read-cache: add invalidate parameter to remove_marked_cache_entries
[git] / unpack-trees.c
1 #define NO_THE_INDEX_COMPATIBILITY_MACROS
2 #include "cache.h"
3 #include "argv-array.h"
4 #include "repository.h"
5 #include "config.h"
6 #include "dir.h"
7 #include "tree.h"
8 #include "tree-walk.h"
9 #include "cache-tree.h"
10 #include "unpack-trees.h"
11 #include "progress.h"
12 #include "refs.h"
13 #include "attr.h"
14 #include "split-index.h"
15 #include "dir.h"
16 #include "submodule.h"
17 #include "submodule-config.h"
18 #include "fsmonitor.h"
19 #include "object-store.h"
20 #include "fetch-object.h"
21
22 /*
23  * Error messages expected by scripts out of plumbing commands such as
24  * read-tree.  Non-scripted Porcelain is not required to use these messages
25  * and in fact are encouraged to reword them to better suit their particular
26  * situation better.  See how "git checkout" and "git merge" replaces
27  * them using setup_unpack_trees_porcelain(), for example.
28  */
29 static const char *unpack_plumbing_errors[NB_UNPACK_TREES_ERROR_TYPES] = {
30         /* ERROR_WOULD_OVERWRITE */
31         "Entry '%s' would be overwritten by merge. Cannot merge.",
32
33         /* ERROR_NOT_UPTODATE_FILE */
34         "Entry '%s' not uptodate. Cannot merge.",
35
36         /* ERROR_NOT_UPTODATE_DIR */
37         "Updating '%s' would lose untracked files in it",
38
39         /* ERROR_WOULD_LOSE_UNTRACKED_OVERWRITTEN */
40         "Untracked working tree file '%s' would be overwritten by merge.",
41
42         /* ERROR_WOULD_LOSE_UNTRACKED_REMOVED */
43         "Untracked working tree file '%s' would be removed by merge.",
44
45         /* ERROR_BIND_OVERLAP */
46         "Entry '%s' overlaps with '%s'.  Cannot bind.",
47
48         /* ERROR_SPARSE_NOT_UPTODATE_FILE */
49         "Entry '%s' not uptodate. Cannot update sparse checkout.",
50
51         /* ERROR_WOULD_LOSE_ORPHANED_OVERWRITTEN */
52         "Working tree file '%s' would be overwritten by sparse checkout update.",
53
54         /* ERROR_WOULD_LOSE_ORPHANED_REMOVED */
55         "Working tree file '%s' would be removed by sparse checkout update.",
56
57         /* ERROR_WOULD_LOSE_SUBMODULE */
58         "Submodule '%s' cannot checkout new HEAD.",
59 };
60
61 #define ERRORMSG(o,type) \
62         ( ((o) && (o)->msgs[(type)]) \
63           ? ((o)->msgs[(type)])      \
64           : (unpack_plumbing_errors[(type)]) )
65
66 static const char *super_prefixed(const char *path)
67 {
68         /*
69          * It is necessary and sufficient to have two static buffers
70          * here, as the return value of this function is fed to
71          * error() using the unpack_*_errors[] templates we see above.
72          */
73         static struct strbuf buf[2] = {STRBUF_INIT, STRBUF_INIT};
74         static int super_prefix_len = -1;
75         static unsigned idx = ARRAY_SIZE(buf) - 1;
76
77         if (super_prefix_len < 0) {
78                 const char *super_prefix = get_super_prefix();
79                 if (!super_prefix) {
80                         super_prefix_len = 0;
81                 } else {
82                         int i;
83                         for (i = 0; i < ARRAY_SIZE(buf); i++)
84                                 strbuf_addstr(&buf[i], super_prefix);
85                         super_prefix_len = buf[0].len;
86                 }
87         }
88
89         if (!super_prefix_len)
90                 return path;
91
92         if (++idx >= ARRAY_SIZE(buf))
93                 idx = 0;
94
95         strbuf_setlen(&buf[idx], super_prefix_len);
96         strbuf_addstr(&buf[idx], path);
97
98         return buf[idx].buf;
99 }
100
101 void setup_unpack_trees_porcelain(struct unpack_trees_options *opts,
102                                   const char *cmd)
103 {
104         int i;
105         const char **msgs = opts->msgs;
106         const char *msg;
107
108         argv_array_init(&opts->msgs_to_free);
109
110         if (!strcmp(cmd, "checkout"))
111                 msg = advice_commit_before_merge
112                       ? _("Your local changes to the following files would be overwritten by checkout:\n%%s"
113                           "Please commit your changes or stash them before you switch branches.")
114                       : _("Your local changes to the following files would be overwritten by checkout:\n%%s");
115         else if (!strcmp(cmd, "merge"))
116                 msg = advice_commit_before_merge
117                       ? _("Your local changes to the following files would be overwritten by merge:\n%%s"
118                           "Please commit your changes or stash them before you merge.")
119                       : _("Your local changes to the following files would be overwritten by merge:\n%%s");
120         else
121                 msg = advice_commit_before_merge
122                       ? _("Your local changes to the following files would be overwritten by %s:\n%%s"
123                           "Please commit your changes or stash them before you %s.")
124                       : _("Your local changes to the following files would be overwritten by %s:\n%%s");
125         msgs[ERROR_WOULD_OVERWRITE] = msgs[ERROR_NOT_UPTODATE_FILE] =
126                 argv_array_pushf(&opts->msgs_to_free, msg, cmd, cmd);
127
128         msgs[ERROR_NOT_UPTODATE_DIR] =
129                 _("Updating the following directories would lose untracked files in them:\n%s");
130
131         if (!strcmp(cmd, "checkout"))
132                 msg = advice_commit_before_merge
133                       ? _("The following untracked working tree files would be removed by checkout:\n%%s"
134                           "Please move or remove them before you switch branches.")
135                       : _("The following untracked working tree files would be removed by checkout:\n%%s");
136         else if (!strcmp(cmd, "merge"))
137                 msg = advice_commit_before_merge
138                       ? _("The following untracked working tree files would be removed by merge:\n%%s"
139                           "Please move or remove them before you merge.")
140                       : _("The following untracked working tree files would be removed by merge:\n%%s");
141         else
142                 msg = advice_commit_before_merge
143                       ? _("The following untracked working tree files would be removed by %s:\n%%s"
144                           "Please move or remove them before you %s.")
145                       : _("The following untracked working tree files would be removed by %s:\n%%s");
146         msgs[ERROR_WOULD_LOSE_UNTRACKED_REMOVED] =
147                 argv_array_pushf(&opts->msgs_to_free, msg, cmd, cmd);
148
149         if (!strcmp(cmd, "checkout"))
150                 msg = advice_commit_before_merge
151                       ? _("The following untracked working tree files would be overwritten by checkout:\n%%s"
152                           "Please move or remove them before you switch branches.")
153                       : _("The following untracked working tree files would be overwritten by checkout:\n%%s");
154         else if (!strcmp(cmd, "merge"))
155                 msg = advice_commit_before_merge
156                       ? _("The following untracked working tree files would be overwritten by merge:\n%%s"
157                           "Please move or remove them before you merge.")
158                       : _("The following untracked working tree files would be overwritten by merge:\n%%s");
159         else
160                 msg = advice_commit_before_merge
161                       ? _("The following untracked working tree files would be overwritten by %s:\n%%s"
162                           "Please move or remove them before you %s.")
163                       : _("The following untracked working tree files would be overwritten by %s:\n%%s");
164         msgs[ERROR_WOULD_LOSE_UNTRACKED_OVERWRITTEN] =
165                 argv_array_pushf(&opts->msgs_to_free, msg, cmd, cmd);
166
167         /*
168          * Special case: ERROR_BIND_OVERLAP refers to a pair of paths, we
169          * cannot easily display it as a list.
170          */
171         msgs[ERROR_BIND_OVERLAP] = _("Entry '%s' overlaps with '%s'.  Cannot bind.");
172
173         msgs[ERROR_SPARSE_NOT_UPTODATE_FILE] =
174                 _("Cannot update sparse checkout: the following entries are not up to date:\n%s");
175         msgs[ERROR_WOULD_LOSE_ORPHANED_OVERWRITTEN] =
176                 _("The following working tree files would be overwritten by sparse checkout update:\n%s");
177         msgs[ERROR_WOULD_LOSE_ORPHANED_REMOVED] =
178                 _("The following working tree files would be removed by sparse checkout update:\n%s");
179         msgs[ERROR_WOULD_LOSE_SUBMODULE] =
180                 _("Cannot update submodule:\n%s");
181
182         opts->show_all_errors = 1;
183         /* rejected paths may not have a static buffer */
184         for (i = 0; i < ARRAY_SIZE(opts->unpack_rejects); i++)
185                 opts->unpack_rejects[i].strdup_strings = 1;
186 }
187
188 void clear_unpack_trees_porcelain(struct unpack_trees_options *opts)
189 {
190         argv_array_clear(&opts->msgs_to_free);
191         memset(opts->msgs, 0, sizeof(opts->msgs));
192 }
193
194 static int do_add_entry(struct unpack_trees_options *o, struct cache_entry *ce,
195                          unsigned int set, unsigned int clear)
196 {
197         clear |= CE_HASHED;
198
199         if (set & CE_REMOVE)
200                 set |= CE_WT_REMOVE;
201
202         ce->ce_flags = (ce->ce_flags & ~clear) | set;
203         return add_index_entry(&o->result, ce,
204                                ADD_CACHE_OK_TO_ADD | ADD_CACHE_OK_TO_REPLACE);
205 }
206
207 static void add_entry(struct unpack_trees_options *o,
208                       const struct cache_entry *ce,
209                       unsigned int set, unsigned int clear)
210 {
211         do_add_entry(o, dup_cache_entry(ce, &o->result), set, clear);
212 }
213
214 /*
215  * add error messages on path <path>
216  * corresponding to the type <e> with the message <msg>
217  * indicating if it should be display in porcelain or not
218  */
219 static int add_rejected_path(struct unpack_trees_options *o,
220                              enum unpack_trees_error_types e,
221                              const char *path)
222 {
223         if (!o->show_all_errors)
224                 return error(ERRORMSG(o, e), super_prefixed(path));
225
226         /*
227          * Otherwise, insert in a list for future display by
228          * display_error_msgs()
229          */
230         string_list_append(&o->unpack_rejects[e], path);
231         return -1;
232 }
233
234 /*
235  * display all the error messages stored in a nice way
236  */
237 static void display_error_msgs(struct unpack_trees_options *o)
238 {
239         int e, i;
240         int something_displayed = 0;
241         for (e = 0; e < NB_UNPACK_TREES_ERROR_TYPES; e++) {
242                 struct string_list *rejects = &o->unpack_rejects[e];
243                 if (rejects->nr > 0) {
244                         struct strbuf path = STRBUF_INIT;
245                         something_displayed = 1;
246                         for (i = 0; i < rejects->nr; i++)
247                                 strbuf_addf(&path, "\t%s\n", rejects->items[i].string);
248                         error(ERRORMSG(o, e), super_prefixed(path.buf));
249                         strbuf_release(&path);
250                 }
251                 string_list_clear(rejects, 0);
252         }
253         if (something_displayed)
254                 fprintf(stderr, _("Aborting\n"));
255 }
256
257 static int check_submodule_move_head(const struct cache_entry *ce,
258                                      const char *old_id,
259                                      const char *new_id,
260                                      struct unpack_trees_options *o)
261 {
262         unsigned flags = SUBMODULE_MOVE_HEAD_DRY_RUN;
263         const struct submodule *sub = submodule_from_ce(ce);
264
265         if (!sub)
266                 return 0;
267
268         if (o->reset)
269                 flags |= SUBMODULE_MOVE_HEAD_FORCE;
270
271         if (submodule_move_head(ce->name, old_id, new_id, flags))
272                 return o->gently ? -1 :
273                                    add_rejected_path(o, ERROR_WOULD_LOSE_SUBMODULE, ce->name);
274         return 0;
275 }
276
277 /*
278  * Preform the loading of the repository's gitmodules file.  This function is
279  * used by 'check_update()' to perform loading of the gitmodules file in two
280  * differnt situations:
281  * (1) before removing entries from the working tree if the gitmodules file has
282  *     been marked for removal.  This situation is specified by 'state' == NULL.
283  * (2) before checking out entries to the working tree if the gitmodules file
284  *     has been marked for update.  This situation is specified by 'state' != NULL.
285  */
286 static void load_gitmodules_file(struct index_state *index,
287                                  struct checkout *state)
288 {
289         int pos = index_name_pos(index, GITMODULES_FILE, strlen(GITMODULES_FILE));
290
291         if (pos >= 0) {
292                 struct cache_entry *ce = index->cache[pos];
293                 if (!state && ce->ce_flags & CE_WT_REMOVE) {
294                         repo_read_gitmodules(the_repository);
295                 } else if (state && (ce->ce_flags & CE_UPDATE)) {
296                         submodule_free(the_repository);
297                         checkout_entry(ce, state, NULL);
298                         repo_read_gitmodules(the_repository);
299                 }
300         }
301 }
302
303 static struct progress *get_progress(struct unpack_trees_options *o)
304 {
305         unsigned cnt = 0, total = 0;
306         struct index_state *index = &o->result;
307
308         if (!o->update || !o->verbose_update)
309                 return NULL;
310
311         for (; cnt < index->cache_nr; cnt++) {
312                 const struct cache_entry *ce = index->cache[cnt];
313                 if (ce->ce_flags & (CE_UPDATE | CE_WT_REMOVE))
314                         total++;
315         }
316
317         return start_delayed_progress(_("Checking out files"), total);
318 }
319
320 static void setup_collided_checkout_detection(struct checkout *state,
321                                               struct index_state *index)
322 {
323         int i;
324
325         state->clone = 1;
326         for (i = 0; i < index->cache_nr; i++)
327                 index->cache[i]->ce_flags &= ~CE_MATCHED;
328 }
329
330 static void report_collided_checkout(struct index_state *index)
331 {
332         struct string_list list = STRING_LIST_INIT_NODUP;
333         int i;
334
335         for (i = 0; i < index->cache_nr; i++) {
336                 struct cache_entry *ce = index->cache[i];
337
338                 if (!(ce->ce_flags & CE_MATCHED))
339                         continue;
340
341                 string_list_append(&list, ce->name);
342                 ce->ce_flags &= ~CE_MATCHED;
343         }
344
345         list.cmp = fspathcmp;
346         string_list_sort(&list);
347
348         if (list.nr) {
349                 warning(_("the following paths have collided (e.g. case-sensitive paths\n"
350                           "on a case-insensitive filesystem) and only one from the same\n"
351                           "colliding group is in the working tree:\n"));
352
353                 for (i = 0; i < list.nr; i++)
354                         fprintf(stderr, "  '%s'\n", list.items[i].string);
355         }
356
357         string_list_clear(&list, 0);
358 }
359
360 static int check_updates(struct unpack_trees_options *o)
361 {
362         unsigned cnt = 0;
363         int errs = 0;
364         struct progress *progress;
365         struct index_state *index = &o->result;
366         struct checkout state = CHECKOUT_INIT;
367         int i;
368
369         trace_performance_enter();
370         state.force = 1;
371         state.quiet = 1;
372         state.refresh_cache = 1;
373         state.istate = index;
374
375         if (o->clone)
376                 setup_collided_checkout_detection(&state, index);
377
378         progress = get_progress(o);
379
380         if (o->update)
381                 git_attr_set_direction(GIT_ATTR_CHECKOUT);
382
383         if (should_update_submodules() && o->update && !o->dry_run)
384                 load_gitmodules_file(index, NULL);
385
386         for (i = 0; i < index->cache_nr; i++) {
387                 const struct cache_entry *ce = index->cache[i];
388
389                 if (ce->ce_flags & CE_WT_REMOVE) {
390                         display_progress(progress, ++cnt);
391                         if (o->update && !o->dry_run)
392                                 unlink_entry(ce);
393                 }
394         }
395         remove_marked_cache_entries(index, 0);
396         remove_scheduled_dirs();
397
398         if (should_update_submodules() && o->update && !o->dry_run)
399                 load_gitmodules_file(index, &state);
400
401         enable_delayed_checkout(&state);
402         if (repository_format_partial_clone && o->update && !o->dry_run) {
403                 /*
404                  * Prefetch the objects that are to be checked out in the loop
405                  * below.
406                  */
407                 struct oid_array to_fetch = OID_ARRAY_INIT;
408                 int fetch_if_missing_store = fetch_if_missing;
409                 fetch_if_missing = 0;
410                 for (i = 0; i < index->cache_nr; i++) {
411                         struct cache_entry *ce = index->cache[i];
412                         if ((ce->ce_flags & CE_UPDATE) &&
413                             !S_ISGITLINK(ce->ce_mode)) {
414                                 if (!has_object_file(&ce->oid))
415                                         oid_array_append(&to_fetch, &ce->oid);
416                         }
417                 }
418                 if (to_fetch.nr)
419                         fetch_objects(repository_format_partial_clone,
420                                       to_fetch.oid, to_fetch.nr);
421                 fetch_if_missing = fetch_if_missing_store;
422                 oid_array_clear(&to_fetch);
423         }
424         for (i = 0; i < index->cache_nr; i++) {
425                 struct cache_entry *ce = index->cache[i];
426
427                 if (ce->ce_flags & CE_UPDATE) {
428                         if (ce->ce_flags & CE_WT_REMOVE)
429                                 BUG("both update and delete flags are set on %s",
430                                     ce->name);
431                         display_progress(progress, ++cnt);
432                         ce->ce_flags &= ~CE_UPDATE;
433                         if (o->update && !o->dry_run) {
434                                 errs |= checkout_entry(ce, &state, NULL);
435                         }
436                 }
437         }
438         stop_progress(&progress);
439         errs |= finish_delayed_checkout(&state);
440         if (o->update)
441                 git_attr_set_direction(GIT_ATTR_CHECKIN);
442
443         if (o->clone)
444                 report_collided_checkout(index);
445
446         trace_performance_leave("check_updates");
447         return errs != 0;
448 }
449
450 static int verify_uptodate_sparse(const struct cache_entry *ce,
451                                   struct unpack_trees_options *o);
452 static int verify_absent_sparse(const struct cache_entry *ce,
453                                 enum unpack_trees_error_types,
454                                 struct unpack_trees_options *o);
455
456 static int apply_sparse_checkout(struct index_state *istate,
457                                  struct cache_entry *ce,
458                                  struct unpack_trees_options *o)
459 {
460         int was_skip_worktree = ce_skip_worktree(ce);
461
462         if (ce->ce_flags & CE_NEW_SKIP_WORKTREE)
463                 ce->ce_flags |= CE_SKIP_WORKTREE;
464         else
465                 ce->ce_flags &= ~CE_SKIP_WORKTREE;
466         if (was_skip_worktree != ce_skip_worktree(ce)) {
467                 ce->ce_flags |= CE_UPDATE_IN_BASE;
468                 mark_fsmonitor_invalid(istate, ce);
469                 istate->cache_changed |= CE_ENTRY_CHANGED;
470         }
471
472         /*
473          * if (!was_skip_worktree && !ce_skip_worktree()) {
474          *      This is perfectly normal. Move on;
475          * }
476          */
477
478         /*
479          * Merge strategies may set CE_UPDATE|CE_REMOVE outside checkout
480          * area as a result of ce_skip_worktree() shortcuts in
481          * verify_absent() and verify_uptodate().
482          * Make sure they don't modify worktree if they are already
483          * outside checkout area
484          */
485         if (was_skip_worktree && ce_skip_worktree(ce)) {
486                 ce->ce_flags &= ~CE_UPDATE;
487
488                 /*
489                  * By default, when CE_REMOVE is on, CE_WT_REMOVE is also
490                  * on to get that file removed from both index and worktree.
491                  * If that file is already outside worktree area, don't
492                  * bother remove it.
493                  */
494                 if (ce->ce_flags & CE_REMOVE)
495                         ce->ce_flags &= ~CE_WT_REMOVE;
496         }
497
498         if (!was_skip_worktree && ce_skip_worktree(ce)) {
499                 /*
500                  * If CE_UPDATE is set, verify_uptodate() must be called already
501                  * also stat info may have lost after merged_entry() so calling
502                  * verify_uptodate() again may fail
503                  */
504                 if (!(ce->ce_flags & CE_UPDATE) && verify_uptodate_sparse(ce, o))
505                         return -1;
506                 ce->ce_flags |= CE_WT_REMOVE;
507                 ce->ce_flags &= ~CE_UPDATE;
508         }
509         if (was_skip_worktree && !ce_skip_worktree(ce)) {
510                 if (verify_absent_sparse(ce, ERROR_WOULD_LOSE_UNTRACKED_OVERWRITTEN, o))
511                         return -1;
512                 ce->ce_flags |= CE_UPDATE;
513         }
514         return 0;
515 }
516
517 static inline int call_unpack_fn(const struct cache_entry * const *src,
518                                  struct unpack_trees_options *o)
519 {
520         int ret = o->fn(src, o);
521         if (ret > 0)
522                 ret = 0;
523         return ret;
524 }
525
526 static void mark_ce_used(struct cache_entry *ce, struct unpack_trees_options *o)
527 {
528         ce->ce_flags |= CE_UNPACKED;
529
530         if (o->cache_bottom < o->src_index->cache_nr &&
531             o->src_index->cache[o->cache_bottom] == ce) {
532                 int bottom = o->cache_bottom;
533                 while (bottom < o->src_index->cache_nr &&
534                        o->src_index->cache[bottom]->ce_flags & CE_UNPACKED)
535                         bottom++;
536                 o->cache_bottom = bottom;
537         }
538 }
539
540 static void mark_all_ce_unused(struct index_state *index)
541 {
542         int i;
543         for (i = 0; i < index->cache_nr; i++)
544                 index->cache[i]->ce_flags &= ~(CE_UNPACKED | CE_ADDED | CE_NEW_SKIP_WORKTREE);
545 }
546
547 static int locate_in_src_index(const struct cache_entry *ce,
548                                struct unpack_trees_options *o)
549 {
550         struct index_state *index = o->src_index;
551         int len = ce_namelen(ce);
552         int pos = index_name_pos(index, ce->name, len);
553         if (pos < 0)
554                 pos = -1 - pos;
555         return pos;
556 }
557
558 /*
559  * We call unpack_index_entry() with an unmerged cache entry
560  * only in diff-index, and it wants a single callback.  Skip
561  * the other unmerged entry with the same name.
562  */
563 static void mark_ce_used_same_name(struct cache_entry *ce,
564                                    struct unpack_trees_options *o)
565 {
566         struct index_state *index = o->src_index;
567         int len = ce_namelen(ce);
568         int pos;
569
570         for (pos = locate_in_src_index(ce, o); pos < index->cache_nr; pos++) {
571                 struct cache_entry *next = index->cache[pos];
572                 if (len != ce_namelen(next) ||
573                     memcmp(ce->name, next->name, len))
574                         break;
575                 mark_ce_used(next, o);
576         }
577 }
578
579 static struct cache_entry *next_cache_entry(struct unpack_trees_options *o)
580 {
581         const struct index_state *index = o->src_index;
582         int pos = o->cache_bottom;
583
584         while (pos < index->cache_nr) {
585                 struct cache_entry *ce = index->cache[pos];
586                 if (!(ce->ce_flags & CE_UNPACKED))
587                         return ce;
588                 pos++;
589         }
590         return NULL;
591 }
592
593 static void add_same_unmerged(const struct cache_entry *ce,
594                               struct unpack_trees_options *o)
595 {
596         struct index_state *index = o->src_index;
597         int len = ce_namelen(ce);
598         int pos = index_name_pos(index, ce->name, len);
599
600         if (0 <= pos)
601                 die("programming error in a caller of mark_ce_used_same_name");
602         for (pos = -pos - 1; pos < index->cache_nr; pos++) {
603                 struct cache_entry *next = index->cache[pos];
604                 if (len != ce_namelen(next) ||
605                     memcmp(ce->name, next->name, len))
606                         break;
607                 add_entry(o, next, 0, 0);
608                 mark_ce_used(next, o);
609         }
610 }
611
612 static int unpack_index_entry(struct cache_entry *ce,
613                               struct unpack_trees_options *o)
614 {
615         const struct cache_entry *src[MAX_UNPACK_TREES + 1] = { NULL, };
616         int ret;
617
618         src[0] = ce;
619
620         mark_ce_used(ce, o);
621         if (ce_stage(ce)) {
622                 if (o->skip_unmerged) {
623                         add_entry(o, ce, 0, 0);
624                         return 0;
625                 }
626         }
627         ret = call_unpack_fn(src, o);
628         if (ce_stage(ce))
629                 mark_ce_used_same_name(ce, o);
630         return ret;
631 }
632
633 static int find_cache_pos(struct traverse_info *, const struct name_entry *);
634
635 static void restore_cache_bottom(struct traverse_info *info, int bottom)
636 {
637         struct unpack_trees_options *o = info->data;
638
639         if (o->diff_index_cached)
640                 return;
641         o->cache_bottom = bottom;
642 }
643
644 static int switch_cache_bottom(struct traverse_info *info)
645 {
646         struct unpack_trees_options *o = info->data;
647         int ret, pos;
648
649         if (o->diff_index_cached)
650                 return 0;
651         ret = o->cache_bottom;
652         pos = find_cache_pos(info->prev, &info->name);
653
654         if (pos < -1)
655                 o->cache_bottom = -2 - pos;
656         else if (pos < 0)
657                 o->cache_bottom = o->src_index->cache_nr;
658         return ret;
659 }
660
661 static inline int are_same_oid(struct name_entry *name_j, struct name_entry *name_k)
662 {
663         return name_j->oid && name_k->oid && oideq(name_j->oid, name_k->oid);
664 }
665
666 static int all_trees_same_as_cache_tree(int n, unsigned long dirmask,
667                                         struct name_entry *names,
668                                         struct traverse_info *info)
669 {
670         struct unpack_trees_options *o = info->data;
671         int i;
672
673         if (!o->merge || dirmask != ((1 << n) - 1))
674                 return 0;
675
676         for (i = 1; i < n; i++)
677                 if (!are_same_oid(names, names + i))
678                         return 0;
679
680         return cache_tree_matches_traversal(o->src_index->cache_tree, names, info);
681 }
682
683 static int index_pos_by_traverse_info(struct name_entry *names,
684                                       struct traverse_info *info)
685 {
686         struct unpack_trees_options *o = info->data;
687         int len = traverse_path_len(info, names);
688         char *name = xmalloc(len + 1 /* slash */ + 1 /* NUL */);
689         int pos;
690
691         make_traverse_path(name, info, names);
692         name[len++] = '/';
693         name[len] = '\0';
694         pos = index_name_pos(o->src_index, name, len);
695         if (pos >= 0)
696                 BUG("This is a directory and should not exist in index");
697         pos = -pos - 1;
698         if (!starts_with(o->src_index->cache[pos]->name, name) ||
699             (pos > 0 && starts_with(o->src_index->cache[pos-1]->name, name)))
700                 BUG("pos must point at the first entry in this directory");
701         free(name);
702         return pos;
703 }
704
705 /*
706  * Fast path if we detect that all trees are the same as cache-tree at this
707  * path. We'll walk these trees in an iterative loop using cache-tree/index
708  * instead of ODB since we already know what these trees contain.
709  */
710 static int traverse_by_cache_tree(int pos, int nr_entries, int nr_names,
711                                   struct name_entry *names,
712                                   struct traverse_info *info)
713 {
714         struct cache_entry *src[MAX_UNPACK_TREES + 1] = { NULL, };
715         struct unpack_trees_options *o = info->data;
716         struct cache_entry *tree_ce = NULL;
717         int ce_len = 0;
718         int i, d;
719
720         if (!o->merge)
721                 BUG("We need cache-tree to do this optimization");
722
723         /*
724          * Do what unpack_callback() and unpack_nondirectories() normally
725          * do. But we walk all paths in an iterative loop instead.
726          *
727          * D/F conflicts and higher stage entries are not a concern
728          * because cache-tree would be invalidated and we would never
729          * get here in the first place.
730          */
731         for (i = 0; i < nr_entries; i++) {
732                 int new_ce_len, len, rc;
733
734                 src[0] = o->src_index->cache[pos + i];
735
736                 len = ce_namelen(src[0]);
737                 new_ce_len = cache_entry_size(len);
738
739                 if (new_ce_len > ce_len) {
740                         new_ce_len <<= 1;
741                         tree_ce = xrealloc(tree_ce, new_ce_len);
742                         memset(tree_ce, 0, new_ce_len);
743                         ce_len = new_ce_len;
744
745                         tree_ce->ce_flags = create_ce_flags(0);
746
747                         for (d = 1; d <= nr_names; d++)
748                                 src[d] = tree_ce;
749                 }
750
751                 tree_ce->ce_mode = src[0]->ce_mode;
752                 tree_ce->ce_namelen = len;
753                 oidcpy(&tree_ce->oid, &src[0]->oid);
754                 memcpy(tree_ce->name, src[0]->name, len + 1);
755
756                 rc = call_unpack_fn((const struct cache_entry * const *)src, o);
757                 if (rc < 0) {
758                         free(tree_ce);
759                         return rc;
760                 }
761
762                 mark_ce_used(src[0], o);
763         }
764         free(tree_ce);
765         if (o->debug_unpack)
766                 printf("Unpacked %d entries from %s to %s using cache-tree\n",
767                        nr_entries,
768                        o->src_index->cache[pos]->name,
769                        o->src_index->cache[pos + nr_entries - 1]->name);
770         return 0;
771 }
772
773 static int traverse_trees_recursive(int n, unsigned long dirmask,
774                                     unsigned long df_conflicts,
775                                     struct name_entry *names,
776                                     struct traverse_info *info)
777 {
778         int i, ret, bottom;
779         int nr_buf = 0;
780         struct tree_desc t[MAX_UNPACK_TREES];
781         void *buf[MAX_UNPACK_TREES];
782         struct traverse_info newinfo;
783         struct name_entry *p;
784         int nr_entries;
785
786         nr_entries = all_trees_same_as_cache_tree(n, dirmask, names, info);
787         if (nr_entries > 0) {
788                 struct unpack_trees_options *o = info->data;
789                 int pos = index_pos_by_traverse_info(names, info);
790
791                 if (!o->merge || df_conflicts)
792                         BUG("Wrong condition to get here buddy");
793
794                 /*
795                  * All entries up to 'pos' must have been processed
796                  * (i.e. marked CE_UNPACKED) at this point. But to be safe,
797                  * save and restore cache_bottom anyway to not miss
798                  * unprocessed entries before 'pos'.
799                  */
800                 bottom = o->cache_bottom;
801                 ret = traverse_by_cache_tree(pos, nr_entries, n, names, info);
802                 o->cache_bottom = bottom;
803                 return ret;
804         }
805
806         p = names;
807         while (!p->mode)
808                 p++;
809
810         newinfo = *info;
811         newinfo.prev = info;
812         newinfo.pathspec = info->pathspec;
813         newinfo.name = *p;
814         newinfo.pathlen += tree_entry_len(p) + 1;
815         newinfo.df_conflicts |= df_conflicts;
816
817         /*
818          * Fetch the tree from the ODB for each peer directory in the
819          * n commits.
820          *
821          * For 2- and 3-way traversals, we try to avoid hitting the
822          * ODB twice for the same OID.  This should yield a nice speed
823          * up in checkouts and merges when the commits are similar.
824          *
825          * We don't bother doing the full O(n^2) search for larger n,
826          * because wider traversals don't happen that often and we
827          * avoid the search setup.
828          *
829          * When 2 peer OIDs are the same, we just copy the tree
830          * descriptor data.  This implicitly borrows the buffer
831          * data from the earlier cell.
832          */
833         for (i = 0; i < n; i++, dirmask >>= 1) {
834                 if (i > 0 && are_same_oid(&names[i], &names[i - 1]))
835                         t[i] = t[i - 1];
836                 else if (i > 1 && are_same_oid(&names[i], &names[i - 2]))
837                         t[i] = t[i - 2];
838                 else {
839                         const struct object_id *oid = NULL;
840                         if (dirmask & 1)
841                                 oid = names[i].oid;
842                         buf[nr_buf++] = fill_tree_descriptor(t + i, oid);
843                 }
844         }
845
846         bottom = switch_cache_bottom(&newinfo);
847         ret = traverse_trees(n, t, &newinfo);
848         restore_cache_bottom(&newinfo, bottom);
849
850         for (i = 0; i < nr_buf; i++)
851                 free(buf[i]);
852
853         return ret;
854 }
855
856 /*
857  * Compare the traverse-path to the cache entry without actually
858  * having to generate the textual representation of the traverse
859  * path.
860  *
861  * NOTE! This *only* compares up to the size of the traverse path
862  * itself - the caller needs to do the final check for the cache
863  * entry having more data at the end!
864  */
865 static int do_compare_entry_piecewise(const struct cache_entry *ce, const struct traverse_info *info, const struct name_entry *n)
866 {
867         int len, pathlen, ce_len;
868         const char *ce_name;
869
870         if (info->prev) {
871                 int cmp = do_compare_entry_piecewise(ce, info->prev,
872                                                      &info->name);
873                 if (cmp)
874                         return cmp;
875         }
876         pathlen = info->pathlen;
877         ce_len = ce_namelen(ce);
878
879         /* If ce_len < pathlen then we must have previously hit "name == directory" entry */
880         if (ce_len < pathlen)
881                 return -1;
882
883         ce_len -= pathlen;
884         ce_name = ce->name + pathlen;
885
886         len = tree_entry_len(n);
887         return df_name_compare(ce_name, ce_len, S_IFREG, n->path, len, n->mode);
888 }
889
890 static int do_compare_entry(const struct cache_entry *ce,
891                             const struct traverse_info *info,
892                             const struct name_entry *n)
893 {
894         int len, pathlen, ce_len;
895         const char *ce_name;
896         int cmp;
897
898         /*
899          * If we have not precomputed the traverse path, it is quicker
900          * to avoid doing so.  But if we have precomputed it,
901          * it is quicker to use the precomputed version.
902          */
903         if (!info->traverse_path)
904                 return do_compare_entry_piecewise(ce, info, n);
905
906         cmp = strncmp(ce->name, info->traverse_path, info->pathlen);
907         if (cmp)
908                 return cmp;
909
910         pathlen = info->pathlen;
911         ce_len = ce_namelen(ce);
912
913         if (ce_len < pathlen)
914                 return -1;
915
916         ce_len -= pathlen;
917         ce_name = ce->name + pathlen;
918
919         len = tree_entry_len(n);
920         return df_name_compare(ce_name, ce_len, S_IFREG, n->path, len, n->mode);
921 }
922
923 static int compare_entry(const struct cache_entry *ce, const struct traverse_info *info, const struct name_entry *n)
924 {
925         int cmp = do_compare_entry(ce, info, n);
926         if (cmp)
927                 return cmp;
928
929         /*
930          * Even if the beginning compared identically, the ce should
931          * compare as bigger than a directory leading up to it!
932          */
933         return ce_namelen(ce) > traverse_path_len(info, n);
934 }
935
936 static int ce_in_traverse_path(const struct cache_entry *ce,
937                                const struct traverse_info *info)
938 {
939         if (!info->prev)
940                 return 1;
941         if (do_compare_entry(ce, info->prev, &info->name))
942                 return 0;
943         /*
944          * If ce (blob) is the same name as the path (which is a tree
945          * we will be descending into), it won't be inside it.
946          */
947         return (info->pathlen < ce_namelen(ce));
948 }
949
950 static struct cache_entry *create_ce_entry(const struct traverse_info *info,
951         const struct name_entry *n,
952         int stage,
953         struct index_state *istate,
954         int is_transient)
955 {
956         int len = traverse_path_len(info, n);
957         struct cache_entry *ce =
958                 is_transient ?
959                 make_empty_transient_cache_entry(len) :
960                 make_empty_cache_entry(istate, len);
961
962         ce->ce_mode = create_ce_mode(n->mode);
963         ce->ce_flags = create_ce_flags(stage);
964         ce->ce_namelen = len;
965         oidcpy(&ce->oid, n->oid);
966         make_traverse_path(ce->name, info, n);
967
968         return ce;
969 }
970
971 /*
972  * Note that traverse_by_cache_tree() duplicates some logic in this function
973  * without actually calling it. If you change the logic here you may need to
974  * check and change there as well.
975  */
976 static int unpack_nondirectories(int n, unsigned long mask,
977                                  unsigned long dirmask,
978                                  struct cache_entry **src,
979                                  const struct name_entry *names,
980                                  const struct traverse_info *info)
981 {
982         int i;
983         struct unpack_trees_options *o = info->data;
984         unsigned long conflicts = info->df_conflicts | dirmask;
985
986         /* Do we have *only* directories? Nothing to do */
987         if (mask == dirmask && !src[0])
988                 return 0;
989
990         /*
991          * Ok, we've filled in up to any potential index entry in src[0],
992          * now do the rest.
993          */
994         for (i = 0; i < n; i++) {
995                 int stage;
996                 unsigned int bit = 1ul << i;
997                 if (conflicts & bit) {
998                         src[i + o->merge] = o->df_conflict_entry;
999                         continue;
1000                 }
1001                 if (!(mask & bit))
1002                         continue;
1003                 if (!o->merge)
1004                         stage = 0;
1005                 else if (i + 1 < o->head_idx)
1006                         stage = 1;
1007                 else if (i + 1 > o->head_idx)
1008                         stage = 3;
1009                 else
1010                         stage = 2;
1011
1012                 /*
1013                  * If the merge bit is set, then the cache entries are
1014                  * discarded in the following block.  In this case,
1015                  * construct "transient" cache_entries, as they are
1016                  * not stored in the index.  otherwise construct the
1017                  * cache entry from the index aware logic.
1018                  */
1019                 src[i + o->merge] = create_ce_entry(info, names + i, stage, &o->result, o->merge);
1020         }
1021
1022         if (o->merge) {
1023                 int rc = call_unpack_fn((const struct cache_entry * const *)src,
1024                                         o);
1025                 for (i = 0; i < n; i++) {
1026                         struct cache_entry *ce = src[i + o->merge];
1027                         if (ce != o->df_conflict_entry)
1028                                 discard_cache_entry(ce);
1029                 }
1030                 return rc;
1031         }
1032
1033         for (i = 0; i < n; i++)
1034                 if (src[i] && src[i] != o->df_conflict_entry)
1035                         if (do_add_entry(o, src[i], 0, 0))
1036                                 return -1;
1037
1038         return 0;
1039 }
1040
1041 static int unpack_failed(struct unpack_trees_options *o, const char *message)
1042 {
1043         discard_index(&o->result);
1044         if (!o->gently && !o->exiting_early) {
1045                 if (message)
1046                         return error("%s", message);
1047                 return -1;
1048         }
1049         return -1;
1050 }
1051
1052 /*
1053  * The tree traversal is looking at name p.  If we have a matching entry,
1054  * return it.  If name p is a directory in the index, do not return
1055  * anything, as we will want to match it when the traversal descends into
1056  * the directory.
1057  */
1058 static int find_cache_pos(struct traverse_info *info,
1059                           const struct name_entry *p)
1060 {
1061         int pos;
1062         struct unpack_trees_options *o = info->data;
1063         struct index_state *index = o->src_index;
1064         int pfxlen = info->pathlen;
1065         int p_len = tree_entry_len(p);
1066
1067         for (pos = o->cache_bottom; pos < index->cache_nr; pos++) {
1068                 const struct cache_entry *ce = index->cache[pos];
1069                 const char *ce_name, *ce_slash;
1070                 int cmp, ce_len;
1071
1072                 if (ce->ce_flags & CE_UNPACKED) {
1073                         /*
1074                          * cache_bottom entry is already unpacked, so
1075                          * we can never match it; don't check it
1076                          * again.
1077                          */
1078                         if (pos == o->cache_bottom)
1079                                 ++o->cache_bottom;
1080                         continue;
1081                 }
1082                 if (!ce_in_traverse_path(ce, info)) {
1083                         /*
1084                          * Check if we can skip future cache checks
1085                          * (because we're already past all possible
1086                          * entries in the traverse path).
1087                          */
1088                         if (info->traverse_path) {
1089                                 if (strncmp(ce->name, info->traverse_path,
1090                                             info->pathlen) > 0)
1091                                         break;
1092                         }
1093                         continue;
1094                 }
1095                 ce_name = ce->name + pfxlen;
1096                 ce_slash = strchr(ce_name, '/');
1097                 if (ce_slash)
1098                         ce_len = ce_slash - ce_name;
1099                 else
1100                         ce_len = ce_namelen(ce) - pfxlen;
1101                 cmp = name_compare(p->path, p_len, ce_name, ce_len);
1102                 /*
1103                  * Exact match; if we have a directory we need to
1104                  * delay returning it.
1105                  */
1106                 if (!cmp)
1107                         return ce_slash ? -2 - pos : pos;
1108                 if (0 < cmp)
1109                         continue; /* keep looking */
1110                 /*
1111                  * ce_name sorts after p->path; could it be that we
1112                  * have files under p->path directory in the index?
1113                  * E.g.  ce_name == "t-i", and p->path == "t"; we may
1114                  * have "t/a" in the index.
1115                  */
1116                 if (p_len < ce_len && !memcmp(ce_name, p->path, p_len) &&
1117                     ce_name[p_len] < '/')
1118                         continue; /* keep looking */
1119                 break;
1120         }
1121         return -1;
1122 }
1123
1124 static struct cache_entry *find_cache_entry(struct traverse_info *info,
1125                                             const struct name_entry *p)
1126 {
1127         int pos = find_cache_pos(info, p);
1128         struct unpack_trees_options *o = info->data;
1129
1130         if (0 <= pos)
1131                 return o->src_index->cache[pos];
1132         else
1133                 return NULL;
1134 }
1135
1136 static void debug_path(struct traverse_info *info)
1137 {
1138         if (info->prev) {
1139                 debug_path(info->prev);
1140                 if (*info->prev->name.path)
1141                         putchar('/');
1142         }
1143         printf("%s", info->name.path);
1144 }
1145
1146 static void debug_name_entry(int i, struct name_entry *n)
1147 {
1148         printf("ent#%d %06o %s\n", i,
1149                n->path ? n->mode : 0,
1150                n->path ? n->path : "(missing)");
1151 }
1152
1153 static void debug_unpack_callback(int n,
1154                                   unsigned long mask,
1155                                   unsigned long dirmask,
1156                                   struct name_entry *names,
1157                                   struct traverse_info *info)
1158 {
1159         int i;
1160         printf("* unpack mask %lu, dirmask %lu, cnt %d ",
1161                mask, dirmask, n);
1162         debug_path(info);
1163         putchar('\n');
1164         for (i = 0; i < n; i++)
1165                 debug_name_entry(i, names + i);
1166 }
1167
1168 /*
1169  * Note that traverse_by_cache_tree() duplicates some logic in this function
1170  * without actually calling it. If you change the logic here you may need to
1171  * check and change there as well.
1172  */
1173 static int unpack_callback(int n, unsigned long mask, unsigned long dirmask, struct name_entry *names, struct traverse_info *info)
1174 {
1175         struct cache_entry *src[MAX_UNPACK_TREES + 1] = { NULL, };
1176         struct unpack_trees_options *o = info->data;
1177         const struct name_entry *p = names;
1178
1179         /* Find first entry with a real name (we could use "mask" too) */
1180         while (!p->mode)
1181                 p++;
1182
1183         if (o->debug_unpack)
1184                 debug_unpack_callback(n, mask, dirmask, names, info);
1185
1186         /* Are we supposed to look at the index too? */
1187         if (o->merge) {
1188                 while (1) {
1189                         int cmp;
1190                         struct cache_entry *ce;
1191
1192                         if (o->diff_index_cached)
1193                                 ce = next_cache_entry(o);
1194                         else
1195                                 ce = find_cache_entry(info, p);
1196
1197                         if (!ce)
1198                                 break;
1199                         cmp = compare_entry(ce, info, p);
1200                         if (cmp < 0) {
1201                                 if (unpack_index_entry(ce, o) < 0)
1202                                         return unpack_failed(o, NULL);
1203                                 continue;
1204                         }
1205                         if (!cmp) {
1206                                 if (ce_stage(ce)) {
1207                                         /*
1208                                          * If we skip unmerged index
1209                                          * entries, we'll skip this
1210                                          * entry *and* the tree
1211                                          * entries associated with it!
1212                                          */
1213                                         if (o->skip_unmerged) {
1214                                                 add_same_unmerged(ce, o);
1215                                                 return mask;
1216                                         }
1217                                 }
1218                                 src[0] = ce;
1219                         }
1220                         break;
1221                 }
1222         }
1223
1224         if (unpack_nondirectories(n, mask, dirmask, src, names, info) < 0)
1225                 return -1;
1226
1227         if (o->merge && src[0]) {
1228                 if (ce_stage(src[0]))
1229                         mark_ce_used_same_name(src[0], o);
1230                 else
1231                         mark_ce_used(src[0], o);
1232         }
1233
1234         /* Now handle any directories.. */
1235         if (dirmask) {
1236                 /* special case: "diff-index --cached" looking at a tree */
1237                 if (o->diff_index_cached &&
1238                     n == 1 && dirmask == 1 && S_ISDIR(names->mode)) {
1239                         int matches;
1240                         matches = cache_tree_matches_traversal(o->src_index->cache_tree,
1241                                                                names, info);
1242                         /*
1243                          * Everything under the name matches; skip the
1244                          * entire hierarchy.  diff_index_cached codepath
1245                          * special cases D/F conflicts in such a way that
1246                          * it does not do any look-ahead, so this is safe.
1247                          */
1248                         if (matches) {
1249                                 o->cache_bottom += matches;
1250                                 return mask;
1251                         }
1252                 }
1253
1254                 if (traverse_trees_recursive(n, dirmask, mask & ~dirmask,
1255                                              names, info) < 0)
1256                         return -1;
1257                 return mask;
1258         }
1259
1260         return mask;
1261 }
1262
1263 static int clear_ce_flags_1(struct index_state *istate,
1264                             struct cache_entry **cache, int nr,
1265                             struct strbuf *prefix,
1266                             int select_mask, int clear_mask,
1267                             struct exclude_list *el, int defval);
1268
1269 /* Whole directory matching */
1270 static int clear_ce_flags_dir(struct index_state *istate,
1271                               struct cache_entry **cache, int nr,
1272                               struct strbuf *prefix,
1273                               char *basename,
1274                               int select_mask, int clear_mask,
1275                               struct exclude_list *el, int defval)
1276 {
1277         struct cache_entry **cache_end;
1278         int dtype = DT_DIR;
1279         int ret = is_excluded_from_list(prefix->buf, prefix->len,
1280                                         basename, &dtype, el, istate);
1281         int rc;
1282
1283         strbuf_addch(prefix, '/');
1284
1285         /* If undecided, use matching result of parent dir in defval */
1286         if (ret < 0)
1287                 ret = defval;
1288
1289         for (cache_end = cache; cache_end != cache + nr; cache_end++) {
1290                 struct cache_entry *ce = *cache_end;
1291                 if (strncmp(ce->name, prefix->buf, prefix->len))
1292                         break;
1293         }
1294
1295         /*
1296          * TODO: check el, if there are no patterns that may conflict
1297          * with ret (iow, we know in advance the incl/excl
1298          * decision for the entire directory), clear flag here without
1299          * calling clear_ce_flags_1(). That function will call
1300          * the expensive is_excluded_from_list() on every entry.
1301          */
1302         rc = clear_ce_flags_1(istate, cache, cache_end - cache,
1303                               prefix,
1304                               select_mask, clear_mask,
1305                               el, ret);
1306         strbuf_setlen(prefix, prefix->len - 1);
1307         return rc;
1308 }
1309
1310 /*
1311  * Traverse the index, find every entry that matches according to
1312  * o->el. Do "ce_flags &= ~clear_mask" on those entries. Return the
1313  * number of traversed entries.
1314  *
1315  * If select_mask is non-zero, only entries whose ce_flags has on of
1316  * those bits enabled are traversed.
1317  *
1318  * cache        : pointer to an index entry
1319  * prefix_len   : an offset to its path
1320  *
1321  * The current path ("prefix") including the trailing '/' is
1322  *   cache[0]->name[0..(prefix_len-1)]
1323  * Top level path has prefix_len zero.
1324  */
1325 static int clear_ce_flags_1(struct index_state *istate,
1326                             struct cache_entry **cache, int nr,
1327                             struct strbuf *prefix,
1328                             int select_mask, int clear_mask,
1329                             struct exclude_list *el, int defval)
1330 {
1331         struct cache_entry **cache_end = cache + nr;
1332
1333         /*
1334          * Process all entries that have the given prefix and meet
1335          * select_mask condition
1336          */
1337         while(cache != cache_end) {
1338                 struct cache_entry *ce = *cache;
1339                 const char *name, *slash;
1340                 int len, dtype, ret;
1341
1342                 if (select_mask && !(ce->ce_flags & select_mask)) {
1343                         cache++;
1344                         continue;
1345                 }
1346
1347                 if (prefix->len && strncmp(ce->name, prefix->buf, prefix->len))
1348                         break;
1349
1350                 name = ce->name + prefix->len;
1351                 slash = strchr(name, '/');
1352
1353                 /* If it's a directory, try whole directory match first */
1354                 if (slash) {
1355                         int processed;
1356
1357                         len = slash - name;
1358                         strbuf_add(prefix, name, len);
1359
1360                         processed = clear_ce_flags_dir(istate, cache, cache_end - cache,
1361                                                        prefix,
1362                                                        prefix->buf + prefix->len - len,
1363                                                        select_mask, clear_mask,
1364                                                        el, defval);
1365
1366                         /* clear_c_f_dir eats a whole dir already? */
1367                         if (processed) {
1368                                 cache += processed;
1369                                 strbuf_setlen(prefix, prefix->len - len);
1370                                 continue;
1371                         }
1372
1373                         strbuf_addch(prefix, '/');
1374                         cache += clear_ce_flags_1(istate, cache, cache_end - cache,
1375                                                   prefix,
1376                                                   select_mask, clear_mask, el, defval);
1377                         strbuf_setlen(prefix, prefix->len - len - 1);
1378                         continue;
1379                 }
1380
1381                 /* Non-directory */
1382                 dtype = ce_to_dtype(ce);
1383                 ret = is_excluded_from_list(ce->name, ce_namelen(ce),
1384                                             name, &dtype, el, istate);
1385                 if (ret < 0)
1386                         ret = defval;
1387                 if (ret > 0)
1388                         ce->ce_flags &= ~clear_mask;
1389                 cache++;
1390         }
1391         return nr - (cache_end - cache);
1392 }
1393
1394 static int clear_ce_flags(struct index_state *istate,
1395                           int select_mask, int clear_mask,
1396                           struct exclude_list *el)
1397 {
1398         static struct strbuf prefix = STRBUF_INIT;
1399
1400         strbuf_reset(&prefix);
1401
1402         return clear_ce_flags_1(istate,
1403                                 istate->cache,
1404                                 istate->cache_nr,
1405                                 &prefix,
1406                                 select_mask, clear_mask,
1407                                 el, 0);
1408 }
1409
1410 /*
1411  * Set/Clear CE_NEW_SKIP_WORKTREE according to $GIT_DIR/info/sparse-checkout
1412  */
1413 static void mark_new_skip_worktree(struct exclude_list *el,
1414                                    struct index_state *istate,
1415                                    int select_flag, int skip_wt_flag)
1416 {
1417         int i;
1418
1419         /*
1420          * 1. Pretend the narrowest worktree: only unmerged entries
1421          * are checked out
1422          */
1423         for (i = 0; i < istate->cache_nr; i++) {
1424                 struct cache_entry *ce = istate->cache[i];
1425
1426                 if (select_flag && !(ce->ce_flags & select_flag))
1427                         continue;
1428
1429                 if (!ce_stage(ce) && !(ce->ce_flags & CE_CONFLICTED))
1430                         ce->ce_flags |= skip_wt_flag;
1431                 else
1432                         ce->ce_flags &= ~skip_wt_flag;
1433         }
1434
1435         /*
1436          * 2. Widen worktree according to sparse-checkout file.
1437          * Matched entries will have skip_wt_flag cleared (i.e. "in")
1438          */
1439         clear_ce_flags(istate, select_flag, skip_wt_flag, el);
1440 }
1441
1442 static int verify_absent(const struct cache_entry *,
1443                          enum unpack_trees_error_types,
1444                          struct unpack_trees_options *);
1445 /*
1446  * N-way merge "len" trees.  Returns 0 on success, -1 on failure to manipulate the
1447  * resulting index, -2 on failure to reflect the changes to the work tree.
1448  *
1449  * CE_ADDED, CE_UNPACKED and CE_NEW_SKIP_WORKTREE are used internally
1450  */
1451 int unpack_trees(unsigned len, struct tree_desc *t, struct unpack_trees_options *o)
1452 {
1453         int i, ret;
1454         static struct cache_entry *dfc;
1455         struct exclude_list el;
1456
1457         if (len > MAX_UNPACK_TREES)
1458                 die("unpack_trees takes at most %d trees", MAX_UNPACK_TREES);
1459
1460         trace_performance_enter();
1461         memset(&el, 0, sizeof(el));
1462         if (!core_apply_sparse_checkout || !o->update)
1463                 o->skip_sparse_checkout = 1;
1464         if (!o->skip_sparse_checkout) {
1465                 char *sparse = git_pathdup("info/sparse-checkout");
1466                 if (add_excludes_from_file_to_list(sparse, "", 0, &el, NULL) < 0)
1467                         o->skip_sparse_checkout = 1;
1468                 else
1469                         o->el = &el;
1470                 free(sparse);
1471         }
1472
1473         memset(&o->result, 0, sizeof(o->result));
1474         o->result.initialized = 1;
1475         o->result.timestamp.sec = o->src_index->timestamp.sec;
1476         o->result.timestamp.nsec = o->src_index->timestamp.nsec;
1477         o->result.version = o->src_index->version;
1478         if (!o->src_index->split_index) {
1479                 o->result.split_index = NULL;
1480         } else if (o->src_index == o->dst_index) {
1481                 /*
1482                  * o->dst_index (and thus o->src_index) will be discarded
1483                  * and overwritten with o->result at the end of this function,
1484                  * so just use src_index's split_index to avoid having to
1485                  * create a new one.
1486                  */
1487                 o->result.split_index = o->src_index->split_index;
1488                 o->result.split_index->refcount++;
1489         } else {
1490                 o->result.split_index = init_split_index(&o->result);
1491         }
1492         oidcpy(&o->result.oid, &o->src_index->oid);
1493         o->merge_size = len;
1494         mark_all_ce_unused(o->src_index);
1495
1496         /*
1497          * Sparse checkout loop #1: set NEW_SKIP_WORKTREE on existing entries
1498          */
1499         if (!o->skip_sparse_checkout)
1500                 mark_new_skip_worktree(o->el, o->src_index, 0, CE_NEW_SKIP_WORKTREE);
1501
1502         if (!dfc)
1503                 dfc = xcalloc(1, cache_entry_size(0));
1504         o->df_conflict_entry = dfc;
1505
1506         if (len) {
1507                 const char *prefix = o->prefix ? o->prefix : "";
1508                 struct traverse_info info;
1509
1510                 setup_traverse_info(&info, prefix);
1511                 info.fn = unpack_callback;
1512                 info.data = o;
1513                 info.show_all_errors = o->show_all_errors;
1514                 info.pathspec = o->pathspec;
1515
1516                 if (o->prefix) {
1517                         /*
1518                          * Unpack existing index entries that sort before the
1519                          * prefix the tree is spliced into.  Note that o->merge
1520                          * is always true in this case.
1521                          */
1522                         while (1) {
1523                                 struct cache_entry *ce = next_cache_entry(o);
1524                                 if (!ce)
1525                                         break;
1526                                 if (ce_in_traverse_path(ce, &info))
1527                                         break;
1528                                 if (unpack_index_entry(ce, o) < 0)
1529                                         goto return_failed;
1530                         }
1531                 }
1532
1533                 trace_performance_enter();
1534                 ret = traverse_trees(len, t, &info);
1535                 trace_performance_leave("traverse_trees");
1536                 if (ret < 0)
1537                         goto return_failed;
1538         }
1539
1540         /* Any left-over entries in the index? */
1541         if (o->merge) {
1542                 while (1) {
1543                         struct cache_entry *ce = next_cache_entry(o);
1544                         if (!ce)
1545                                 break;
1546                         if (unpack_index_entry(ce, o) < 0)
1547                                 goto return_failed;
1548                 }
1549         }
1550         mark_all_ce_unused(o->src_index);
1551
1552         if (o->trivial_merges_only && o->nontrivial_merge) {
1553                 ret = unpack_failed(o, "Merge requires file-level merging");
1554                 goto done;
1555         }
1556
1557         if (!o->skip_sparse_checkout) {
1558                 int empty_worktree = 1;
1559
1560                 /*
1561                  * Sparse checkout loop #2: set NEW_SKIP_WORKTREE on entries not in loop #1
1562                  * If the will have NEW_SKIP_WORKTREE, also set CE_SKIP_WORKTREE
1563                  * so apply_sparse_checkout() won't attempt to remove it from worktree
1564                  */
1565                 mark_new_skip_worktree(o->el, &o->result, CE_ADDED, CE_SKIP_WORKTREE | CE_NEW_SKIP_WORKTREE);
1566
1567                 ret = 0;
1568                 for (i = 0; i < o->result.cache_nr; i++) {
1569                         struct cache_entry *ce = o->result.cache[i];
1570
1571                         /*
1572                          * Entries marked with CE_ADDED in merged_entry() do not have
1573                          * verify_absent() check (the check is effectively disabled
1574                          * because CE_NEW_SKIP_WORKTREE is set unconditionally).
1575                          *
1576                          * Do the real check now because we have had
1577                          * correct CE_NEW_SKIP_WORKTREE
1578                          */
1579                         if (ce->ce_flags & CE_ADDED &&
1580                             verify_absent(ce, ERROR_WOULD_LOSE_UNTRACKED_OVERWRITTEN, o)) {
1581                                 if (!o->show_all_errors)
1582                                         goto return_failed;
1583                                 ret = -1;
1584                         }
1585
1586                         if (apply_sparse_checkout(&o->result, ce, o)) {
1587                                 if (!o->show_all_errors)
1588                                         goto return_failed;
1589                                 ret = -1;
1590                         }
1591                         if (!ce_skip_worktree(ce))
1592                                 empty_worktree = 0;
1593
1594                 }
1595                 if (ret < 0)
1596                         goto return_failed;
1597                 /*
1598                  * Sparse checkout is meant to narrow down checkout area
1599                  * but it does not make sense to narrow down to empty working
1600                  * tree. This is usually a mistake in sparse checkout rules.
1601                  * Do not allow users to do that.
1602                  */
1603                 if (o->result.cache_nr && empty_worktree) {
1604                         ret = unpack_failed(o, "Sparse checkout leaves no entry on working directory");
1605                         goto done;
1606                 }
1607         }
1608
1609         ret = check_updates(o) ? (-2) : 0;
1610         if (o->dst_index) {
1611                 move_index_extensions(&o->result, o->src_index);
1612                 if (!ret) {
1613                         if (git_env_bool("GIT_TEST_CHECK_CACHE_TREE", 0))
1614                                 cache_tree_verify(&o->result);
1615                         if (!o->result.cache_tree)
1616                                 o->result.cache_tree = cache_tree();
1617                         if (!cache_tree_fully_valid(o->result.cache_tree))
1618                                 cache_tree_update(&o->result,
1619                                                   WRITE_TREE_SILENT |
1620                                                   WRITE_TREE_REPAIR);
1621                 }
1622                 discard_index(o->dst_index);
1623                 *o->dst_index = o->result;
1624         } else {
1625                 discard_index(&o->result);
1626         }
1627         o->src_index = NULL;
1628
1629 done:
1630         trace_performance_leave("unpack_trees");
1631         clear_exclude_list(&el);
1632         return ret;
1633
1634 return_failed:
1635         if (o->show_all_errors)
1636                 display_error_msgs(o);
1637         mark_all_ce_unused(o->src_index);
1638         ret = unpack_failed(o, NULL);
1639         if (o->exiting_early)
1640                 ret = 0;
1641         goto done;
1642 }
1643
1644 /* Here come the merge functions */
1645
1646 static int reject_merge(const struct cache_entry *ce,
1647                         struct unpack_trees_options *o)
1648 {
1649         return o->gently ? -1 :
1650                 add_rejected_path(o, ERROR_WOULD_OVERWRITE, ce->name);
1651 }
1652
1653 static int same(const struct cache_entry *a, const struct cache_entry *b)
1654 {
1655         if (!!a != !!b)
1656                 return 0;
1657         if (!a && !b)
1658                 return 1;
1659         if ((a->ce_flags | b->ce_flags) & CE_CONFLICTED)
1660                 return 0;
1661         return a->ce_mode == b->ce_mode &&
1662                oideq(&a->oid, &b->oid);
1663 }
1664
1665
1666 /*
1667  * When a CE gets turned into an unmerged entry, we
1668  * want it to be up-to-date
1669  */
1670 static int verify_uptodate_1(const struct cache_entry *ce,
1671                              struct unpack_trees_options *o,
1672                              enum unpack_trees_error_types error_type)
1673 {
1674         struct stat st;
1675
1676         if (o->index_only)
1677                 return 0;
1678
1679         /*
1680          * CE_VALID and CE_SKIP_WORKTREE cheat, we better check again
1681          * if this entry is truly up-to-date because this file may be
1682          * overwritten.
1683          */
1684         if ((ce->ce_flags & CE_VALID) || ce_skip_worktree(ce))
1685                 ; /* keep checking */
1686         else if (o->reset || ce_uptodate(ce))
1687                 return 0;
1688
1689         if (!lstat(ce->name, &st)) {
1690                 int flags = CE_MATCH_IGNORE_VALID|CE_MATCH_IGNORE_SKIP_WORKTREE;
1691                 unsigned changed = ie_match_stat(o->src_index, ce, &st, flags);
1692
1693                 if (submodule_from_ce(ce)) {
1694                         int r = check_submodule_move_head(ce,
1695                                 "HEAD", oid_to_hex(&ce->oid), o);
1696                         if (r)
1697                                 return o->gently ? -1 :
1698                                         add_rejected_path(o, error_type, ce->name);
1699                         return 0;
1700                 }
1701
1702                 if (!changed)
1703                         return 0;
1704                 /*
1705                  * Historic default policy was to allow submodule to be out
1706                  * of sync wrt the superproject index. If the submodule was
1707                  * not considered interesting above, we don't care here.
1708                  */
1709                 if (S_ISGITLINK(ce->ce_mode))
1710                         return 0;
1711
1712                 errno = 0;
1713         }
1714         if (errno == ENOENT)
1715                 return 0;
1716         return o->gently ? -1 :
1717                 add_rejected_path(o, error_type, ce->name);
1718 }
1719
1720 int verify_uptodate(const struct cache_entry *ce,
1721                     struct unpack_trees_options *o)
1722 {
1723         if (!o->skip_sparse_checkout && (ce->ce_flags & CE_NEW_SKIP_WORKTREE))
1724                 return 0;
1725         return verify_uptodate_1(ce, o, ERROR_NOT_UPTODATE_FILE);
1726 }
1727
1728 static int verify_uptodate_sparse(const struct cache_entry *ce,
1729                                   struct unpack_trees_options *o)
1730 {
1731         return verify_uptodate_1(ce, o, ERROR_SPARSE_NOT_UPTODATE_FILE);
1732 }
1733
1734 /*
1735  * TODO: We should actually invalidate o->result, not src_index [1].
1736  * But since cache tree and untracked cache both are not copied to
1737  * o->result until unpacking is complete, we invalidate them on
1738  * src_index instead with the assumption that they will be copied to
1739  * dst_index at the end.
1740  *
1741  * [1] src_index->cache_tree is also used in unpack_callback() so if
1742  * we invalidate o->result, we need to update it to use
1743  * o->result.cache_tree as well.
1744  */
1745 static void invalidate_ce_path(const struct cache_entry *ce,
1746                                struct unpack_trees_options *o)
1747 {
1748         if (!ce)
1749                 return;
1750         cache_tree_invalidate_path(o->src_index, ce->name);
1751         untracked_cache_invalidate_path(o->src_index, ce->name, 1);
1752 }
1753
1754 /*
1755  * Check that checking out ce->sha1 in subdir ce->name is not
1756  * going to overwrite any working files.
1757  *
1758  * Currently, git does not checkout subprojects during a superproject
1759  * checkout, so it is not going to overwrite anything.
1760  */
1761 static int verify_clean_submodule(const char *old_sha1,
1762                                   const struct cache_entry *ce,
1763                                   enum unpack_trees_error_types error_type,
1764                                   struct unpack_trees_options *o)
1765 {
1766         if (!submodule_from_ce(ce))
1767                 return 0;
1768
1769         return check_submodule_move_head(ce, old_sha1,
1770                                          oid_to_hex(&ce->oid), o);
1771 }
1772
1773 static int verify_clean_subdirectory(const struct cache_entry *ce,
1774                                      enum unpack_trees_error_types error_type,
1775                                      struct unpack_trees_options *o)
1776 {
1777         /*
1778          * we are about to extract "ce->name"; we would not want to lose
1779          * anything in the existing directory there.
1780          */
1781         int namelen;
1782         int i;
1783         struct dir_struct d;
1784         char *pathbuf;
1785         int cnt = 0;
1786
1787         if (S_ISGITLINK(ce->ce_mode)) {
1788                 struct object_id oid;
1789                 int sub_head = resolve_gitlink_ref(ce->name, "HEAD", &oid);
1790                 /*
1791                  * If we are not going to update the submodule, then
1792                  * we don't care.
1793                  */
1794                 if (!sub_head && oideq(&oid, &ce->oid))
1795                         return 0;
1796                 return verify_clean_submodule(sub_head ? NULL : oid_to_hex(&oid),
1797                                               ce, error_type, o);
1798         }
1799
1800         /*
1801          * First let's make sure we do not have a local modification
1802          * in that directory.
1803          */
1804         namelen = ce_namelen(ce);
1805         for (i = locate_in_src_index(ce, o);
1806              i < o->src_index->cache_nr;
1807              i++) {
1808                 struct cache_entry *ce2 = o->src_index->cache[i];
1809                 int len = ce_namelen(ce2);
1810                 if (len < namelen ||
1811                     strncmp(ce->name, ce2->name, namelen) ||
1812                     ce2->name[namelen] != '/')
1813                         break;
1814                 /*
1815                  * ce2->name is an entry in the subdirectory to be
1816                  * removed.
1817                  */
1818                 if (!ce_stage(ce2)) {
1819                         if (verify_uptodate(ce2, o))
1820                                 return -1;
1821                         add_entry(o, ce2, CE_REMOVE, 0);
1822                         invalidate_ce_path(ce, o);
1823                         mark_ce_used(ce2, o);
1824                 }
1825                 cnt++;
1826         }
1827
1828         /*
1829          * Then we need to make sure that we do not lose a locally
1830          * present file that is not ignored.
1831          */
1832         pathbuf = xstrfmt("%.*s/", namelen, ce->name);
1833
1834         memset(&d, 0, sizeof(d));
1835         if (o->dir)
1836                 d.exclude_per_dir = o->dir->exclude_per_dir;
1837         i = read_directory(&d, o->src_index, pathbuf, namelen+1, NULL);
1838         if (i)
1839                 return o->gently ? -1 :
1840                         add_rejected_path(o, ERROR_NOT_UPTODATE_DIR, ce->name);
1841         free(pathbuf);
1842         return cnt;
1843 }
1844
1845 /*
1846  * This gets called when there was no index entry for the tree entry 'dst',
1847  * but we found a file in the working tree that 'lstat()' said was fine,
1848  * and we're on a case-insensitive filesystem.
1849  *
1850  * See if we can find a case-insensitive match in the index that also
1851  * matches the stat information, and assume it's that other file!
1852  */
1853 static int icase_exists(struct unpack_trees_options *o, const char *name, int len, struct stat *st)
1854 {
1855         const struct cache_entry *src;
1856
1857         src = index_file_exists(o->src_index, name, len, 1);
1858         return src && !ie_match_stat(o->src_index, src, st, CE_MATCH_IGNORE_VALID|CE_MATCH_IGNORE_SKIP_WORKTREE);
1859 }
1860
1861 static int check_ok_to_remove(const char *name, int len, int dtype,
1862                               const struct cache_entry *ce, struct stat *st,
1863                               enum unpack_trees_error_types error_type,
1864                               struct unpack_trees_options *o)
1865 {
1866         const struct cache_entry *result;
1867
1868         /*
1869          * It may be that the 'lstat()' succeeded even though
1870          * target 'ce' was absent, because there is an old
1871          * entry that is different only in case..
1872          *
1873          * Ignore that lstat() if it matches.
1874          */
1875         if (ignore_case && icase_exists(o, name, len, st))
1876                 return 0;
1877
1878         if (o->dir &&
1879             is_excluded(o->dir, o->src_index, name, &dtype))
1880                 /*
1881                  * ce->name is explicitly excluded, so it is Ok to
1882                  * overwrite it.
1883                  */
1884                 return 0;
1885         if (S_ISDIR(st->st_mode)) {
1886                 /*
1887                  * We are checking out path "foo" and
1888                  * found "foo/." in the working tree.
1889                  * This is tricky -- if we have modified
1890                  * files that are in "foo/" we would lose
1891                  * them.
1892                  */
1893                 if (verify_clean_subdirectory(ce, error_type, o) < 0)
1894                         return -1;
1895                 return 0;
1896         }
1897
1898         /*
1899          * The previous round may already have decided to
1900          * delete this path, which is in a subdirectory that
1901          * is being replaced with a blob.
1902          */
1903         result = index_file_exists(&o->result, name, len, 0);
1904         if (result) {
1905                 if (result->ce_flags & CE_REMOVE)
1906                         return 0;
1907         }
1908
1909         return o->gently ? -1 :
1910                 add_rejected_path(o, error_type, name);
1911 }
1912
1913 /*
1914  * We do not want to remove or overwrite a working tree file that
1915  * is not tracked, unless it is ignored.
1916  */
1917 static int verify_absent_1(const struct cache_entry *ce,
1918                            enum unpack_trees_error_types error_type,
1919                            struct unpack_trees_options *o)
1920 {
1921         int len;
1922         struct stat st;
1923
1924         if (o->index_only || o->reset || !o->update)
1925                 return 0;
1926
1927         len = check_leading_path(ce->name, ce_namelen(ce));
1928         if (!len)
1929                 return 0;
1930         else if (len > 0) {
1931                 char *path;
1932                 int ret;
1933
1934                 path = xmemdupz(ce->name, len);
1935                 if (lstat(path, &st))
1936                         ret = error_errno("cannot stat '%s'", path);
1937                 else {
1938                         if (submodule_from_ce(ce))
1939                                 ret = check_submodule_move_head(ce,
1940                                                                 oid_to_hex(&ce->oid),
1941                                                                 NULL, o);
1942                         else
1943                                 ret = check_ok_to_remove(path, len, DT_UNKNOWN, NULL,
1944                                                          &st, error_type, o);
1945                 }
1946                 free(path);
1947                 return ret;
1948         } else if (lstat(ce->name, &st)) {
1949                 if (errno != ENOENT)
1950                         return error_errno("cannot stat '%s'", ce->name);
1951                 return 0;
1952         } else {
1953                 if (submodule_from_ce(ce))
1954                         return check_submodule_move_head(ce, oid_to_hex(&ce->oid),
1955                                                          NULL, o);
1956
1957                 return check_ok_to_remove(ce->name, ce_namelen(ce),
1958                                           ce_to_dtype(ce), ce, &st,
1959                                           error_type, o);
1960         }
1961 }
1962
1963 static int verify_absent(const struct cache_entry *ce,
1964                          enum unpack_trees_error_types error_type,
1965                          struct unpack_trees_options *o)
1966 {
1967         if (!o->skip_sparse_checkout && (ce->ce_flags & CE_NEW_SKIP_WORKTREE))
1968                 return 0;
1969         return verify_absent_1(ce, error_type, o);
1970 }
1971
1972 static int verify_absent_sparse(const struct cache_entry *ce,
1973                                 enum unpack_trees_error_types error_type,
1974                                 struct unpack_trees_options *o)
1975 {
1976         enum unpack_trees_error_types orphaned_error = error_type;
1977         if (orphaned_error == ERROR_WOULD_LOSE_UNTRACKED_OVERWRITTEN)
1978                 orphaned_error = ERROR_WOULD_LOSE_ORPHANED_OVERWRITTEN;
1979
1980         return verify_absent_1(ce, orphaned_error, o);
1981 }
1982
1983 static int merged_entry(const struct cache_entry *ce,
1984                         const struct cache_entry *old,
1985                         struct unpack_trees_options *o)
1986 {
1987         int update = CE_UPDATE;
1988         struct cache_entry *merge = dup_cache_entry(ce, &o->result);
1989
1990         if (!old) {
1991                 /*
1992                  * New index entries. In sparse checkout, the following
1993                  * verify_absent() will be delayed until after
1994                  * traverse_trees() finishes in unpack_trees(), then:
1995                  *
1996                  *  - CE_NEW_SKIP_WORKTREE will be computed correctly
1997                  *  - verify_absent() be called again, this time with
1998                  *    correct CE_NEW_SKIP_WORKTREE
1999                  *
2000                  * verify_absent() call here does nothing in sparse
2001                  * checkout (i.e. o->skip_sparse_checkout == 0)
2002                  */
2003                 update |= CE_ADDED;
2004                 merge->ce_flags |= CE_NEW_SKIP_WORKTREE;
2005
2006                 if (verify_absent(merge,
2007                                   ERROR_WOULD_LOSE_UNTRACKED_OVERWRITTEN, o)) {
2008                         discard_cache_entry(merge);
2009                         return -1;
2010                 }
2011                 invalidate_ce_path(merge, o);
2012
2013                 if (submodule_from_ce(ce)) {
2014                         int ret = check_submodule_move_head(ce, NULL,
2015                                                             oid_to_hex(&ce->oid),
2016                                                             o);
2017                         if (ret)
2018                                 return ret;
2019                 }
2020
2021         } else if (!(old->ce_flags & CE_CONFLICTED)) {
2022                 /*
2023                  * See if we can re-use the old CE directly?
2024                  * That way we get the uptodate stat info.
2025                  *
2026                  * This also removes the UPDATE flag on a match; otherwise
2027                  * we will end up overwriting local changes in the work tree.
2028                  */
2029                 if (same(old, merge)) {
2030                         copy_cache_entry(merge, old);
2031                         update = 0;
2032                 } else {
2033                         if (verify_uptodate(old, o)) {
2034                                 discard_cache_entry(merge);
2035                                 return -1;
2036                         }
2037                         /* Migrate old flags over */
2038                         update |= old->ce_flags & (CE_SKIP_WORKTREE | CE_NEW_SKIP_WORKTREE);
2039                         invalidate_ce_path(old, o);
2040                 }
2041
2042                 if (submodule_from_ce(ce)) {
2043                         int ret = check_submodule_move_head(ce, oid_to_hex(&old->oid),
2044                                                             oid_to_hex(&ce->oid),
2045                                                             o);
2046                         if (ret)
2047                                 return ret;
2048                 }
2049         } else {
2050                 /*
2051                  * Previously unmerged entry left as an existence
2052                  * marker by read_index_unmerged();
2053                  */
2054                 invalidate_ce_path(old, o);
2055         }
2056
2057         do_add_entry(o, merge, update, CE_STAGEMASK);
2058         return 1;
2059 }
2060
2061 static int deleted_entry(const struct cache_entry *ce,
2062                          const struct cache_entry *old,
2063                          struct unpack_trees_options *o)
2064 {
2065         /* Did it exist in the index? */
2066         if (!old) {
2067                 if (verify_absent(ce, ERROR_WOULD_LOSE_UNTRACKED_REMOVED, o))
2068                         return -1;
2069                 return 0;
2070         }
2071         if (!(old->ce_flags & CE_CONFLICTED) && verify_uptodate(old, o))
2072                 return -1;
2073         add_entry(o, ce, CE_REMOVE, 0);
2074         invalidate_ce_path(ce, o);
2075         return 1;
2076 }
2077
2078 static int keep_entry(const struct cache_entry *ce,
2079                       struct unpack_trees_options *o)
2080 {
2081         add_entry(o, ce, 0, 0);
2082         if (ce_stage(ce))
2083                 invalidate_ce_path(ce, o);
2084         return 1;
2085 }
2086
2087 #if DBRT_DEBUG
2088 static void show_stage_entry(FILE *o,
2089                              const char *label, const struct cache_entry *ce)
2090 {
2091         if (!ce)
2092                 fprintf(o, "%s (missing)\n", label);
2093         else
2094                 fprintf(o, "%s%06o %s %d\t%s\n",
2095                         label,
2096                         ce->ce_mode,
2097                         oid_to_hex(&ce->oid),
2098                         ce_stage(ce),
2099                         ce->name);
2100 }
2101 #endif
2102
2103 int threeway_merge(const struct cache_entry * const *stages,
2104                    struct unpack_trees_options *o)
2105 {
2106         const struct cache_entry *index;
2107         const struct cache_entry *head;
2108         const struct cache_entry *remote = stages[o->head_idx + 1];
2109         int count;
2110         int head_match = 0;
2111         int remote_match = 0;
2112
2113         int df_conflict_head = 0;
2114         int df_conflict_remote = 0;
2115
2116         int any_anc_missing = 0;
2117         int no_anc_exists = 1;
2118         int i;
2119
2120         for (i = 1; i < o->head_idx; i++) {
2121                 if (!stages[i] || stages[i] == o->df_conflict_entry)
2122                         any_anc_missing = 1;
2123                 else
2124                         no_anc_exists = 0;
2125         }
2126
2127         index = stages[0];
2128         head = stages[o->head_idx];
2129
2130         if (head == o->df_conflict_entry) {
2131                 df_conflict_head = 1;
2132                 head = NULL;
2133         }
2134
2135         if (remote == o->df_conflict_entry) {
2136                 df_conflict_remote = 1;
2137                 remote = NULL;
2138         }
2139
2140         /*
2141          * First, if there's a #16 situation, note that to prevent #13
2142          * and #14.
2143          */
2144         if (!same(remote, head)) {
2145                 for (i = 1; i < o->head_idx; i++) {
2146                         if (same(stages[i], head)) {
2147                                 head_match = i;
2148                         }
2149                         if (same(stages[i], remote)) {
2150                                 remote_match = i;
2151                         }
2152                 }
2153         }
2154
2155         /*
2156          * We start with cases where the index is allowed to match
2157          * something other than the head: #14(ALT) and #2ALT, where it
2158          * is permitted to match the result instead.
2159          */
2160         /* #14, #14ALT, #2ALT */
2161         if (remote && !df_conflict_head && head_match && !remote_match) {
2162                 if (index && !same(index, remote) && !same(index, head))
2163                         return reject_merge(index, o);
2164                 return merged_entry(remote, index, o);
2165         }
2166         /*
2167          * If we have an entry in the index cache, then we want to
2168          * make sure that it matches head.
2169          */
2170         if (index && !same(index, head))
2171                 return reject_merge(index, o);
2172
2173         if (head) {
2174                 /* #5ALT, #15 */
2175                 if (same(head, remote))
2176                         return merged_entry(head, index, o);
2177                 /* #13, #3ALT */
2178                 if (!df_conflict_remote && remote_match && !head_match)
2179                         return merged_entry(head, index, o);
2180         }
2181
2182         /* #1 */
2183         if (!head && !remote && any_anc_missing)
2184                 return 0;
2185
2186         /*
2187          * Under the "aggressive" rule, we resolve mostly trivial
2188          * cases that we historically had git-merge-one-file resolve.
2189          */
2190         if (o->aggressive) {
2191                 int head_deleted = !head;
2192                 int remote_deleted = !remote;
2193                 const struct cache_entry *ce = NULL;
2194
2195                 if (index)
2196                         ce = index;
2197                 else if (head)
2198                         ce = head;
2199                 else if (remote)
2200                         ce = remote;
2201                 else {
2202                         for (i = 1; i < o->head_idx; i++) {
2203                                 if (stages[i] && stages[i] != o->df_conflict_entry) {
2204                                         ce = stages[i];
2205                                         break;
2206                                 }
2207                         }
2208                 }
2209
2210                 /*
2211                  * Deleted in both.
2212                  * Deleted in one and unchanged in the other.
2213                  */
2214                 if ((head_deleted && remote_deleted) ||
2215                     (head_deleted && remote && remote_match) ||
2216                     (remote_deleted && head && head_match)) {
2217                         if (index)
2218                                 return deleted_entry(index, index, o);
2219                         if (ce && !head_deleted) {
2220                                 if (verify_absent(ce, ERROR_WOULD_LOSE_UNTRACKED_REMOVED, o))
2221                                         return -1;
2222                         }
2223                         return 0;
2224                 }
2225                 /*
2226                  * Added in both, identically.
2227                  */
2228                 if (no_anc_exists && head && remote && same(head, remote))
2229                         return merged_entry(head, index, o);
2230
2231         }
2232
2233         /* Below are "no merge" cases, which require that the index be
2234          * up-to-date to avoid the files getting overwritten with
2235          * conflict resolution files.
2236          */
2237         if (index) {
2238                 if (verify_uptodate(index, o))
2239                         return -1;
2240         }
2241
2242         o->nontrivial_merge = 1;
2243
2244         /* #2, #3, #4, #6, #7, #9, #10, #11. */
2245         count = 0;
2246         if (!head_match || !remote_match) {
2247                 for (i = 1; i < o->head_idx; i++) {
2248                         if (stages[i] && stages[i] != o->df_conflict_entry) {
2249                                 keep_entry(stages[i], o);
2250                                 count++;
2251                                 break;
2252                         }
2253                 }
2254         }
2255 #if DBRT_DEBUG
2256         else {
2257                 fprintf(stderr, "read-tree: warning #16 detected\n");
2258                 show_stage_entry(stderr, "head   ", stages[head_match]);
2259                 show_stage_entry(stderr, "remote ", stages[remote_match]);
2260         }
2261 #endif
2262         if (head) { count += keep_entry(head, o); }
2263         if (remote) { count += keep_entry(remote, o); }
2264         return count;
2265 }
2266
2267 /*
2268  * Two-way merge.
2269  *
2270  * The rule is to "carry forward" what is in the index without losing
2271  * information across a "fast-forward", favoring a successful merge
2272  * over a merge failure when it makes sense.  For details of the
2273  * "carry forward" rule, please see <Documentation/git-read-tree.txt>.
2274  *
2275  */
2276 int twoway_merge(const struct cache_entry * const *src,
2277                  struct unpack_trees_options *o)
2278 {
2279         const struct cache_entry *current = src[0];
2280         const struct cache_entry *oldtree = src[1];
2281         const struct cache_entry *newtree = src[2];
2282
2283         if (o->merge_size != 2)
2284                 return error("Cannot do a twoway merge of %d trees",
2285                              o->merge_size);
2286
2287         if (oldtree == o->df_conflict_entry)
2288                 oldtree = NULL;
2289         if (newtree == o->df_conflict_entry)
2290                 newtree = NULL;
2291
2292         if (current) {
2293                 if (current->ce_flags & CE_CONFLICTED) {
2294                         if (same(oldtree, newtree) || o->reset) {
2295                                 if (!newtree)
2296                                         return deleted_entry(current, current, o);
2297                                 else
2298                                         return merged_entry(newtree, current, o);
2299                         }
2300                         return reject_merge(current, o);
2301                 } else if ((!oldtree && !newtree) || /* 4 and 5 */
2302                          (!oldtree && newtree &&
2303                           same(current, newtree)) || /* 6 and 7 */
2304                          (oldtree && newtree &&
2305                           same(oldtree, newtree)) || /* 14 and 15 */
2306                          (oldtree && newtree &&
2307                           !same(oldtree, newtree) && /* 18 and 19 */
2308                           same(current, newtree))) {
2309                         return keep_entry(current, o);
2310                 } else if (oldtree && !newtree && same(current, oldtree)) {
2311                         /* 10 or 11 */
2312                         return deleted_entry(oldtree, current, o);
2313                 } else if (oldtree && newtree &&
2314                          same(current, oldtree) && !same(current, newtree)) {
2315                         /* 20 or 21 */
2316                         return merged_entry(newtree, current, o);
2317                 } else
2318                         return reject_merge(current, o);
2319         }
2320         else if (newtree) {
2321                 if (oldtree && !o->initial_checkout) {
2322                         /*
2323                          * deletion of the path was staged;
2324                          */
2325                         if (same(oldtree, newtree))
2326                                 return 1;
2327                         return reject_merge(oldtree, o);
2328                 }
2329                 return merged_entry(newtree, current, o);
2330         }
2331         return deleted_entry(oldtree, current, o);
2332 }
2333
2334 /*
2335  * Bind merge.
2336  *
2337  * Keep the index entries at stage0, collapse stage1 but make sure
2338  * stage0 does not have anything there.
2339  */
2340 int bind_merge(const struct cache_entry * const *src,
2341                struct unpack_trees_options *o)
2342 {
2343         const struct cache_entry *old = src[0];
2344         const struct cache_entry *a = src[1];
2345
2346         if (o->merge_size != 1)
2347                 return error("Cannot do a bind merge of %d trees",
2348                              o->merge_size);
2349         if (a && old)
2350                 return o->gently ? -1 :
2351                         error(ERRORMSG(o, ERROR_BIND_OVERLAP),
2352                               super_prefixed(a->name),
2353                               super_prefixed(old->name));
2354         if (!a)
2355                 return keep_entry(old, o);
2356         else
2357                 return merged_entry(a, NULL, o);
2358 }
2359
2360 /*
2361  * One-way merge.
2362  *
2363  * The rule is:
2364  * - take the stat information from stage0, take the data from stage1
2365  */
2366 int oneway_merge(const struct cache_entry * const *src,
2367                  struct unpack_trees_options *o)
2368 {
2369         const struct cache_entry *old = src[0];
2370         const struct cache_entry *a = src[1];
2371
2372         if (o->merge_size != 1)
2373                 return error("Cannot do a oneway merge of %d trees",
2374                              o->merge_size);
2375
2376         if (!a || a == o->df_conflict_entry)
2377                 return deleted_entry(old, old, o);
2378
2379         if (old && same(old, a)) {
2380                 int update = 0;
2381                 if (o->reset && o->update && !ce_uptodate(old) && !ce_skip_worktree(old)) {
2382                         struct stat st;
2383                         if (lstat(old->name, &st) ||
2384                             ie_match_stat(o->src_index, old, &st, CE_MATCH_IGNORE_VALID|CE_MATCH_IGNORE_SKIP_WORKTREE))
2385                                 update |= CE_UPDATE;
2386                 }
2387                 if (o->update && S_ISGITLINK(old->ce_mode) &&
2388                     should_update_submodules() && !verify_uptodate(old, o))
2389                         update |= CE_UPDATE;
2390                 add_entry(o, old, update, 0);
2391                 return 0;
2392         }
2393         return merged_entry(a, old, o);
2394 }