is incompatible with --left-only");
1958                 revs->cherry_mark = 1;
1959                 revs->right_only = 1;
1960                 revs->max_parents = 1;
1961                 revs->limited = 1;
1962         } else if (!strcmp(arg, "--count")) {
1963                 revs->count = 1;
1964         } else if (!strcmp(arg, "--cherry-mark")) {
1965                 if (revs->cherry_pick)
1966                         die("--cherry-mark is incompatible with --cherry-pick");
1967                 revs->cherry_mark = 1;
1968                 revs->limited = 1; /* needs limit_list() */
1969         } else if (!strcmp(arg, "--cherry-pick")) {
1970                 if (revs->cherry_mark)
1971                         die("--cherry-pick is incompatible with --cherry-mark");
1972                 revs->cherry_pick = 1;
1973                 revs->limited = 1;
1974         } else if (!strcmp(arg, "--objects")) {
1975                 revs->tag_objects = 1;
1976                 revs->tree_objects = 1;
1977                 revs->blob_objects = 1;
1978         } else if (!strcmp(arg, "--objects-edge")) {
1979                 revs->tag_objects = 1;
1980                 revs->tree_objects = 1;
1981                 revs->blob_objects = 1;
1982                 revs->edge_hint = 1;
1983         } else if (!strcmp(arg, "--objects-edge-aggressive")) {
1984                 revs->tag_objects = 1;
1985                 revs->tree_objects = 1;
1986                 revs->blob_objects = 1;
1987                 revs->edge_hint = 1;
1988                 revs->edge_hint_aggressive = 1;
1989         } else if (!strcmp(arg, "--verify-objects")) {
1990                 revs->tag_objects = 1;
1991                 revs->tree_objects = 1;
1992                 revs->blob_objects = 1;
1993                 revs->verify_objects = 1;
1994         } else if (!strcmp(arg, "--unpacked")) {
1995                 revs->unpacked = 1;
1996         } else if (starts_with(arg, "--unpacked=")) {
1997                 die("--unpacked=<packfile> no longer supported.");
1998         } else if (!strcmp(arg, "-r")) {
1999                 revs->diff = 1;
2000                 revs->diffopt.flags.recursive = 1;
2001         } else if (!strcmp(arg, "-t")) {
2002                 revs->diff = 1;
2003                 revs->diffopt.flags.recursive = 1;
2004                 revs->diffopt.flags.tree_in_recursive = 1;
2005         } else if (!strcmp(arg, "-m")) {
2006                 revs->ignore_merges = 0;
2007         } else if (!strcmp(arg, "-c")) {
2008                 revs->diff = 1;
2009                 revs->dense_combined_merges = 0;
2010                 revs->combine_merges = 1;
2011         } else if (!strcmp(arg, "--cc")) {
2012                 revs->diff = 1;
2013                 revs->dense_combined_merges = 1;
2014                 revs->combine_merges = 1;
2015         } else if (!strcmp(arg, "-v")) {
2016                 revs->verbose_header = 1;
2017         } else if (!strcmp(arg, "--pretty")) {
2018                 revs->verbose_header = 1;
2019                 revs->pretty_given = 1;
2020                 get_commit_format(NULL, revs);
2021         } else if (skip_prefix(arg, "--pretty=", &optarg) ||
2022                    skip_prefix(arg, "--format=", &optarg)) {
2023                 /*
2024                  * Detached form ("--pretty X" as opposed to "--pretty=X")
2025                  * not allowed, since the argument is optional.
2026                  */
2027                 revs->verbose_header = 1;
2028                 revs->pretty_given = 1;
2029                 get_commit_format(optarg, revs);
2030         } else if (!strcmp(arg, "--expand-tabs")) {
2031                 revs->expand_tabs_in_log = 8;
2032         } else if (!strcmp(arg, "--no-expand-tabs")) {
2033                 revs->expand_tabs_in_log = 0;
2034         } else if (skip_prefix(arg, "--expand-tabs=", &arg)) {
2035                 int val;
2036                 if (strtol_i(arg, 10, &val) < 0 || val < 0)
2037                         die("'%s': not a non-negative integer", arg);
2038                 revs->expand_tabs_in_log = val;
2039         } else if (!strcmp(arg, "--show-notes") || !strcmp(arg, "--notes")) {
2040                 revs->show_notes = 1;
2041                 revs->show_notes_given = 1;
2042                 revs->notes_opt.use_default_notes = 1;
2043         } else if (!strcmp(arg, "--show-signature")) {
2044                 revs->show_signature = 1;
2045         } else if (!strcmp(arg, "--no-show-signature")) {
2046                 revs->show_signature = 0;
2047         } else if (!strcmp(arg, "--show-linear-break")) {
2048                 revs->break_bar = "                    ..........";
2049                 revs->track_linear = 1;
2050                 revs->track_first_time = 1;
2051         } else if (skip_prefix(arg, "--show-linear-break=", &optarg)) {
2052                 revs->break_bar = xstrdup(optarg);
2053                 revs->track_linear = 1;
2054                 revs->track_first_time = 1;
2055         } else if (skip_prefix(arg, "--show-notes=", &optarg) ||
2056                    skip_prefix(arg, "--notes=", &optarg)) {
2057                 struct strbuf buf = STRBUF_INIT;
2058                 revs->show_notes = 1;
2059                 revs->show_notes_given = 1;
2060                 if (starts_with(arg, "--show-notes=") &&
2061                     revs->notes_opt.use_default_notes < 0)
2062                         revs->notes_opt.use_default_notes = 1;
2063                 strbuf_addstr(&buf, optarg);
2064                 expand_notes_ref(&buf);
2065                 string_list_append(&revs->notes_opt.extra_notes_refs,
2066                                    strbuf_detach(&buf, NULL));
2067         } else if (!strcmp(arg, "--no-notes")) {
2068                 revs->show_notes = 0;
2069                 revs->show_notes_given = 1;
2070                 revs->notes_opt.use_default_notes = -1;
2071                 /* we have been strdup'ing ourselves, so trick
2072                  * string_list into free()ing strings */
2073                 revs->notes_opt.extra_notes_refs.strdup_strings = 1;
2074                 string_list_clear(&revs->notes_opt.extra_notes_refs, 0);
2075                 revs->notes_opt.extra_notes_refs.strdup_strings = 0;
2076         } else if (!strcmp(arg, "--standard-notes")) {
2077                 revs->show_notes_given = 1;
2078                 revs->notes_opt.use_default_notes = 1;
2079         } else if (!strcmp(arg, "--no-standard-notes")) {
2080                 revs->notes_opt.use_default_notes = 0;
2081         } else if (!strcmp(arg, "--oneline")) {
2082                 revs->verbose_header = 1;
2083                 get_commit_format("oneline", revs);
2084                 revs->pretty_given = 1;
2085                 revs->abbrev_commit = 1;
2086         } else if (!strcmp(arg, "--graph")) {
2087                 revs->topo_order = 1;
2088                 revs->rewrite_parents = 1;
2089                 revs->graph = graph_init(revs);
2090         } else if (!strcmp(arg, "--root")) {
2091                 revs->show_root_diff = 1;
2092         } else if (!strcmp(arg, "--no-commit-id")) {
2093                 revs->no_commit_id = 1;
2094         } else if (!strcmp(arg, "--always")) {
2095                 revs->always_show_header = 1;
2096         } else if (!strcmp(arg, "--no-abbrev")) {
2097                 revs->abbrev = 0;
2098         } else if (!strcmp(arg, "--abbrev")) {
2099                 revs->abbrev = DEFAULT_ABBREV;
2100         } else if (skip_prefix(arg, "--abbrev=", &optarg)) {
2101                 revs->abbrev = strtoul(optarg, NULL, 10);
2102                 if (revs->abbrev < MINIMUM_ABBREV)
2103                         revs->abbrev = MINIMUM_ABBREV;
2104                 else if (revs->abbrev > hexsz)
2105                         revs->abbrev = hexsz;
2106         } else if (!strcmp(arg, "--abbrev-commit")) {
2107                 revs->abbrev_commit = 1;
2108                 revs->abbrev_commit_given = 1;
2109         } else if (!strcmp(arg, "--no-abbrev-commit")) {
2110                 revs->abbrev_commit = 0;
2111         } else if (!strcmp(arg, "--full-diff")) {
2112                 revs->diff = 1;
2113                 revs->full_diff = 1;
2114         } else if (!strcmp(arg, "--full-history")) {
2115                 revs->simplify_history = 0;
2116         } else if (!strcmp(arg, "--relative-date")) {
2117                 revs->date_mode.type = DATE_RELATIVE;
2118                 revs->date_mode_explicit = 1;
2119         } else if ((argcount = parse_long_opt("date", argv, &optarg))) {
2120                 parse_date_format(optarg, &revs->date_mode);
2121                 revs->date_mode_explicit = 1;
2122                 return argcount;
2123         } else if (!strcmp(arg, "--log-size")) {
2124                 revs->show_log_size = 1;
2125         }
2126         /*
2127          * Grepping the commit log
2128          */
2129         else if ((argcount = parse_long_opt("author", argv, &optarg))) {
2130                 add_header_grep(revs, GREP_HEADER_AUTHOR, optarg);
2131                 return argcount;
2132         } else if ((argcount = parse_long_opt("committer", argv, &optarg))) {
2133                 add_header_grep(revs, GREP_HEADER_COMMITTER, optarg);
2134                 return argcount;
2135         } else if ((argcount = parse_long_opt("grep-reflog", argv, &optarg))) {
2136                 add_header_grep(revs, GREP_HEADER_REFLOG, optarg);
2137                 return argcount;
2138         } else if ((argcount = parse_long_opt("grep", argv, &optarg))) {
2139                 add_message_grep(revs, optarg);
2140                 return argcount;
2141         } else if (!strcmp(arg, "--grep-debug")) {
2142                 revs->grep_filter.debug = 1;
2143         } else if (!strcmp(arg, "--basic-regexp")) {
2144                 revs->grep_filter.pattern_type_option = GREP_PATTERN_TYPE_BRE;
2145         } else if (!strcmp(arg, "--extended-regexp") || !strcmp(arg, "-E")) {
2146                 revs->grep_filter.pattern_type_option = GREP_PATTERN_TYPE_ERE;
2147         } else if (!strcmp(arg, "--regexp-ignore-case") || !strcmp(arg, "-i")) {
2148                 revs->grep_filter.ignore_case = 1;
2149                 revs->diffopt.pickaxe_opts |= DIFF_PICKAXE_IGNORE_CASE;
2150         } else if (!strcmp(arg, "--fixed-strings") || !strcmp(arg, "-F")) {
2151                 revs->grep_filter.pattern_type_option = GREP_PATTERN_TYPE_FIXED;
2152         } else if (!strcmp(arg, "--perl-regexp") || !strcmp(arg, "-P")) {
2153                 revs->grep_filter.pattern_type_option = GREP_PATTERN_TYPE_PCRE;
2154         } else if (!strcmp(arg, "--all-match")) {
2155                 revs->grep_filter.all_match = 1;
2156         } else if (!strcmp(arg, "--invert-grep")) {
2157                 revs->invert_grep = 1;
2158         } else if ((argcount = parse_long_opt("encoding", argv, &optarg))) {
2159                 if (strcmp(optarg, "none"))
2160                         git_log_output_encoding = xstrdup(optarg);
2161                 else
2162                         git_log_output_encoding = "";
2163                 return argcount;
2164         } else if (!strcmp(arg, "--reverse")) {
2165                 revs->reverse ^= 1;
2166         } else if (!strcmp(arg, "--children")) {
2167                 revs->children.name = "children";
2168                 revs->limited = 1;
2169         } else if (!strcmp(arg, "--ignore-missing")) {
2170                 revs->ignore_missing = 1;
2171         } else if (opt && opt->allow_exclude_promisor_objects &&
2172                    !strcmp(arg, "--exclude-promisor-objects")) {
2173                 if (fetch_if_missing)
2174                         BUG("exclude_promisor_objects can only be used when fetch_if_missing is 0");
2175                 revs->exclude_promisor_objects = 1;
2176         } else {
2177                 int opts = diff_opt_parse(&revs->diffopt, argv, argc, revs->prefix);
2178                 if (!opts)
2179                         unkv[(*unkc)++] = arg;
2180                 return opts;
2181         }
2182         if (revs->graph && revs->track_linear)
2183                 die("--show-linear-break and --graph are incompatible");
2184
2185         return 1;
2186 }
2187
2188 void parse_revision_opt(struct rev_info *revs, struct parse_opt_ctx_t *ctx,
2189                         const struct option *options,
2190                         const char * const usagestr[])
2191 {
2192         int n = handle_revision_opt(revs, ctx->argc, ctx->argv,
2193                                     &ctx->cpidx, ctx->out, NULL);
2194         if (n <= 0) {
2195                 error("unknown option `%s'", ctx->argv[0]);
2196                 usage_with_options(usagestr, options);
2197         }
2198         ctx->argv += n;
2199         ctx->argc -= n;
2200 }
2201
2202 static int for_each_bisect_ref(struct ref_store *refs, each_ref_fn fn,
2203                                void *cb_data, const char *term)
2204 {
2205         struct strbuf bisect_refs = STRBUF_INIT;
2206         int status;
2207         strbuf_addf(&bisect_refs, "refs/bisect/%s", term);
2208         status = refs_for_each_fullref_in(refs, bisect_refs.buf, fn, cb_data, 0);
2209         strbuf_release(&bisect_refs);
2210         return status;
2211 }
2212
2213 static int for_each_bad_bisect_ref(struct ref_store *refs, each_ref_fn fn, void *cb_data)
2214 {
2215         return for_each_bisect_ref(refs, fn, cb_data, term_bad);
2216 }
2217
2218 static int for_each_good_bisect_ref(struct ref_store *refs, each_ref_fn fn, void *cb_data)
2219 {
2220         return for_each_bisect_ref(refs, fn, cb_data, term_good);
2221 }
2222
2223 static int handle_revision_pseudo_opt(const char *submodule,
2224                                 struct rev_info *revs,
2225                                 int argc, const char **argv, int *flags)
2226 {
2227         const char *arg = argv[0];
2228         const char *optarg;
2229         struct ref_store *refs;
2230         int argcount;
2231
2232         if (submodule) {
2233                 /*
2234                  * We need some something like get_submodule_worktrees()
2235                  * before we can go through all worktrees of a submodule,
2236                  * .e.g with adding all HEADs from --all, which is not
2237                  * supported right now, so stick to single worktree.
2238                  */
2239                 if (!revs->single_worktree)
2240                         BUG("--single-worktree cannot be used together with submodule");
2241                 refs = get_submodule_ref_store(submodule);
2242         } else
2243                 refs = get_main_ref_store(revs->repo);
2244
2245         /*
2246          * NOTE!
2247          *
2248          * Commands like "git shortlog" will not accept the options below
2249          * unless parse_revision_opt queues them (as opposed to erroring
2250          * out).
2251          *
2252          * When implementing your new pseudo-option, remember to
2253          * register it in the list at the top of handle_revision_opt.
2254          */
2255         if (!strcmp(arg, "--all")) {
2256                 handle_refs(refs, revs, *flags, refs_for_each_ref);
2257                 handle_refs(refs, revs, *flags, refs_head_ref);
2258                 if (!revs->single_worktree) {
2259                         struct all_refs_cb cb;
2260
2261                         init_all_refs_cb(&cb, revs, *flags);
2262                         other_head_refs(handle_one_ref, &cb);
2263                 }
2264                 clear_ref_exclusion(&revs->ref_excludes);
2265         } else if (!strcmp(arg, "--branches")) {
2266                 handle_refs(refs, revs, *flags, refs_for_each_branch_ref);
2267                 clear_ref_exclusion(&revs->ref_excludes);
2268         } else if (!strcmp(arg, "--bisect")) {
2269                 read_bisect_terms(&term_bad, &term_good);
2270                 handle_refs(refs, revs, *flags, for_each_bad_bisect_ref);
2271                 handle_refs(refs, revs, *flags ^ (UNINTERESTING | BOTTOM),
2272                             for_each_good_bisect_ref);
2273                 revs->bisect = 1;
2274         } else if (!strcmp(arg, "--tags")) {
2275                 handle_refs(refs, revs, *flags, refs_for_each_tag_ref);
2276                 clear_ref_exclusion(&revs->ref_excludes);
2277         } else if (!strcmp(arg, "--remotes")) {
2278                 handle_refs(refs, revs, *flags, refs_for_each_remote_ref);
2279                 clear_ref_exclusion(&revs->ref_excludes);
2280         } else if ((argcount = parse_long_opt("glob", argv, &optarg))) {
2281                 struct all_refs_cb cb;
2282                 init_all_refs_cb(&cb, revs, *flags);
2283                 for_each_glob_ref(handle_one_ref, optarg, &cb);
2284                 clear_ref_exclusion(&revs->ref_excludes);
2285                 return argcount;
2286         } else if ((argcount = parse_long_opt("exclude", argv, &optarg))) {
2287                 add_ref_exclusion(&revs->ref_excludes, optarg);
2288                 return argcount;
2289         } else if (skip_prefix(arg, "--branches=", &optarg)) {
2290                 struct all_refs_cb cb;
2291                 init_all_refs_cb(&cb, revs, *flags);
2292                 for_each_glob_ref_in(handle_one_ref, optarg, "refs/heads/", &cb);
2293                 clear_ref_exclusion(&revs->ref_excludes);
2294         } else if (skip_prefix(arg, "--tags=", &optarg)) {
2295                 struct all_refs_cb cb;
2296                 init_all_refs_cb(&cb, revs, *flags);
2297                 for_each_glob_ref_in(handle_one_ref, optarg, "refs/tags/", &cb);
2298                 clear_ref_exclusion(&revs->ref_excludes);
2299         } else if (skip_prefix(arg, "--remotes=", &optarg)) {
2300                 struct all_refs_cb cb;
2301                 init_all_refs_cb(&cb, revs, *flags);
2302                 for_each_glob_ref_in(handle_one_ref, optarg, "refs/remotes/", &cb);
2303                 clear_ref_exclusion(&revs->ref_excludes);
2304         } else if (!strcmp(arg, "--reflog")) {
2305                 add_reflogs_to_pending(revs, *flags);
2306         } else if (!strcmp(arg, "--indexed-objects")) {
2307                 add_index_objects_to_pending(revs, *flags);
2308         } else if (!strcmp(arg, "--not")) {
2309                 *flags ^= UNINTERESTING | BOTTOM;
2310         } else if (!strcmp(arg, "--no-walk")) {
2311                 revs->no_walk = REVISION_WALK_NO_WALK_SORTED;
2312         } else if (skip_prefix(arg, "--no-walk=", &optarg)) {
2313                 /*
2314                  * Detached form ("--no-walk X" as opposed to "--no-walk=X")
2315                  * not allowed, since the argument is optional.
2316                  */
2317                 if (!strcmp(optarg, "sorted"))
2318                         revs->no_walk = REVISION_WALK_NO_WALK_SORTED;
2319                 else if (!strcmp(optarg, "unsorted"))
2320                         revs->no_walk = REVISION_WALK_NO_WALK_UNSORTED;
2321                 else
2322                         return error("invalid argument to --no-walk");
2323         } else if (!strcmp(arg, "--do-walk")) {
2324                 revs->no_walk = 0;
2325         } else if (!strcmp(arg, "--single-worktree")) {
2326                 revs->single_worktree = 1;
2327         } else {
2328                 return 0;
2329         }
2330
2331         return 1;
2332 }
2333
2334 static void NORETURN diagnose_missing_default(const char *def)
2335 {
2336         int flags;
2337         const char *refname;
2338
2339         refname = resolve_ref_unsafe(def, 0, NULL, &flags);
2340         if (!refname || !(flags & REF_ISSYMREF) || (flags & REF_ISBROKEN))
2341                 die(_("your current branch appears to be broken"));
2342
2343         skip_prefix(refname, "refs/heads/", &refname);
2344         die(_("your current branch '%s' does not have any commits yet"),
2345             refname);
2346 }
2347
2348 /*
2349  * Parse revision information, filling in the "rev_info" structure,
2350  * and removing the used arguments from the argument list.
2351  *
2352  * Returns the number of arguments left that weren't recognized
2353  * (which are also moved to the head of the argument list)
2354  */
2355 int setup_revisions(int argc, const char **argv, struct rev_info *revs, struct setup_revision_opt *opt)
2356 {
2357         int i, flags, left, seen_dashdash, got_rev_arg = 0, revarg_opt;
2358         struct argv_array prune_data = ARGV_ARRAY_INIT;
2359         const char *submodule = NULL;
2360
2361         if (opt)
2362                 submodule = opt->submodule;
2363
2364         /* First, search for "--" */
2365         if (opt && opt->assume_dashdash) {
2366                 seen_dashdash = 1;
2367         } else {
2368                 seen_dashdash = 0;
2369                 for (i = 1; i < argc; i++) {
2370                         const char *arg = argv[i];
2371                         if (strcmp(arg, "--"))
2372                                 continue;
2373                         argv[i] = NULL;
2374                         argc = i;
2375                         if (argv[i + 1])
2376                                 argv_array_pushv(&prune_data, argv + i + 1);
2377                         seen_dashdash = 1;
2378                         break;
2379                 }
2380         }
2381
2382         /* Second, deal with arguments and options */
2383         flags = 0;
2384         revarg_opt = opt ? opt->revarg_opt : 0;
2385         if (seen_dashdash)
2386                 revarg_opt |= REVARG_CANNOT_BE_FILENAME;
2387         for (left = i = 1; i < argc; i++) {
2388                 const char *arg = argv[i];
2389                 if (*arg == '-') {
2390                         int opts;
2391
2392                         opts = handle_revision_pseudo_opt(submodule,
2393                                                 revs, argc - i, argv + i,
2394                                                 &flags);
2395                         if (opts > 0) {
2396                                 i += opts - 1;
2397                                 continue;
2398                         }
2399
2400                         if (!strcmp(arg, "--stdin")) {
2401                                 if (revs->disable_stdin) {
2402                                         argv[left++] = arg;
2403                                         continue;
2404                                 }
2405                                 if (revs->read_from_stdin++)
2406                                         die("--stdin given twice?");
2407                                 read_revisions_from_stdin(revs, &prune_data);
2408                                 continue;
2409                         }
2410
2411                         opts = handle_revision_opt(revs, argc - i, argv + i,
2412                                                    &left, argv, opt);
2413                         if (opts > 0) {
2414                                 i += opts - 1;
2415                                 continue;
2416                         }
2417                         if (opts < 0)
2418                                 exit(128);
2419                         continue;
2420                 }
2421
2422
2423                 if (handle_revision_arg(arg, revs, flags, revarg_opt)) {
2424                         int j;
2425                         if (seen_dashdash || *arg == '^')
2426                                 die("bad revision '%s'", arg);
2427
2428                         /* If we didn't have a "--":
2429                          * (1) all filenames must exist;
2430                          * (2) all rev-args must not be interpretable
2431                          *     as a valid filename.
2432                          * but the latter we have checked in the main loop.
2433                          */
2434                         for (j = i; j < argc; j++)
2435                                 verify_filename(revs->prefix, argv[j], j == i);
2436
2437                         argv_array_pushv(&prune_data, argv + i);
2438                         break;
2439                 }
2440                 else
2441                         got_rev_arg = 1;
2442         }
2443
2444         if (prune_data.argc) {
2445                 /*
2446                  * If we need to introduce the magic "a lone ':' means no
2447                  * pathspec whatsoever", here is the place to do so.
2448                  *
2449                  * if (prune_data.nr == 1 && !strcmp(prune_data[0], ":")) {
2450                  *      prune_data.nr = 0;
2451                  *      prune_data.alloc = 0;
2452                  *      free(prune_data.path);
2453                  *      prune_data.path = NULL;
2454                  * } else {
2455                  *      terminate prune_data.alloc with NULL and
2456                  *      call init_pathspec() to set revs->prune_data here.
2457                  * }
2458                  */
2459                 parse_pathspec(&revs->prune_data, 0, 0,
2460                                revs->prefix, prune_data.argv);
2461         }
2462         argv_array_clear(&prune_data);
2463
2464         if (revs->def == NULL)
2465                 revs->def = opt ? opt->def : NULL;
2466         if (opt && opt->tweak)
2467                 opt->tweak(revs, opt);
2468         if (revs->show_merge)
2469                 prepare_show_merge(revs);
2470         if (revs->def && !revs->pending.nr && !revs->rev_input_given && !got_rev_arg) {
2471                 struct object_id oid;
2472                 struct object *object;
2473                 struct object_context oc;
2474                 if (get_oid_with_context(revs->def, 0, &oid, &oc))
2475                         diagnose_missing_default(revs->def);
2476                 object = get_reference(revs, revs->def, &oid, 0);
2477                 add_pending_object_with_mode(revs, object, revs->def, oc.mode);
2478         }
2479
2480         /* Did the user ask for any diff output? Run the diff! */
2481         if (revs->diffopt.output_format & ~DIFF_FORMAT_NO_OUTPUT)
2482                 revs->diff = 1;
2483
2484         /* Pickaxe, diff-filter and rename following need diffs */
2485         if ((revs->diffopt.pickaxe_opts & DIFF_PICKAXE_KINDS_MASK) ||
2486             revs->diffopt.filter ||
2487             revs->diffopt.flags.follow_renames)
2488                 revs->diff = 1;
2489
2490         if (revs->diffopt.objfind)
2491                 revs->simplify_history = 0;
2492
2493         if (revs->topo_order && !generation_numbers_enabled(the_repository))
2494                 revs->limited = 1;
2495
2496         if (revs->prune_data.nr) {
2497                 copy_pathspec(&revs->pruning.pathspec, &revs->prune_data);
2498                 /* Can't prune commits with rename following: the paths change.. */
2499                 if (!revs->diffopt.flags.follow_renames)
2500                         revs->prune = 1;
2501                 if (!revs->full_diff)
2502                         copy_pathspec(&revs->diffopt.pathspec,
2503                                       &revs->prune_data);
2504         }
2505         if (revs->combine_merges)
2506                 revs->ignore_merges = 0;
2507         revs->diffopt.abbrev = revs->abbrev;
2508
2509         if (revs->line_level_traverse) {
2510                 revs->limited = 1;
2511                 revs->topo_order = 1;
2512         }
2513
2514         diff_setup_done(&revs->diffopt);
2515
2516         grep_commit_pattern_type(GREP_PATTERN_TYPE_UNSPECIFIED,
2517                                  &revs->grep_filter);
2518         compile_grep_patterns(&revs->grep_filter);
2519
2520         if (revs->reverse && revs->reflog_info)
2521                 die("cannot combine --reverse with --walk-reflogs");
2522         if (revs->reflog_info && revs->limited)
2523                 die("cannot combine --walk-reflogs with history-limiting options");
2524         if (revs->rewrite_parents && revs->children.name)
2525                 die("cannot combine --parents and --children");
2526
2527         /*
2528          * Limitations on the graph functionality
2529          */
2530         if (revs->reverse && revs->graph)
2531                 die("cannot combine --reverse with --graph");
2532
2533         if (revs->reflog_info && revs->graph)
2534                 die("cannot combine --walk-reflogs with --graph");
2535         if (revs->no_walk && revs->graph)
2536                 die("cannot combine --no-walk with --graph");
2537         if (!revs->reflog_info && revs->grep_filter.use_reflog_filter)
2538                 die("cannot use --grep-reflog without --walk-reflogs");
2539
2540         if (revs->first_parent_only && revs->bisect)
2541                 die(_("--first-parent is incompatible with --bisect"));
2542
2543         if (revs->expand_tabs_in_log < 0)
2544                 revs->expand_tabs_in_log = revs->expand_tabs_in_log_default;
2545
2546         return left;
2547 }
2548
2549 static void add_child(struct rev_info *revs, struct commit *parent, struct commit *child)
2550 {
2551         struct commit_list *l = xcalloc(1, sizeof(*l));
2552
2553         l->item = child;
2554         l->next = add_decoration(&revs->children, &parent->object, l);
2555 }
2556
2557 static int remove_duplicate_parents(struct rev_info *revs, struct commit *commit)
2558 {
2559         struct treesame_state *ts = lookup_decoration(&revs->treesame, &commit->object);
2560         struct commit_list **pp, *p;
2561         int surviving_parents;
2562
2563         /* Examine existing parents while marking ones we have seen... */
2564         pp = &commit->parents;
2565         surviving_parents = 0;
2566         while ((p = *pp) != NULL) {
2567                 struct commit *parent = p->item;
2568                 if (parent->object.flags & TMP_MARK) {
2569                         *pp = p->next;
2570                         if (ts)
2571                                 compact_treesame(revs, commit, surviving_parents);
2572                         continue;
2573                 }
2574                 parent->object.flags |= TMP_MARK;
2575                 surviving_parents++;
2576                 pp = &p->next;
2577         }
2578         /* clear the temporary mark */
2579         for (p = commit->parents; p; p = p->next) {
2580                 p->item->object.flags &= ~TMP_MARK;
2581         }
2582         /* no update_treesame() - removing duplicates can't affect TREESAME */
2583         return surviving_parents;
2584 }
2585
2586 struct merge_simplify_state {
2587         struct commit *simplified;
2588 };
2589
2590 static struct merge_simplify_state *locate_simplify_state(struct rev_info *revs, struct commit *commit)
2591 {
2592         struct merge_simplify_state *st;
2593
2594         st = lookup_decoration(&revs->merge_simplification, &commit->object);
2595         if (!st) {
2596                 st = xcalloc(1, sizeof(*st));
2597                 add_decoration(&revs->merge_simplification, &commit->object, st);
2598         }
2599         return st;
2600 }
2601
2602 static int mark_redundant_parents(struct rev_info *revs, struct commit *commit)
2603 {
2604         struct commit_list *h = reduce_heads(commit->parents);
2605         int i = 0, marked = 0;
2606         struct commit_list *po, *pn;
2607
2608         /* Want these for sanity-checking only */
2609         int orig_cnt = commit_list_count(commit->parents);
2610         int cnt = commit_list_count(h);
2611
2612         /*
2613          * Not ready to remove items yet, just mark them for now, based
2614          * on the output of reduce_heads(). reduce_heads outputs the reduced
2615          * set in its original order, so this isn't too hard.
2616          */
2617         po = commit->parents;
2618         pn = h;
2619         while (po) {
2620                 if (pn && po->item == pn->item) {
2621                         pn = pn->next;
2622                         i++;
2623                 } else {
2624                         po->item->object.flags |= TMP_MARK;
2625                         marked++;
2626                 }
2627                 po=po->next;
2628         }
2629
2630         if (i != cnt || cnt+marked != orig_cnt)
2631                 die("mark_redundant_parents %d %d %d %d", orig_cnt, cnt, i, marked);
2632
2633         free_commit_list(h);
2634
2635         return marked;
2636 }
2637
2638 static int mark_treesame_root_parents(struct rev_info *revs, struct commit *commit)
2639 {
2640         struct commit_list *p;
2641         int marked = 0;
2642
2643         for (p = commit->parents; p; p = p->next) {
2644                 struct commit *parent = p->item;
2645                 if (!parent->parents && (parent->object.flags & TREESAME)) {
2646                         parent->object.flags |= TMP_MARK;
2647                         marked++;
2648                 }
2649         }
2650
2651         return marked;
2652 }
2653
2654 /*
2655  * Awkward naming - this means one parent we are TREESAME to.
2656  * cf mark_treesame_root_parents: root parents that are TREESAME (to an
2657  * empty tree). Better name suggestions?
2658  */
2659 static int leave_one_treesame_to_parent(struct rev_info *revs, struct commit *commit)
2660 {
2661         struct treesame_state *ts = lookup_decoration(&revs->treesame, &commit->object);
2662         struct commit *unmarked = NULL, *marked = NULL;
2663         struct commit_list *p;
2664         unsigned n;
2665
2666         for (p = commit->parents, n = 0; p; p = p->next, n++) {
2667                 if (ts->treesame[n]) {
2668                         if (p->item->object.flags & TMP_MARK) {
2669                                 if (!marked)
2670                                         marked = p->item;
2671                         } else {
2672                                 if (!unmarked) {
2673                                         unmarked = p->item;
2674                                         break;
2675                                 }
2676                         }
2677                 }
2678         }
2679
2680         /*
2681          * If we are TREESAME to a marked-for-deletion parent, but not to any
2682          * unmarked parents, unmark the first TREESAME parent. This is the
2683          * parent that the default simplify_history==1 scan would have followed,
2684          * and it doesn't make sense to omit that path when asking for a
2685          * simplified full history. Retaining it improves the chances of
2686          * understanding odd missed merges that took an old version of a file.
2687          *
2688          * Example:
2689          *
2690          *   I--------*X       A modified the file, but mainline merge X used
2691          *    \       /        "-s ours", so took the version from I. X is
2692          *     `-*A--'         TREESAME to I and !TREESAME to A.
2693          *
2694          * Default log from X would produce "I". Without this check,
2695          * --full-history --simplify-merges would produce "I-A-X", showing
2696          * the merge commit X and that it changed A, but not making clear that
2697          * it had just taken the I version. With this check, the topology above
2698          * is retained.
2699          *
2700          * Note that it is possible that the simplification chooses a different
2701          * TREESAME parent from the default, in which case this test doesn't
2702          * activate, and we _do_ drop the default parent. Example:
2703          *
2704          *   I------X         A modified the file, but it was reverted in B,
2705          *    \    /          meaning mainline merge X is TREESAME to both
2706          *    *A-*B           parents.
2707          *
2708          * Default log would produce "I" by following the first parent;
2709          * --full-history --simplify-merges will produce "I-A-B". But this is a
2710          * reasonable result - it presents a logical full history leading from
2711          * I to X, and X is not an important merge.
2712          */
2713         if (!unmarked && marked) {
2714                 marked->object.flags &= ~TMP_MARK;
2715                 return 1;
2716         }
2717
2718         return 0;
2719 }
2720
2721 static int remove_marked_parents(struct rev_info *revs, struct commit *commit)
2722 {
2723         struct commit_list **pp, *p;
2724         int nth_parent, removed = 0;
2725
2726         pp = &commit->parents;
2727         nth_parent = 0;
2728         while ((p = *pp) != NULL) {
2729                 struct commit *parent = p->item;
2730                 if (parent->object.flags & TMP_MARK) {
2731                         parent->object.flags &= ~TMP_MARK;
2732                         *pp = p->next;
2733                         free(p);
2734                         removed++;
2735                         compact_treesame(revs, commit, nth_parent);
2736                         continue;
2737                 }
2738                 pp = &p->next;
2739                 nth_parent++;
2740         }
2741
2742         /* Removing parents can only increase TREESAMEness */
2743         if (removed && !(commit->object.flags & TREESAME))
2744                 update_treesame(revs, commit);
2745
2746         return nth_parent;
2747 }
2748
2749 static struct commit_list **simplify_one(struct rev_info *revs, struct commit *commit, struct commit_list **tail)
2750 {
2751         struct commit_list *p;
2752         struct commit *parent;
2753         struct merge_simplify_state *st, *pst;
2754         int cnt;
2755
2756         st = locate_simplify_state(revs, commit);
2757
2758         /*
2759          * Have we handled this one?
2760          */
2761         if (st->simplified)
2762                 return tail;
2763
2764         /*
2765          * An UNINTERESTING commit simplifies to itself, so does a
2766          * root commit.  We do not rewrite parents of such commit
2767          * anyway.
2768          */
2769         if ((commit->object.flags & UNINTERESTING) || !commit->parents) {
2770                 st->simplified = commit;
2771                 return tail;
2772         }
2773
2774         /*
2775          * Do we know what commit all of our parents that matter
2776          * should be rewritten to?  Otherwise we are not ready to
2777          * rewrite this one yet.
2778          */
2779         for (cnt = 0, p = commit->parents; p; p = p->next) {
2780                 pst = locate_simplify_state(revs, p->item);
2781                 if (!pst->simplified) {
2782                         tail = &commit_list_insert(p->item, tail)->next;
2783                         cnt++;
2784                 }
2785                 if (revs->first_parent_only)
2786                         break;
2787         }
2788         if (cnt) {
2789                 tail = &commit_list_insert(commit, tail)->next;
2790                 return tail;
2791         }
2792
2793         /*
2794          * Rewrite our list of parents. Note that this cannot
2795          * affect our TREESAME flags in any way - a commit is
2796          * always TREESAME to its simplification.
2797          */
2798         for (p = commit->parents; p; p = p->next) {
2799                 pst = locate_simplify_state(revs, p->item);
2800                 p->item = pst->simplified;
2801                 if (revs->first_parent_only)
2802                         break;
2803         }
2804
2805         if (revs->first_parent_only)
2806                 cnt = 1;
2807         else
2808                 cnt = remove_duplicate_parents(revs, commit);
2809
2810         /*
2811          * It is possible that we are a merge and one side branch
2812          * does not have any commit that touches the given paths;
2813          * in such a case, the immediate parent from that branch
2814          * will be rewritten to be the merge base.
2815          *
2816          *      o----X          X: the commit we are looking at;
2817          *     /    /           o: a commit that touches the paths;
2818          * ---o----'
2819          *
2820          * Further, a merge of an independent branch that doesn't
2821          * touch the path will reduce to a treesame root parent:
2822          *
2823          *  ----o----X          X: the commit we are looking at;
2824          *          /           o: a commit that touches the paths;
2825          *         r            r: a root commit not touching the paths
2826          *
2827          * Detect and simplify both cases.
2828          */
2829         if (1 < cnt) {
2830                 int marked = mark_redundant_parents(revs, commit);
2831                 marked += mark_treesame_root_parents(revs, commit);
2832                 if (marked)
2833                         marked -= leave_one_treesame_to_parent(revs, commit);
2834                 if (marked)
2835                         cnt = remove_marked_parents(revs, commit);
2836         }
2837
2838         /*
2839          * A commit simplifies to itself if it is a root, if it is
2840          * UNINTERESTING, if it touches the given paths, or if it is a
2841          * merge and its parents don't simplify to one relevant commit
2842          * (the first two cases are already handled at the beginning of
2843          * this function).
2844          *
2845          * Otherwise, it simplifies to what its sole relevant parent
2846          * simplifies to.
2847          */
2848         if (!cnt ||
2849             (commit->object.flags & UNINTERESTING) ||
2850             !(commit->object.flags & TREESAME) ||
2851             (parent = one_relevant_parent(revs, commit->parents)) == NULL)
2852                 st->simplified = commit;
2853         else {
2854                 pst = locate_simplify_state(revs, parent);
2855                 st->simplified = pst->simplified;
2856         }
2857         return tail;
2858 }
2859
2860 static void simplify_merges(struct rev_info *revs)
2861 {
2862         struct commit_list *list, *next;
2863         struct commit_list *yet_to_do, **tail;
2864         struct commit *commit;
2865
2866         if (!revs->prune)
2867                 return;
2868
2869         /* feed the list reversed */
2870         yet_to_do = NULL;
2871         for (list = revs->commits; list; list = next) {
2872                 commit = list->item;
2873                 next = list->next;
2874                 /*
2875                  * Do not free(list) here yet; the original list
2876                  * is used later in this function.
2877                  */
2878                 commit_list_insert(commit, &yet_to_do);
2879         }
2880         while (yet_to_do) {
2881                 list = yet_to_do;
2882                 yet_to_do = NULL;
2883                 tail = &yet_to_do;
2884                 while (list) {
2885                         commit = pop_commit(&list);
2886                         tail = simplify_one(revs, commit, tail);
2887                 }
2888         }
2889
2890         /* clean up the result, removing the simplified ones */
2891         list = revs->commits;
2892         revs->commits = NULL;
2893         tail = &revs->commits;
2894         while (list) {
2895                 struct merge_simplify_state *st;
2896
2897                 commit = pop_commit(&list);
2898                 st = locate_simplify_state(revs, commit);
2899                 if (st->simplified == commit)
2900                         tail = &commit_list_insert(commit, tail)->next;
2901         }
2902 }
2903
2904 static void set_children(struct rev_info *revs)
2905 {
2906         struct commit_list *l;
2907         for (l = revs->commits; l; l = l->next) {
2908                 struct commit *commit = l->item;
2909                 struct commit_list *p;
2910
2911                 for (p = commit->parents; p; p = p->next)
2912                         add_child(revs, p->item, commit);
2913         }
2914 }
2915
2916 void reset_revision_walk(void)
2917 {
2918         clear_object_flags(SEEN | ADDED | SHOWN);
2919 }
2920
2921 static int mark_uninteresting(const struct object_id *oid,
2922                               struct packed_git *pack,
2923                               uint32_t pos,
2924                               void *cb)
2925 {
2926         struct rev_info *revs = cb;
2927         struct object *o = parse_object(revs->repo, oid);
2928         o->flags |= UNINTERESTING | SEEN;
2929         return 0;
2930 }
2931
2932 define_commit_slab(indegree_slab, int);
2933 define_commit_slab(author_date_slab, timestamp_t);
2934
2935 struct topo_walk_info {
2936         uint32_t min_generation;
2937         struct prio_queue explore_queue;
2938         struct prio_queue indegree_queue;
2939         struct prio_queue topo_queue;
2940         struct indegree_slab indegree;
2941         struct author_date_slab author_date;
2942 };
2943
2944 static inline void test_flag_and_insert(struct prio_queue *q, struct commit *c, int flag)
2945 {
2946         if (c->object.flags & flag)
2947                 return;
2948
2949         c->object.flags |= flag;
2950         prio_queue_put(q, c);
2951 }
2952
2953 static void explore_walk_step(struct rev_info *revs)
2954 {
2955         struct topo_walk_info *info = revs->topo_walk_info;
2956         struct commit_list *p;
2957         struct commit *c = prio_queue_get(&info->explore_queue);
2958
2959         if (!c)
2960                 return;
2961
2962         if (parse_commit_gently(c, 1) < 0)
2963                 return;
2964
2965         if (revs->sort_order == REV_SORT_BY_AUTHOR_DATE)
2966                 record_author_date(&info->author_date, c);
2967
2968         if (revs->max_age != -1 && (c->date < revs->max_age))
2969                 c->object.flags |= UNINTERESTING;
2970
2971         if (process_parents(revs, c, NULL, NULL) < 0)
2972                 return;
2973
2974         if (c->object.flags & UNINTERESTING)
2975                 mark_parents_uninteresting(c);
2976
2977         for (p = c->parents; p; p = p->next)
2978                 test_flag_and_insert(&info->explore_queue, p->item, TOPO_WALK_EXPLORED);
2979 }
2980
2981 static void explore_to_depth(struct rev_info *revs,
2982                              uint32_t gen_cutoff)
2983 {
2984         struct topo_walk_info *info = revs->topo_walk_info;
2985         struct commit *c;
2986         while ((c = prio_queue_peek(&info->explore_queue)) &&
2987                c->generation >= gen_cutoff)
2988                 explore_walk_step(revs);
2989 }
2990
2991 static void indegree_walk_step(struct rev_info *revs)
2992 {
2993         struct commit_list *p;
2994         struct topo_walk_info *info = revs->topo_walk_info;
2995         struct commit *c = prio_queue_get(&info->indegree_queue);
2996
2997         if (!c)
2998                 return;
2999
3000         if (parse_commit_gently(c, 1) < 0)
3001                 return;
3002
3003         explore_to_depth(revs, c->generation);
3004
3005         for (p = c->parents; p; p = p->next) {
3006                 struct commit *parent = p->item;
3007                 int *pi = indegree_slab_at(&info->indegree, parent);
3008
3009                 if (*pi)
3010                         (*pi)++;
3011                 else
3012                         *pi = 2;
3013
3014                 test_flag_and_insert(&info->indegree_queue, parent, TOPO_WALK_INDEGREE);
3015
3016                 if (revs->first_parent_only)
3017                         return;
3018         }
3019 }
3020
3021 static void compute_indegrees_to_depth(struct rev_info *revs,
3022                                        uint32_t gen_cutoff)
3023 {
3024         struct topo_walk_info *info = revs->topo_walk_info;
3025         struct commit *c;
3026         while ((c = prio_queue_peek(&info->indegree_queue)) &&
3027                c->generation >= gen_cutoff)
3028                 indegree_walk_step(revs);
3029 }
3030
3031 static void init_topo_walk(struct rev_info *revs)
3032 {
3033         struct topo_walk_info *info;
3034         struct commit_list *list;
3035         revs->topo_walk_info = xmalloc(sizeof(struct topo_walk_info));
3036         info = revs->topo_walk_info;
3037         memset(info, 0, sizeof(struct topo_walk_info));
3038
3039         init_indegree_slab(&info->indegree);
3040         memset(&info->explore_queue, 0, sizeof(info->explore_queue));
3041         memset(&info->indegree_queue, 0, sizeof(info->indegree_queue));
3042         memset(&info->topo_queue, 0, sizeof(info->topo_queue));
3043
3044         switch (revs->sort_order) {
3045         default: /* REV_SORT_IN_GRAPH_ORDER */
3046                 info->topo_queue.compare = NULL;
3047                 break;
3048         case REV_SORT_BY_COMMIT_DATE:
3049                 info->topo_queue.compare = compare_commits_by_commit_date;
3050                 break;
3051         case REV_SORT_BY_AUTHOR_DATE:
3052                 init_author_date_slab(&info->author_date);
3053                 info->topo_queue.compare = compare_commits_by_author_date;
3054                 info->topo_queue.cb_data = &info->author_date;
3055                 break;
3056         }
3057
3058         info->explore_queue.compare = compare_commits_by_gen_then_commit_date;
3059         info->indegree_queue.compare = compare_commits_by_gen_then_commit_date;
3060
3061         info->min_generation = GENERATION_NUMBER_INFINITY;
3062         for (list = revs->commits; list; list = list->next) {
3063                 struct commit *c = list->item;
3064
3065                 if (parse_commit_gently(c, 1))
3066                         continue;
3067
3068                 test_flag_and_insert(&info->explore_queue, c, TOPO_WALK_EXPLORED);
3069                 test_flag_and_insert(&info->indegree_queue, c, TOPO_WALK_INDEGREE);
3070
3071                 if (c->generation < info->min_generation)
3072                         info->min_generation = c->generation;
3073
3074                 *(indegree_slab_at(&info->indegree, c)) = 1;
3075
3076                 if (revs->sort_order == REV_SORT_BY_AUTHOR_DATE)
3077                         record_author_date(&info->author_date, c);
3078         }
3079         compute_indegrees_to_depth(revs, info->min_generation);
3080
3081         for (list = revs->commits; list; list = list->next) {
3082                 struct commit *c = list->item;
3083
3084                 if (*(indegree_slab_at(&info->indegree, c)) == 1)
3085                         prio_queue_put(&info->topo_queue, c);
3086         }
3087
3088         /*
3089          * This is unfortunate; the initial tips need to be shown
3090          * in the order given from the revision traversal machinery.
3091          */
3092         if (revs->sort_order == REV_SORT_IN_GRAPH_ORDER)
3093                 prio_queue_reverse(&info->topo_queue);
3094 }
3095
3096 static struct commit *next_topo_commit(struct rev_info *revs)
3097 {
3098         struct commit *c;
3099         struct topo_walk_info *info = revs->topo_walk_info;
3100
3101         /* pop next off of topo_queue */
3102         c = prio_queue_get(&info->topo_queue);
3103
3104         if (c)
3105                 *(indegree_slab_at(&info->indegree, c)) = 0;
3106
3107         return c;
3108 }
3109
3110 static void expand_topo_walk(struct rev_info *revs, struct commit *commit)
3111 {
3112         struct commit_list *p;
3113         struct topo_walk_info *info = revs->topo_walk_info;
3114         if (process_parents(revs, commit, NULL, NULL) < 0) {
3115                 if (!revs->ignore_missing_links)
3116                         die("Failed to traverse parents of commit %s",
3117                             oid_to_hex(&commit->object.oid));
3118         }
3119
3120         for (p = commit->parents; p; p = p->next) {
3121                 struct commit *parent = p->item;
3122                 int *pi;
3123
3124                 if (parse_commit_gently(parent, 1) < 0)
3125                         continue;
3126
3127                 if (parent->generation < info->min_generation) {
3128                         info->min_generation = parent->generation;
3129                         compute_indegrees_to_depth(revs, info->min_generation);
3130                 }
3131
3132                 pi = indegree_slab_at(&info->indegree, parent);
3133
3134                 (*pi)--;
3135                 if (*pi == 1)
3136                         prio_queue_put(&info->topo_queue, parent);
3137
3138                 if (revs->first_parent_only)
3139                         return;
3140         }
3141 }
3142
3143 int prepare_revision_walk(struct rev_info *revs)
3144 {
3145         int i;
3146         struct object_array old_pending;
3147         struct commit_list **next = &revs->commits;
3148
3149         memcpy(&old_pending, &revs->pending, sizeof(old_pending));
3150         revs->pending.nr = 0;
3151         revs->pending.alloc = 0;
3152         revs->pending.objects = NULL;
3153         for (i = 0; i < old_pending.nr; i++) {
3154                 struct object_array_entry *e = old_pending.objects + i;
3155                 struct commit *commit = handle_commit(revs, e);
3156                 if (commit) {
3157                         if (!(commit->object.flags & SEEN)) {
3158                                 commit->object.flags |= SEEN;
3159                                 next = commit_list_append(commit, next);
3160                         }
3161                 }
3162         }
3163         object_array_clear(&old_pending);
3164
3165         /* Signal whether we need per-parent treesame decoration */
3166         if (revs->simplify_merges ||
3167             (revs->limited && limiting_can_increase_treesame(revs)))
3168                 revs->treesame.name = "treesame";
3169
3170         if (revs->exclude_promisor_objects) {
3171                 for_each_packed_object(mark_uninteresting, revs,
3172                                        FOR_EACH_OBJECT_PROMISOR_ONLY);
3173         }
3174
3175         if (revs->no_walk != REVISION_WALK_NO_WALK_UNSORTED)
3176                 commit_list_sort_by_date(&revs->commits);
3177         if (revs->no_walk)
3178                 return 0;
3179         if (revs->limited) {
3180                 if (limit_list(revs) < 0)
3181                         return -1;
3182                 if (revs->topo_order)
3183                         sort_in_topological_order(&revs->commits, revs->sort_order);
3184         } else if (revs->topo_order)
3185                 init_topo_walk(revs);
3186         if (revs->line_level_traverse)
3187                 line_log_filter(revs);
3188         if (revs->simplify_merges)
3189                 simplify_merges(revs);
3190         if (revs->children.name)
3191                 set_children(revs);
3192         return 0;
3193 }
3194
3195 static enum rewrite_result rewrite_one(struct rev_info *revs, struct commit **pp)
3196 {
3197         struct commit_list *cache = NULL;
3198
3199         for (;;) {
3200                 struct commit *p = *pp;
3201                 if (!revs->limited)
3202                         if (process_parents(revs, p, &revs->commits, &cache) < 0)
3203                                 return rewrite_one_error;
3204                 if (p->object.flags & UNINTERESTING)
3205                         return rewrite_one_ok;
3206                 if (!(p->object.flags & TREESAME))
3207                         return rewrite_one_ok;
3208                 if (!p->parents)
3209                         return rewrite_one_noparents;
3210                 if ((p = one_relevant_parent(revs, p->parents)) == NULL)
3211                         return rewrite_one_ok;
3212                 *pp = p;
3213         }
3214 }
3215
3216 int rewrite_parents(struct rev_info *revs, struct commit *commit,
3217         rewrite_parent_fn_t rewrite_parent)
3218 {
3219         struct commit_list **pp = &commit->parents;
3220         while (*pp) {
3221                 struct commit_list *parent = *pp;
3222                 switch (rewrite_parent(revs, &parent->item)) {
3223                 case rewrite_one_ok:
3224                         break;
3225                 case rewrite_one_noparents:
3226                         *pp = parent->next;
3227                         continue;
3228                 case rewrite_one_error:
3229                         return -1;
3230                 }
3231                 pp = &parent->next;
3232         }
3233         remove_duplicate_parents(revs, commit);
3234         return 0;
3235 }
3236
3237 static int commit_rewrite_person(struct strbuf *buf, const char *what, struct string_list *mailmap)
3238 {
3239         char *person, *endp;
3240         size_t len, namelen, maillen;
3241         const char *name;
3242         const char *mail;
3243         struct ident_split ident;
3244
3245         person = strstr(buf->buf, what);
3246         if (!person)
3247                 return 0;
3248
3249         person += strlen(what);
3250         endp = strchr(person, '\n');
3251         if (!endp)
3252                 return 0;
3253
3254         len = endp - person;
3255
3256         if (split_ident_line(&ident, person, len))
3257                 return 0;
3258
3259         mail = ident.mail_begin;
3260         maillen = ident.mail_end - ident.mail_begin;
3261         name = ident.name_begin;
3262         namelen = ident.name_end - ident.name_begin;
3263
3264         if (map_user(mailmap, &mail, &maillen, &name, &namelen)) {
3265                 struct strbuf namemail = STRBUF_INIT;
3266
3267                 strbuf_addf(&namemail, "%.*s <%.*s>",
3268                             (int)namelen, name, (int)maillen, mail);
3269
3270                 strbuf_splice(buf, ident.name_begin - buf->buf,
3271                               ident.mail_end - ident.name_begin + 1,
3272                               namemail.buf, namemail.len);
3273
3274                 strbuf_release(&namemail);
3275
3276                 return 1;
3277         }
3278
3279         return 0;
3280 }
3281
3282 static int commit_match(struct commit *commit, struct rev_info *opt)
3283 {
3284         int retval;
3285         const char *encoding;
3286         const char *message;
3287         struct strbuf buf = STRBUF_INIT;
3288
3289         if (!opt->grep_filter.pattern_list && !opt->grep_filter.header_list)
3290                 return 1;
3291
3292         /* Prepend "fake" headers as needed */
3293         if (opt->grep_filter.use_reflog_filter) {
3294                 strbuf_addstr(&buf, "reflog ");
3295                 get_reflog_message(&buf, opt->reflog_info);
3296                 strbuf_addch(&buf, '\n');
3297         }
3298
3299         /*
3300          * We grep in the user's output encoding, under the assumption that it
3301          * is the encoding they are most likely to write their grep pattern
3302          * for. In addition, it means we will match the "notes" encoding below,
3303          * so we will not end up with a buffer that has two different encodings
3304          * in it.
3305          */
3306         encoding = get_log_output_encoding();
3307         message = logmsg_reencode(commit, NULL, encoding);
3308
3309         /* Copy the commit to temporary if we are using "fake" headers */
3310         if (buf.len)
3311                 strbuf_addstr(&buf, message);
3312
3313         if (opt->grep_filter.header_list && opt->mailmap) {
3314                 if (!buf.len)
3315                         strbuf_addstr(&buf, message);
3316
3317                 commit_rewrite_person(&buf, "\nauthor ", opt->mailmap);
3318                 commit_rewrite_person(&buf, "\ncommitter ", opt->mailmap);
3319         }
3320
3321         /* Append "fake" message parts as needed */
3322         if (opt->show_notes) {
3323                 if (!buf.len)
3324                         strbuf_addstr(&buf, message);
3325                 format_display_notes(&commit->object.oid, &buf, encoding, 1);
3326         }
3327
3328         /*
3329          * Find either in the original commit message, or in the temporary.
3330          * Note that we cast away the constness of "message" here. It is
3331          * const because it may come from the cached commit buffer. That's OK,
3332          * because we know that it is modifiable heap memory, and that while
3333          * grep_buffer may modify it for speed, it will restore any
3334          * changes before returning.
3335          */
3336         if (buf.len)
3337                 retval = grep_buffer(&opt->grep_filter, buf.buf, buf.len);
3338         else
3339                 retval = grep_buffer(&opt->grep_filter,
3340                                      (char *)message, strlen(message));
3341         strbuf_release(&buf);
3342         unuse_commit_buffer(commit, message);
3343         return opt->invert_grep ? !retval : retval;
3344 }
3345
3346 static inline int want_ancestry(const struct rev_info *revs)
3347 {
3348         return (revs->rewrite_parents || revs->children.name);
3349 }
3350
3351 /*
3352  * Return a timestamp to be used for --since/--until comparisons for this
3353  * commit, based on the revision options.
3354  */
3355 static timestamp_t comparison_date(const struct rev_info *revs,
3356                                    struct commit *commit)
3357 {
3358         return revs->reflog_info ?
3359                 get_reflog_timestamp(revs->reflog_info) :
3360                 commit->date;
3361 }
3362
3363 enum commit_action get_commit_action(struct rev_info *revs, struct commit *commit)
3364 {
3365         if (commit->object.flags & SHOWN)
3366                 return commit_ignore;
3367         if (revs->unpacked && has_object_pack(&commit->object.oid))
3368                 return commit_ignore;
3369         if (commit->object.flags & UNINTERESTING)
3370                 return commit_ignore;
3371         if (revs->min_age != -1 &&
3372             comparison_date(revs, commit) > revs->min_age)
3373                         return commit_ignore;
3374         if (revs->min_parents || (revs->max_parents >= 0)) {
3375                 int n = commit_list_count(commit->parents);
3376                 if ((n < revs->min_parents) ||
3377                     ((revs->max_parents >= 0) && (n > revs->max_parents)))
3378                         return commit_ignore;
3379         }
3380         if (!commit_match(commit, revs))
3381                 return commit_ignore;
3382         if (revs->prune && revs->dense) {
3383                 /* Commit without changes? */
3384                 if (commit->object.flags & TREESAME) {
3385                         int n;
3386                         struct commit_list *p;
3387                         /* drop merges unless we want parenthood */
3388                         if (!want_ancestry(revs))
3389                                 return commit_ignore;
3390                         /*
3391                          * If we want ancestry, then need to keep any merges
3392                          * between relevant commits to tie together topology.
3393                          * For consistency with TREESAME and simplification
3394                          * use "relevant" here rather than just INTERESTING,
3395                          * to treat bottom commit(s) as part of the topology.
3396                          */
3397                         for (n = 0, p = commit->parents; p; p = p->next)
3398                                 if (relevant_commit(p->item))
3399                                         if (++n >= 2)
3400                                                 return commit_show;
3401                         return commit_ignore;
3402                 }
3403         }
3404         return commit_show;
3405 }
3406
3407 define_commit_slab(saved_parents, struct commit_list *);
3408
3409 #define EMPTY_PARENT_LIST ((struct commit_list *)-1)
3410
3411 /*
3412  * You may only call save_parents() once per commit (this is checked
3413  * for non-root commits).
3414  */
3415 static void save_parents(struct rev_info *revs, struct commit *commit)
3416 {
3417         struct commit_list **pp;
3418
3419         if (!revs->saved_parents_slab) {
3420                 revs->saved_parents_slab = xmalloc(sizeof(struct saved_parents));
3421                 init_saved_parents(revs->saved_parents_slab);
3422         }
3423
3424         pp = saved_parents_at(revs->saved_parents_slab, commit);
3425
3426         /*
3427          * When walking with reflogs, we may visit the same commit
3428          * several times: once for each appearance in the reflog.
3429          *
3430          * In this case, save_parents() will be called multiple times.
3431          * We want to keep only the first set of parents.  We need to
3432          * store a sentinel value for an empty (i.e., NULL) parent
3433          * list to distinguish it from a not-yet-saved list, however.
3434          */
3435         if (*pp)
3436                 return;
3437         if (commit->parents)
3438                 *pp = copy_commit_list(commit->parents);
3439         else
3440                 *pp = EMPTY_PARENT_LIST;
3441 }
3442
3443 static void free_saved_parents(struct rev_info *revs)
3444 {
3445         if (revs->saved_parents_slab)
3446                 clear_saved_parents(revs->saved_parents_slab);
3447 }
3448
3449 struct commit_list *get_saved_parents(struct rev_info *revs, const struct commit *commit)
3450 {
3451         struct commit_list *parents;
3452
3453         if (!revs->saved_parents_slab)
3454                 return commit->parents;
3455
3456         parents = *saved_parents_at(revs->saved_parents_slab, commit);
3457         if (parents == EMPTY_PARENT_LIST)
3458                 return NULL;
3459         return parents;
3460 }
3461
3462 enum commit_action simplify_commit(struct rev_info *revs, struct commit *commit)
3463 {
3464         enum commit_action action = get_commit_action(revs, commit);
3465
3466         if (action == commit_show &&
3467             revs->prune && revs->dense && want_ancestry(revs)) {
3468                 /*
3469                  * --full-diff on simplified parents is no good: it
3470                  * will show spurious changes from the commits that
3471                  * were elided.  So we save the parents on the side
3472                  * when --full-diff is in effect.
3473                  */
3474                 if (revs->full_diff)
3475                         save_parents(revs, commit);
3476                 if (rewrite_parents(revs, commit, rewrite_one) < 0)
3477                         return commit_error;
3478         }
3479         return action;
3480 }
3481
3482 static void track_linear(struct rev_info *revs, struct commit *commit)
3483 {
3484         if (revs->track_first_time) {
3485                 revs->linear = 1;
3486                 revs->track_first_time = 0;
3487         } else {
3488                 struct commit_list *p;
3489                 for (p = revs->previous_parents; p; p = p->next)
3490                         if (p->item == NULL || /* first commit */
3491                             oideq(&p->item->object.oid, &commit->object.oid))
3492                                 break;
3493                 revs->linear = p != NULL;
3494         }
3495         if (revs->reverse) {
3496                 if (revs->linear)
3497                         commit->object.flags |= TRACK_LINEAR;
3498         }
3499         free_commit_list(revs->previous_parents);
3500         revs->previous_parents = copy_commit_list(commit->parents);
3501 }
3502
3503 static struct commit *get_revision_1(struct rev_info *revs)
3504 {
3505         while (1) {
3506                 struct commit *commit;
3507
3508                 if (revs->reflog_info)
3509                         commit = next_reflog_entry(revs->reflog_info);
3510                 else if (revs->topo_walk_info)
3511                         commit = next_topo_commit(revs);
3512                 else
3513                         commit = pop_commit(&revs->commits);
3514
3515                 if (!commit)
3516                         return NULL;
3517
3518                 if (revs->reflog_info)
3519                         commit->object.flags &= ~(ADDED | SEEN | SHOWN);
3520
3521                 /*
3522                  * If we haven't done the list limiting, we need to look at
3523                  * the parents here. We also need to do the date-based limiting
3524                  * that we'd otherwise have done in limit_list().
3525                  */
3526                 if (!revs->limited) {
3527                         if (revs->max_age != -1 &&
3528                             comparison_date(revs, commit) < revs->max_age)
3529                                 continue;
3530
3531                         if (revs->reflog_info)
3532                                 try_to_simplify_commit(revs, commit);
3533                         else if (revs->topo_walk_info)
3534                                 expand_topo_walk(revs, commit);
3535                         else if (process_parents(revs, commit, &revs->commits, NULL) < 0) {
3536                                 if (!revs->ignore_missing_links)
3537                                         die("Failed to traverse parents of commit %s",
3538                                                 oid_to_hex(&commit->object.oid));
3539                         }
3540                 }
3541
3542                 switch (simplify_commit(revs, commit)) {
3543                 case commit_ignore:
3544                         continue;
3545                 case commit_error:
3546                         die("Failed to simplify parents of commit %s",
3547                             oid_to_hex(&commit->object.oid));
3548                 default:
3549                         if (revs->track_linear)
3550                                 track_linear(revs, commit);
3551                         return commit;
3552                 }
3553         }
3554 }
3555
3556 /*
3557  * Return true for entries that have not yet been shown.  (This is an
3558  * object_array_each_func_t.)
3559  */
3560 static int entry_unshown(struct object_array_entry *entry, void *cb_data_unused)
3561 {
3562         return !(entry->item->flags & SHOWN);
3563 }
3564
3565 /*
3566  * If array is on the verge of a realloc, garbage-collect any entries
3567  * that have already been shown to try to free up some space.
3568  */
3569 static void gc_boundary(struct object_array *array)
3570 {
3571         if (array->nr == array->alloc)
3572                 object_array_filter(array, entry_unshown, NULL);
3573 }
3574
3575 static void create_boundary_commit_list(struct rev_info *revs)
3576 {
3577         unsigned i;
3578         struct commit *c;
3579         struct object_array *array = &revs->boundary_commits;
3580         struct object_array_entry *objects = array->objects;
3581
3582         /*
3583          * If revs->commits is non-NULL at this point, an error occurred in
3584          * get_revision_1().  Ignore the error and continue printing the
3585          * boundary commits anyway.  (This is what the code has always
3586          * done.)
3587          */
3588         if (revs->commits) {
3589                 free_commit_list(revs->commits);
3590                 revs->commits = NULL;
3591         }
3592
3593         /*
3594          * Put all of the actual boundary commits from revs->boundary_commits
3595          * into revs->commits
3596          */
3597         for (i = 0; i < array->nr; i++) {
3598                 c = (struct commit *)(objects[i].item);
3599                 if (!c)
3600                         continue;
3601                 if (!(c->object.flags & CHILD_SHOWN))
3602                         continue;
3603                 if (c->object.flags & (SHOWN | BOUNDARY))
3604                         continue;
3605                 c->object.flags |= BOUNDARY;
3606                 commit_list_insert(c, &revs->commits);
3607         }
3608
3609         /*
3610          * If revs->topo_order is set, sort the boundary commits
3611          * in topological order
3612          */
3613         sort_in_topological_order(&revs->commits, revs->sort_order);
3614 }
3615
3616 static struct commit *get_revision_internal(struct rev_info *revs)
3617 {
3618         struct commit *c = NULL;
3619         struct commit_list *l;
3620
3621         if (revs->boundary == 2) {
3622                 /*
3623                  * All of the normal commits have already been returned,
3624                  * and we are now returning boundary commits.
3625                  * create_boundary_commit_list() has populated
3626                  * revs->commits with the remaining commits to return.
3627                  */
3628                 c = pop_commit(&revs->commits);
3629                 if (c)
3630                         c->object.flags |= SHOWN;
3631                 return c;
3632         }
3633
3634         /*
3635          * If our max_count counter has reached zero, then we are done. We
3636          * don't simply return NULL because we still might need to show
3637          * boundary commits. But we want to avoid calling get_revision_1, which
3638          * might do a considerable amount of work finding the next commit only
3639          * for us to throw it away.
3640          *
3641          * If it is non-zero, then either we don't have a max_count at all
3642          * (-1), or it is still counting, in which case we decrement.
3643          */
3644         if (revs->max_count) {
3645                 c = get_revision_1(revs);
3646                 if (c) {
3647                         while (revs->skip_count > 0) {
3648                                 revs->skip_count--;
3649                                 c = get_revision_1(revs);
3650                                 if (!c)
3651                                         break;
3652                         }
3653                 }
3654
3655                 if (revs->max_count > 0)
3656                         revs->max_count--;
3657         }
3658
3659         if (c)
3660                 c->object.flags |= SHOWN;
3661
3662         if (!revs->boundary)
3663                 return c;
3664
3665         if (!c) {
3666                 /*
3667                  * get_revision_1() runs out the commits, and
3668                  * we are done computing the boundaries.
3669                  * switch to boundary commits output mode.
3670                  */
3671                 revs->boundary = 2;
3672
3673                 /*
3674                  * Update revs->commits to contain the list of
3675                  * boundary commits.
3676                  */
3677                 create_boundary_commit_list(revs);
3678
3679                 return get_revision_internal(revs);
3680         }
3681
3682         /*
3683          * boundary commits are the commits that are parents of the
3684          * ones we got from get_revision_1() but they themselves are
3685          * not returned from get_revision_1().  Before returning
3686          * 'c', we need to mark its parents that they could be boundaries.
3687          */
3688
3689         for (l = c->parents; l; l = l->next) {
3690                 struct object *p;
3691                 p = &(l->item->object);
3692                 if (p->flags & (CHILD_SHOWN | SHOWN))
3693                         continue;
3694                 p->flags |= CHILD_SHOWN;
3695                 gc_boundary(&revs->boundary_commits);
3696                 add_object_array(p, NULL, &revs->boundary_commits);
3697         }
3698
3699         return c;
3700 }
3701
3702 struct commit *get_revision(struct rev_info *revs)
3703 {
3704         struct commit *c;
3705         struct commit_list *reversed;
3706
3707         if (revs->reverse) {
3708                 reversed = NULL;
3709                 while ((c = get_revision_internal(revs)))
3710                         commit_list_insert(c, &reversed);
3711                 revs->commits = reversed;
3712                 revs->reverse = 0;
3713                 revs->reverse_output_stage = 1;
3714         }
3715
3716         if (revs->reverse_output_stage) {
3717                 c = pop_commit(&revs->commits);
3718                 if (revs->track_linear)
3719                         revs->linear = !!(c && c->object.flags & TRACK_LINEAR);
3720                 return c;
3721         }
3722
3723         c = get_revision_internal(revs);
3724         if (c && revs->graph)
3725                 graph_update(revs->graph, c);
3726         if (!c) {
3727                 free_saved_parents(revs);
3728                 if (revs->previous_parents) {
3729                         free_commit_list(revs->previous_parents);
3730                         revs->previous_parents = NULL;
3731                 }
3732         }
3733         return c;
3734 }
3735
3736 char *get_revision_mark(const struct rev_info *revs, const struct commit *commit)
3737 {
3738         if (commit->object.flags & BOUNDARY)
3739                 return "-";
3740         else if (commit->object.flags & UNINTERESTING)
3741                 return "^";
3742         else if (commit->object.flags & PATCHSAME)
3743                 return "=";
3744         else if (!revs || revs->left_right) {
3745                 if (commit->object.flags & SYMMETRIC_LEFT)
3746                         return "<";
3747                 else
3748                         return ">";
3749         } else if (revs->graph)
3750                 return "*";
3751         else if (revs->cherry_mark)
3752                 return "+";
3753         return "";
3754 }
3755
3756 void put_revision_mark(const struct rev_info *revs, const struct commit *commit)
3757 {
3758         char *mark = get_revision_mark(revs, commit);
3759         if (!strlen(mark))
3760                 return;
3761         fputs(mark, stdout);
3762         putchar(' ');
3763 }