harden REALLOC_ARRAY and xcalloc against size_t overflow
[git] / diff-lib.c
1 /*
2  * Copyright (C) 2005 Junio C Hamano
3  */
4 #include "cache.h"
5 #include "quote.h"
6 #include "commit.h"
7 #include "diff.h"
8 #include "diffcore.h"
9 #include "revision.h"
10 #include "cache-tree.h"
11 #include "unpack-trees.h"
12 #include "refs.h"
13 #include "submodule.h"
14 #include "dir.h"
15
16 /*
17  * diff-files
18  */
19
20 /*
21  * Has the work tree entity been removed?
22  *
23  * Return 1 if it was removed from the work tree, 0 if an entity to be
24  * compared with the cache entry ce still exists (the latter includes
25  * the case where a directory that is not a submodule repository
26  * exists for ce that is a submodule -- it is a submodule that is not
27  * checked out).  Return negative for an error.
28  */
29 static int check_removed(const struct cache_entry *ce, struct stat *st)
30 {
31         if (lstat(ce->name, st) < 0) {
32                 if (errno != ENOENT && errno != ENOTDIR)
33                         return -1;
34                 return 1;
35         }
36         if (has_symlink_leading_path(ce->name, ce_namelen(ce)))
37                 return 1;
38         if (S_ISDIR(st->st_mode)) {
39                 unsigned char sub[20];
40
41                 /*
42                  * If ce is already a gitlink, we can have a plain
43                  * directory (i.e. the submodule is not checked out),
44                  * or a checked out submodule.  Either case this is not
45                  * a case where something was removed from the work tree,
46                  * so we will return 0.
47                  *
48                  * Otherwise, if the directory is not a submodule
49                  * repository, that means ce which was a blob turned into
50                  * a directory --- the blob was removed!
51                  */
52                 if (!S_ISGITLINK(ce->ce_mode) &&
53                     resolve_gitlink_ref(ce->name, "HEAD", sub))
54                         return 1;
55         }
56         return 0;
57 }
58
59 /*
60  * Has a file changed or has a submodule new commits or a dirty work tree?
61  *
62  * Return 1 when changes are detected, 0 otherwise. If the DIRTY_SUBMODULES
63  * option is set, the caller does not only want to know if a submodule is
64  * modified at all but wants to know all the conditions that are met (new
65  * commits, untracked content and/or modified content).
66  */
67 static int match_stat_with_submodule(struct diff_options *diffopt,
68                                      const struct cache_entry *ce,
69                                      struct stat *st, unsigned ce_option,
70                                      unsigned *dirty_submodule)
71 {
72         int changed = ce_match_stat(ce, st, ce_option);
73         if (S_ISGITLINK(ce->ce_mode)) {
74                 unsigned orig_flags = diffopt->flags;
75                 if (!DIFF_OPT_TST(diffopt, OVERRIDE_SUBMODULE_CONFIG))
76                         set_diffopt_flags_from_submodule_config(diffopt, ce->name);
77                 if (DIFF_OPT_TST(diffopt, IGNORE_SUBMODULES))
78                         changed = 0;
79                 else if (!DIFF_OPT_TST(diffopt, IGNORE_DIRTY_SUBMODULES)
80                     && (!changed || DIFF_OPT_TST(diffopt, DIRTY_SUBMODULES)))
81                         *dirty_submodule = is_submodule_modified(ce->name, DIFF_OPT_TST(diffopt, IGNORE_UNTRACKED_IN_SUBMODULES));
82                 diffopt->flags = orig_flags;
83         }
84         return changed;
85 }
86
87 int run_diff_files(struct rev_info *revs, unsigned int option)
88 {
89         int entries, i;
90         int diff_unmerged_stage = revs->max_count;
91         unsigned ce_option = ((option & DIFF_RACY_IS_MODIFIED)
92                               ? CE_MATCH_RACY_IS_DIRTY : 0);
93
94         diff_set_mnemonic_prefix(&revs->diffopt, "i/", "w/");
95
96         if (diff_unmerged_stage < 0)
97                 diff_unmerged_stage = 2;
98         entries = active_nr;
99         for (i = 0; i < entries; i++) {
100                 unsigned int oldmode, newmode;
101                 struct cache_entry *ce = active_cache[i];
102                 int changed;
103                 unsigned dirty_submodule = 0;
104                 const unsigned char *old_sha1, *new_sha1;
105
106                 if (diff_can_quit_early(&revs->diffopt))
107                         break;
108
109                 if (!ce_path_match(ce, &revs->prune_data, NULL))
110                         continue;
111
112                 if (ce_stage(ce)) {
113                         struct combine_diff_path *dpath;
114                         struct diff_filepair *pair;
115                         unsigned int wt_mode = 0;
116                         int num_compare_stages = 0;
117                         size_t path_len;
118                         struct stat st;
119
120                         path_len = ce_namelen(ce);
121
122                         dpath = xmalloc(combine_diff_path_size(5, path_len));
123                         dpath->path = (char *) &(dpath->parent[5]);
124
125                         dpath->next = NULL;
126                         memcpy(dpath->path, ce->name, path_len);
127                         dpath->path[path_len] = '\0';
128                         oidclr(&dpath->oid);
129                         memset(&(dpath->parent[0]), 0,
130                                sizeof(struct combine_diff_parent)*5);
131
132                         changed = check_removed(ce, &st);
133                         if (!changed)
134                                 wt_mode = ce_mode_from_stat(ce, st.st_mode);
135                         else {
136                                 if (changed < 0) {
137                                         perror(ce->name);
138                                         continue;
139                                 }
140                                 wt_mode = 0;
141                         }
142                         dpath->mode = wt_mode;
143
144                         while (i < entries) {
145                                 struct cache_entry *nce = active_cache[i];
146                                 int stage;
147
148                                 if (strcmp(ce->name, nce->name))
149                                         break;
150
151                                 /* Stage #2 (ours) is the first parent,
152                                  * stage #3 (theirs) is the second.
153                                  */
154                                 stage = ce_stage(nce);
155                                 if (2 <= stage) {
156                                         int mode = nce->ce_mode;
157                                         num_compare_stages++;
158                                         hashcpy(dpath->parent[stage-2].oid.hash, nce->sha1);
159                                         dpath->parent[stage-2].mode = ce_mode_from_stat(nce, mode);
160                                         dpath->parent[stage-2].status =
161                                                 DIFF_STATUS_MODIFIED;
162                                 }
163
164                                 /* diff against the proper unmerged stage */
165                                 if (stage == diff_unmerged_stage)
166                                         ce = nce;
167                                 i++;
168                         }
169                         /*
170                          * Compensate for loop update
171                          */
172                         i--;
173
174                         if (revs->combine_merges && num_compare_stages == 2) {
175                                 show_combined_diff(dpath, 2,
176                                                    revs->dense_combined_merges,
177                                                    revs);
178                                 free(dpath);
179                                 continue;
180                         }
181                         free(dpath);
182                         dpath = NULL;
183
184                         /*
185                          * Show the diff for the 'ce' if we found the one
186                          * from the desired stage.
187                          */
188                         pair = diff_unmerge(&revs->diffopt, ce->name);
189                         if (wt_mode)
190                                 pair->two->mode = wt_mode;
191                         if (ce_stage(ce) != diff_unmerged_stage)
192                                 continue;
193                 }
194
195                 if (ce_uptodate(ce) || ce_skip_worktree(ce))
196                         continue;
197
198                 /* If CE_VALID is set, don't look at workdir for file removal */
199                 if (ce->ce_flags & CE_VALID) {
200                         changed = 0;
201                         newmode = ce->ce_mode;
202                 } else {
203                         struct stat st;
204
205                         changed = check_removed(ce, &st);
206                         if (changed) {
207                                 if (changed < 0) {
208                                         perror(ce->name);
209                                         continue;
210                                 }
211                                 diff_addremove(&revs->diffopt, '-', ce->ce_mode,
212                                                ce->sha1, !is_null_sha1(ce->sha1),
213                                                ce->name, 0);
214                                 continue;
215                         }
216
217                         changed = match_stat_with_submodule(&revs->diffopt, ce, &st,
218                                                             ce_option, &dirty_submodule);
219                         newmode = ce_mode_from_stat(ce, st.st_mode);
220                 }
221
222                 if (!changed && !dirty_submodule) {
223                         ce_mark_uptodate(ce);
224                         if (!DIFF_OPT_TST(&revs->diffopt, FIND_COPIES_HARDER))
225                                 continue;
226                 }
227                 oldmode = ce->ce_mode;
228                 old_sha1 = ce->sha1;
229                 new_sha1 = changed ? null_sha1 : ce->sha1;
230                 diff_change(&revs->diffopt, oldmode, newmode,
231                             old_sha1, new_sha1,
232                             !is_null_sha1(old_sha1),
233                             !is_null_sha1(new_sha1),
234                             ce->name, 0, dirty_submodule);
235
236         }
237         diffcore_std(&revs->diffopt);
238         diff_flush(&revs->diffopt);
239         return 0;
240 }
241
242 /*
243  * diff-index
244  */
245
246 /* A file entry went away or appeared */
247 static void diff_index_show_file(struct rev_info *revs,
248                                  const char *prefix,
249                                  const struct cache_entry *ce,
250                                  const unsigned char *sha1, int sha1_valid,
251                                  unsigned int mode,
252                                  unsigned dirty_submodule)
253 {
254         diff_addremove(&revs->diffopt, prefix[0], mode,
255                        sha1, sha1_valid, ce->name, dirty_submodule);
256 }
257
258 static int get_stat_data(const struct cache_entry *ce,
259                          const unsigned char **sha1p,
260                          unsigned int *modep,
261                          int cached, int match_missing,
262                          unsigned *dirty_submodule, struct diff_options *diffopt)
263 {
264         const unsigned char *sha1 = ce->sha1;
265         unsigned int mode = ce->ce_mode;
266
267         if (!cached && !ce_uptodate(ce)) {
268                 int changed;
269                 struct stat st;
270                 changed = check_removed(ce, &st);
271                 if (changed < 0)
272                         return -1;
273                 else if (changed) {
274                         if (match_missing) {
275                                 *sha1p = sha1;
276                                 *modep = mode;
277                                 return 0;
278                         }
279                         return -1;
280                 }
281                 changed = match_stat_with_submodule(diffopt, ce, &st,
282                                                     0, dirty_submodule);
283                 if (changed) {
284                         mode = ce_mode_from_stat(ce, st.st_mode);
285                         sha1 = null_sha1;
286                 }
287         }
288
289         *sha1p = sha1;
290         *modep = mode;
291         return 0;
292 }
293
294 static void show_new_file(struct rev_info *revs,
295                           const struct cache_entry *new,
296                           int cached, int match_missing)
297 {
298         const unsigned char *sha1;
299         unsigned int mode;
300         unsigned dirty_submodule = 0;
301
302         /*
303          * New file in the index: it might actually be different in
304          * the working tree.
305          */
306         if (get_stat_data(new, &sha1, &mode, cached, match_missing,
307             &dirty_submodule, &revs->diffopt) < 0)
308                 return;
309
310         diff_index_show_file(revs, "+", new, sha1, !is_null_sha1(sha1), mode, dirty_submodule);
311 }
312
313 static int show_modified(struct rev_info *revs,
314                          const struct cache_entry *old,
315                          const struct cache_entry *new,
316                          int report_missing,
317                          int cached, int match_missing)
318 {
319         unsigned int mode, oldmode;
320         const unsigned char *sha1;
321         unsigned dirty_submodule = 0;
322
323         if (get_stat_data(new, &sha1, &mode, cached, match_missing,
324                           &dirty_submodule, &revs->diffopt) < 0) {
325                 if (report_missing)
326                         diff_index_show_file(revs, "-", old,
327                                              old->sha1, 1, old->ce_mode, 0);
328                 return -1;
329         }
330
331         if (revs->combine_merges && !cached &&
332             (hashcmp(sha1, old->sha1) || hashcmp(old->sha1, new->sha1))) {
333                 struct combine_diff_path *p;
334                 int pathlen = ce_namelen(new);
335
336                 p = xmalloc(combine_diff_path_size(2, pathlen));
337                 p->path = (char *) &p->parent[2];
338                 p->next = NULL;
339                 memcpy(p->path, new->name, pathlen);
340                 p->path[pathlen] = 0;
341                 p->mode = mode;
342                 oidclr(&p->oid);
343                 memset(p->parent, 0, 2 * sizeof(struct combine_diff_parent));
344                 p->parent[0].status = DIFF_STATUS_MODIFIED;
345                 p->parent[0].mode = new->ce_mode;
346                 hashcpy(p->parent[0].oid.hash, new->sha1);
347                 p->parent[1].status = DIFF_STATUS_MODIFIED;
348                 p->parent[1].mode = old->ce_mode;
349                 hashcpy(p->parent[1].oid.hash, old->sha1);
350                 show_combined_diff(p, 2, revs->dense_combined_merges, revs);
351                 free(p);
352                 return 0;
353         }
354
355         oldmode = old->ce_mode;
356         if (mode == oldmode && !hashcmp(sha1, old->sha1) && !dirty_submodule &&
357             !DIFF_OPT_TST(&revs->diffopt, FIND_COPIES_HARDER))
358                 return 0;
359
360         diff_change(&revs->diffopt, oldmode, mode,
361                     old->sha1, sha1, 1, !is_null_sha1(sha1),
362                     old->name, 0, dirty_submodule);
363         return 0;
364 }
365
366 /*
367  * This gets a mix of an existing index and a tree, one pathname entry
368  * at a time. The index entry may be a single stage-0 one, but it could
369  * also be multiple unmerged entries (in which case idx_pos/idx_nr will
370  * give you the position and number of entries in the index).
371  */
372 static void do_oneway_diff(struct unpack_trees_options *o,
373                            const struct cache_entry *idx,
374                            const struct cache_entry *tree)
375 {
376         struct rev_info *revs = o->unpack_data;
377         int match_missing, cached;
378
379         /* if the entry is not checked out, don't examine work tree */
380         cached = o->index_only ||
381                 (idx && ((idx->ce_flags & CE_VALID) || ce_skip_worktree(idx)));
382         /*
383          * Backward compatibility wart - "diff-index -m" does
384          * not mean "do not ignore merges", but "match_missing".
385          *
386          * But with the revision flag parsing, that's found in
387          * "!revs->ignore_merges".
388          */
389         match_missing = !revs->ignore_merges;
390
391         if (cached && idx && ce_stage(idx)) {
392                 struct diff_filepair *pair;
393                 pair = diff_unmerge(&revs->diffopt, idx->name);
394                 if (tree)
395                         fill_filespec(pair->one, tree->sha1, 1, tree->ce_mode);
396                 return;
397         }
398
399         /*
400          * Something added to the tree?
401          */
402         if (!tree) {
403                 show_new_file(revs, idx, cached, match_missing);
404                 return;
405         }
406
407         /*
408          * Something removed from the tree?
409          */
410         if (!idx) {
411                 diff_index_show_file(revs, "-", tree, tree->sha1, 1, tree->ce_mode, 0);
412                 return;
413         }
414
415         /* Show difference between old and new */
416         show_modified(revs, tree, idx, 1, cached, match_missing);
417 }
418
419 /*
420  * The unpack_trees() interface is designed for merging, so
421  * the different source entries are designed primarily for
422  * the source trees, with the old index being really mainly
423  * used for being replaced by the result.
424  *
425  * For diffing, the index is more important, and we only have a
426  * single tree.
427  *
428  * We're supposed to advance o->pos to skip what we have already processed.
429  *
430  * This wrapper makes it all more readable, and takes care of all
431  * the fairly complex unpack_trees() semantic requirements, including
432  * the skipping, the path matching, the type conflict cases etc.
433  */
434 static int oneway_diff(const struct cache_entry * const *src,
435                        struct unpack_trees_options *o)
436 {
437         const struct cache_entry *idx = src[0];
438         const struct cache_entry *tree = src[1];
439         struct rev_info *revs = o->unpack_data;
440
441         /*
442          * Unpack-trees generates a DF/conflict entry if
443          * there was a directory in the index and a tree
444          * in the tree. From a diff standpoint, that's a
445          * delete of the tree and a create of the file.
446          */
447         if (tree == o->df_conflict_entry)
448                 tree = NULL;
449
450         if (ce_path_match(idx ? idx : tree, &revs->prune_data, NULL)) {
451                 do_oneway_diff(o, idx, tree);
452                 if (diff_can_quit_early(&revs->diffopt)) {
453                         o->exiting_early = 1;
454                         return -1;
455                 }
456         }
457
458         return 0;
459 }
460
461 static int diff_cache(struct rev_info *revs,
462                       const unsigned char *tree_sha1,
463                       const char *tree_name,
464                       int cached)
465 {
466         struct tree *tree;
467         struct tree_desc t;
468         struct unpack_trees_options opts;
469
470         tree = parse_tree_indirect(tree_sha1);
471         if (!tree)
472                 return error("bad tree object %s",
473                              tree_name ? tree_name : sha1_to_hex(tree_sha1));
474         memset(&opts, 0, sizeof(opts));
475         opts.head_idx = 1;
476         opts.index_only = cached;
477         opts.diff_index_cached = (cached &&
478                                   !DIFF_OPT_TST(&revs->diffopt, FIND_COPIES_HARDER));
479         opts.merge = 1;
480         opts.fn = oneway_diff;
481         opts.unpack_data = revs;
482         opts.src_index = &the_index;
483         opts.dst_index = NULL;
484         opts.pathspec = &revs->diffopt.pathspec;
485         opts.pathspec->recursive = 1;
486
487         init_tree_desc(&t, tree->buffer, tree->size);
488         return unpack_trees(1, &t, &opts);
489 }
490
491 int run_diff_index(struct rev_info *revs, int cached)
492 {
493         struct object_array_entry *ent;
494
495         ent = revs->pending.objects;
496         if (diff_cache(revs, ent->item->oid.hash, ent->name, cached))
497                 exit(128);
498
499         diff_set_mnemonic_prefix(&revs->diffopt, "c/", cached ? "i/" : "w/");
500         diffcore_fix_diff_index(&revs->diffopt);
501         diffcore_std(&revs->diffopt);
502         diff_flush(&revs->diffopt);
503         return 0;
504 }
505
506 int do_diff_cache(const unsigned char *tree_sha1, struct diff_options *opt)
507 {
508         struct rev_info revs;
509
510         init_revisions(&revs, NULL);
511         copy_pathspec(&revs.prune_data, &opt->pathspec);
512         revs.diffopt = *opt;
513
514         if (diff_cache(&revs, tree_sha1, NULL, 1))
515                 exit(128);
516         return 0;
517 }
518
519 int index_differs_from(const char *def, int diff_flags)
520 {
521         struct rev_info rev;
522         struct setup_revision_opt opt;
523
524         init_revisions(&rev, NULL);
525         memset(&opt, 0, sizeof(opt));
526         opt.def = def;
527         setup_revisions(0, NULL, &rev, &opt);
528         DIFF_OPT_SET(&rev.diffopt, QUICK);
529         DIFF_OPT_SET(&rev.diffopt, EXIT_WITH_STATUS);
530         rev.diffopt.flags |= diff_flags;
531         run_diff_index(&rev, 1);
532         if (rev.pending.alloc)
533                 free(rev.pending.objects);
534         return (DIFF_OPT_TST(&rev.diffopt, HAS_CHANGES) != 0);
535 }