Merge branch 'ab/pcre-jit-fixes'
[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 #include "hashmap.h"
31 #include "utf8.h"
32
33 volatile show_early_output_fn_t show_early_output;
34
35 static const char *term_bad;
36 static const char *term_good;
37
38 implement_shared_commit_slab(revision_sources, char *);
39
40 void show_object_with_name(FILE *out, struct object *obj, const char *name)
41 {
42         const char *p;
43
44         fprintf(out, "%s ", oid_to_hex(&obj->oid));
45         for (p = name; *p && *p != '\n'; p++)
46                 fputc(*p, out);
47         fputc('\n', out);
48 }
49
50 static void mark_blob_uninteresting(struct blob *blob)
51 {
52         if (!blob)
53                 return;
54         if (blob->object.flags & UNINTERESTING)
55                 return;
56         blob->object.flags |= UNINTERESTING;
57 }
58
59 static void mark_tree_contents_uninteresting(struct repository *r,
60                                              struct tree *tree)
61 {
62         struct tree_desc desc;
63         struct name_entry entry;
64
65         if (parse_tree_gently(tree, 1) < 0)
66                 return;
67
68         init_tree_desc(&desc, tree->buffer, tree->size);
69         while (tree_entry(&desc, &entry)) {
70                 switch (object_type(entry.mode)) {
71                 case OBJ_TREE:
72                         mark_tree_uninteresting(r, lookup_tree(r, &entry.oid));
73                         break;
74                 case OBJ_BLOB:
75                         mark_blob_uninteresting(lookup_blob(r, &entry.oid));
76                         break;
77                 default:
78                         /* Subproject commit - not in this repository */
79                         break;
80                 }
81         }
82
83         /*
84          * We don't care about the tree any more
85          * after it has been marked uninteresting.
86          */
87         free_tree_buffer(tree);
88 }
89
90 void mark_tree_uninteresting(struct repository *r, struct tree *tree)
91 {
92         struct object *obj;
93
94         if (!tree)
95                 return;
96
97         obj = &tree->object;
98         if (obj->flags & UNINTERESTING)
99                 return;
100         obj->flags |= UNINTERESTING;
101         mark_tree_contents_uninteresting(r, tree);
102 }
103
104 struct path_and_oids_entry {
105         struct hashmap_entry ent;
106         char *path;
107         struct oidset trees;
108 };
109
110 static int path_and_oids_cmp(const void *hashmap_cmp_fn_data,
111                              const struct path_and_oids_entry *e1,
112                              const struct path_and_oids_entry *e2,
113                              const void *keydata)
114 {
115         return strcmp(e1->path, e2->path);
116 }
117
118 static void paths_and_oids_init(struct hashmap *map)
119 {
120         hashmap_init(map, (hashmap_cmp_fn) path_and_oids_cmp, NULL, 0);
121 }
122
123 static void paths_and_oids_clear(struct hashmap *map)
124 {
125         struct hashmap_iter iter;
126         struct path_and_oids_entry *entry;
127         hashmap_iter_init(map, &iter);
128
129         while ((entry = (struct path_and_oids_entry *)hashmap_iter_next(&iter))) {
130                 oidset_clear(&entry->trees);
131                 free(entry->path);
132         }
133
134         hashmap_free(map, 1);
135 }
136
137 static void paths_and_oids_insert(struct hashmap *map,
138                                   const char *path,
139                                   const struct object_id *oid)
140 {
141         int hash = strhash(path);
142         struct path_and_oids_entry key;
143         struct path_and_oids_entry *entry;
144
145         hashmap_entry_init(&key, hash);
146
147         /* use a shallow copy for the lookup */
148         key.path = (char *)path;
149         oidset_init(&key.trees, 0);
150
151         if (!(entry = (struct path_and_oids_entry *)hashmap_get(map, &key, NULL))) {
152                 entry = xcalloc(1, sizeof(struct path_and_oids_entry));
153                 hashmap_entry_init(entry, hash);
154                 entry->path = xstrdup(key.path);
155                 oidset_init(&entry->trees, 16);
156                 hashmap_put(map, entry);
157         }
158
159         oidset_insert(&entry->trees, oid);
160 }
161
162 static void add_children_by_path(struct repository *r,
163                                  struct tree *tree,
164                                  struct hashmap *map)
165 {
166         struct tree_desc desc;
167         struct name_entry entry;
168
169         if (!tree)
170                 return;
171
172         if (parse_tree_gently(tree, 1) < 0)
173                 return;
174
175         init_tree_desc(&desc, tree->buffer, tree->size);
176         while (tree_entry(&desc, &entry)) {
177                 switch (object_type(entry.mode)) {
178                 case OBJ_TREE:
179                         paths_and_oids_insert(map, entry.path, &entry.oid);
180
181                         if (tree->object.flags & UNINTERESTING) {
182                                 struct tree *child = lookup_tree(r, &entry.oid);
183                                 if (child)
184                                         child->object.flags |= UNINTERESTING;
185                         }
186                         break;
187                 case OBJ_BLOB:
188                         if (tree->object.flags & UNINTERESTING) {
189                                 struct blob *child = lookup_blob(r, &entry.oid);
190                                 if (child)
191                                         child->object.flags |= UNINTERESTING;
192                         }
193                         break;
194                 default:
195                         /* Subproject commit - not in this repository */
196                         break;
197                 }
198         }
199
200         free_tree_buffer(tree);
201 }
202
203 void mark_trees_uninteresting_sparse(struct repository *r,
204                                      struct oidset *trees)
205 {
206         unsigned has_interesting = 0, has_uninteresting = 0;
207         struct hashmap map;
208         struct hashmap_iter map_iter;
209         struct path_and_oids_entry *entry;
210         struct object_id *oid;
211         struct oidset_iter iter;
212
213         oidset_iter_init(trees, &iter);
214         while ((!has_interesting || !has_uninteresting) &&
215                (oid = oidset_iter_next(&iter))) {
216                 struct tree *tree = lookup_tree(r, oid);
217
218                 if (!tree)
219                         continue;
220
221                 if (tree->object.flags & UNINTERESTING)
222                         has_uninteresting = 1;
223                 else
224                         has_interesting = 1;
225         }
226
227         /* Do not walk unless we have both types of trees. */
228         if (!has_uninteresting || !has_interesting)
229                 return;
230
231         paths_and_oids_init(&map);
232
233         oidset_iter_init(trees, &iter);
234         while ((oid = oidset_iter_next(&iter))) {
235                 struct tree *tree = lookup_tree(r, oid);
236                 add_children_by_path(r, tree, &map);
237         }
238
239         hashmap_iter_init(&map, &map_iter);
240         while ((entry = hashmap_iter_next(&map_iter)))
241                 mark_trees_uninteresting_sparse(r, &entry->trees);
242
243         paths_and_oids_clear(&map);
244 }
245
246 struct commit_stack {
247         struct commit **items;
248         size_t nr, alloc;
249 };
250 #define COMMIT_STACK_INIT { NULL, 0, 0 }
251
252 static void commit_stack_push(struct commit_stack *stack, struct commit *commit)
253 {
254         ALLOC_GROW(stack->items, stack->nr + 1, stack->alloc);
255         stack->items[stack->nr++] = commit;
256 }
257
258 static struct commit *commit_stack_pop(struct commit_stack *stack)
259 {
260         return stack->nr ? stack->items[--stack->nr] : NULL;
261 }
262
263 static void commit_stack_clear(struct commit_stack *stack)
264 {
265         FREE_AND_NULL(stack->items);
266         stack->nr = stack->alloc = 0;
267 }
268
269 static void mark_one_parent_uninteresting(struct commit *commit,
270                                           struct commit_stack *pending)
271 {
272         struct commit_list *l;
273
274         if (commit->object.flags & UNINTERESTING)
275                 return;
276         commit->object.flags |= UNINTERESTING;
277
278         /*
279          * Normally we haven't parsed the parent
280          * yet, so we won't have a parent of a parent
281          * here. However, it may turn out that we've
282          * reached this commit some other way (where it
283          * wasn't uninteresting), in which case we need
284          * to mark its parents recursively too..
285          */
286         for (l = commit->parents; l; l = l->next)
287                 commit_stack_push(pending, l->item);
288 }
289
290 void mark_parents_uninteresting(struct commit *commit)
291 {
292         struct commit_stack pending = COMMIT_STACK_INIT;
293         struct commit_list *l;
294
295         for (l = commit->parents; l; l = l->next)
296                 mark_one_parent_uninteresting(l->item, &pending);
297
298         while (pending.nr > 0)
299                 mark_one_parent_uninteresting(commit_stack_pop(&pending),
300                                               &pending);
301
302         commit_stack_clear(&pending);
303 }
304
305 static void add_pending_object_with_path(struct rev_info *revs,
306                                          struct object *obj,
307                                          const char *name, unsigned mode,
308                                          const char *path)
309 {
310         if (!obj)
311                 return;
312         if (revs->no_walk && (obj->flags & UNINTERESTING))
313                 revs->no_walk = 0;
314         if (revs->reflog_info && obj->type == OBJ_COMMIT) {
315                 struct strbuf buf = STRBUF_INIT;
316                 int len = interpret_branch_name(name, 0, &buf, 0);
317
318                 if (0 < len && name[len] && buf.len)
319                         strbuf_addstr(&buf, name + len);
320                 add_reflog_for_walk(revs->reflog_info,
321                                     (struct commit *)obj,
322                                     buf.buf[0] ? buf.buf: name);
323                 strbuf_release(&buf);
324                 return; /* do not add the commit itself */
325         }
326         add_object_array_with_path(obj, name, &revs->pending, mode, path);
327 }
328
329 static void add_pending_object_with_mode(struct rev_info *revs,
330                                          struct object *obj,
331                                          const char *name, unsigned mode)
332 {
333         add_pending_object_with_path(revs, obj, name, mode, NULL);
334 }
335
336 void add_pending_object(struct rev_info *revs,
337                         struct object *obj, const char *name)
338 {
339         add_pending_object_with_mode(revs, obj, name, S_IFINVALID);
340 }
341
342 void add_head_to_pending(struct rev_info *revs)
343 {
344         struct object_id oid;
345         struct object *obj;
346         if (get_oid("HEAD", &oid))
347                 return;
348         obj = parse_object(revs->repo, &oid);
349         if (!obj)
350                 return;
351         add_pending_object(revs, obj, "HEAD");
352 }
353
354 static struct object *get_reference(struct rev_info *revs, const char *name,
355                                     const struct object_id *oid,
356                                     unsigned int flags)
357 {
358         struct object *object;
359
360         /*
361          * If the repository has commit graphs, repo_parse_commit() avoids
362          * reading the object buffer, so use it whenever possible.
363          */
364         if (oid_object_info(revs->repo, oid, NULL) == OBJ_COMMIT) {
365                 struct commit *c = lookup_commit(revs->repo, oid);
366                 if (!repo_parse_commit(revs->repo, c))
367                         object = (struct object *) c;
368                 else
369                         object = NULL;
370         } else {
371                 object = parse_object(revs->repo, oid);
372         }
373
374         if (!object) {
375                 if (revs->ignore_missing)
376                         return object;
377                 if (revs->exclude_promisor_objects && is_promisor_object(oid))
378                         return NULL;
379                 die("bad object %s", name);
380         }
381         object->flags |= flags;
382         return object;
383 }
384
385 void add_pending_oid(struct rev_info *revs, const char *name,
386                       const struct object_id *oid, unsigned int flags)
387 {
388         struct object *object = get_reference(revs, name, oid, flags);
389         add_pending_object(revs, object, name);
390 }
391
392 static struct commit *handle_commit(struct rev_info *revs,
393                                     struct object_array_entry *entry)
394 {
395         struct object *object = entry->item;
396         const char *name = entry->name;
397         const char *path = entry->path;
398         unsigned int mode = entry->mode;
399         unsigned long flags = object->flags;
400
401         /*
402          * Tag object? Look what it points to..
403          */
404         while (object->type == OBJ_TAG) {
405                 struct tag *tag = (struct tag *) object;
406                 if (revs->tag_objects && !(flags & UNINTERESTING))
407                         add_pending_object(revs, object, tag->tag);
408                 object = parse_object(revs->repo, get_tagged_oid(tag));
409                 if (!object) {
410                         if (revs->ignore_missing_links || (flags & UNINTERESTING))
411                                 return NULL;
412                         if (revs->exclude_promisor_objects &&
413                             is_promisor_object(&tag->tagged->oid))
414                                 return NULL;
415                         die("bad object %s", oid_to_hex(&tag->tagged->oid));
416                 }
417                 object->flags |= flags;
418                 /*
419                  * We'll handle the tagged object by looping or dropping
420                  * through to the non-tag handlers below. Do not
421                  * propagate path data from the tag's pending entry.
422                  */
423                 path = NULL;
424                 mode = 0;
425         }
426
427         /*
428          * Commit object? Just return it, we'll do all the complex
429          * reachability crud.
430          */
431         if (object->type == OBJ_COMMIT) {
432                 struct commit *commit = (struct commit *)object;
433
434                 if (parse_commit(commit) < 0)
435                         die("unable to parse commit %s", name);
436                 if (flags & UNINTERESTING) {
437                         mark_parents_uninteresting(commit);
438
439                         if (!revs->topo_order || !generation_numbers_enabled(the_repository))
440                                 revs->limited = 1;
441                 }
442                 if (revs->sources) {
443                         char **slot = revision_sources_at(revs->sources, commit);
444
445                         if (!*slot)
446                                 *slot = xstrdup(name);
447                 }
448                 return commit;
449         }
450
451         /*
452          * Tree object? Either mark it uninteresting, or add it
453          * to the list of objects to look at later..
454          */
455         if (object->type == OBJ_TREE) {
456                 struct tree *tree = (struct tree *)object;
457                 if (!revs->tree_objects)
458                         return NULL;
459                 if (flags & UNINTERESTING) {
460                         mark_tree_contents_uninteresting(revs->repo, tree);
461                         return NULL;
462                 }
463                 add_pending_object_with_path(revs, object, name, mode, path);
464                 return NULL;
465         }
466
467         /*
468          * Blob object? You know the drill by now..
469          */
470         if (object->type == OBJ_BLOB) {
471                 if (!revs->blob_objects)
472                         return NULL;
473                 if (flags & UNINTERESTING)
474                         return NULL;
475                 add_pending_object_with_path(revs, object, name, mode, path);
476                 return NULL;
477         }
478         die("%s is unknown object", name);
479 }
480
481 static int everybody_uninteresting(struct commit_list *orig,
482                                    struct commit **interesting_cache)
483 {
484         struct commit_list *list = orig;
485
486         if (*interesting_cache) {
487                 struct commit *commit = *interesting_cache;
488                 if (!(commit->object.flags & UNINTERESTING))
489                         return 0;
490         }
491
492         while (list) {
493                 struct commit *commit = list->item;
494                 list = list->next;
495                 if (commit->object.flags & UNINTERESTING)
496                         continue;
497
498                 *interesting_cache = commit;
499                 return 0;
500         }
501         return 1;
502 }
503
504 /*
505  * A definition of "relevant" commit that we can use to simplify limited graphs
506  * by eliminating side branches.
507  *
508  * A "relevant" commit is one that is !UNINTERESTING (ie we are including it
509  * in our list), or that is a specified BOTTOM commit. Then after computing
510  * a limited list, during processing we can generally ignore boundary merges
511  * coming from outside the graph, (ie from irrelevant parents), and treat
512  * those merges as if they were single-parent. TREESAME is defined to consider
513  * only relevant parents, if any. If we are TREESAME to our on-graph parents,
514  * we don't care if we were !TREESAME to non-graph parents.
515  *
516  * Treating bottom commits as relevant ensures that a limited graph's
517  * connection to the actual bottom commit is not viewed as a side branch, but
518  * treated as part of the graph. For example:
519  *
520  *   ....Z...A---X---o---o---B
521  *        .     /
522  *         W---Y
523  *
524  * When computing "A..B", the A-X connection is at least as important as
525  * Y-X, despite A being flagged UNINTERESTING.
526  *
527  * And when computing --ancestry-path "A..B", the A-X connection is more
528  * important than Y-X, despite both A and Y being flagged UNINTERESTING.
529  */
530 static inline int relevant_commit(struct commit *commit)
531 {
532         return (commit->object.flags & (UNINTERESTING | BOTTOM)) != UNINTERESTING;
533 }
534
535 /*
536  * Return a single relevant commit from a parent list. If we are a TREESAME
537  * commit, and this selects one of our parents, then we can safely simplify to
538  * that parent.
539  */
540 static struct commit *one_relevant_parent(const struct rev_info *revs,
541                                           struct commit_list *orig)
542 {
543         struct commit_list *list = orig;
544         struct commit *relevant = NULL;
545
546         if (!orig)
547                 return NULL;
548
549         /*
550          * For 1-parent commits, or if first-parent-only, then return that
551          * first parent (even if not "relevant" by the above definition).
552          * TREESAME will have been set purely on that parent.
553          */
554         if (revs->first_parent_only || !orig->next)
555                 return orig->item;
556
557         /*
558          * For multi-parent commits, identify a sole relevant parent, if any.
559          * If we have only one relevant parent, then TREESAME will be set purely
560          * with regard to that parent, and we can simplify accordingly.
561          *
562          * If we have more than one relevant parent, or no relevant parents
563          * (and multiple irrelevant ones), then we can't select a parent here
564          * and return NULL.
565          */
566         while (list) {
567                 struct commit *commit = list->item;
568                 list = list->next;
569                 if (relevant_commit(commit)) {
570                         if (relevant)
571                                 return NULL;
572                         relevant = commit;
573                 }
574         }
575         return relevant;
576 }
577
578 /*
579  * The goal is to get REV_TREE_NEW as the result only if the
580  * diff consists of all '+' (and no other changes), REV_TREE_OLD
581  * if the whole diff is removal of old data, and otherwise
582  * REV_TREE_DIFFERENT (of course if the trees are the same we
583  * want REV_TREE_SAME).
584  *
585  * The only time we care about the distinction is when
586  * remove_empty_trees is in effect, in which case we care only about
587  * whether the whole change is REV_TREE_NEW, or if there's another type
588  * of change. Which means we can stop the diff early in either of these
589  * cases:
590  *
591  *   1. We're not using remove_empty_trees at all.
592  *
593  *   2. We saw anything except REV_TREE_NEW.
594  */
595 static int tree_difference = REV_TREE_SAME;
596
597 static void file_add_remove(struct diff_options *options,
598                     int addremove, unsigned mode,
599                     const struct object_id *oid,
600                     int oid_valid,
601                     const char *fullpath, unsigned dirty_submodule)
602 {
603         int diff = addremove == '+' ? REV_TREE_NEW : REV_TREE_OLD;
604         struct rev_info *revs = options->change_fn_data;
605
606         tree_difference |= diff;
607         if (!revs->remove_empty_trees || tree_difference != REV_TREE_NEW)
608                 options->flags.has_changes = 1;
609 }
610
611 static void file_change(struct diff_options *options,
612                  unsigned old_mode, unsigned new_mode,
613                  const struct object_id *old_oid,
614                  const struct object_id *new_oid,
615                  int old_oid_valid, int new_oid_valid,
616                  const char *fullpath,
617                  unsigned old_dirty_submodule, unsigned new_dirty_submodule)
618 {
619         tree_difference = REV_TREE_DIFFERENT;
620         options->flags.has_changes = 1;
621 }
622
623 static int rev_compare_tree(struct rev_info *revs,
624                             struct commit *parent, struct commit *commit)
625 {
626         struct tree *t1 = get_commit_tree(parent);
627         struct tree *t2 = get_commit_tree(commit);
628
629         if (!t1)
630                 return REV_TREE_NEW;
631         if (!t2)
632                 return REV_TREE_OLD;
633
634         if (revs->simplify_by_decoration) {
635                 /*
636                  * If we are simplifying by decoration, then the commit
637                  * is worth showing if it has a tag pointing at it.
638                  */
639                 if (get_name_decoration(&commit->object))
640                         return REV_TREE_DIFFERENT;
641                 /*
642                  * A commit that is not pointed by a tag is uninteresting
643                  * if we are not limited by path.  This means that you will
644                  * see the usual "commits that touch the paths" plus any
645                  * tagged commit by specifying both --simplify-by-decoration
646                  * and pathspec.
647                  */
648                 if (!revs->prune_data.nr)
649                         return REV_TREE_SAME;
650         }
651
652         tree_difference = REV_TREE_SAME;
653         revs->pruning.flags.has_changes = 0;
654         if (diff_tree_oid(&t1->object.oid, &t2->object.oid, "",
655                            &revs->pruning) < 0)
656                 return REV_TREE_DIFFERENT;
657         return tree_difference;
658 }
659
660 static int rev_same_tree_as_empty(struct rev_info *revs, struct commit *commit)
661 {
662         int retval;
663         struct tree *t1 = get_commit_tree(commit);
664
665         if (!t1)
666                 return 0;
667
668         tree_difference = REV_TREE_SAME;
669         revs->pruning.flags.has_changes = 0;
670         retval = diff_tree_oid(NULL, &t1->object.oid, "", &revs->pruning);
671
672         return retval >= 0 && (tree_difference == REV_TREE_SAME);
673 }
674
675 struct treesame_state {
676         unsigned int nparents;
677         unsigned char treesame[FLEX_ARRAY];
678 };
679
680 static struct treesame_state *initialise_treesame(struct rev_info *revs, struct commit *commit)
681 {
682         unsigned n = commit_list_count(commit->parents);
683         struct treesame_state *st = xcalloc(1, st_add(sizeof(*st), n));
684         st->nparents = n;
685         add_decoration(&revs->treesame, &commit->object, st);
686         return st;
687 }
688
689 /*
690  * Must be called immediately after removing the nth_parent from a commit's
691  * parent list, if we are maintaining the per-parent treesame[] decoration.
692  * This does not recalculate the master TREESAME flag - update_treesame()
693  * should be called to update it after a sequence of treesame[] modifications
694  * that may have affected it.
695  */
696 static int compact_treesame(struct rev_info *revs, struct commit *commit, unsigned nth_parent)
697 {
698         struct treesame_state *st;
699         int old_same;
700
701         if (!commit->parents) {
702                 /*
703                  * Have just removed the only parent from a non-merge.
704                  * Different handling, as we lack decoration.
705                  */
706                 if (nth_parent != 0)
707                         die("compact_treesame %u", nth_parent);
708                 old_same = !!(commit->object.flags & TREESAME);
709                 if (rev_same_tree_as_empty(revs, commit))
710                         commit->object.flags |= TREESAME;
711                 else
712                         commit->object.flags &= ~TREESAME;
713                 return old_same;
714         }
715
716         st = lookup_decoration(&revs->treesame, &commit->object);
717         if (!st || nth_parent >= st->nparents)
718                 die("compact_treesame %u", nth_parent);
719
720         old_same = st->treesame[nth_parent];
721         memmove(st->treesame + nth_parent,
722                 st->treesame + nth_parent + 1,
723                 st->nparents - nth_parent - 1);
724
725         /*
726          * If we've just become a non-merge commit, update TREESAME
727          * immediately, and remove the no-longer-needed decoration.
728          * If still a merge, defer update until update_treesame().
729          */
730         if (--st->nparents == 1) {
731                 if (commit->parents->next)
732                         die("compact_treesame parents mismatch");
733                 if (st->treesame[0] && revs->dense)
734                         commit->object.flags |= TREESAME;
735                 else
736                         commit->object.flags &= ~TREESAME;
737                 free(add_decoration(&revs->treesame, &commit->object, NULL));
738         }
739
740         return old_same;
741 }
742
743 static unsigned update_treesame(struct rev_info *revs, struct commit *commit)
744 {
745         if (commit->parents && commit->parents->next) {
746                 unsigned n;
747                 struct treesame_state *st;
748                 struct commit_list *p;
749                 unsigned relevant_parents;
750                 unsigned relevant_change, irrelevant_change;
751
752                 st = lookup_decoration(&revs->treesame, &commit->object);
753                 if (!st)
754                         die("update_treesame %s", oid_to_hex(&commit->object.oid));
755                 relevant_parents = 0;
756                 relevant_change = irrelevant_change = 0;
757                 for (p = commit->parents, n = 0; p; n++, p = p->next) {
758                         if (relevant_commit(p->item)) {
759                                 relevant_change |= !st->treesame[n];
760                                 relevant_parents++;
761                         } else
762                                 irrelevant_change |= !st->treesame[n];
763                 }
764                 if (relevant_parents ? relevant_change : irrelevant_change)
765                         commit->object.flags &= ~TREESAME;
766                 else
767                         commit->object.flags |= TREESAME;
768         }
769
770         return commit->object.flags & TREESAME;
771 }
772
773 static inline int limiting_can_increase_treesame(const struct rev_info *revs)
774 {
775         /*
776          * TREESAME is irrelevant unless prune && dense;
777          * if simplify_history is set, we can't have a mixture of TREESAME and
778          *    !TREESAME INTERESTING parents (and we don't have treesame[]
779          *    decoration anyway);
780          * if first_parent_only is set, then the TREESAME flag is locked
781          *    against the first parent (and again we lack treesame[] decoration).
782          */
783         return revs->prune && revs->dense &&
784                !revs->simplify_history &&
785                !revs->first_parent_only;
786 }
787
788 static void try_to_simplify_commit(struct rev_info *revs, struct commit *commit)
789 {
790         struct commit_list **pp, *parent;
791         struct treesame_state *ts = NULL;
792         int relevant_change = 0, irrelevant_change = 0;
793         int relevant_parents, nth_parent;
794
795         /*
796          * If we don't do pruning, everything is interesting
797          */
798         if (!revs->prune)
799                 return;
800
801         if (!get_commit_tree(commit))
802                 return;
803
804         if (!commit->parents) {
805                 if (rev_same_tree_as_empty(revs, commit))
806                         commit->object.flags |= TREESAME;
807                 return;
808         }
809
810         /*
811          * Normal non-merge commit? If we don't want to make the
812          * history dense, we consider it always to be a change..
813          */
814         if (!revs->dense && !commit->parents->next)
815                 return;
816
817         for (pp = &commit->parents, nth_parent = 0, relevant_parents = 0;
818              (parent = *pp) != NULL;
819              pp = &parent->next, nth_parent++) {
820                 struct commit *p = parent->item;
821                 if (relevant_commit(p))
822                         relevant_parents++;
823
824                 if (nth_parent == 1) {
825                         /*
826                          * This our second loop iteration - so we now know
827                          * we're dealing with a merge.
828                          *
829                          * Do not compare with later parents when we care only about
830                          * the first parent chain, in order to avoid derailing the
831                          * traversal to follow a side branch that brought everything
832                          * in the path we are limited to by the pathspec.
833                          */
834                         if (revs->first_parent_only)
835                                 break;
836                         /*
837                          * If this will remain a potentially-simplifiable
838                          * merge, remember per-parent treesame if needed.
839                          * Initialise the array with the comparison from our
840                          * first iteration.
841                          */
842                         if (revs->treesame.name &&
843                             !revs->simplify_history &&
844                             !(commit->object.flags & UNINTERESTING)) {
845                                 ts = initialise_treesame(revs, commit);
846                                 if (!(irrelevant_change || relevant_change))
847                                         ts->treesame[0] = 1;
848                         }
849                 }
850                 if (parse_commit(p) < 0)
851                         die("cannot simplify commit %s (because of %s)",
852                             oid_to_hex(&commit->object.oid),
853                             oid_to_hex(&p->object.oid));
854                 switch (rev_compare_tree(revs, p, commit)) {
855                 case REV_TREE_SAME:
856                         if (!revs->simplify_history || !relevant_commit(p)) {
857                                 /* Even if a merge with an uninteresting
858                                  * side branch brought the entire change
859                                  * we are interested in, we do not want
860                                  * to lose the other branches of this
861                                  * merge, so we just keep going.
862                                  */
863                                 if (ts)
864                                         ts->treesame[nth_parent] = 1;
865                                 continue;
866                         }
867                         parent->next = NULL;
868                         commit->parents = parent;
869                         commit->object.flags |= TREESAME;
870                         return;
871
872                 case REV_TREE_NEW:
873                         if (revs->remove_empty_trees &&
874                             rev_same_tree_as_empty(revs, p)) {
875                                 /* We are adding all the specified
876                                  * paths from this parent, so the
877                                  * history beyond this parent is not
878                                  * interesting.  Remove its parents
879                                  * (they are grandparents for us).
880                                  * IOW, we pretend this parent is a
881                                  * "root" commit.
882                                  */
883                                 if (parse_commit(p) < 0)
884                                         die("cannot simplify commit %s (invalid %s)",
885                                             oid_to_hex(&commit->object.oid),
886                                             oid_to_hex(&p->object.oid));
887                                 p->parents = NULL;
888                         }
889                 /* fallthrough */
890                 case REV_TREE_OLD:
891                 case REV_TREE_DIFFERENT:
892                         if (relevant_commit(p))
893                                 relevant_change = 1;
894                         else
895                                 irrelevant_change = 1;
896                         continue;
897                 }
898                 die("bad tree compare for commit %s", oid_to_hex(&commit->object.oid));
899         }
900
901         /*
902          * TREESAME is straightforward for single-parent commits. For merge
903          * commits, it is most useful to define it so that "irrelevant"
904          * parents cannot make us !TREESAME - if we have any relevant
905          * parents, then we only consider TREESAMEness with respect to them,
906          * allowing irrelevant merges from uninteresting branches to be
907          * simplified away. Only if we have only irrelevant parents do we
908          * base TREESAME on them. Note that this logic is replicated in
909          * update_treesame, which should be kept in sync.
910          */
911         if (relevant_parents ? !relevant_change : !irrelevant_change)
912                 commit->object.flags |= TREESAME;
913 }
914
915 static int process_parents(struct rev_info *revs, struct commit *commit,
916                            struct commit_list **list, struct prio_queue *queue)
917 {
918         struct commit_list *parent = commit->parents;
919         unsigned left_flag;
920
921         if (commit->object.flags & ADDED)
922                 return 0;
923         commit->object.flags |= ADDED;
924
925         if (revs->include_check &&
926             !revs->include_check(commit, revs->include_check_data))
927                 return 0;
928
929         /*
930          * If the commit is uninteresting, don't try to
931          * prune parents - we want the maximal uninteresting
932          * set.
933          *
934          * Normally we haven't parsed the parent
935          * yet, so we won't have a parent of a parent
936          * here. However, it may turn out that we've
937          * reached this commit some other way (where it
938          * wasn't uninteresting), in which case we need
939          * to mark its parents recursively too..
940          */
941         if (commit->object.flags & UNINTERESTING) {
942                 while (parent) {
943                         struct commit *p = parent->item;
944                         parent = parent->next;
945                         if (p)
946                                 p->object.flags |= UNINTERESTING;
947                         if (parse_commit_gently(p, 1) < 0)
948                                 continue;
949                         if (p->parents)
950                                 mark_parents_uninteresting(p);
951                         if (p->object.flags & SEEN)
952                                 continue;
953                         p->object.flags |= SEEN;
954                         if (list)
955                                 commit_list_insert_by_date(p, list);
956                         if (queue)
957                                 prio_queue_put(queue, p);
958                 }
959                 return 0;
960         }
961
962         /*
963          * Ok, the commit wasn't uninteresting. Try to
964          * simplify the commit history and find the parent
965          * that has no differences in the path set if one exists.
966          */
967         try_to_simplify_commit(revs, commit);
968
969         if (revs->no_walk)
970                 return 0;
971
972         left_flag = (commit->object.flags & SYMMETRIC_LEFT);
973
974         for (parent = commit->parents; parent; parent = parent->next) {
975                 struct commit *p = parent->item;
976                 int gently = revs->ignore_missing_links ||
977                              revs->exclude_promisor_objects;
978                 if (parse_commit_gently(p, gently) < 0) {
979                         if (revs->exclude_promisor_objects &&
980                             is_promisor_object(&p->object.oid)) {
981                                 if (revs->first_parent_only)
982                                         break;
983                                 continue;
984                         }
985                         return -1;
986                 }
987                 if (revs->sources) {
988                         char **slot = revision_sources_at(revs->sources, p);
989
990                         if (!*slot)
991                                 *slot = *revision_sources_at(revs->sources, commit);
992                 }
993                 p->object.flags |= left_flag;
994                 if (!(p->object.flags & SEEN)) {
995                         p->object.flags |= SEEN;
996                         if (list)
997                                 commit_list_insert_by_date(p, list);
998                         if (queue)
999                                 prio_queue_put(queue, p);
1000                 }
1001                 if (revs->first_parent_only)
1002                         break;
1003         }
1004         return 0;
1005 }
1006
1007 static void cherry_pick_list(struct commit_list *list, struct rev_info *revs)
1008 {
1009         struct commit_list *p;
1010         int left_count = 0, right_count = 0;
1011         int left_first;
1012         struct patch_ids ids;
1013         unsigned cherry_flag;
1014
1015         /* First count the commits on the left and on the right */
1016         for (p = list; p; p = p->next) {
1017                 struct commit *commit = p->item;
1018                 unsigned flags = commit->object.flags;
1019                 if (flags & BOUNDARY)
1020                         ;
1021                 else if (flags & SYMMETRIC_LEFT)
1022                         left_count++;
1023                 else
1024                         right_count++;
1025         }
1026
1027         if (!left_count || !right_count)
1028                 return;
1029
1030         left_first = left_count < right_count;
1031         init_patch_ids(revs->repo, &ids);
1032         ids.diffopts.pathspec = revs->diffopt.pathspec;
1033
1034         /* Compute patch-ids for one side */
1035         for (p = list; p; p = p->next) {
1036                 struct commit *commit = p->item;
1037                 unsigned flags = commit->object.flags;
1038
1039                 if (flags & BOUNDARY)
1040                         continue;
1041                 /*
1042                  * If we have fewer left, left_first is set and we omit
1043                  * commits on the right branch in this loop.  If we have
1044                  * fewer right, we skip the left ones.
1045                  */
1046                 if (left_first != !!(flags & SYMMETRIC_LEFT))
1047                         continue;
1048                 add_commit_patch_id(commit, &ids);
1049         }
1050
1051         /* either cherry_mark or cherry_pick are true */
1052         cherry_flag = revs->cherry_mark ? PATCHSAME : SHOWN;
1053
1054         /* Check the other side */
1055         for (p = list; p; p = p->next) {
1056                 struct commit *commit = p->item;
1057                 struct patch_id *id;
1058                 unsigned flags = commit->object.flags;
1059
1060                 if (flags & BOUNDARY)
1061                         continue;
1062                 /*
1063                  * If we have fewer left, left_first is set and we omit
1064                  * commits on the left branch in this loop.
1065                  */
1066                 if (left_first == !!(flags & SYMMETRIC_LEFT))
1067                         continue;
1068
1069                 /*
1070                  * Have we seen the same patch id?
1071                  */
1072                 id = has_commit_patch_id(commit, &ids);
1073                 if (!id)
1074                         continue;
1075
1076                 commit->object.flags |= cherry_flag;
1077                 id->commit->object.flags |= cherry_flag;
1078         }
1079
1080         free_patch_ids(&ids);
1081 }
1082
1083 /* How many extra uninteresting commits we want to see.. */
1084 #define SLOP 5
1085
1086 static int still_interesting(struct commit_list *src, timestamp_t date, int slop,
1087                              struct commit **interesting_cache)
1088 {
1089         /*
1090          * No source list at all? We're definitely done..
1091          */
1092         if (!src)
1093                 return 0;
1094
1095         /*
1096          * Does the destination list contain entries with a date
1097          * before the source list? Definitely _not_ done.
1098          */
1099         if (date <= src->item->date)
1100                 return SLOP;
1101
1102         /*
1103          * Does the source list still have interesting commits in
1104          * it? Definitely not done..
1105          */
1106         if (!everybody_uninteresting(src, interesting_cache))
1107                 return SLOP;
1108
1109         /* Ok, we're closing in.. */
1110         return slop-1;
1111 }
1112
1113 /*
1114  * "rev-list --ancestry-path A..B" computes commits that are ancestors
1115  * of B but not ancestors of A but further limits the result to those
1116  * that are descendants of A.  This takes the list of bottom commits and
1117  * the result of "A..B" without --ancestry-path, and limits the latter
1118  * further to the ones that can reach one of the commits in "bottom".
1119  */
1120 static void limit_to_ancestry(struct commit_list *bottom, struct commit_list *list)
1121 {
1122         struct commit_list *p;
1123         struct commit_list *rlist = NULL;
1124         int made_progress;
1125
1126         /*
1127          * Reverse the list so that it will be likely that we would
1128          * process parents before children.
1129          */
1130         for (p = list; p; p = p->next)
1131                 commit_list_insert(p->item, &rlist);
1132
1133         for (p = bottom; p; p = p->next)
1134                 p->item->object.flags |= TMP_MARK;
1135
1136         /*
1137          * Mark the ones that can reach bottom commits in "list",
1138          * in a bottom-up fashion.
1139          */
1140         do {
1141                 made_progress = 0;
1142                 for (p = rlist; p; p = p->next) {
1143                         struct commit *c = p->item;
1144                         struct commit_list *parents;
1145                         if (c->object.flags & (TMP_MARK | UNINTERESTING))
1146                                 continue;
1147                         for (parents = c->parents;
1148                              parents;
1149                              parents = parents->next) {
1150                                 if (!(parents->item->object.flags & TMP_MARK))
1151                                         continue;
1152                                 c->object.flags |= TMP_MARK;
1153                                 made_progress = 1;
1154                                 break;
1155                         }
1156                 }
1157         } while (made_progress);
1158
1159         /*
1160          * NEEDSWORK: decide if we want to remove parents that are
1161          * not marked with TMP_MARK from commit->parents for commits
1162          * in the resulting list.  We may not want to do that, though.
1163          */
1164
1165         /*
1166          * The ones that are not marked with TMP_MARK are uninteresting
1167          */
1168         for (p = list; p; p = p->next) {
1169                 struct commit *c = p->item;
1170                 if (c->object.flags & TMP_MARK)
1171                         continue;
1172                 c->object.flags |= UNINTERESTING;
1173         }
1174
1175         /* We are done with the TMP_MARK */
1176         for (p = list; p; p = p->next)
1177                 p->item->object.flags &= ~TMP_MARK;
1178         for (p = bottom; p; p = p->next)
1179                 p->item->object.flags &= ~TMP_MARK;
1180         free_commit_list(rlist);
1181 }
1182
1183 /*
1184  * Before walking the history, keep the set of "negative" refs the
1185  * caller has asked to exclude.
1186  *
1187  * This is used to compute "rev-list --ancestry-path A..B", as we need
1188  * to filter the result of "A..B" further to the ones that can actually
1189  * reach A.
1190  */
1191 static struct commit_list *collect_bottom_commits(struct commit_list *list)
1192 {
1193         struct commit_list *elem, *bottom = NULL;
1194         for (elem = list; elem; elem = elem->next)
1195                 if (elem->item->object.flags & BOTTOM)
1196                         commit_list_insert(elem->item, &bottom);
1197         return bottom;
1198 }
1199
1200 /* Assumes either left_only or right_only is set */
1201 static void limit_left_right(struct commit_list *list, struct rev_info *revs)
1202 {
1203         struct commit_list *p;
1204
1205         for (p = list; p; p = p->next) {
1206                 struct commit *commit = p->item;
1207
1208                 if (revs->right_only) {
1209                         if (commit->object.flags & SYMMETRIC_LEFT)
1210                                 commit->object.flags |= SHOWN;
1211                 } else  /* revs->left_only is set */
1212                         if (!(commit->object.flags & SYMMETRIC_LEFT))
1213                                 commit->object.flags |= SHOWN;
1214         }
1215 }
1216
1217 static int limit_list(struct rev_info *revs)
1218 {
1219         int slop = SLOP;
1220         timestamp_t date = TIME_MAX;
1221         struct commit_list *list = revs->commits;
1222         struct commit_list *newlist = NULL;
1223         struct commit_list **p = &newlist;
1224         struct commit_list *bottom = NULL;
1225         struct commit *interesting_cache = NULL;
1226
1227         if (revs->ancestry_path) {
1228                 bottom = collect_bottom_commits(list);
1229                 if (!bottom)
1230                         die("--ancestry-path given but there are no bottom commits");
1231         }
1232
1233         while (list) {
1234                 struct commit *commit = pop_commit(&list);
1235                 struct object *obj = &commit->object;
1236                 show_early_output_fn_t show;
1237
1238                 if (commit == interesting_cache)
1239                         interesting_cache = NULL;
1240
1241                 if (revs->max_age != -1 && (commit->date < revs->max_age))
1242                         obj->flags |= UNINTERESTING;
1243                 if (process_parents(revs, commit, &list, NULL) < 0)
1244                         return -1;
1245                 if (obj->flags & UNINTERESTING) {
1246                         mark_parents_uninteresting(commit);
1247                         slop = still_interesting(list, date, slop, &interesting_cache);
1248                         if (slop)
1249                                 continue;
1250                         break;
1251                 }
1252                 if (revs->min_age != -1 && (commit->date > revs->min_age))
1253                         continue;
1254                 date = commit->date;
1255                 p = &commit_list_insert(commit, p)->next;
1256
1257                 show = show_early_output;
1258                 if (!show)
1259                         continue;
1260
1261                 show(revs, newlist);
1262                 show_early_output = NULL;
1263         }
1264         if (revs->cherry_pick || revs->cherry_mark)
1265                 cherry_pick_list(newlist, revs);
1266
1267         if (revs->left_only || revs->right_only)
1268                 limit_left_right(newlist, revs);
1269
1270         if (bottom) {
1271                 limit_to_ancestry(bottom, newlist);
1272                 free_commit_list(bottom);
1273         }
1274
1275         /*
1276          * Check if any commits have become TREESAME by some of their parents
1277          * becoming UNINTERESTING.
1278          */
1279         if (limiting_can_increase_treesame(revs))
1280                 for (list = newlist; list; list = list->next) {
1281                         struct commit *c = list->item;
1282                         if (c->object.flags & (UNINTERESTING | TREESAME))
1283                                 continue;
1284                         update_treesame(revs, c);
1285                 }
1286
1287         revs->commits = newlist;
1288         return 0;
1289 }
1290
1291 /*
1292  * Add an entry to refs->cmdline with the specified information.
1293  * *name is copied.
1294  */
1295 static void add_rev_cmdline(struct rev_info *revs,
1296                             struct object *item,
1297                             const char *name,
1298                             int whence,
1299                             unsigned flags)
1300 {
1301         struct rev_cmdline_info *info = &revs->cmdline;
1302         unsigned int nr = info->nr;
1303
1304         ALLOC_GROW(info->rev, nr + 1, info->alloc);
1305         info->rev[nr].item = item;
1306         info->rev[nr].name = xstrdup(name);
1307         info->rev[nr].whence = whence;
1308         info->rev[nr].flags = flags;
1309         info->nr++;
1310 }
1311
1312 static void add_rev_cmdline_list(struct rev_info *revs,
1313                                  struct commit_list *commit_list,
1314                                  int whence,
1315                                  unsigned flags)
1316 {
1317         while (commit_list) {
1318                 struct object *object = &commit_list->item->object;
1319                 add_rev_cmdline(revs, object, oid_to_hex(&object->oid),
1320                                 whence, flags);
1321                 commit_list = commit_list->next;
1322         }
1323 }
1324
1325 struct all_refs_cb {
1326         int all_flags;
1327         int warned_bad_reflog;
1328         struct rev_info *all_revs;
1329         const char *name_for_errormsg;
1330         struct worktree *wt;
1331 };
1332
1333 int ref_excluded(struct string_list *ref_excludes, const char *path)
1334 {
1335         struct string_list_item *item;
1336
1337         if (!ref_excludes)
1338                 return 0;
1339         for_each_string_list_item(item, ref_excludes) {
1340                 if (!wildmatch(item->string, path, 0))
1341                         return 1;
1342         }
1343         return 0;
1344 }
1345
1346 static int handle_one_ref(const char *path, const struct object_id *oid,
1347                           int flag, void *cb_data)
1348 {
1349         struct all_refs_cb *cb = cb_data;
1350         struct object *object;
1351
1352         if (ref_excluded(cb->all_revs->ref_excludes, path))
1353             return 0;
1354
1355         object = get_reference(cb->all_revs, path, oid, cb->all_flags);
1356         add_rev_cmdline(cb->all_revs, object, path, REV_CMD_REF, cb->all_flags);
1357         add_pending_oid(cb->all_revs, path, oid, cb->all_flags);
1358         return 0;
1359 }
1360
1361 static void init_all_refs_cb(struct all_refs_cb *cb, struct rev_info *revs,
1362         unsigned flags)
1363 {
1364         cb->all_revs = revs;
1365         cb->all_flags = flags;
1366         revs->rev_input_given = 1;
1367         cb->wt = NULL;
1368 }
1369
1370 void clear_ref_exclusion(struct string_list **ref_excludes_p)
1371 {
1372         if (*ref_excludes_p) {
1373                 string_list_clear(*ref_excludes_p, 0);
1374                 free(*ref_excludes_p);
1375         }
1376         *ref_excludes_p = NULL;
1377 }
1378
1379 void add_ref_exclusion(struct string_list **ref_excludes_p, const char *exclude)
1380 {
1381         if (!*ref_excludes_p) {
1382                 *ref_excludes_p = xcalloc(1, sizeof(**ref_excludes_p));
1383                 (*ref_excludes_p)->strdup_strings = 1;
1384         }
1385         string_list_append(*ref_excludes_p, exclude);
1386 }
1387
1388 static void handle_refs(struct ref_store *refs,
1389                         struct rev_info *revs, unsigned flags,
1390                         int (*for_each)(struct ref_store *, each_ref_fn, void *))
1391 {
1392         struct all_refs_cb cb;
1393
1394         if (!refs) {
1395                 /* this could happen with uninitialized submodules */
1396                 return;
1397         }
1398
1399         init_all_refs_cb(&cb, revs, flags);
1400         for_each(refs, handle_one_ref, &cb);
1401 }
1402
1403 static void handle_one_reflog_commit(struct object_id *oid, void *cb_data)
1404 {
1405         struct all_refs_cb *cb = cb_data;
1406         if (!is_null_oid(oid)) {
1407                 struct object *o = parse_object(cb->all_revs->repo, oid);
1408                 if (o) {
1409                         o->flags |= cb->all_flags;
1410                         /* ??? CMDLINEFLAGS ??? */
1411                         add_pending_object(cb->all_revs, o, "");
1412                 }
1413                 else if (!cb->warned_bad_reflog) {
1414                         warning("reflog of '%s' references pruned commits",
1415                                 cb->name_for_errormsg);
1416                         cb->warned_bad_reflog = 1;
1417                 }
1418         }
1419 }
1420
1421 static int handle_one_reflog_ent(struct object_id *ooid, struct object_id *noid,
1422                 const char *email, timestamp_t timestamp, int tz,
1423                 const char *message, void *cb_data)
1424 {
1425         handle_one_reflog_commit(ooid, cb_data);
1426         handle_one_reflog_commit(noid, cb_data);
1427         return 0;
1428 }
1429
1430 static int handle_one_reflog(const char *refname_in_wt,
1431                              const struct object_id *oid,
1432                              int flag, void *cb_data)
1433 {
1434         struct all_refs_cb *cb = cb_data;
1435         struct strbuf refname = STRBUF_INIT;
1436
1437         cb->warned_bad_reflog = 0;
1438         strbuf_worktree_ref(cb->wt, &refname, refname_in_wt);
1439         cb->name_for_errormsg = refname.buf;
1440         refs_for_each_reflog_ent(get_main_ref_store(the_repository),
1441                                  refname.buf,
1442                                  handle_one_reflog_ent, cb_data);
1443         strbuf_release(&refname);
1444         return 0;
1445 }
1446
1447 static void add_other_reflogs_to_pending(struct all_refs_cb *cb)
1448 {
1449         struct worktree **worktrees, **p;
1450
1451         worktrees = get_worktrees(0);
1452         for (p = worktrees; *p; p++) {
1453                 struct worktree *wt = *p;
1454
1455                 if (wt->is_current)
1456                         continue;
1457
1458                 cb->wt = wt;
1459                 refs_for_each_reflog(get_worktree_ref_store(wt),
1460                                      handle_one_reflog,
1461                                      cb);
1462         }
1463         free_worktrees(worktrees);
1464 }
1465
1466 void add_reflogs_to_pending(struct rev_info *revs, unsigned flags)
1467 {
1468         struct all_refs_cb cb;
1469
1470         cb.all_revs = revs;
1471         cb.all_flags = flags;
1472         cb.wt = NULL;
1473         for_each_reflog(handle_one_reflog, &cb);
1474
1475         if (!revs->single_worktree)
1476                 add_other_reflogs_to_pending(&cb);
1477 }
1478
1479 static void add_cache_tree(struct cache_tree *it, struct rev_info *revs,
1480                            struct strbuf *path, unsigned int flags)
1481 {
1482         size_t baselen = path->len;
1483         int i;
1484
1485         if (it->entry_count >= 0) {
1486                 struct tree *tree = lookup_tree(revs->repo, &it->oid);
1487                 tree->object.flags |= flags;
1488                 add_pending_object_with_path(revs, &tree->object, "",
1489                                              040000, path->buf);
1490         }
1491
1492         for (i = 0; i < it->subtree_nr; i++) {
1493                 struct cache_tree_sub *sub = it->down[i];
1494                 strbuf_addf(path, "%s%s", baselen ? "/" : "", sub->name);
1495                 add_cache_tree(sub->cache_tree, revs, path, flags);
1496                 strbuf_setlen(path, baselen);
1497         }
1498
1499 }
1500
1501 static void do_add_index_objects_to_pending(struct rev_info *revs,
1502                                             struct index_state *istate,
1503                                             unsigned int flags)
1504 {
1505         int i;
1506
1507         for (i = 0; i < istate->cache_nr; i++) {
1508                 struct cache_entry *ce = istate->cache[i];
1509                 struct blob *blob;
1510
1511                 if (S_ISGITLINK(ce->ce_mode))
1512                         continue;
1513
1514                 blob = lookup_blob(revs->repo, &ce->oid);
1515                 if (!blob)
1516                         die("unable to add index blob to traversal");
1517                 blob->object.flags |= flags;
1518                 add_pending_object_with_path(revs, &blob->object, "",
1519                                              ce->ce_mode, ce->name);
1520         }
1521
1522         if (istate->cache_tree) {
1523                 struct strbuf path = STRBUF_INIT;
1524                 add_cache_tree(istate->cache_tree, revs, &path, flags);
1525                 strbuf_release(&path);
1526         }
1527 }
1528
1529 void add_index_objects_to_pending(struct rev_info *revs, unsigned int flags)
1530 {
1531         struct worktree **worktrees, **p;
1532
1533         repo_read_index(revs->repo);
1534         do_add_index_objects_to_pending(revs, revs->repo->index, flags);
1535
1536         if (revs->single_worktree)
1537                 return;
1538
1539         worktrees = get_worktrees(0);
1540         for (p = worktrees; *p; p++) {
1541                 struct worktree *wt = *p;
1542                 struct index_state istate = { NULL };
1543
1544                 if (wt->is_current)
1545                         continue; /* current index already taken care of */
1546
1547                 if (read_index_from(&istate,
1548                                     worktree_git_path(wt, "index"),
1549                                     get_worktree_git_dir(wt)) > 0)
1550                         do_add_index_objects_to_pending(revs, &istate, flags);
1551                 discard_index(&istate);
1552         }
1553         free_worktrees(worktrees);
1554 }
1555
1556 struct add_alternate_refs_data {
1557         struct rev_info *revs;
1558         unsigned int flags;
1559 };
1560
1561 static void add_one_alternate_ref(const struct object_id *oid,
1562                                   void *vdata)
1563 {
1564         const char *name = ".alternate";
1565         struct add_alternate_refs_data *data = vdata;
1566         struct object *obj;
1567
1568         obj = get_reference(data->revs, name, oid, data->flags);
1569         add_rev_cmdline(data->revs, obj, name, REV_CMD_REV, data->flags);
1570         add_pending_object(data->revs, obj, name);
1571 }
1572
1573 static void add_alternate_refs_to_pending(struct rev_info *revs,
1574                                           unsigned int flags)
1575 {
1576         struct add_alternate_refs_data data;
1577         data.revs = revs;
1578         data.flags = flags;
1579         for_each_alternate_ref(add_one_alternate_ref, &data);
1580 }
1581
1582 static int add_parents_only(struct rev_info *revs, const char *arg_, int flags,
1583                             int exclude_parent)
1584 {
1585         struct object_id oid;
1586         struct object *it;
1587         struct commit *commit;
1588         struct commit_list *parents;
1589         int parent_number;
1590         const char *arg = arg_;
1591
1592         if (*arg == '^') {
1593                 flags ^= UNINTERESTING | BOTTOM;
1594                 arg++;
1595         }
1596         if (get_oid_committish(arg, &oid))
1597                 return 0;
1598         while (1) {
1599                 it = get_reference(revs, arg, &oid, 0);
1600                 if (!it && revs->ignore_missing)
1601                         return 0;
1602                 if (it->type != OBJ_TAG)
1603                         break;
1604                 if (!((struct tag*)it)->tagged)
1605                         return 0;
1606                 oidcpy(&oid, &((struct tag*)it)->tagged->oid);
1607         }
1608         if (it->type != OBJ_COMMIT)
1609                 return 0;
1610         commit = (struct commit *)it;
1611         if (exclude_parent &&
1612             exclude_parent > commit_list_count(commit->parents))
1613                 return 0;
1614         for (parents = commit->parents, parent_number = 1;
1615              parents;
1616              parents = parents->next, parent_number++) {
1617                 if (exclude_parent && parent_number != exclude_parent)
1618                         continue;
1619
1620                 it = &parents->item->object;
1621                 it->flags |= flags;
1622                 add_rev_cmdline(revs, it, arg_, REV_CMD_PARENTS_ONLY, flags);
1623                 add_pending_object(revs, it, arg);
1624         }
1625         return 1;
1626 }
1627
1628 void repo_init_revisions(struct repository *r,
1629                          struct rev_info *revs,
1630                          const char *prefix)
1631 {
1632         memset(revs, 0, sizeof(*revs));
1633
1634         revs->repo = r;
1635         revs->abbrev = DEFAULT_ABBREV;
1636         revs->ignore_merges = 1;
1637         revs->simplify_history = 1;
1638         revs->pruning.repo = r;
1639         revs->pruning.flags.recursive = 1;
1640         revs->pruning.flags.quick = 1;
1641         revs->pruning.add_remove = file_add_remove;
1642         revs->pruning.change = file_change;
1643         revs->pruning.change_fn_data = revs;
1644         revs->sort_order = REV_SORT_IN_GRAPH_ORDER;
1645         revs->dense = 1;
1646         revs->prefix = prefix;
1647         revs->max_age = -1;
1648         revs->min_age = -1;
1649         revs->skip_count = -1;
1650         revs->max_count = -1;
1651         revs->max_parents = -1;
1652         revs->expand_tabs_in_log = -1;
1653
1654         revs->commit_format = CMIT_FMT_DEFAULT;
1655         revs->expand_tabs_in_log_default = 8;
1656
1657         init_grep_defaults(revs->repo);
1658         grep_init(&revs->grep_filter, revs->repo, prefix);
1659         revs->grep_filter.status_only = 1;
1660
1661         repo_diff_setup(revs->repo, &revs->diffopt);
1662         if (prefix && !revs->diffopt.prefix) {
1663                 revs->diffopt.prefix = prefix;
1664                 revs->diffopt.prefix_length = strlen(prefix);
1665         }
1666
1667         revs->notes_opt.use_default_notes = -1;
1668 }
1669
1670 static void add_pending_commit_list(struct rev_info *revs,
1671                                     struct commit_list *commit_list,
1672                                     unsigned int flags)
1673 {
1674         while (commit_list) {
1675                 struct object *object = &commit_list->item->object;
1676                 object->flags |= flags;
1677                 add_pending_object(revs, object, oid_to_hex(&object->oid));
1678                 commit_list = commit_list->next;
1679         }
1680 }
1681
1682 static void prepare_show_merge(struct rev_info *revs)
1683 {
1684         struct commit_list *bases;
1685         struct commit *head, *other;
1686         struct object_id oid;
1687         const char **prune = NULL;
1688         int i, prune_num = 1; /* counting terminating NULL */
1689         struct index_state *istate = revs->repo->index;
1690
1691         if (get_oid("HEAD", &oid))
1692                 die("--merge without HEAD?");
1693         head = lookup_commit_or_die(&oid, "HEAD");
1694         if (get_oid("MERGE_HEAD", &oid))
1695                 die("--merge without MERGE_HEAD?");
1696         other = lookup_commit_or_die(&oid, "MERGE_HEAD");
1697         add_pending_object(revs, &head->object, "HEAD");
1698         add_pending_object(revs, &other->object, "MERGE_HEAD");
1699         bases = get_merge_bases(head, other);
1700         add_rev_cmdline_list(revs, bases, REV_CMD_MERGE_BASE, UNINTERESTING | BOTTOM);
1701         add_pending_commit_list(revs, bases, UNINTERESTING | BOTTOM);
1702         free_commit_list(bases);
1703         head->object.flags |= SYMMETRIC_LEFT;
1704
1705         if (!istate->cache_nr)
1706                 repo_read_index(revs->repo);
1707         for (i = 0; i < istate->cache_nr; i++) {
1708                 const struct cache_entry *ce = istate->cache[i];
1709                 if (!ce_stage(ce))
1710                         continue;
1711                 if (ce_path_match(istate, ce, &revs->prune_data, NULL)) {
1712                         prune_num++;
1713                         REALLOC_ARRAY(prune, prune_num);
1714                         prune[prune_num-2] = ce->name;
1715                         prune[prune_num-1] = NULL;
1716                 }
1717                 while ((i+1 < istate->cache_nr) &&
1718                        ce_same_name(ce, istate->cache[i+1]))
1719                         i++;
1720         }
1721         clear_pathspec(&revs->prune_data);
1722         parse_pathspec(&revs->prune_data, PATHSPEC_ALL_MAGIC & ~PATHSPEC_LITERAL,
1723                        PATHSPEC_PREFER_FULL | PATHSPEC_LITERAL_PATH, "", prune);
1724         revs->limited = 1;
1725 }
1726
1727 static int dotdot_missing(const char *arg, char *dotdot,
1728                           struct rev_info *revs, int symmetric)
1729 {
1730         if (revs->ignore_missing)
1731                 return 0;
1732         /* de-munge so we report the full argument */
1733         *dotdot = '.';
1734         die(symmetric
1735             ? "Invalid symmetric difference expression %s"
1736             : "Invalid revision range %s", arg);
1737 }
1738
1739 static int handle_dotdot_1(const char *arg, char *dotdot,
1740                            struct rev_info *revs, int flags,
1741                            int cant_be_filename,
1742                            struct object_context *a_oc,
1743                            struct object_context *b_oc)
1744 {
1745         const char *a_name, *b_name;
1746         struct object_id a_oid, b_oid;
1747         struct object *a_obj, *b_obj;
1748         unsigned int a_flags, b_flags;
1749         int symmetric = 0;
1750         unsigned int flags_exclude = flags ^ (UNINTERESTING | BOTTOM);
1751         unsigned int oc_flags = GET_OID_COMMITTISH | GET_OID_RECORD_PATH;
1752
1753         a_name = arg;
1754         if (!*a_name)
1755                 a_name = "HEAD";
1756
1757         b_name = dotdot + 2;
1758         if (*b_name == '.') {
1759                 symmetric = 1;
1760                 b_name++;
1761         }
1762         if (!*b_name)
1763                 b_name = "HEAD";
1764
1765         if (get_oid_with_context(revs->repo, a_name, oc_flags, &a_oid, a_oc) ||
1766             get_oid_with_context(revs->repo, b_name, oc_flags, &b_oid, b_oc))
1767                 return -1;
1768
1769         if (!cant_be_filename) {
1770                 *dotdot = '.';
1771                 verify_non_filename(revs->prefix, arg);
1772                 *dotdot = '\0';
1773         }
1774
1775         a_obj = parse_object(revs->repo, &a_oid);
1776         b_obj = parse_object(revs->repo, &b_oid);
1777         if (!a_obj || !b_obj)
1778                 return dotdot_missing(arg, dotdot, revs, symmetric);
1779
1780         if (!symmetric) {
1781                 /* just A..B */
1782                 b_flags = flags;
1783                 a_flags = flags_exclude;
1784         } else {
1785                 /* A...B -- find merge bases between the two */
1786                 struct commit *a, *b;
1787                 struct commit_list *exclude;
1788
1789                 a = lookup_commit_reference(revs->repo, &a_obj->oid);
1790                 b = lookup_commit_reference(revs->repo, &b_obj->oid);
1791                 if (!a || !b)
1792                         return dotdot_missing(arg, dotdot, revs, symmetric);
1793
1794                 exclude = get_merge_bases(a, b);
1795                 add_rev_cmdline_list(revs, exclude, REV_CMD_MERGE_BASE,
1796                                      flags_exclude);
1797                 add_pending_commit_list(revs, exclude, flags_exclude);
1798                 free_commit_list(exclude);
1799
1800                 b_flags = flags;
1801                 a_flags = flags | SYMMETRIC_LEFT;
1802         }
1803
1804         a_obj->flags |= a_flags;
1805         b_obj->flags |= b_flags;
1806         add_rev_cmdline(revs, a_obj, a_name, REV_CMD_LEFT, a_flags);
1807         add_rev_cmdline(revs, b_obj, b_name, REV_CMD_RIGHT, b_flags);
1808         add_pending_object_with_path(revs, a_obj, a_name, a_oc->mode, a_oc->path);
1809         add_pending_object_with_path(revs, b_obj, b_name, b_oc->mode, b_oc->path);
1810         return 0;
1811 }
1812
1813 static int handle_dotdot(const char *arg,
1814                          struct rev_info *revs, int flags,
1815                          int cant_be_filename)
1816 {
1817         struct object_context a_oc, b_oc;
1818         char *dotdot = strstr(arg, "..");
1819         int ret;
1820
1821         if (!dotdot)
1822                 return -1;
1823
1824         memset(&a_oc, 0, sizeof(a_oc));
1825         memset(&b_oc, 0, sizeof(b_oc));
1826
1827         *dotdot = '\0';
1828         ret = handle_dotdot_1(arg, dotdot, revs, flags, cant_be_filename,
1829                               &a_oc, &b_oc);
1830         *dotdot = '.';
1831
1832         free(a_oc.path);
1833         free(b_oc.path);
1834
1835         return ret;
1836 }
1837
1838 int handle_revision_arg(const char *arg_, struct rev_info *revs, int flags, unsigned revarg_opt)
1839 {
1840         struct object_context oc;
1841         char *mark;
1842         struct object *object;
1843         struct object_id oid;
1844         int local_flags;
1845         const char *arg = arg_;
1846         int cant_be_filename = revarg_opt & REVARG_CANNOT_BE_FILENAME;
1847         unsigned get_sha1_flags = GET_OID_RECORD_PATH;
1848
1849         flags = flags & UNINTERESTING ? flags | BOTTOM : flags & ~BOTTOM;
1850
1851         if (!cant_be_filename && !strcmp(arg, "..")) {
1852                 /*
1853                  * Just ".."?  That is not a range but the
1854                  * pathspec for the parent directory.
1855                  */
1856                 return -1;
1857         }
1858
1859         if (!handle_dotdot(arg, revs, flags, revarg_opt))
1860                 return 0;
1861
1862         mark = strstr(arg, "^@");
1863         if (mark && !mark[2]) {
1864                 *mark = 0;
1865                 if (add_parents_only(revs, arg, flags, 0))
1866                         return 0;
1867                 *mark = '^';
1868         }
1869         mark = strstr(arg, "^!");
1870         if (mark && !mark[2]) {
1871                 *mark = 0;
1872                 if (!add_parents_only(revs, arg, flags ^ (UNINTERESTING | BOTTOM), 0))
1873                         *mark = '^';
1874         }
1875         mark = strstr(arg, "^-");
1876         if (mark) {
1877                 int exclude_parent = 1;
1878
1879                 if (mark[2]) {
1880                         char *end;
1881                         exclude_parent = strtoul(mark + 2, &end, 10);
1882                         if (*end != '\0' || !exclude_parent)
1883                                 return -1;
1884                 }
1885
1886                 *mark = 0;
1887                 if (!add_parents_only(revs, arg, flags ^ (UNINTERESTING | BOTTOM), exclude_parent))
1888                         *mark = '^';
1889         }
1890
1891         local_flags = 0;
1892         if (*arg == '^') {
1893                 local_flags = UNINTERESTING | BOTTOM;
1894                 arg++;
1895         }
1896
1897         if (revarg_opt & REVARG_COMMITTISH)
1898                 get_sha1_flags |= GET_OID_COMMITTISH;
1899
1900         if (get_oid_with_context(revs->repo, arg, get_sha1_flags, &oid, &oc))
1901                 return revs->ignore_missing ? 0 : -1;
1902         if (!cant_be_filename)
1903                 verify_non_filename(revs->prefix, arg);
1904         object = get_reference(revs, arg, &oid, flags ^ local_flags);
1905         if (!object)
1906                 return revs->ignore_missing ? 0 : -1;
1907         add_rev_cmdline(revs, object, arg_, REV_CMD_REV, flags ^ local_flags);
1908         add_pending_object_with_path(revs, object, arg, oc.mode, oc.path);
1909         free(oc.path);
1910         return 0;
1911 }
1912
1913 static void read_pathspec_from_stdin(struct strbuf *sb,
1914                                      struct argv_array *prune)
1915 {
1916         while (strbuf_getline(sb, stdin) != EOF)
1917                 argv_array_push(prune, sb->buf);
1918 }
1919
1920 static void read_revisions_from_stdin(struct rev_info *revs,
1921                                       struct argv_array *prune)
1922 {
1923         struct strbuf sb;
1924         int seen_dashdash = 0;
1925         int save_warning;
1926
1927         save_warning = warn_on_object_refname_ambiguity;
1928         warn_on_object_refname_ambiguity = 0;
1929
1930         strbuf_init(&sb, 1000);
1931         while (strbuf_getline(&sb, stdin) != EOF) {
1932                 int len = sb.len;
1933                 if (!len)
1934                         break;
1935                 if (sb.buf[0] == '-') {
1936                         if (len == 2 && sb.buf[1] == '-') {
1937                                 seen_dashdash = 1;
1938                                 break;
1939                         }
1940                         die("options not supported in --stdin mode");
1941                 }
1942                 if (handle_revision_arg(sb.buf, revs, 0,
1943                                         REVARG_CANNOT_BE_FILENAME))
1944                         die("bad revision '%s'", sb.buf);
1945         }
1946         if (seen_dashdash)
1947                 read_pathspec_from_stdin(&sb, prune);
1948
1949         strbuf_release(&sb);
1950         warn_on_object_refname_ambiguity = save_warning;
1951 }
1952
1953 static void add_grep(struct rev_info *revs, const char *ptn, enum grep_pat_token what)
1954 {
1955         append_grep_pattern(&revs->grep_filter, ptn, "command line", 0, what);
1956 }
1957
1958 static void add_header_grep(struct rev_info *revs, enum grep_header_field field, const char *pattern)
1959 {
1960         append_header_grep_pattern(&revs->grep_filter, field, pattern);
1961 }
1962
1963 static void add_message_grep(struct rev_info *revs, const char *pattern)
1964 {
1965         add_grep(revs, pattern, GREP_PATTERN_BODY);
1966 }
1967
1968 static int handle_revision_opt(struct rev_info *revs, int argc, const char **argv,
1969                                int *unkc, const char **unkv,
1970                                const struct setup_revision_opt* opt)
1971 {
1972         const char *arg = argv[0];
1973         const char *optarg;
1974         int argcount;
1975         const unsigned hexsz = the_hash_algo->hexsz;
1976
1977         /* pseudo revision arguments */
1978         if (!strcmp(arg, "--all") || !strcmp(arg, "--branches") ||
1979             !strcmp(arg, "--tags") || !strcmp(arg, "--remotes") ||
1980             !strcmp(arg, "--reflog") || !strcmp(arg, "--not") ||
1981             !strcmp(arg, "--no-walk") || !strcmp(arg, "--do-walk") ||
1982             !strcmp(arg, "--bisect") || starts_with(arg, "--glob=") ||
1983             !strcmp(arg, "--indexed-objects") ||
1984             !strcmp(arg, "--alternate-refs") ||
1985             starts_with(arg, "--exclude=") ||
1986             starts_with(arg, "--branches=") || starts_with(arg, "--tags=") ||
1987             starts_with(arg, "--remotes=") || starts_with(arg, "--no-walk="))
1988         {
1989                 unkv[(*unkc)++] = arg;
1990                 return 1;
1991         }
1992
1993         if ((argcount = parse_long_opt("max-count", argv, &optarg))) {
1994                 revs->max_count = atoi(optarg);
1995                 revs->no_walk = 0;
1996                 return argcount;
1997         } else if ((argcount = parse_long_opt("skip", argv, &optarg))) {
1998                 revs->skip_count = atoi(optarg);
1999                 return argcount;
2000         } else if ((*arg == '-') && isdigit(arg[1])) {
2001                 /* accept -<digit>, like traditional "head" */
2002                 if (strtol_i(arg + 1, 10, &revs->max_count) < 0 ||
2003                     revs->max_count < 0)
2004                         die("'%s': not a non-negative integer", arg + 1);
2005                 revs->no_walk = 0;
2006         } else if (!strcmp(arg, "-n")) {
2007                 if (argc <= 1)
2008                         return error("-n requires an argument");
2009                 revs->max_count = atoi(argv[1]);
2010                 revs->no_walk = 0;
2011                 return 2;
2012         } else if (skip_prefix(arg, "-n", &optarg)) {
2013                 revs->max_count = atoi(optarg);
2014                 revs->no_walk = 0;
2015         } else if ((argcount = parse_long_opt("max-age", argv, &optarg))) {
2016                 revs->max_age = atoi(optarg);
2017                 return argcount;
2018         } else if ((argcount = parse_long_opt("since", argv, &optarg))) {
2019                 revs->max_age = approxidate(optarg);
2020                 return argcount;
2021         } else if ((argcount = parse_long_opt("after", argv, &optarg))) {
2022                 revs->max_age = approxidate(optarg);
2023                 return argcount;
2024         } else if ((argcount = parse_long_opt("min-age", argv, &optarg))) {
2025                 revs->min_age = atoi(optarg);
2026                 return argcount;
2027         } else if ((argcount = parse_long_opt("before", argv, &optarg))) {
2028                 revs->min_age = approxidate(optarg);
2029                 return argcount;
2030         } else if ((argcount = parse_long_opt("until", argv, &optarg))) {
2031                 revs->min_age = approxidate(optarg);
2032                 return argcount;
2033         } else if (!strcmp(arg, "--first-parent")) {
2034                 revs->first_parent_only = 1;
2035         } else if (!strcmp(arg, "--ancestry-path")) {
2036                 revs->ancestry_path = 1;
2037                 revs->simplify_history = 0;
2038                 revs->limited = 1;
2039         } else if (!strcmp(arg, "-g") || !strcmp(arg, "--walk-reflogs")) {
2040                 init_reflog_walk(&revs->reflog_info);
2041         } else if (!strcmp(arg, "--default")) {
2042                 if (argc <= 1)
2043                         return error("bad --default argument");
2044                 revs->def = argv[1];
2045                 return 2;
2046         } else if (!strcmp(arg, "--merge")) {
2047                 revs->show_merge = 1;
2048         } else if (!strcmp(arg, "--topo-order")) {
2049                 revs->sort_order = REV_SORT_IN_GRAPH_ORDER;
2050                 revs->topo_order = 1;
2051         } else if (!strcmp(arg, "--simplify-merges")) {
2052                 revs->simplify_merges = 1;
2053                 revs->topo_order = 1;
2054                 revs->rewrite_parents = 1;
2055                 revs->simplify_history = 0;
2056                 revs->limited = 1;
2057         } else if (!strcmp(arg, "--simplify-by-decoration")) {
2058                 revs->simplify_merges = 1;
2059                 revs->topo_order = 1;
2060                 revs->rewrite_parents = 1;
2061                 revs->simplify_history = 0;
2062                 revs->simplify_by_decoration = 1;
2063                 revs->limited = 1;
2064                 revs->prune = 1;
2065         } else if (!strcmp(arg, "--date-order")) {
2066                 revs->sort_order = REV_SORT_BY_COMMIT_DATE;
2067                 revs->topo_order = 1;
2068         } else if (!strcmp(arg, "--author-date-order")) {
2069                 revs->sort_order = REV_SORT_BY_AUTHOR_DATE;
2070                 revs->topo_order = 1;
2071         } else if (!strcmp(arg, "--early-output")) {
2072                 revs->early_output = 100;
2073                 revs->topo_order = 1;
2074         } else if (skip_prefix(arg, "--early-output=", &optarg)) {
2075                 if (strtoul_ui(optarg, 10, &revs->early_output) < 0)
2076                         die("'%s': not a non-negative integer", optarg);
2077                 revs->topo_order = 1;
2078         } else if (!strcmp(arg, "--parents")) {
2079                 revs->rewrite_parents = 1;
2080                 revs->print_parents = 1;
2081         } else if (!strcmp(arg, "--dense")) {
2082                 revs->dense = 1;
2083         } else if (!strcmp(arg, "--sparse")) {
2084                 revs->dense = 0;
2085         } else if (!strcmp(arg, "--in-commit-order")) {
2086                 revs->tree_blobs_in_commit_order = 1;
2087         } else if (!strcmp(arg, "--remove-empty")) {
2088                 revs->remove_empty_trees = 1;
2089         } else if (!strcmp(arg, "--merges")) {
2090                 revs->min_parents = 2;
2091         } else if (!strcmp(arg, "--no-merges")) {
2092                 revs->max_parents = 1;
2093         } else if (skip_prefix(arg, "--min-parents=", &optarg)) {
2094                 revs->min_parents = atoi(optarg);
2095         } else if (!strcmp(arg, "--no-min-parents")) {
2096                 revs->min_parents = 0;
2097         } else if (skip_prefix(arg, "--max-parents=", &optarg)) {
2098                 revs->max_parents = atoi(optarg);
2099         } else if (!strcmp(arg, "--no-max-parents")) {
2100                 revs->max_parents = -1;
2101         } else if (!strcmp(arg, "--boundary")) {
2102                 revs->boundary = 1;
2103         } else if (!strcmp(arg, "--left-right")) {
2104                 revs->left_right = 1;
2105         } else if (!strcmp(arg, "--left-only")) {
2106                 if (revs->right_only)
2107                         die("--left-only is incompatible with --right-only"
2108                             " or --cherry");
2109                 revs->left_only = 1;
2110         } else if (!strcmp(arg, "--right-only")) {
2111                 if (revs->left_only)
2112                         die("--right-only is incompatible with --left-only");
2113                 revs->right_only = 1;
2114         } else if (!strcmp(arg, "--cherry")) {
2115                 if (revs->left_only)
2116                         die("--cherry is incompatible with --left-only");
2117                 revs->cherry_mark = 1;
2118                 revs->right_only = 1;
2119                 revs->max_parents = 1;
2120                 revs->limited = 1;
2121         } else if (!strcmp(arg, "--count")) {
2122                 revs->count = 1;
2123         } else if (!strcmp(arg, "--cherry-mark")) {
2124                 if (revs->cherry_pick)
2125                         die("--cherry-mark is incompatible with --cherry-pick");
2126                 revs->cherry_mark = 1;
2127                 revs->limited = 1; /* needs limit_list() */
2128         } else if (!strcmp(arg, "--cherry-pick")) {
2129                 if (revs->cherry_mark)
2130                         die("--cherry-pick is incompatible with --cherry-mark");
2131                 revs->cherry_pick = 1;
2132                 revs->limited = 1;
2133         } else if (!strcmp(arg, "--objects")) {
2134                 revs->tag_objects = 1;
2135                 revs->tree_objects = 1;
2136                 revs->blob_objects = 1;
2137         } else if (!strcmp(arg, "--objects-edge")) {
2138                 revs->tag_objects = 1;
2139                 revs->tree_objects = 1;
2140                 revs->blob_objects = 1;
2141                 revs->edge_hint = 1;
2142         } else if (!strcmp(arg, "--objects-edge-aggressive")) {
2143                 revs->tag_objects = 1;
2144                 revs->tree_objects = 1;
2145                 revs->blob_objects = 1;
2146                 revs->edge_hint = 1;
2147                 revs->edge_hint_aggressive = 1;
2148         } else if (!strcmp(arg, "--verify-objects")) {
2149                 revs->tag_objects = 1;
2150                 revs->tree_objects = 1;
2151                 revs->blob_objects = 1;
2152                 revs->verify_objects = 1;
2153         } else if (!strcmp(arg, "--unpacked")) {
2154                 revs->unpacked = 1;
2155         } else if (starts_with(arg, "--unpacked=")) {
2156                 die("--unpacked=<packfile> no longer supported.");
2157         } else if (!strcmp(arg, "-r")) {
2158                 revs->diff = 1;
2159                 revs->diffopt.flags.recursive = 1;
2160         } else if (!strcmp(arg, "-t")) {
2161                 revs->diff = 1;
2162                 revs->diffopt.flags.recursive = 1;
2163                 revs->diffopt.flags.tree_in_recursive = 1;
2164         } else if (!strcmp(arg, "-m")) {
2165                 revs->ignore_merges = 0;
2166         } else if (!strcmp(arg, "-c")) {
2167                 revs->diff = 1;
2168                 revs->dense_combined_merges = 0;
2169                 revs->combine_merges = 1;
2170         } else if (!strcmp(arg, "--combined-all-paths")) {
2171                 revs->diff = 1;
2172                 revs->combined_all_paths = 1;
2173         } else if (!strcmp(arg, "--cc")) {
2174                 revs->diff = 1;
2175                 revs->dense_combined_merges = 1;
2176                 revs->combine_merges = 1;
2177         } else if (!strcmp(arg, "-v")) {
2178                 revs->verbose_header = 1;
2179         } else if (!strcmp(arg, "--pretty")) {
2180                 revs->verbose_header = 1;
2181                 revs->pretty_given = 1;
2182                 get_commit_format(NULL, revs);
2183         } else if (skip_prefix(arg, "--pretty=", &optarg) ||
2184                    skip_prefix(arg, "--format=", &optarg)) {
2185                 /*
2186                  * Detached form ("--pretty X" as opposed to "--pretty=X")
2187                  * not allowed, since the argument is optional.
2188                  */
2189                 revs->verbose_header = 1;
2190                 revs->pretty_given = 1;
2191                 get_commit_format(optarg, revs);
2192         } else if (!strcmp(arg, "--expand-tabs")) {
2193                 revs->expand_tabs_in_log = 8;
2194         } else if (!strcmp(arg, "--no-expand-tabs")) {
2195                 revs->expand_tabs_in_log = 0;
2196         } else if (skip_prefix(arg, "--expand-tabs=", &arg)) {
2197                 int val;
2198                 if (strtol_i(arg, 10, &val) < 0 || val < 0)
2199                         die("'%s': not a non-negative integer", arg);
2200                 revs->expand_tabs_in_log = val;
2201         } else if (!strcmp(arg, "--show-notes") || !strcmp(arg, "--notes")) {
2202                 revs->show_notes = 1;
2203                 revs->show_notes_given = 1;
2204                 revs->notes_opt.use_default_notes = 1;
2205         } else if (!strcmp(arg, "--show-signature")) {
2206                 revs->show_signature = 1;
2207         } else if (!strcmp(arg, "--no-show-signature")) {
2208                 revs->show_signature = 0;
2209         } else if (!strcmp(arg, "--show-linear-break")) {
2210                 revs->break_bar = "                    ..........";
2211                 revs->track_linear = 1;
2212                 revs->track_first_time = 1;
2213         } else if (skip_prefix(arg, "--show-linear-break=", &optarg)) {
2214                 revs->break_bar = xstrdup(optarg);
2215                 revs->track_linear = 1;
2216                 revs->track_first_time = 1;
2217         } else if (skip_prefix(arg, "--show-notes=", &optarg) ||
2218                    skip_prefix(arg, "--notes=", &optarg)) {
2219                 struct strbuf buf = STRBUF_INIT;
2220                 revs->show_notes = 1;
2221                 revs->show_notes_given = 1;
2222                 if (starts_with(arg, "--show-notes=") &&
2223                     revs->notes_opt.use_default_notes < 0)
2224                         revs->notes_opt.use_default_notes = 1;
2225                 strbuf_addstr(&buf, optarg);
2226                 expand_notes_ref(&buf);
2227                 string_list_append(&revs->notes_opt.extra_notes_refs,
2228                                    strbuf_detach(&buf, NULL));
2229         } else if (!strcmp(arg, "--no-notes")) {
2230                 revs->show_notes = 0;
2231                 revs->show_notes_given = 1;
2232                 revs->notes_opt.use_default_notes = -1;
2233                 /* we have been strdup'ing ourselves, so trick
2234                  * string_list into free()ing strings */
2235                 revs->notes_opt.extra_notes_refs.strdup_strings = 1;
2236                 string_list_clear(&revs->notes_opt.extra_notes_refs, 0);
2237                 revs->notes_opt.extra_notes_refs.strdup_strings = 0;
2238         } else if (!strcmp(arg, "--standard-notes")) {
2239                 revs->show_notes_given = 1;
2240                 revs->notes_opt.use_default_notes = 1;
2241         } else if (!strcmp(arg, "--no-standard-notes")) {
2242                 revs->notes_opt.use_default_notes = 0;
2243         } else if (!strcmp(arg, "--oneline")) {
2244                 revs->verbose_header = 1;
2245                 get_commit_format("oneline", revs);
2246                 revs->pretty_given = 1;
2247                 revs->abbrev_commit = 1;
2248         } else if (!strcmp(arg, "--graph")) {
2249                 revs->topo_order = 1;
2250                 revs->rewrite_parents = 1;
2251                 revs->graph = graph_init(revs);
2252         } else if (!strcmp(arg, "--root")) {
2253                 revs->show_root_diff = 1;
2254         } else if (!strcmp(arg, "--no-commit-id")) {
2255                 revs->no_commit_id = 1;
2256         } else if (!strcmp(arg, "--always")) {
2257                 revs->always_show_header = 1;
2258         } else if (!strcmp(arg, "--no-abbrev")) {
2259                 revs->abbrev = 0;
2260         } else if (!strcmp(arg, "--abbrev")) {
2261                 revs->abbrev = DEFAULT_ABBREV;
2262         } else if (skip_prefix(arg, "--abbrev=", &optarg)) {
2263                 revs->abbrev = strtoul(optarg, NULL, 10);
2264                 if (revs->abbrev < MINIMUM_ABBREV)
2265                         revs->abbrev = MINIMUM_ABBREV;
2266                 else if (revs->abbrev > hexsz)
2267                         revs->abbrev = hexsz;
2268         } else if (!strcmp(arg, "--abbrev-commit")) {
2269                 revs->abbrev_commit = 1;
2270                 revs->abbrev_commit_given = 1;
2271         } else if (!strcmp(arg, "--no-abbrev-commit")) {
2272                 revs->abbrev_commit = 0;
2273         } else if (!strcmp(arg, "--full-diff")) {
2274                 revs->diff = 1;
2275                 revs->full_diff = 1;
2276         } else if (!strcmp(arg, "--full-history")) {
2277                 revs->simplify_history = 0;
2278         } else if (!strcmp(arg, "--relative-date")) {
2279                 revs->date_mode.type = DATE_RELATIVE;
2280                 revs->date_mode_explicit = 1;
2281         } else if ((argcount = parse_long_opt("date", argv, &optarg))) {
2282                 parse_date_format(optarg, &revs->date_mode);
2283                 revs->date_mode_explicit = 1;
2284                 return argcount;
2285         } else if (!strcmp(arg, "--log-size")) {
2286                 revs->show_log_size = 1;
2287         }
2288         /*
2289          * Grepping the commit log
2290          */
2291         else if ((argcount = parse_long_opt("author", argv, &optarg))) {
2292                 add_header_grep(revs, GREP_HEADER_AUTHOR, optarg);
2293                 return argcount;
2294         } else if ((argcount = parse_long_opt("committer", argv, &optarg))) {
2295                 add_header_grep(revs, GREP_HEADER_COMMITTER, optarg);
2296                 return argcount;
2297         } else if ((argcount = parse_long_opt("grep-reflog", argv, &optarg))) {
2298                 add_header_grep(revs, GREP_HEADER_REFLOG, optarg);
2299                 return argcount;
2300         } else if ((argcount = parse_long_opt("grep", argv, &optarg))) {
2301                 add_message_grep(revs, optarg);
2302                 return argcount;
2303         } else if (!strcmp(arg, "--grep-debug")) {
2304                 revs->grep_filter.debug = 1;
2305         } else if (!strcmp(arg, "--basic-regexp")) {
2306                 revs->grep_filter.pattern_type_option = GREP_PATTERN_TYPE_BRE;
2307         } else if (!strcmp(arg, "--extended-regexp") || !strcmp(arg, "-E")) {
2308                 revs->grep_filter.pattern_type_option = GREP_PATTERN_TYPE_ERE;
2309         } else if (!strcmp(arg, "--regexp-ignore-case") || !strcmp(arg, "-i")) {
2310                 revs->grep_filter.ignore_case = 1;
2311                 revs->diffopt.pickaxe_opts |= DIFF_PICKAXE_IGNORE_CASE;
2312         } else if (!strcmp(arg, "--fixed-strings") || !strcmp(arg, "-F")) {
2313                 revs->grep_filter.pattern_type_option = GREP_PATTERN_TYPE_FIXED;
2314         } else if (!strcmp(arg, "--perl-regexp") || !strcmp(arg, "-P")) {
2315                 revs->grep_filter.pattern_type_option = GREP_PATTERN_TYPE_PCRE;
2316         } else if (!strcmp(arg, "--all-match")) {
2317                 revs->grep_filter.all_match = 1;
2318         } else if (!strcmp(arg, "--invert-grep")) {
2319                 revs->invert_grep = 1;
2320         } else if ((argcount = parse_long_opt("encoding", argv, &optarg))) {
2321                 if (strcmp(optarg, "none"))
2322                         git_log_output_encoding = xstrdup(optarg);
2323                 else
2324                         git_log_output_encoding = "";
2325                 return argcount;
2326         } else if (!strcmp(arg, "--reverse")) {
2327                 revs->reverse ^= 1;
2328         } else if (!strcmp(arg, "--children")) {
2329                 revs->children.name = "children";
2330                 revs->limited = 1;
2331         } else if (!strcmp(arg, "--ignore-missing")) {
2332                 revs->ignore_missing = 1;
2333         } else if (opt && opt->allow_exclude_promisor_objects &&
2334                    !strcmp(arg, "--exclude-promisor-objects")) {
2335                 if (fetch_if_missing)
2336                         BUG("exclude_promisor_objects can only be used when fetch_if_missing is 0");
2337                 revs->exclude_promisor_objects = 1;
2338         } else {
2339                 int opts = diff_opt_parse(&revs->diffopt, argv, argc, revs->prefix);
2340                 if (!opts)
2341                         unkv[(*unkc)++] = arg;
2342                 return opts;
2343         }
2344         if (revs->graph && revs->track_linear)
2345                 die("--show-linear-break and --graph are incompatible");
2346
2347         return 1;
2348 }
2349
2350 void parse_revision_opt(struct rev_info *revs, struct parse_opt_ctx_t *ctx,
2351                         const struct option *options,
2352                         const char * const usagestr[])
2353 {
2354         int n = handle_revision_opt(revs, ctx->argc, ctx->argv,
2355                                     &ctx->cpidx, ctx->out, NULL);
2356         if (n <= 0) {
2357                 error("unknown option `%s'", ctx->argv[0]);
2358                 usage_with_options(usagestr, options);
2359         }
2360         ctx->argv += n;
2361         ctx->argc -= n;
2362 }
2363
2364 static int for_each_bisect_ref(struct ref_store *refs, each_ref_fn fn,
2365                                void *cb_data, const char *term)
2366 {
2367         struct strbuf bisect_refs = STRBUF_INIT;
2368         int status;
2369         strbuf_addf(&bisect_refs, "refs/bisect/%s", term);
2370         status = refs_for_each_fullref_in(refs, bisect_refs.buf, fn, cb_data, 0);
2371         strbuf_release(&bisect_refs);
2372         return status;
2373 }
2374
2375 static int for_each_bad_bisect_ref(struct ref_store *refs, each_ref_fn fn, void *cb_data)
2376 {
2377         return for_each_bisect_ref(refs, fn, cb_data, term_bad);
2378 }
2379
2380 static int for_each_good_bisect_ref(struct ref_store *refs, each_ref_fn fn, void *cb_data)
2381 {
2382         return for_each_bisect_ref(refs, fn, cb_data, term_good);
2383 }
2384
2385 static int handle_revision_pseudo_opt(const char *submodule,
2386                                 struct rev_info *revs,
2387                                 int argc, const char **argv, int *flags)
2388 {
2389         const char *arg = argv[0];
2390         const char *optarg;
2391         struct ref_store *refs;
2392         int argcount;
2393
2394         if (submodule) {
2395                 /*
2396                  * We need some something like get_submodule_worktrees()
2397                  * before we can go through all worktrees of a submodule,
2398                  * .e.g with adding all HEADs from --all, which is not
2399                  * supported right now, so stick to single worktree.
2400                  */
2401                 if (!revs->single_worktree)
2402                         BUG("--single-worktree cannot be used together with submodule");
2403                 refs = get_submodule_ref_store(submodule);
2404         } else
2405                 refs = get_main_ref_store(revs->repo);
2406
2407         /*
2408          * NOTE!
2409          *
2410          * Commands like "git shortlog" will not accept the options below
2411          * unless parse_revision_opt queues them (as opposed to erroring
2412          * out).
2413          *
2414          * When implementing your new pseudo-option, remember to
2415          * register it in the list at the top of handle_revision_opt.
2416          */
2417         if (!strcmp(arg, "--all")) {
2418                 handle_refs(refs, revs, *flags, refs_for_each_ref);
2419                 handle_refs(refs, revs, *flags, refs_head_ref);
2420                 if (!revs->single_worktree) {
2421                         struct all_refs_cb cb;
2422
2423                         init_all_refs_cb(&cb, revs, *flags);
2424                         other_head_refs(handle_one_ref, &cb);
2425                 }
2426                 clear_ref_exclusion(&revs->ref_excludes);
2427         } else if (!strcmp(arg, "--branches")) {
2428                 handle_refs(refs, revs, *flags, refs_for_each_branch_ref);
2429                 clear_ref_exclusion(&revs->ref_excludes);
2430         } else if (!strcmp(arg, "--bisect")) {
2431                 read_bisect_terms(&term_bad, &term_good);
2432                 handle_refs(refs, revs, *flags, for_each_bad_bisect_ref);
2433                 handle_refs(refs, revs, *flags ^ (UNINTERESTING | BOTTOM),
2434                             for_each_good_bisect_ref);
2435                 revs->bisect = 1;
2436         } else if (!strcmp(arg, "--tags")) {
2437                 handle_refs(refs, revs, *flags, refs_for_each_tag_ref);
2438                 clear_ref_exclusion(&revs->ref_excludes);
2439         } else if (!strcmp(arg, "--remotes")) {
2440                 handle_refs(refs, revs, *flags, refs_for_each_remote_ref);
2441                 clear_ref_exclusion(&revs->ref_excludes);
2442         } else if ((argcount = parse_long_opt("glob", argv, &optarg))) {
2443                 struct all_refs_cb cb;
2444                 init_all_refs_cb(&cb, revs, *flags);
2445                 for_each_glob_ref(handle_one_ref, optarg, &cb);
2446                 clear_ref_exclusion(&revs->ref_excludes);
2447                 return argcount;
2448         } else if ((argcount = parse_long_opt("exclude", argv, &optarg))) {
2449                 add_ref_exclusion(&revs->ref_excludes, optarg);
2450                 return argcount;
2451         } else if (skip_prefix(arg, "--branches=", &optarg)) {
2452                 struct all_refs_cb cb;
2453                 init_all_refs_cb(&cb, revs, *flags);
2454                 for_each_glob_ref_in(handle_one_ref, optarg, "refs/heads/", &cb);
2455                 clear_ref_exclusion(&revs->ref_excludes);
2456         } else if (skip_prefix(arg, "--tags=", &optarg)) {
2457                 struct all_refs_cb cb;
2458                 init_all_refs_cb(&cb, revs, *flags);
2459                 for_each_glob_ref_in(handle_one_ref, optarg, "refs/tags/", &cb);
2460                 clear_ref_exclusion(&revs->ref_excludes);
2461         } else if (skip_prefix(arg, "--remotes=", &optarg)) {
2462                 struct all_refs_cb cb;
2463                 init_all_refs_cb(&cb, revs, *flags);
2464                 for_each_glob_ref_in(handle_one_ref, optarg, "refs/remotes/", &cb);
2465                 clear_ref_exclusion(&revs->ref_excludes);
2466         } else if (!strcmp(arg, "--reflog")) {
2467                 add_reflogs_to_pending(revs, *flags);
2468         } else if (!strcmp(arg, "--indexed-objects")) {
2469                 add_index_objects_to_pending(revs, *flags);
2470         } else if (!strcmp(arg, "--alternate-refs")) {
2471                 add_alternate_refs_to_pending(revs, *flags);
2472         } else if (!strcmp(arg, "--not")) {
2473                 *flags ^= UNINTERESTING | BOTTOM;
2474         } else if (!strcmp(arg, "--no-walk")) {
2475                 revs->no_walk = REVISION_WALK_NO_WALK_SORTED;
2476         } else if (skip_prefix(arg, "--no-walk=", &optarg)) {
2477                 /*
2478                  * Detached form ("--no-walk X" as opposed to "--no-walk=X")
2479                  * not allowed, since the argument is optional.
2480                  */
2481                 if (!strcmp(optarg, "sorted"))
2482                         revs->no_walk = REVISION_WALK_NO_WALK_SORTED;
2483                 else if (!strcmp(optarg, "unsorted"))
2484                         revs->no_walk = REVISION_WALK_NO_WALK_UNSORTED;
2485                 else
2486                         return error("invalid argument to --no-walk");
2487         } else if (!strcmp(arg, "--do-walk")) {
2488                 revs->no_walk = 0;
2489         } else if (!strcmp(arg, "--single-worktree")) {
2490                 revs->single_worktree = 1;
2491         } else {
2492                 return 0;
2493         }
2494
2495         return 1;
2496 }
2497
2498 static void NORETURN diagnose_missing_default(const char *def)
2499 {
2500         int flags;
2501         const char *refname;
2502
2503         refname = resolve_ref_unsafe(def, 0, NULL, &flags);
2504         if (!refname || !(flags & REF_ISSYMREF) || (flags & REF_ISBROKEN))
2505                 die(_("your current branch appears to be broken"));
2506
2507         skip_prefix(refname, "refs/heads/", &refname);
2508         die(_("your current branch '%s' does not have any commits yet"),
2509             refname);
2510 }
2511
2512 /*
2513  * Parse revision information, filling in the "rev_info" structure,
2514  * and removing the used arguments from the argument list.
2515  *
2516  * Returns the number of arguments left that weren't recognized
2517  * (which are also moved to the head of the argument list)
2518  */
2519 int setup_revisions(int argc, const char **argv, struct rev_info *revs, struct setup_revision_opt *opt)
2520 {
2521         int i, flags, left, seen_dashdash, got_rev_arg = 0, revarg_opt;
2522         struct argv_array prune_data = ARGV_ARRAY_INIT;
2523         const char *submodule = NULL;
2524         int seen_end_of_options = 0;
2525
2526         if (opt)
2527                 submodule = opt->submodule;
2528
2529         /* First, search for "--" */
2530         if (opt && opt->assume_dashdash) {
2531                 seen_dashdash = 1;
2532         } else {
2533                 seen_dashdash = 0;
2534                 for (i = 1; i < argc; i++) {
2535                         const char *arg = argv[i];
2536                         if (strcmp(arg, "--"))
2537                                 continue;
2538                         argv[i] = NULL;
2539                         argc = i;
2540                         if (argv[i + 1])
2541                                 argv_array_pushv(&prune_data, argv + i + 1);
2542                         seen_dashdash = 1;
2543                         break;
2544                 }
2545         }
2546
2547         /* Second, deal with arguments and options */
2548         flags = 0;
2549         revarg_opt = opt ? opt->revarg_opt : 0;
2550         if (seen_dashdash)
2551                 revarg_opt |= REVARG_CANNOT_BE_FILENAME;
2552         for (left = i = 1; i < argc; i++) {
2553                 const char *arg = argv[i];
2554                 if (!seen_end_of_options && *arg == '-') {
2555                         int opts;
2556
2557                         opts = handle_revision_pseudo_opt(submodule,
2558                                                 revs, argc - i, argv + i,
2559                                                 &flags);
2560                         if (opts > 0) {
2561                                 i += opts - 1;
2562                                 continue;
2563                         }
2564
2565                         if (!strcmp(arg, "--stdin")) {
2566                                 if (revs->disable_stdin) {
2567                                         argv[left++] = arg;
2568                                         continue;
2569                                 }
2570                                 if (revs->read_from_stdin++)
2571                                         die("--stdin given twice?");
2572                                 read_revisions_from_stdin(revs, &prune_data);
2573                                 continue;
2574                         }
2575
2576                         if (!strcmp(arg, "--end-of-options")) {
2577                                 seen_end_of_options = 1;
2578                                 continue;
2579                         }
2580
2581                         opts = handle_revision_opt(revs, argc - i, argv + i,
2582                                                    &left, argv, opt);
2583                         if (opts > 0) {
2584                                 i += opts - 1;
2585                                 continue;
2586                         }
2587                         if (opts < 0)
2588                                 exit(128);
2589                         continue;
2590                 }
2591
2592
2593                 if (handle_revision_arg(arg, revs, flags, revarg_opt)) {
2594                         int j;
2595                         if (seen_dashdash || *arg == '^')
2596                                 die("bad revision '%s'", arg);
2597
2598                         /* If we didn't have a "--":
2599                          * (1) all filenames must exist;
2600                          * (2) all rev-args must not be interpretable
2601                          *     as a valid filename.
2602                          * but the latter we have checked in the main loop.
2603                          */
2604                         for (j = i; j < argc; j++)
2605                                 verify_filename(revs->prefix, argv[j], j == i);
2606
2607                         argv_array_pushv(&prune_data, argv + i);
2608                         break;
2609                 }
2610                 else
2611                         got_rev_arg = 1;
2612         }
2613
2614         if (prune_data.argc) {
2615                 /*
2616                  * If we need to introduce the magic "a lone ':' means no
2617                  * pathspec whatsoever", here is the place to do so.
2618                  *
2619                  * if (prune_data.nr == 1 && !strcmp(prune_data[0], ":")) {
2620                  *      prune_data.nr = 0;
2621                  *      prune_data.alloc = 0;
2622                  *      free(prune_data.path);
2623                  *      prune_data.path = NULL;
2624                  * } else {
2625                  *      terminate prune_data.alloc with NULL and
2626                  *      call init_pathspec() to set revs->prune_data here.
2627                  * }
2628                  */
2629                 parse_pathspec(&revs->prune_data, 0, 0,
2630                                revs->prefix, prune_data.argv);
2631         }
2632         argv_array_clear(&prune_data);
2633
2634         if (revs->def == NULL)
2635                 revs->def = opt ? opt->def : NULL;
2636         if (opt && opt->tweak)
2637                 opt->tweak(revs, opt);
2638         if (revs->show_merge)
2639                 prepare_show_merge(revs);
2640         if (revs->def && !revs->pending.nr && !revs->rev_input_given && !got_rev_arg) {
2641                 struct object_id oid;
2642                 struct object *object;
2643                 struct object_context oc;
2644                 if (get_oid_with_context(revs->repo, revs->def, 0, &oid, &oc))
2645                         diagnose_missing_default(revs->def);
2646                 object = get_reference(revs, revs->def, &oid, 0);
2647                 add_pending_object_with_mode(revs, object, revs->def, oc.mode);
2648         }
2649
2650         /* Did the user ask for any diff output? Run the diff! */
2651         if (revs->diffopt.output_format & ~DIFF_FORMAT_NO_OUTPUT)
2652                 revs->diff = 1;
2653
2654         /* Pickaxe, diff-filter and rename following need diffs */
2655         if ((revs->diffopt.pickaxe_opts & DIFF_PICKAXE_KINDS_MASK) ||
2656             revs->diffopt.filter ||
2657             revs->diffopt.flags.follow_renames)
2658                 revs->diff = 1;
2659
2660         if (revs->diffopt.objfind)
2661                 revs->simplify_history = 0;
2662
2663         if (revs->topo_order && !generation_numbers_enabled(the_repository))
2664                 revs->limited = 1;
2665
2666         if (revs->prune_data.nr) {
2667                 copy_pathspec(&revs->pruning.pathspec, &revs->prune_data);
2668                 /* Can't prune commits with rename following: the paths change.. */
2669                 if (!revs->diffopt.flags.follow_renames)
2670                         revs->prune = 1;
2671                 if (!revs->full_diff)
2672                         copy_pathspec(&revs->diffopt.pathspec,
2673                                       &revs->prune_data);
2674         }
2675         if (revs->combine_merges)
2676                 revs->ignore_merges = 0;
2677         if (revs->combined_all_paths && !revs->combine_merges)
2678                 die("--combined-all-paths makes no sense without -c or --cc");
2679
2680         revs->diffopt.abbrev = revs->abbrev;
2681
2682         if (revs->line_level_traverse) {
2683                 revs->limited = 1;
2684                 revs->topo_order = 1;
2685         }
2686
2687         diff_setup_done(&revs->diffopt);
2688
2689         grep_commit_pattern_type(GREP_PATTERN_TYPE_UNSPECIFIED,
2690                                  &revs->grep_filter);
2691         if (!is_encoding_utf8(get_log_output_encoding()))
2692                 revs->grep_filter.ignore_locale = 1;
2693         compile_grep_patterns(&revs->grep_filter);
2694
2695         if (revs->reverse && revs->reflog_info)
2696                 die("cannot combine --reverse with --walk-reflogs");
2697         if (revs->reflog_info && revs->limited)
2698                 die("cannot combine --walk-reflogs with history-limiting options");
2699         if (revs->rewrite_parents && revs->children.name)
2700                 die("cannot combine --parents and --children");
2701
2702         /*
2703          * Limitations on the graph functionality
2704          */
2705         if (revs->reverse && revs->graph)
2706                 die("cannot combine --reverse with --graph");
2707
2708         if (revs->reflog_info && revs->graph)
2709                 die("cannot combine --walk-reflogs with --graph");
2710         if (revs->no_walk && revs->graph)
2711                 die("cannot combine --no-walk with --graph");
2712         if (!revs->reflog_info && revs->grep_filter.use_reflog_filter)
2713                 die("cannot use --grep-reflog without --walk-reflogs");
2714
2715         if (revs->first_parent_only && revs->bisect)
2716                 die(_("--first-parent is incompatible with --bisect"));
2717
2718         if (revs->line_level_traverse &&
2719             (revs->diffopt.output_format & ~(DIFF_FORMAT_PATCH | DIFF_FORMAT_NO_OUTPUT)))
2720                 die(_("-L does not yet support diff formats besides -p and -s"));
2721
2722         if (revs->expand_tabs_in_log < 0)
2723                 revs->expand_tabs_in_log = revs->expand_tabs_in_log_default;
2724
2725         return left;
2726 }
2727
2728 static void add_child(struct rev_info *revs, struct commit *parent, struct commit *child)
2729 {
2730         struct commit_list *l = xcalloc(1, sizeof(*l));
2731
2732         l->item = child;
2733         l->next = add_decoration(&revs->children, &parent->object, l);
2734 }
2735
2736 static int remove_duplicate_parents(struct rev_info *revs, struct commit *commit)
2737 {
2738         struct treesame_state *ts = lookup_decoration(&revs->treesame, &commit->object);
2739         struct commit_list **pp, *p;
2740         int surviving_parents;
2741
2742         /* Examine existing parents while marking ones we have seen... */
2743         pp = &commit->parents;
2744         surviving_parents = 0;
2745         while ((p = *pp) != NULL) {
2746                 struct commit *parent = p->item;
2747                 if (parent->object.flags & TMP_MARK) {
2748                         *pp = p->next;
2749                         if (ts)
2750                                 compact_treesame(revs, commit, surviving_parents);
2751                         continue;
2752                 }
2753                 parent->object.flags |= TMP_MARK;
2754                 surviving_parents++;
2755                 pp = &p->next;
2756         }
2757         /* clear the temporary mark */
2758         for (p = commit->parents; p; p = p->next) {
2759                 p->item->object.flags &= ~TMP_MARK;
2760         }
2761         /* no update_treesame() - removing duplicates can't affect TREESAME */
2762         return surviving_parents;
2763 }
2764
2765 struct merge_simplify_state {
2766         struct commit *simplified;
2767 };
2768
2769 static struct merge_simplify_state *locate_simplify_state(struct rev_info *revs, struct commit *commit)
2770 {
2771         struct merge_simplify_state *st;
2772
2773         st = lookup_decoration(&revs->merge_simplification, &commit->object);
2774         if (!st) {
2775                 st = xcalloc(1, sizeof(*st));
2776                 add_decoration(&revs->merge_simplification, &commit->object, st);
2777         }
2778         return st;
2779 }
2780
2781 static int mark_redundant_parents(struct commit *commit)
2782 {
2783         struct commit_list *h = reduce_heads(commit->parents);
2784         int i = 0, marked = 0;
2785         struct commit_list *po, *pn;
2786
2787         /* Want these for sanity-checking only */
2788         int orig_cnt = commit_list_count(commit->parents);
2789         int cnt = commit_list_count(h);
2790
2791         /*
2792          * Not ready to remove items yet, just mark them for now, based
2793          * on the output of reduce_heads(). reduce_heads outputs the reduced
2794          * set in its original order, so this isn't too hard.
2795          */
2796         po = commit->parents;
2797         pn = h;
2798         while (po) {
2799                 if (pn && po->item == pn->item) {
2800                         pn = pn->next;
2801                         i++;
2802                 } else {
2803                         po->item->object.flags |= TMP_MARK;
2804                         marked++;
2805                 }
2806                 po=po->next;
2807         }
2808
2809         if (i != cnt || cnt+marked != orig_cnt)
2810                 die("mark_redundant_parents %d %d %d %d", orig_cnt, cnt, i, marked);
2811
2812         free_commit_list(h);
2813
2814         return marked;
2815 }
2816
2817 static int mark_treesame_root_parents(struct commit *commit)
2818 {
2819         struct commit_list *p;
2820         int marked = 0;
2821
2822         for (p = commit->parents; p; p = p->next) {
2823                 struct commit *parent = p->item;
2824                 if (!parent->parents && (parent->object.flags & TREESAME)) {
2825                         parent->object.flags |= TMP_MARK;
2826                         marked++;
2827                 }
2828         }
2829
2830         return marked;
2831 }
2832
2833 /*
2834  * Awkward naming - this means one parent we are TREESAME to.
2835  * cf mark_treesame_root_parents: root parents that are TREESAME (to an
2836  * empty tree). Better name suggestions?
2837  */
2838 static int leave_one_treesame_to_parent(struct rev_info *revs, struct commit *commit)
2839 {
2840         struct treesame_state *ts = lookup_decoration(&revs->treesame, &commit->object);
2841         struct commit *unmarked = NULL, *marked = NULL;
2842         struct commit_list *p;
2843         unsigned n;
2844
2845         for (p = commit->parents, n = 0; p; p = p->next, n++) {
2846                 if (ts->treesame[n]) {
2847                         if (p->item->object.flags & TMP_MARK) {
2848                                 if (!marked)
2849                                         marked = p->item;
2850                         } else {
2851                                 if (!unmarked) {
2852                                         unmarked = p->item;
2853                                         break;
2854                                 }
2855                         }
2856                 }
2857         }
2858
2859         /*
2860          * If we are TREESAME to a marked-for-deletion parent, but not to any
2861          * unmarked parents, unmark the first TREESAME parent. This is the
2862          * parent that the default simplify_history==1 scan would have followed,
2863          * and it doesn't make sense to omit that path when asking for a
2864          * simplified full history. Retaining it improves the chances of
2865          * understanding odd missed merges that took an old version of a file.
2866          *
2867          * Example:
2868          *
2869          *   I--------*X       A modified the file, but mainline merge X used
2870          *    \       /        "-s ours", so took the version from I. X is
2871          *     `-*A--'         TREESAME to I and !TREESAME to A.
2872          *
2873          * Default log from X would produce "I". Without this check,
2874          * --full-history --simplify-merges would produce "I-A-X", showing
2875          * the merge commit X and that it changed A, but not making clear that
2876          * it had just taken the I version. With this check, the topology above
2877          * is retained.
2878          *
2879          * Note that it is possible that the simplification chooses a different
2880          * TREESAME parent from the default, in which case this test doesn't
2881          * activate, and we _do_ drop the default parent. Example:
2882          *
2883          *   I------X         A modified the file, but it was reverted in B,
2884          *    \    /          meaning mainline merge X is TREESAME to both
2885          *    *A-*B           parents.
2886          *
2887          * Default log would produce "I" by following the first parent;
2888          * --full-history --simplify-merges will produce "I-A-B". But this is a
2889          * reasonable result - it presents a logical full history leading from
2890          * I to X, and X is not an important merge.
2891          */
2892         if (!unmarked && marked) {
2893                 marked->object.flags &= ~TMP_MARK;
2894                 return 1;
2895         }
2896
2897         return 0;
2898 }
2899
2900 static int remove_marked_parents(struct rev_info *revs, struct commit *commit)
2901 {
2902         struct commit_list **pp, *p;
2903         int nth_parent, removed = 0;
2904
2905         pp = &commit->parents;
2906         nth_parent = 0;
2907         while ((p = *pp) != NULL) {
2908                 struct commit *parent = p->item;
2909                 if (parent->object.flags & TMP_MARK) {
2910                         parent->object.flags &= ~TMP_MARK;
2911                         *pp = p->next;
2912                         free(p);
2913                         removed++;
2914                         compact_treesame(revs, commit, nth_parent);
2915                         continue;
2916                 }
2917                 pp = &p->next;
2918                 nth_parent++;
2919         }
2920
2921         /* Removing parents can only increase TREESAMEness */
2922         if (removed && !(commit->object.flags & TREESAME))
2923                 update_treesame(revs, commit);
2924
2925         return nth_parent;
2926 }
2927
2928 static struct commit_list **simplify_one(struct rev_info *revs, struct commit *commit, struct commit_list **tail)
2929 {
2930         struct commit_list *p;
2931         struct commit *parent;
2932         struct merge_simplify_state *st, *pst;
2933         int cnt;
2934
2935         st = locate_simplify_state(revs, commit);
2936
2937         /*
2938          * Have we handled this one?
2939          */
2940         if (st->simplified)
2941                 return tail;
2942
2943         /*
2944          * An UNINTERESTING commit simplifies to itself, so does a
2945          * root commit.  We do not rewrite parents of such commit
2946          * anyway.
2947          */
2948         if ((commit->object.flags & UNINTERESTING) || !commit->parents) {
2949                 st->simplified = commit;
2950                 return tail;
2951         }
2952
2953         /*
2954          * Do we know what commit all of our parents that matter
2955          * should be rewritten to?  Otherwise we are not ready to
2956          * rewrite this one yet.
2957          */
2958         for (cnt = 0, p = commit->parents; p; p = p->next) {
2959                 pst = locate_simplify_state(revs, p->item);
2960                 if (!pst->simplified) {
2961                         tail = &commit_list_insert(p->item, tail)->next;
2962                         cnt++;
2963                 }
2964                 if (revs->first_parent_only)
2965                         break;
2966         }
2967         if (cnt) {
2968                 tail = &commit_list_insert(commit, tail)->next;
2969                 return tail;
2970         }
2971
2972         /*
2973          * Rewrite our list of parents. Note that this cannot
2974          * affect our TREESAME flags in any way - a commit is
2975          * always TREESAME to its simplification.
2976          */
2977         for (p = commit->parents; p; p = p->next) {
2978                 pst = locate_simplify_state(revs, p->item);
2979                 p->item = pst->simplified;
2980                 if (revs->first_parent_only)
2981                         break;
2982         }
2983
2984         if (revs->first_parent_only)
2985                 cnt = 1;
2986         else
2987                 cnt = remove_duplicate_parents(revs, commit);
2988
2989         /*
2990          * It is possible that we are a merge and one side branch
2991          * does not have any commit that touches the given paths;
2992          * in such a case, the immediate parent from that branch
2993          * will be rewritten to be the merge base.
2994          *
2995          *      o----X          X: the commit we are looking at;
2996          *     /    /           o: a commit that touches the paths;
2997          * ---o----'
2998          *
2999          * Further, a merge of an independent branch that doesn't
3000          * touch the path will reduce to a treesame root parent:
3001          *
3002          *  ----o----X          X: the commit we are looking at;
3003          *          /           o: a commit that touches the paths;
3004          *         r            r: a root commit not touching the paths
3005          *
3006          * Detect and simplify both cases.
3007          */
3008         if (1 < cnt) {
3009                 int marked = mark_redundant_parents(commit);
3010                 marked += mark_treesame_root_parents(commit);
3011                 if (marked)
3012                         marked -= leave_one_treesame_to_parent(revs, commit);
3013                 if (marked)
3014                         cnt = remove_marked_parents(revs, commit);
3015         }
3016
3017         /*
3018          * A commit simplifies to itself if it is a root, if it is
3019          * UNINTERESTING, if it touches the given paths, or if it is a
3020          * merge and its parents don't simplify to one relevant commit
3021          * (the first two cases are already handled at the beginning of
3022          * this function).
3023          *
3024          * Otherwise, it simplifies to what its sole relevant parent
3025          * simplifies to.
3026          */
3027         if (!cnt ||
3028             (commit->object.flags & UNINTERESTING) ||
3029             !(commit->object.flags & TREESAME) ||
3030             (parent = one_relevant_parent(revs, commit->parents)) == NULL)
3031                 st->simplified = commit;
3032         else {
3033                 pst = locate_simplify_state(revs, parent);
3034                 st->simplified = pst->simplified;
3035         }
3036         return tail;
3037 }
3038
3039 static void simplify_merges(struct rev_info *revs)
3040 {
3041         struct commit_list *list, *next;
3042         struct commit_list *yet_to_do, **tail;
3043         struct commit *commit;
3044
3045         if (!revs->prune)
3046                 return;
3047
3048         /* feed the list reversed */
3049         yet_to_do = NULL;
3050         for (list = revs->commits; list; list = next) {
3051                 commit = list->item;
3052                 next = list->next;
3053                 /*
3054                  * Do not free(list) here yet; the original list
3055                  * is used later in this function.
3056                  */
3057                 commit_list_insert(commit, &yet_to_do);
3058         }
3059         while (yet_to_do) {
3060                 list = yet_to_do;
3061                 yet_to_do = NULL;
3062                 tail = &yet_to_do;
3063                 while (list) {
3064                         commit = pop_commit(&list);
3065                         tail = simplify_one(revs, commit, tail);
3066                 }
3067         }
3068
3069         /* clean up the result, removing the simplified ones */
3070         list = revs->commits;
3071         revs->commits = NULL;
3072         tail = &revs->commits;
3073         while (list) {
3074                 struct merge_simplify_state *st;
3075
3076                 commit = pop_commit(&list);
3077                 st = locate_simplify_state(revs, commit);
3078                 if (st->simplified == commit)
3079                         tail = &commit_list_insert(commit, tail)->next;
3080         }
3081 }
3082
3083 static void set_children(struct rev_info *revs)
3084 {
3085         struct commit_list *l;
3086         for (l = revs->commits; l; l = l->next) {
3087                 struct commit *commit = l->item;
3088                 struct commit_list *p;
3089
3090                 for (p = commit->parents; p; p = p->next)
3091                         add_child(revs, p->item, commit);
3092         }
3093 }
3094
3095 void reset_revision_walk(void)
3096 {
3097         clear_object_flags(SEEN | ADDED | SHOWN);
3098 }
3099
3100 static int mark_uninteresting(const struct object_id *oid,
3101                               struct packed_git *pack,
3102                               uint32_t pos,
3103                               void *cb)
3104 {
3105         struct rev_info *revs = cb;
3106         struct object *o = parse_object(revs->repo, oid);
3107         o->flags |= UNINTERESTING | SEEN;
3108         return 0;
3109 }
3110
3111 define_commit_slab(indegree_slab, int);
3112 define_commit_slab(author_date_slab, timestamp_t);
3113
3114 struct topo_walk_info {
3115         uint32_t min_generation;
3116         struct prio_queue explore_queue;
3117         struct prio_queue indegree_queue;
3118         struct prio_queue topo_queue;
3119         struct indegree_slab indegree;
3120         struct author_date_slab author_date;
3121 };
3122
3123 static inline void test_flag_and_insert(struct prio_queue *q, struct commit *c, int flag)
3124 {
3125         if (c->object.flags & flag)
3126                 return;
3127
3128         c->object.flags |= flag;
3129         prio_queue_put(q, c);
3130 }
3131
3132 static void explore_walk_step(struct rev_info *revs)
3133 {
3134         struct topo_walk_info *info = revs->topo_walk_info;
3135         struct commit_list *p;
3136         struct commit *c = prio_queue_get(&info->explore_queue);
3137
3138         if (!c)
3139                 return;
3140
3141         if (parse_commit_gently(c, 1) < 0)
3142                 return;
3143
3144         if (revs->sort_order == REV_SORT_BY_AUTHOR_DATE)
3145                 record_author_date(&info->author_date, c);
3146
3147         if (revs->max_age != -1 && (c->date < revs->max_age))
3148                 c->object.flags |= UNINTERESTING;
3149
3150         if (process_parents(revs, c, NULL, NULL) < 0)
3151                 return;
3152
3153         if (c->object.flags & UNINTERESTING)
3154                 mark_parents_uninteresting(c);
3155
3156         for (p = c->parents; p; p = p->next)
3157                 test_flag_and_insert(&info->explore_queue, p->item, TOPO_WALK_EXPLORED);
3158 }
3159
3160 static void explore_to_depth(struct rev_info *revs,
3161                              uint32_t gen_cutoff)
3162 {
3163         struct topo_walk_info *info = revs->topo_walk_info;
3164         struct commit *c;
3165         while ((c = prio_queue_peek(&info->explore_queue)) &&
3166                c->generation >= gen_cutoff)
3167                 explore_walk_step(revs);
3168 }
3169
3170 static void indegree_walk_step(struct rev_info *revs)
3171 {
3172         struct commit_list *p;
3173         struct topo_walk_info *info = revs->topo_walk_info;
3174         struct commit *c = prio_queue_get(&info->indegree_queue);
3175
3176         if (!c)
3177                 return;
3178
3179         if (parse_commit_gently(c, 1) < 0)
3180                 return;
3181
3182         explore_to_depth(revs, c->generation);
3183
3184         for (p = c->parents; p; p = p->next) {
3185                 struct commit *parent = p->item;
3186                 int *pi = indegree_slab_at(&info->indegree, parent);
3187
3188                 if (*pi)
3189                         (*pi)++;
3190                 else
3191                         *pi = 2;
3192
3193                 test_flag_and_insert(&info->indegree_queue, parent, TOPO_WALK_INDEGREE);
3194
3195                 if (revs->first_parent_only)
3196                         return;
3197         }
3198 }
3199
3200 static void compute_indegrees_to_depth(struct rev_info *revs,
3201                                        uint32_t gen_cutoff)
3202 {
3203         struct topo_walk_info *info = revs->topo_walk_info;
3204         struct commit *c;
3205         while ((c = prio_queue_peek(&info->indegree_queue)) &&
3206                c->generation >= gen_cutoff)
3207                 indegree_walk_step(revs);
3208 }
3209
3210 static void init_topo_walk(struct rev_info *revs)
3211 {
3212         struct topo_walk_info *info;
3213         struct commit_list *list;
3214         revs->topo_walk_info = xmalloc(sizeof(struct topo_walk_info));
3215         info = revs->topo_walk_info;
3216         memset(info, 0, sizeof(struct topo_walk_info));
3217
3218         init_indegree_slab(&info->indegree);
3219         memset(&info->explore_queue, 0, sizeof(info->explore_queue));
3220         memset(&info->indegree_queue, 0, sizeof(info->indegree_queue));
3221         memset(&info->topo_queue, 0, sizeof(info->topo_queue));
3222
3223         switch (revs->sort_order) {
3224         default: /* REV_SORT_IN_GRAPH_ORDER */
3225                 info->topo_queue.compare = NULL;
3226                 break;
3227         case REV_SORT_BY_COMMIT_DATE:
3228                 info->topo_queue.compare = compare_commits_by_commit_date;
3229                 break;
3230         case REV_SORT_BY_AUTHOR_DATE:
3231                 init_author_date_slab(&info->author_date);
3232                 info->topo_queue.compare = compare_commits_by_author_date;
3233                 info->topo_queue.cb_data = &info->author_date;
3234                 break;
3235         }
3236
3237         info->explore_queue.compare = compare_commits_by_gen_then_commit_date;
3238         info->indegree_queue.compare = compare_commits_by_gen_then_commit_date;
3239
3240         info->min_generation = GENERATION_NUMBER_INFINITY;
3241         for (list = revs->commits; list; list = list->next) {
3242                 struct commit *c = list->item;
3243
3244                 if (parse_commit_gently(c, 1))
3245                         continue;
3246
3247                 test_flag_and_insert(&info->explore_queue, c, TOPO_WALK_EXPLORED);
3248                 test_flag_and_insert(&info->indegree_queue, c, TOPO_WALK_INDEGREE);
3249
3250                 if (c->generation < info->min_generation)
3251                         info->min_generation = c->generation;
3252
3253                 *(indegree_slab_at(&info->indegree, c)) = 1;
3254
3255                 if (revs->sort_order == REV_SORT_BY_AUTHOR_DATE)
3256                         record_author_date(&info->author_date, c);
3257         }
3258         compute_indegrees_to_depth(revs, info->min_generation);
3259
3260         for (list = revs->commits; list; list = list->next) {
3261                 struct commit *c = list->item;
3262
3263                 if (*(indegree_slab_at(&info->indegree, c)) == 1)
3264                         prio_queue_put(&info->topo_queue, c);
3265         }
3266
3267         /*
3268          * This is unfortunate; the initial tips need to be shown
3269          * in the order given from the revision traversal machinery.
3270          */
3271         if (revs->sort_order == REV_SORT_IN_GRAPH_ORDER)
3272                 prio_queue_reverse(&info->topo_queue);
3273 }
3274
3275 static struct commit *next_topo_commit(struct rev_info *revs)
3276 {
3277         struct commit *c;
3278         struct topo_walk_info *info = revs->topo_walk_info;
3279
3280         /* pop next off of topo_queue */
3281         c = prio_queue_get(&info->topo_queue);
3282
3283         if (c)
3284                 *(indegree_slab_at(&info->indegree, c)) = 0;
3285
3286         return c;
3287 }
3288
3289 static void expand_topo_walk(struct rev_info *revs, struct commit *commit)
3290 {
3291         struct commit_list *p;
3292         struct topo_walk_info *info = revs->topo_walk_info;
3293         if (process_parents(revs, commit, NULL, NULL) < 0) {
3294                 if (!revs->ignore_missing_links)
3295                         die("Failed to traverse parents of commit %s",
3296                             oid_to_hex(&commit->object.oid));
3297         }
3298
3299         for (p = commit->parents; p; p = p->next) {
3300                 struct commit *parent = p->item;
3301                 int *pi;
3302
3303                 if (parent->object.flags & UNINTERESTING)
3304                         continue;
3305
3306                 if (parse_commit_gently(parent, 1) < 0)
3307                         continue;
3308
3309                 if (parent->generation < info->min_generation) {
3310                         info->min_generation = parent->generation;
3311                         compute_indegrees_to_depth(revs, info->min_generation);
3312                 }
3313
3314                 pi = indegree_slab_at(&info->indegree, parent);
3315
3316                 (*pi)--;
3317                 if (*pi == 1)
3318                         prio_queue_put(&info->topo_queue, parent);
3319
3320                 if (revs->first_parent_only)
3321                         return;
3322         }
3323 }
3324
3325 int prepare_revision_walk(struct rev_info *revs)
3326 {
3327         int i;
3328         struct object_array old_pending;
3329         struct commit_list **next = &revs->commits;
3330
3331         memcpy(&old_pending, &revs->pending, sizeof(old_pending));
3332         revs->pending.nr = 0;
3333         revs->pending.alloc = 0;
3334         revs->pending.objects = NULL;
3335         for (i = 0; i < old_pending.nr; i++) {
3336                 struct object_array_entry *e = old_pending.objects + i;
3337                 struct commit *commit = handle_commit(revs, e);
3338                 if (commit) {
3339                         if (!(commit->object.flags & SEEN)) {
3340                                 commit->object.flags |= SEEN;
3341                                 next = commit_list_append(commit, next);
3342                         }
3343                 }
3344         }
3345         object_array_clear(&old_pending);
3346
3347         /* Signal whether we need per-parent treesame decoration */
3348         if (revs->simplify_merges ||
3349             (revs->limited && limiting_can_increase_treesame(revs)))
3350                 revs->treesame.name = "treesame";
3351
3352         if (revs->exclude_promisor_objects) {
3353                 for_each_packed_object(mark_uninteresting, revs,
3354                                        FOR_EACH_OBJECT_PROMISOR_ONLY);
3355         }
3356
3357         if (revs->no_walk != REVISION_WALK_NO_WALK_UNSORTED)
3358                 commit_list_sort_by_date(&revs->commits);
3359         if (revs->no_walk)
3360                 return 0;
3361         if (revs->limited) {
3362                 if (limit_list(revs) < 0)
3363                         return -1;
3364                 if (revs->topo_order)
3365                         sort_in_topological_order(&revs->commits, revs->sort_order);
3366         } else if (revs->topo_order)
3367                 init_topo_walk(revs);
3368         if (revs->line_level_traverse)
3369                 line_log_filter(revs);
3370         if (revs->simplify_merges)
3371                 simplify_merges(revs);
3372         if (revs->children.name)
3373                 set_children(revs);
3374         return 0;
3375 }
3376
3377 static enum rewrite_result rewrite_one_1(struct rev_info *revs,
3378                                          struct commit **pp,
3379                                          struct prio_queue *queue)
3380 {
3381         for (;;) {
3382                 struct commit *p = *pp;
3383                 if (!revs->limited)
3384                         if (process_parents(revs, p, NULL, queue) < 0)
3385                                 return rewrite_one_error;
3386                 if (p->object.flags & UNINTERESTING)
3387                         return rewrite_one_ok;
3388                 if (!(p->object.flags & TREESAME))
3389                         return rewrite_one_ok;
3390                 if (!p->parents)
3391                         return rewrite_one_noparents;
3392                 if ((p = one_relevant_parent(revs, p->parents)) == NULL)
3393                         return rewrite_one_ok;
3394                 *pp = p;
3395         }
3396 }
3397
3398 static void merge_queue_into_list(struct prio_queue *q, struct commit_list **list)
3399 {
3400         while (q->nr) {
3401                 struct commit *item = prio_queue_peek(q);
3402                 struct commit_list *p = *list;
3403
3404                 if (p && p->item->date >= item->date)
3405                         list = &p->next;
3406                 else {
3407                         p = commit_list_insert(item, list);
3408                         list = &p->next; /* skip newly added item */
3409                         prio_queue_get(q); /* pop item */
3410                 }
3411         }
3412 }
3413
3414 static enum rewrite_result rewrite_one(struct rev_info *revs, struct commit **pp)
3415 {
3416         struct prio_queue queue = { compare_commits_by_commit_date };
3417         enum rewrite_result ret = rewrite_one_1(revs, pp, &queue);
3418         merge_queue_into_list(&queue, &revs->commits);
3419         clear_prio_queue(&queue);
3420         return ret;
3421 }
3422
3423 int rewrite_parents(struct rev_info *revs, struct commit *commit,
3424         rewrite_parent_fn_t rewrite_parent)
3425 {
3426         struct commit_list **pp = &commit->parents;
3427         while (*pp) {
3428                 struct commit_list *parent = *pp;
3429                 switch (rewrite_parent(revs, &parent->item)) {
3430                 case rewrite_one_ok:
3431                         break;
3432                 case rewrite_one_noparents:
3433                         *pp = parent->next;
3434                         continue;
3435                 case rewrite_one_error:
3436                         return -1;
3437                 }
3438                 pp = &parent->next;
3439         }
3440         remove_duplicate_parents(revs, commit);
3441         return 0;
3442 }
3443
3444 static int commit_rewrite_person(struct strbuf *buf, const char *what, struct string_list *mailmap)
3445 {
3446         char *person, *endp;
3447         size_t len, namelen, maillen;
3448         const char *name;
3449         const char *mail;
3450         struct ident_split ident;
3451
3452         person = strstr(buf->buf, what);
3453         if (!person)
3454                 return 0;
3455
3456         person += strlen(what);
3457         endp = strchr(person, '\n');
3458         if (!endp)
3459                 return 0;
3460
3461         len = endp - person;
3462
3463         if (split_ident_line(&ident, person, len))
3464                 return 0;
3465
3466         mail = ident.mail_begin;
3467         maillen = ident.mail_end - ident.mail_begin;
3468         name = ident.name_begin;
3469         namelen = ident.name_end - ident.name_begin;
3470
3471         if (map_user(mailmap, &mail, &maillen, &name, &namelen)) {
3472                 struct strbuf namemail = STRBUF_INIT;
3473
3474                 strbuf_addf(&namemail, "%.*s <%.*s>",
3475                             (int)namelen, name, (int)maillen, mail);
3476
3477                 strbuf_splice(buf, ident.name_begin - buf->buf,
3478                               ident.mail_end - ident.name_begin + 1,
3479                               namemail.buf, namemail.len);
3480
3481                 strbuf_release(&namemail);
3482
3483                 return 1;
3484         }
3485
3486         return 0;
3487 }
3488
3489 static int commit_match(struct commit *commit, struct rev_info *opt)
3490 {
3491         int retval;
3492         const char *encoding;
3493         const char *message;
3494         struct strbuf buf = STRBUF_INIT;
3495
3496         if (!opt->grep_filter.pattern_list && !opt->grep_filter.header_list)
3497                 return 1;
3498
3499         /* Prepend "fake" headers as needed */
3500         if (opt->grep_filter.use_reflog_filter) {
3501                 strbuf_addstr(&buf, "reflog ");
3502                 get_reflog_message(&buf, opt->reflog_info);
3503                 strbuf_addch(&buf, '\n');
3504         }
3505
3506         /*
3507          * We grep in the user's output encoding, under the assumption that it
3508          * is the encoding they are most likely to write their grep pattern
3509          * for. In addition, it means we will match the "notes" encoding below,
3510          * so we will not end up with a buffer that has two different encodings
3511          * in it.
3512          */
3513         encoding = get_log_output_encoding();
3514         message = logmsg_reencode(commit, NULL, encoding);
3515
3516         /* Copy the commit to temporary if we are using "fake" headers */
3517         if (buf.len)
3518                 strbuf_addstr(&buf, message);
3519
3520         if (opt->grep_filter.header_list && opt->mailmap) {
3521                 if (!buf.len)
3522                         strbuf_addstr(&buf, message);
3523
3524                 commit_rewrite_person(&buf, "\nauthor ", opt->mailmap);
3525                 commit_rewrite_person(&buf, "\ncommitter ", opt->mailmap);
3526         }
3527
3528         /* Append "fake" message parts as needed */
3529         if (opt->show_notes) {
3530                 if (!buf.len)
3531                         strbuf_addstr(&buf, message);
3532                 format_display_notes(&commit->object.oid, &buf, encoding, 1);
3533         }
3534
3535         /*
3536          * Find either in the original commit message, or in the temporary.
3537          * Note that we cast away the constness of "message" here. It is
3538          * const because it may come from the cached commit buffer. That's OK,
3539          * because we know that it is modifiable heap memory, and that while
3540          * grep_buffer may modify it for speed, it will restore any
3541          * changes before returning.
3542          */
3543         if (buf.len)
3544                 retval = grep_buffer(&opt->grep_filter, buf.buf, buf.len);
3545         else
3546                 retval = grep_buffer(&opt->grep_filter,
3547                                      (char *)message, strlen(message));
3548         strbuf_release(&buf);
3549         unuse_commit_buffer(commit, message);
3550         return opt->invert_grep ? !retval : retval;
3551 }
3552
3553 static inline int want_ancestry(const struct rev_info *revs)
3554 {
3555         return (revs->rewrite_parents || revs->children.name);
3556 }
3557
3558 /*
3559  * Return a timestamp to be used for --since/--until comparisons for this
3560  * commit, based on the revision options.
3561  */
3562 static timestamp_t comparison_date(const struct rev_info *revs,
3563                                    struct commit *commit)
3564 {
3565         return revs->reflog_info ?
3566                 get_reflog_timestamp(revs->reflog_info) :
3567                 commit->date;
3568 }
3569
3570 enum commit_action get_commit_action(struct rev_info *revs, struct commit *commit)
3571 {
3572         if (commit->object.flags & SHOWN)
3573                 return commit_ignore;
3574         if (revs->unpacked && has_object_pack(&commit->object.oid))
3575                 return commit_ignore;
3576         if (commit->object.flags & UNINTERESTING)
3577                 return commit_ignore;
3578         if (revs->min_age != -1 &&
3579             comparison_date(revs, commit) > revs->min_age)
3580                         return commit_ignore;
3581         if (revs->min_parents || (revs->max_parents >= 0)) {
3582                 int n = commit_list_count(commit->parents);
3583                 if ((n < revs->min_parents) ||
3584                     ((revs->max_parents >= 0) && (n > revs->max_parents)))
3585                         return commit_ignore;
3586         }
3587         if (!commit_match(commit, revs))
3588                 return commit_ignore;
3589         if (revs->prune && revs->dense) {
3590                 /* Commit without changes? */
3591                 if (commit->object.flags & TREESAME) {
3592                         int n;
3593                         struct commit_list *p;
3594                         /* drop merges unless we want parenthood */
3595                         if (!want_ancestry(revs))
3596                                 return commit_ignore;
3597                         /*
3598                          * If we want ancestry, then need to keep any merges
3599                          * between relevant commits to tie together topology.
3600                          * For consistency with TREESAME and simplification
3601                          * use "relevant" here rather than just INTERESTING,
3602                          * to treat bottom commit(s) as part of the topology.
3603                          */
3604                         for (n = 0, p = commit->parents; p; p = p->next)
3605                                 if (relevant_commit(p->item))
3606                                         if (++n >= 2)
3607                                                 return commit_show;
3608                         return commit_ignore;
3609                 }
3610         }
3611         return commit_show;
3612 }
3613
3614 define_commit_slab(saved_parents, struct commit_list *);
3615
3616 #define EMPTY_PARENT_LIST ((struct commit_list *)-1)
3617
3618 /*
3619  * You may only call save_parents() once per commit (this is checked
3620  * for non-root commits).
3621  */
3622 static void save_parents(struct rev_info *revs, struct commit *commit)
3623 {
3624         struct commit_list **pp;
3625
3626         if (!revs->saved_parents_slab) {
3627                 revs->saved_parents_slab = xmalloc(sizeof(struct saved_parents));
3628                 init_saved_parents(revs->saved_parents_slab);
3629         }
3630
3631         pp = saved_parents_at(revs->saved_parents_slab, commit);
3632
3633         /*
3634          * When walking with reflogs, we may visit the same commit
3635          * several times: once for each appearance in the reflog.
3636          *
3637          * In this case, save_parents() will be called multiple times.
3638          * We want to keep only the first set of parents.  We need to
3639          * store a sentinel value for an empty (i.e., NULL) parent
3640          * list to distinguish it from a not-yet-saved list, however.
3641          */
3642         if (*pp)
3643                 return;
3644         if (commit->parents)
3645                 *pp = copy_commit_list(commit->parents);
3646         else
3647                 *pp = EMPTY_PARENT_LIST;
3648 }
3649
3650 static void free_saved_parents(struct rev_info *revs)
3651 {
3652         if (revs->saved_parents_slab)
3653                 clear_saved_parents(revs->saved_parents_slab);
3654 }
3655
3656 struct commit_list *get_saved_parents(struct rev_info *revs, const struct commit *commit)
3657 {
3658         struct commit_list *parents;
3659
3660         if (!revs->saved_parents_slab)
3661                 return commit->parents;
3662
3663         parents = *saved_parents_at(revs->saved_parents_slab, commit);
3664         if (parents == EMPTY_PARENT_LIST)
3665                 return NULL;
3666         return parents;
3667 }
3668
3669 enum commit_action simplify_commit(struct rev_info *revs, struct commit *commit)
3670 {
3671         enum commit_action action = get_commit_action(revs, commit);
3672
3673         if (action == commit_show &&
3674             revs->prune && revs->dense && want_ancestry(revs)) {
3675                 /*
3676                  * --full-diff on simplified parents is no good: it
3677                  * will show spurious changes from the commits that
3678                  * were elided.  So we save the parents on the side
3679                  * when --full-diff is in effect.
3680                  */
3681                 if (revs->full_diff)
3682                         save_parents(revs, commit);
3683                 if (rewrite_parents(revs, commit, rewrite_one) < 0)
3684                         return commit_error;
3685         }
3686         return action;
3687 }
3688
3689 static void track_linear(struct rev_info *revs, struct commit *commit)
3690 {
3691         if (revs->track_first_time) {
3692                 revs->linear = 1;
3693                 revs->track_first_time = 0;
3694         } else {
3695                 struct commit_list *p;
3696                 for (p = revs->previous_parents; p; p = p->next)
3697                         if (p->item == NULL || /* first commit */
3698                             oideq(&p->item->object.oid, &commit->object.oid))
3699                                 break;
3700                 revs->linear = p != NULL;
3701         }
3702         if (revs->reverse) {
3703                 if (revs->linear)
3704                         commit->object.flags |= TRACK_LINEAR;
3705         }
3706         free_commit_list(revs->previous_parents);
3707         revs->previous_parents = copy_commit_list(commit->parents);
3708 }
3709
3710 static struct commit *get_revision_1(struct rev_info *revs)
3711 {
3712         while (1) {
3713                 struct commit *commit;
3714
3715                 if (revs->reflog_info)
3716                         commit = next_reflog_entry(revs->reflog_info);
3717                 else if (revs->topo_walk_info)
3718                         commit = next_topo_commit(revs);
3719                 else
3720                         commit = pop_commit(&revs->commits);
3721
3722                 if (!commit)
3723                         return NULL;
3724
3725                 if (revs->reflog_info)
3726                         commit->object.flags &= ~(ADDED | SEEN | SHOWN);
3727
3728                 /*
3729                  * If we haven't done the list limiting, we need to look at
3730                  * the parents here. We also need to do the date-based limiting
3731                  * that we'd otherwise have done in limit_list().
3732                  */
3733                 if (!revs->limited) {
3734                         if (revs->max_age != -1 &&
3735                             comparison_date(revs, commit) < revs->max_age)
3736                                 continue;
3737
3738                         if (revs->reflog_info)
3739                                 try_to_simplify_commit(revs, commit);
3740                         else if (revs->topo_walk_info)
3741                                 expand_topo_walk(revs, commit);
3742                         else if (process_parents(revs, commit, &revs->commits, NULL) < 0) {
3743                                 if (!revs->ignore_missing_links)
3744                                         die("Failed to traverse parents of commit %s",
3745                                                 oid_to_hex(&commit->object.oid));
3746                         }
3747                 }
3748
3749                 switch (simplify_commit(revs, commit)) {
3750                 case commit_ignore:
3751                         continue;
3752                 case commit_error:
3753                         die("Failed to simplify parents of commit %s",
3754                             oid_to_hex(&commit->object.oid));
3755                 default:
3756                         if (revs->track_linear)
3757                                 track_linear(revs, commit);
3758                         return commit;
3759                 }
3760         }
3761 }
3762
3763 /*
3764  * Return true for entries that have not yet been shown.  (This is an
3765  * object_array_each_func_t.)
3766  */
3767 static int entry_unshown(struct object_array_entry *entry, void *cb_data_unused)
3768 {
3769         return !(entry->item->flags & SHOWN);
3770 }
3771
3772 /*
3773  * If array is on the verge of a realloc, garbage-collect any entries
3774  * that have already been shown to try to free up some space.
3775  */
3776 static void gc_boundary(struct object_array *array)
3777 {
3778         if (array->nr == array->alloc)
3779                 object_array_filter(array, entry_unshown, NULL);
3780 }
3781
3782 static void create_boundary_commit_list(struct rev_info *revs)
3783 {
3784         unsigned i;
3785         struct commit *c;
3786         struct object_array *array = &revs->boundary_commits;
3787         struct object_array_entry *objects = array->objects;
3788
3789         /*
3790          * If revs->commits is non-NULL at this point, an error occurred in
3791          * get_revision_1().  Ignore the error and continue printing the
3792          * boundary commits anyway.  (This is what the code has always
3793          * done.)
3794          */
3795         if (revs->commits) {
3796                 free_commit_list(revs->commits);
3797                 revs->commits = NULL;
3798         }
3799
3800         /*
3801          * Put all of the actual boundary commits from revs->boundary_commits
3802          * into revs->commits
3803          */
3804         for (i = 0; i < array->nr; i++) {
3805                 c = (struct commit *)(objects[i].item);
3806                 if (!c)
3807                         continue;
3808                 if (!(c->object.flags & CHILD_SHOWN))
3809                         continue;
3810                 if (c->object.flags & (SHOWN | BOUNDARY))
3811                         continue;
3812                 c->object.flags |= BOUNDARY;
3813                 commit_list_insert(c, &revs->commits);
3814         }
3815
3816         /*
3817          * If revs->topo_order is set, sort the boundary commits
3818          * in topological order
3819          */
3820         sort_in_topological_order(&revs->commits, revs->sort_order);
3821 }
3822
3823 static struct commit *get_revision_internal(struct rev_info *revs)
3824 {
3825         struct commit *c = NULL;
3826         struct commit_list *l;
3827
3828         if (revs->boundary == 2) {
3829                 /*
3830                  * All of the normal commits have already been returned,
3831                  * and we are now returning boundary commits.
3832                  * create_boundary_commit_list() has populated
3833                  * revs->commits with the remaining commits to return.
3834                  */
3835                 c = pop_commit(&revs->commits);
3836                 if (c)
3837                         c->object.flags |= SHOWN;
3838                 return c;
3839         }
3840
3841         /*
3842          * If our max_count counter has reached zero, then we are done. We
3843          * don't simply return NULL because we still might need to show
3844          * boundary commits. But we want to avoid calling get_revision_1, which
3845          * might do a considerable amount of work finding the next commit only
3846          * for us to throw it away.
3847          *
3848          * If it is non-zero, then either we don't have a max_count at all
3849          * (-1), or it is still counting, in which case we decrement.
3850          */
3851         if (revs->max_count) {
3852                 c = get_revision_1(revs);
3853                 if (c) {
3854                         while (revs->skip_count > 0) {
3855                                 revs->skip_count--;
3856                                 c = get_revision_1(revs);
3857                                 if (!c)
3858                                         break;
3859                         }
3860                 }
3861
3862                 if (revs->max_count > 0)
3863                         revs->max_count--;
3864         }
3865
3866         if (c)
3867                 c->object.flags |= SHOWN;
3868
3869         if (!revs->boundary)
3870                 return c;
3871
3872         if (!c) {
3873                 /*
3874                  * get_revision_1() runs out the commits, and
3875                  * we are done computing the boundaries.
3876                  * switch to boundary commits output mode.
3877                  */
3878                 revs->boundary = 2;
3879
3880                 /*
3881                  * Update revs->commits to contain the list of
3882                  * boundary commits.
3883                  */
3884                 create_boundary_commit_list(revs);
3885
3886                 return get_revision_internal(revs);
3887         }
3888
3889         /*
3890          * boundary commits are the commits that are parents of the
3891          * ones we got from get_revision_1() but they themselves are
3892          * not returned from get_revision_1().  Before returning
3893          * 'c', we need to mark its parents that they could be boundaries.
3894          */
3895
3896         for (l = c->parents; l; l = l->next) {
3897                 struct object *p;
3898                 p = &(l->item->object);
3899                 if (p->flags & (CHILD_SHOWN | SHOWN))
3900                         continue;
3901                 p->flags |= CHILD_SHOWN;
3902                 gc_boundary(&revs->boundary_commits);
3903                 add_object_array(p, NULL, &revs->boundary_commits);
3904         }
3905
3906         return c;
3907 }
3908
3909 struct commit *get_revision(struct rev_info *revs)
3910 {
3911         struct commit *c;
3912         struct commit_list *reversed;
3913
3914         if (revs->reverse) {
3915                 reversed = NULL;
3916                 while ((c = get_revision_internal(revs)))
3917                         commit_list_insert(c, &reversed);
3918                 revs->commits = reversed;
3919                 revs->reverse = 0;
3920                 revs->reverse_output_stage = 1;
3921         }
3922
3923         if (revs->reverse_output_stage) {
3924                 c = pop_commit(&revs->commits);
3925                 if (revs->track_linear)
3926                         revs->linear = !!(c && c->object.flags & TRACK_LINEAR);
3927                 return c;
3928         }
3929
3930         c = get_revision_internal(revs);
3931         if (c && revs->graph)
3932                 graph_update(revs->graph, c);
3933         if (!c) {
3934                 free_saved_parents(revs);
3935                 if (revs->previous_parents) {
3936                         free_commit_list(revs->previous_parents);
3937                         revs->previous_parents = NULL;
3938                 }
3939         }
3940         return c;
3941 }
3942
3943 char *get_revision_mark(const struct rev_info *revs, const struct commit *commit)
3944 {
3945         if (commit->object.flags & BOUNDARY)
3946                 return "-";
3947         else if (commit->object.flags & UNINTERESTING)
3948                 return "^";
3949         else if (commit->object.flags & PATCHSAME)
3950                 return "=";
3951         else if (!revs || revs->left_right) {
3952                 if (commit->object.flags & SYMMETRIC_LEFT)
3953                         return "<";
3954                 else
3955                         return ">";
3956         } else if (revs->graph)
3957                 return "*";
3958         else if (revs->cherry_mark)
3959                 return "+";
3960         return "";
3961 }
3962
3963 void put_revision_mark(const struct rev_info *revs, const struct commit *commit)
3964 {
3965         char *mark = get_revision_mark(revs, commit);
3966         if (!strlen(mark))
3967                 return;
3968         fputs(mark, stdout);
3969         putchar(' ');
3970 }