merge-recursive.c

   1 /*
   2  * Recursive Merge algorithm stolen from git-merge-recursive.py by
   3  * Fredrik Kuivinen.
   4  * The thieves were Alex Riesen and Johannes Schindelin, in June/July 2006
   5  */
   6 #include "advice.h"
   7 #include "cache.h"
   8 #include "cache-tree.h"
   9 #include "commit.h"
  10 #include "blob.h"
  11 #include "builtin.h"
  12 #include "tree-walk.h"
  13 #include "diff.h"
  14 #include "diffcore.h"
  15 #include "tag.h"
  16 #include "unpack-trees.h"
  17 #include "string-list.h"
  18 #include "xdiff-interface.h"
  19 #include "ll-merge.h"
  20 #include "attr.h"
  21 #include "merge-recursive.h"
  22 #include "dir.h"
  23 #include "submodule.h"
  24
  25 static struct tree *shift_tree_object(struct tree *one, struct tree *two,
  26                                       const char *subtree_shift)
  27 {
  28         unsigned char shifted[20];
  29
  30         if (!*subtree_shift) {
  31                 shift_tree(one->object.sha1, two->object.sha1, shifted, 0);
  32         } else {
  33                 shift_tree_by(one->object.sha1, two->object.sha1, shifted,
  34                               subtree_shift);
  35         }
  36         if (!hashcmp(two->object.sha1, shifted))
  37                 return two;
  38         return lookup_tree(shifted);
  39 }
  40
  41 /*
  42  * A virtual commit has (const char *)commit->util set to the name.
  43  */
  44
  45 static struct commit *make_virtual_commit(struct tree *tree, const char *comment)
  46 {
  47         struct commit *commit = xcalloc(1, sizeof(struct commit));
  48         commit->tree = tree;
  49         commit->util = (void*)comment;
  50         /* avoid warnings */
  51         commit->object.parsed = 1;
  52         return commit;
  53 }
  54
  55 /*
  56  * Since we use get_tree_entry(), which does not put the read object into
  57  * the object pool, we cannot rely on a == b.
  58  */
  59 static int sha_eq(const unsigned char *a, const unsigned char *b)
  60 {
  61         if (!a && !b)
  62                 return 2;
  63         return a && b && hashcmp(a, b) == 0;
  64 }
  65
  66 enum rename_type {
  67         RENAME_NORMAL = 0,
  68         RENAME_DELETE,
  69         RENAME_ONE_FILE_TO_ONE,
  70         RENAME_ONE_FILE_TO_TWO
  71 };
  72
  73 struct rename_conflict_info {
  74         enum rename_type rename_type;
  75         struct diff_filepair *pair1;
  76         struct diff_filepair *pair2;
  77         const char *branch1;
  78         const char *branch2;
  79         struct stage_data *dst_entry1;
  80         struct stage_data *dst_entry2;
  81 };
  82
  83 /*
  84  * Since we want to write the index eventually, we cannot reuse the index
  85  * for these (temporary) data.
  86  */
  87 struct stage_data {
  88         struct {
  89                 unsigned mode;
  90                 unsigned char sha[20];
  91         } stages[4];
  92         struct rename_conflict_info *rename_conflict_info;
  93         unsigned processed:1;
  94 };
  95
  96 static inline void setup_rename_conflict_info(enum rename_type rename_type,
  97                                               struct diff_filepair *pair1,
  98                                               struct diff_filepair *pair2,
  99                                               const char *branch1,
 100                                               const char *branch2,
 101                                               struct stage_data *dst_entry1,
 102                                               struct stage_data *dst_entry2)
 103 {
 104         struct rename_conflict_info *ci = xcalloc(1, sizeof(struct rename_conflict_info));
 105         ci->rename_type = rename_type;
 106         ci->pair1 = pair1;
 107         ci->branch1 = branch1;
 108         ci->branch2 = branch2;
 109
 110         ci->dst_entry1 = dst_entry1;
 111         dst_entry1->rename_conflict_info = ci;
 112         dst_entry1->processed = 0;
 113
 114         assert(!pair2 == !dst_entry2);
 115         if (dst_entry2) {
 116                 ci->dst_entry2 = dst_entry2;
 117                 ci->pair2 = pair2;
 118                 dst_entry2->rename_conflict_info = ci;
 119         }
 120 }
 121
 122 static int show(struct merge_options *o, int v)
 123 {
 124         return (!o->call_depth && o->verbosity >= v) || o->verbosity >= 5;
 125 }
 126
 127 static void flush_output(struct merge_options *o)
 128 {
 129         if (o->obuf.len) {
 130                 fputs(o->obuf.buf, stdout);
 131                 strbuf_reset(&o->obuf);
 132         }
 133 }
 134
 135 __attribute__((format (printf, 3, 4)))
 136 static void output(struct merge_options *o, int v, const char *fmt, ...)
 137 {
 138         va_list ap;
 139
 140         if (!show(o, v))
 141                 return;
 142
 143         strbuf_grow(&o->obuf, o->call_depth * 2 + 2);
 144         memset(o->obuf.buf + o->obuf.len, ' ', o->call_depth * 2);
 145         strbuf_setlen(&o->obuf, o->obuf.len + o->call_depth * 2);
 146
 147         va_start(ap, fmt);
 148         strbuf_vaddf(&o->obuf, fmt, ap);
 149         va_end(ap);
 150
 151         strbuf_add(&o->obuf, "\n", 1);
 152         if (!o->buffer_output)
 153                 flush_output(o);
 154 }
 155
 156 static void output_commit_title(struct merge_options *o, struct commit *commit)
 157 {
 158         int i;
 159         flush_output(o);
 160         for (i = o->call_depth; i--;)
 161                 fputs("  ", stdout);
 162         if (commit->util)
 163                 printf("virtual %s\n", (char *)commit->util);
 164         else {
 165                 printf("%s ", find_unique_abbrev(commit->object.sha1, DEFAULT_ABBREV));
 166                 if (parse_commit(commit) != 0)
 167                         printf("(bad commit)\n");
 168                 else {
 169                         const char *title;
 170                         int len = find_commit_subject(commit->buffer, &title);
 171                         if (len)
 172                                 printf("%.*s\n", len, title);
 173                 }
 174         }
 175 }
 176
 177 static int add_cacheinfo(unsigned int mode, const unsigned char *sha1,
 178                 const char *path, int stage, int refresh, int options)
 179 {
 180         struct cache_entry *ce;
 181         ce = make_cache_entry(mode, sha1 ? sha1 : null_sha1, path, stage, refresh);
 182         if (!ce)
 183                 return error("addinfo_cache failed for path '%s'", path);
 184         return add_cache_entry(ce, options);
 185 }
 186
 187 static void init_tree_desc_from_tree(struct tree_desc *desc, struct tree *tree)
 188 {
 189         parse_tree(tree);
 190         init_tree_desc(desc, tree->buffer, tree->size);
 191 }
 192
 193 static int git_merge_trees(int index_only,
 194                            struct tree *common,
 195                            struct tree *head,
 196                            struct tree *merge)
 197 {
 198         int rc;
 199         struct tree_desc t[3];
 200         struct unpack_trees_options opts;
 201
 202         memset(&opts, 0, sizeof(opts));
 203         if (index_only)
 204                 opts.index_only = 1;
 205         else
 206                 opts.update = 1;
 207         opts.merge = 1;
 208         opts.head_idx = 2;
 209         opts.fn = threeway_merge;
 210         opts.src_index = &the_index;
 211         opts.dst_index = &the_index;
 212         setup_unpack_trees_porcelain(&opts, "merge");
 213
 214         init_tree_desc_from_tree(t+0, common);
 215         init_tree_desc_from_tree(t+1, head);
 216         init_tree_desc_from_tree(t+2, merge);
 217
 218         rc = unpack_trees(3, t, &opts);
 219         cache_tree_free(&active_cache_tree);
 220         return rc;
 221 }
 222
 223 struct tree *write_tree_from_memory(struct merge_options *o)
 224 {
 225         struct tree *result = NULL;
 226
 227         if (unmerged_cache()) {
 228                 int i;
 229                 fprintf(stderr, "BUG: There are unmerged index entries:\n");
 230                 for (i = 0; i < active_nr; i++) {
 231                         struct cache_entry *ce = active_cache[i];
 232                         if (ce_stage(ce))
 233                                 fprintf(stderr, "BUG: %d %.*s\n", ce_stage(ce),
 234                                         (int)ce_namelen(ce), ce->name);
 235                 }
 236                 die("Bug in merge-recursive.c");
 237         }
 238
 239         if (!active_cache_tree)
 240                 active_cache_tree = cache_tree();
 241
 242         if (!cache_tree_fully_valid(active_cache_tree) &&
 243             cache_tree_update(active_cache_tree,
 244                               active_cache, active_nr, 0, 0) < 0)
 245                 die("error building trees");
 246
 247         result = lookup_tree(active_cache_tree->sha1);
 248
 249         return result;
 250 }
 251
 252 static int save_files_dirs(const unsigned char *sha1,
 253                 const char *base, int baselen, const char *path,
 254                 unsigned int mode, int stage, void *context)
 255 {
 256         int len = strlen(path);
 257         char *newpath = xmalloc(baselen + len + 1);
 258         struct merge_options *o = context;
 259
 260         memcpy(newpath, base, baselen);
 261         memcpy(newpath + baselen, path, len);
 262         newpath[baselen + len] = '\0';
 263
 264         if (S_ISDIR(mode))
 265                 string_list_insert(&o->current_directory_set, newpath);
 266         else
 267                 string_list_insert(&o->current_file_set, newpath);
 268         free(newpath);
 269
 270         return (S_ISDIR(mode) ? READ_TREE_RECURSIVE : 0);
 271 }
 272
 273 static int get_files_dirs(struct merge_options *o, struct tree *tree)
 274 {
 275         int n;
 276         if (read_tree_recursive(tree, "", 0, 0, NULL, save_files_dirs, o))
 277                 return 0;
 278         n = o->current_file_set.nr + o->current_directory_set.nr;
 279         return n;
 280 }
 281
 282 /*
 283  * Returns an index_entry instance which doesn't have to correspond to
 284  * a real cache entry in Git's index.
 285  */
 286 static struct stage_data *insert_stage_data(const char *path,
 287                 struct tree *o, struct tree *a, struct tree *b,
 288                 struct string_list *entries)
 289 {
 290         struct string_list_item *item;
 291         struct stage_data *e = xcalloc(1, sizeof(struct stage_data));
 292         get_tree_entry(o->object.sha1, path,
 293                         e->stages[1].sha, &e->stages[1].mode);
 294         get_tree_entry(a->object.sha1, path,
 295                         e->stages[2].sha, &e->stages[2].mode);
 296         get_tree_entry(b->object.sha1, path,
 297                         e->stages[3].sha, &e->stages[3].mode);
 298         item = string_list_insert(entries, path);
 299         item->util = e;
 300         return e;
 301 }
 302
 303 /*
 304  * Create a dictionary mapping file names to stage_data objects. The
 305  * dictionary contains one entry for every path with a non-zero stage entry.
 306  */
 307 static struct string_list *get_unmerged(void)
 308 {
 309         struct string_list *unmerged = xcalloc(1, sizeof(struct string_list));
 310         int i;
 311
 312         unmerged->strdup_strings = 1;
 313
 314         for (i = 0; i < active_nr; i++) {
 315                 struct string_list_item *item;
 316                 struct stage_data *e;
 317                 struct cache_entry *ce = active_cache[i];
 318                 if (!ce_stage(ce))
 319                         continue;
 320
 321                 item = string_list_lookup(unmerged, ce->name);
 322                 if (!item) {
 323                         item = string_list_insert(unmerged, ce->name);
 324                         item->util = xcalloc(1, sizeof(struct stage_data));
 325                 }
 326                 e = item->util;
 327                 e->stages[ce_stage(ce)].mode = ce->ce_mode;
 328                 hashcpy(e->stages[ce_stage(ce)].sha, ce->sha1);
 329         }
 330
 331         return unmerged;
 332 }
 333
 334 static int string_list_df_name_compare(const void *a, const void *b)
 335 {
 336         const struct string_list_item *one = a;
 337         const struct string_list_item *two = b;
 338         int onelen = strlen(one->string);
 339         int twolen = strlen(two->string);
 340         /*
 341          * Here we only care that entries for D/F conflicts are
 342          * adjacent, in particular with the file of the D/F conflict
 343          * appearing before files below the corresponding directory.
 344          * The order of the rest of the list is irrelevant for us.
 345          *
 346          * To achieve this, we sort with df_name_compare and provide
 347          * the mode S_IFDIR so that D/F conflicts will sort correctly.
 348          * We use the mode S_IFDIR for everything else for simplicity,
 349          * since in other cases any changes in their order due to
 350          * sorting cause no problems for us.
 351          */
 352         int cmp = df_name_compare(one->string, onelen, S_IFDIR,
 353                                   two->string, twolen, S_IFDIR);
 354         /*
 355          * Now that 'foo' and 'foo/bar' compare equal, we have to make sure
 356          * that 'foo' comes before 'foo/bar'.
 357          */
 358         if (cmp)
 359                 return cmp;
 360         return onelen - twolen;
 361 }
 362
 363 static void record_df_conflict_files(struct merge_options *o,
 364                                      struct string_list *entries)
 365 {
 366         /* If there is a D/F conflict and the file for such a conflict
 367          * currently exist in the working copy, we want to allow it to be
 368          * removed to make room for the corresponding directory if needed.
 369          * The files underneath the directories of such D/F conflicts will
 370          * be processed before the corresponding file involved in the D/F
 371          * conflict.  If the D/F directory ends up being removed by the
 372          * merge, then we won't have to touch the D/F file.  If the D/F
 373          * directory needs to be written to the working copy, then the D/F
 374          * file will simply be removed (in make_room_for_path()) to make
 375          * room for the necessary paths.  Note that if both the directory
 376          * and the file need to be present, then the D/F file will be
 377          * reinstated with a new unique name at the time it is processed.
 378          */
 379         const char *last_file = NULL;
 380         int last_len = 0;
 381         int i;
 382
 383         /*
 384          * If we're merging merge-bases, we don't want to bother with
 385          * any working directory changes.
 386          */
 387         if (o->call_depth)
 388                 return;
 389
 390         /* Ensure D/F conflicts are adjacent in the entries list. */
 391         qsort(entries->items, entries->nr, sizeof(*entries->items),
 392               string_list_df_name_compare);
 393
 394         string_list_clear(&o->df_conflict_file_set, 1);
 395         for (i = 0; i < entries->nr; i++) {
 396                 const char *path = entries->items[i].string;
 397                 int len = strlen(path);
 398                 struct stage_data *e = entries->items[i].util;
 399
 400                 /*
 401                  * Check if last_file & path correspond to a D/F conflict;
 402                  * i.e. whether path is last_file+'/'+<something>.
 403                  * If so, record that it's okay to remove last_file to make
 404                  * room for path and friends if needed.
 405                  */
 406                 if (last_file &&
 407                     len > last_len &&
 408                     memcmp(path, last_file, last_len) == 0 &&
 409                     path[last_len] == '/') {
 410                         string_list_insert(&o->df_conflict_file_set, last_file);
 411                 }
 412
 413                 /*
 414                  * Determine whether path could exist as a file in the
 415                  * working directory as a possible D/F conflict.  This
 416                  * will only occur when it exists in stage 2 as a
 417                  * file.
 418                  */
 419                 if (S_ISREG(e->stages[2].mode) || S_ISLNK(e->stages[2].mode)) {
 420                         last_file = path;
 421                         last_len = len;
 422                 } else {
 423                         last_file = NULL;
 424                 }
 425         }
 426 }
 427
 428 struct rename {
 429         struct diff_filepair *pair;
 430         struct stage_data *src_entry;
 431         struct stage_data *dst_entry;
 432         unsigned processed:1;
 433 };
 434
 435 /*
 436  * Get information of all renames which occurred between 'o_tree' and
 437  * 'tree'. We need the three trees in the merge ('o_tree', 'a_tree' and
 438  * 'b_tree') to be able to associate the correct cache entries with
 439  * the rename information. 'tree' is always equal to either a_tree or b_tree.
 440  */
 441 static struct string_list *get_renames(struct merge_options *o,
 442                                        struct tree *tree,
 443                                        struct tree *o_tree,
 444                                        struct tree *a_tree,
 445                                        struct tree *b_tree,
 446                                        struct string_list *entries)
 447 {
 448         int i;
 449         struct string_list *renames;
 450         struct diff_options opts;
 451
 452         renames = xcalloc(1, sizeof(struct string_list));
 453         diff_setup(&opts);
 454         DIFF_OPT_SET(&opts, RECURSIVE);
 455         opts.detect_rename = DIFF_DETECT_RENAME;
 456         opts.rename_limit = o->merge_rename_limit >= 0 ? o->merge_rename_limit :
 457                             o->diff_rename_limit >= 0 ? o->diff_rename_limit :
 458                             1000;
 459         opts.rename_score = o->rename_score;
 460         opts.show_rename_progress = o->show_rename_progress;
 461         opts.output_format = DIFF_FORMAT_NO_OUTPUT;
 462         if (diff_setup_done(&opts) < 0)
 463                 die("diff setup failed");
 464         diff_tree_sha1(o_tree->object.sha1, tree->object.sha1, "", &opts);
 465         diffcore_std(&opts);
 466         if (opts.needed_rename_limit > o->needed_rename_limit)
 467                 o->needed_rename_limit = opts.needed_rename_limit;
 468         for (i = 0; i < diff_queued_diff.nr; ++i) {
 469                 struct string_list_item *item;
 470                 struct rename *re;
 471                 struct diff_filepair *pair = diff_queued_diff.queue[i];
 472                 if (pair->status != 'R') {
 473                         diff_free_filepair(pair);
 474                         continue;
 475                 }
 476                 re = xmalloc(sizeof(*re));
 477                 re->processed = 0;
 478                 re->pair = pair;
 479                 item = string_list_lookup(entries, re->pair->one->path);
 480                 if (!item)
 481                         re->src_entry = insert_stage_data(re->pair->one->path,
 482                                         o_tree, a_tree, b_tree, entries);
 483                 else
 484                         re->src_entry = item->util;
 485
 486                 item = string_list_lookup(entries, re->pair->two->path);
 487                 if (!item)
 488                         re->dst_entry = insert_stage_data(re->pair->two->path,
 489                                         o_tree, a_tree, b_tree, entries);
 490                 else
 491                         re->dst_entry = item->util;
 492                 item = string_list_insert(renames, pair->one->path);
 493                 item->util = re;
 494         }
 495         opts.output_format = DIFF_FORMAT_NO_OUTPUT;
 496         diff_queued_diff.nr = 0;
 497         diff_flush(&opts);
 498         return renames;
 499 }
 500
 501 static int update_stages(const char *path, const struct diff_filespec *o,
 502                          const struct diff_filespec *a,
 503                          const struct diff_filespec *b)
 504 {
 505         int clear = 1;
 506         int options = ADD_CACHE_OK_TO_ADD | ADD_CACHE_SKIP_DFCHECK;
 507         if (clear)
 508                 if (remove_file_from_cache(path))
 509                         return -1;
 510         if (o)
 511                 if (add_cacheinfo(o->mode, o->sha1, path, 1, 0, options))
 512                         return -1;
 513         if (a)
 514                 if (add_cacheinfo(a->mode, a->sha1, path, 2, 0, options))
 515                         return -1;
 516         if (b)
 517                 if (add_cacheinfo(b->mode, b->sha1, path, 3, 0, options))
 518                         return -1;
 519         return 0;
 520 }
 521
 522 static void update_entry(struct stage_data *entry,
 523                          struct diff_filespec *o,
 524                          struct diff_filespec *a,
 525                          struct diff_filespec *b)
 526 {
 527         entry->processed = 0;
 528         entry->stages[1].mode = o->mode;
 529         entry->stages[2].mode = a->mode;
 530         entry->stages[3].mode = b->mode;
 531         hashcpy(entry->stages[1].sha, o->sha1);
 532         hashcpy(entry->stages[2].sha, a->sha1);
 533         hashcpy(entry->stages[3].sha, b->sha1);
 534 }
 535
 536 static int remove_file(struct merge_options *o, int clean,
 537                        const char *path, int no_wd)
 538 {
 539         int update_cache = o->call_depth || clean;
 540         int update_working_directory = !o->call_depth && !no_wd;
 541
 542         if (update_cache) {
 543                 if (remove_file_from_cache(path))
 544                         return -1;
 545         }
 546         if (update_working_directory) {
 547                 if (remove_path(path))
 548                         return -1;
 549         }
 550         return 0;
 551 }
 552
 553 static char *unique_path(struct merge_options *o, const char *path, const char *branch)
 554 {
 555         char *newpath = xmalloc(strlen(path) + 1 + strlen(branch) + 8 + 1);
 556         int suffix = 0;
 557         struct stat st;
 558         char *p = newpath + strlen(path);
 559         strcpy(newpath, path);
 560         *(p++) = '~';
 561         strcpy(p, branch);
 562         for (; *p; ++p)
 563                 if ('/' == *p)
 564                         *p = '_';
 565         while (string_list_has_string(&o->current_file_set, newpath) ||
 566                string_list_has_string(&o->current_directory_set, newpath) ||
 567                lstat(newpath, &st) == 0)
 568                 sprintf(p, "_%d", suffix++);
 569
 570         string_list_insert(&o->current_file_set, newpath);
 571         return newpath;
 572 }
 573
 574 static void flush_buffer(int fd, const char *buf, unsigned long size)
 575 {
 576         while (size > 0) {
 577                 long ret = write_in_full(fd, buf, size);
 578                 if (ret < 0) {
 579                         /* Ignore epipe */
 580                         if (errno == EPIPE)
 581                                 break;
 582                         die_errno("merge-recursive");
 583                 } else if (!ret) {
 584                         die("merge-recursive: disk full?");
 585                 }
 586                 size -= ret;
 587                 buf += ret;
 588         }
 589 }
 590
 591 static int dir_in_way(const char *path, int check_working_copy)
 592 {
 593         int pos, pathlen = strlen(path);
 594         char *dirpath = xmalloc(pathlen + 2);
 595         struct stat st;
 596
 597         strcpy(dirpath, path);
 598         dirpath[pathlen] = '/';
 599         dirpath[pathlen+1] = '\0';
 600
 601         pos = cache_name_pos(dirpath, pathlen+1);
 602
 603         if (pos < 0)
 604                 pos = -1 - pos;
 605         if (pos < active_nr &&
 606             !strncmp(dirpath, active_cache[pos]->name, pathlen+1)) {
 607                 free(dirpath);
 608                 return 1;
 609         }
 610
 611         free(dirpath);
 612         return check_working_copy && !lstat(path, &st) && S_ISDIR(st.st_mode);
 613 }
 614
 615 static int was_tracked(const char *path)
 616 {
 617         int pos = cache_name_pos(path, strlen(path));
 618
 619         if (pos < 0)
 620                 pos = -1 - pos;
 621         while (pos < active_nr &&
 622                !strcmp(path, active_cache[pos]->name)) {
 623                 /*
 624                  * If stage #0, it is definitely tracked.
 625                  * If it has stage #2 then it was tracked
 626                  * before this merge started.  All other
 627                  * cases the path was not tracked.
 628                  */
 629                 switch (ce_stage(active_cache[pos])) {
 630                 case 0:
 631                 case 2:
 632                         return 1;
 633                 }
 634                 pos++;
 635         }
 636         return 0;
 637 }
 638
 639 static int would_lose_untracked(const char *path)
 640 {
 641         return !was_tracked(path) && file_exists(path);
 642 }
 643
 644 static int make_room_for_path(struct merge_options *o, const char *path)
 645 {
 646         int status, i;
 647         const char *msg = "failed to create path '%s'%s";
 648
 649         /* Unlink any D/F conflict files that are in the way */
 650         for (i = 0; i < o->df_conflict_file_set.nr; i++) {
 651                 const char *df_path = o->df_conflict_file_set.items[i].string;
 652                 size_t pathlen = strlen(path);
 653                 size_t df_pathlen = strlen(df_path);
 654                 if (df_pathlen < pathlen &&
 655                     path[df_pathlen] == '/' &&
 656                     strncmp(path, df_path, df_pathlen) == 0) {
 657                         output(o, 3,
 658                                "Removing %s to make room for subdirectory\n",
 659                                df_path);
 660                         unlink(df_path);
 661                         unsorted_string_list_delete_item(&o->df_conflict_file_set,
 662                                                          i, 0);
 663                         break;
 664                 }
 665         }
 666
 667         /* Make sure leading directories are created */
 668         status = safe_create_leading_directories_const(path);
 669         if (status) {
 670                 if (status == -3) {
 671                         /* something else exists */
 672                         error(msg, path, ": perhaps a D/F conflict?");
 673                         return -1;
 674                 }
 675                 die(msg, path, "");
 676         }
 677
 678         /*
 679          * Do not unlink a file in the work tree if we are not
 680          * tracking it.
 681          */
 682         if (would_lose_untracked(path))
 683                 return error("refusing to lose untracked file at '%s'",
 684                              path);
 685
 686         /* Successful unlink is good.. */
 687         if (!unlink(path))
 688                 return 0;
 689         /* .. and so is no existing file */
 690         if (errno == ENOENT)
 691                 return 0;
 692         /* .. but not some other error (who really cares what?) */
 693         return error(msg, path, ": perhaps a D/F conflict?");
 694 }
 695
 696 static void update_file_flags(struct merge_options *o,
 697                               const unsigned char *sha,
 698                               unsigned mode,
 699                               const char *path,
 700                               int update_cache,
 701                               int update_wd)
 702 {
 703         if (o->call_depth)
 704                 update_wd = 0;
 705
 706         if (update_wd) {
 707                 enum object_type type;
 708                 void *buf;
 709                 unsigned long size;
 710
 711                 if (S_ISGITLINK(mode)) {
 712                         /*
 713                          * We may later decide to recursively descend into
 714                          * the submodule directory and update its index
 715                          * and/or work tree, but we do not do that now.
 716                          */
 717                         update_wd = 0;
 718                         goto update_index;
 719                 }
 720
 721                 buf = read_sha1_file(sha, &type, &size);
 722                 if (!buf)
 723                         die("cannot read object %s '%s'", sha1_to_hex(sha), path);
 724                 if (type != OBJ_BLOB)
 725                         die("blob expected for %s '%s'", sha1_to_hex(sha), path);
 726                 if (S_ISREG(mode)) {
 727                         struct strbuf strbuf = STRBUF_INIT;
 728                         if (convert_to_working_tree(path, buf, size, &strbuf)) {
 729                                 free(buf);
 730                                 size = strbuf.len;
 731                                 buf = strbuf_detach(&strbuf, NULL);
 732                         }
 733                 }
 734
 735                 if (make_room_for_path(o, path) < 0) {
 736                         update_wd = 0;
 737                         free(buf);
 738                         goto update_index;
 739                 }
 740                 if (S_ISREG(mode) || (!has_symlinks && S_ISLNK(mode))) {
 741                         int fd;
 742                         if (mode & 0100)
 743                                 mode = 0777;
 744                         else
 745                                 mode = 0666;
 746                         fd = open(path, O_WRONLY | O_TRUNC | O_CREAT, mode);
 747                         if (fd < 0)
 748                                 die_errno("failed to open '%s'", path);
 749                         flush_buffer(fd, buf, size);
 750                         close(fd);
 751                 } else if (S_ISLNK(mode)) {
 752                         char *lnk = xmemdupz(buf, size);
 753                         safe_create_leading_directories_const(path);
 754                         unlink(path);
 755                         if (symlink(lnk, path))
 756                                 die_errno("failed to symlink '%s'", path);
 757                         free(lnk);
 758                 } else
 759                         die("do not know what to do with %06o %s '%s'",
 760                             mode, sha1_to_hex(sha), path);
 761                 free(buf);
 762         }
 763  update_index:
 764         if (update_cache)
 765                 add_cacheinfo(mode, sha, path, 0, update_wd, ADD_CACHE_OK_TO_ADD);
 766 }
 767
 768 static void update_file(struct merge_options *o,
 769                         int clean,
 770                         const unsigned char *sha,
 771                         unsigned mode,
 772                         const char *path)
 773 {
 774         update_file_flags(o, sha, mode, path, o->call_depth || clean, !o->call_depth);
 775 }
 776
 777 /* Low level file merging, update and removal */
 778
 779 struct merge_file_info {
 780         unsigned char sha[20];
 781         unsigned mode;
 782         unsigned clean:1,
 783                  merge:1;
 784 };
 785
 786 static int merge_3way(struct merge_options *o,
 787                       mmbuffer_t *result_buf,
 788                       const struct diff_filespec *one,
 789                       const struct diff_filespec *a,
 790                       const struct diff_filespec *b,
 791                       const char *branch1,
 792                       const char *branch2)
 793 {
 794         mmfile_t orig, src1, src2;
 795         struct ll_merge_options ll_opts = {0};
 796         char *base_name, *name1, *name2;
 797         int merge_status;
 798
 799         ll_opts.renormalize = o->renormalize;
 800         ll_opts.xdl_opts = o->xdl_opts;
 801
 802         if (o->call_depth) {
 803                 ll_opts.virtual_ancestor = 1;
 804                 ll_opts.variant = 0;
 805         } else {
 806                 switch (o->recursive_variant) {
 807                 case MERGE_RECURSIVE_OURS:
 808                         ll_opts.variant = XDL_MERGE_FAVOR_OURS;
 809                         break;
 810                 case MERGE_RECURSIVE_THEIRS:
 811                         ll_opts.variant = XDL_MERGE_FAVOR_THEIRS;
 812                         break;
 813                 default:
 814                         ll_opts.variant = 0;
 815                         break;
 816                 }
 817         }
 818
 819         if (strcmp(a->path, b->path) ||
 820             (o->ancestor != NULL && strcmp(a->path, one->path) != 0)) {
 821                 base_name = o->ancestor == NULL ? NULL :
 822                         xstrdup(mkpath("%s:%s", o->ancestor, one->path));
 823                 name1 = xstrdup(mkpath("%s:%s", branch1, a->path));
 824                 name2 = xstrdup(mkpath("%s:%s", branch2, b->path));
 825         } else {
 826                 base_name = o->ancestor == NULL ? NULL :
 827                         xstrdup(mkpath("%s", o->ancestor));
 828                 name1 = xstrdup(mkpath("%s", branch1));
 829                 name2 = xstrdup(mkpath("%s", branch2));
 830         }
 831
 832         read_mmblob(&orig, one->sha1);
 833         read_mmblob(&src1, a->sha1);
 834         read_mmblob(&src2, b->sha1);
 835
 836         merge_status = ll_merge(result_buf, a->path, &orig, base_name,
 837                                 &src1, name1, &src2, name2, &ll_opts);
 838
 839         free(name1);
 840         free(name2);
 841         free(orig.ptr);
 842         free(src1.ptr);
 843         free(src2.ptr);
 844         return merge_status;
 845 }
 846
 847 static struct merge_file_info merge_file(struct merge_options *o,
 848                                          const struct diff_filespec *one,
 849                                          const struct diff_filespec *a,
 850                                          const struct diff_filespec *b,
 851                                          const char *branch1,
 852                                          const char *branch2)
 853 {
 854         struct merge_file_info result;
 855         result.merge = 0;
 856         result.clean = 1;
 857
 858         if ((S_IFMT & a->mode) != (S_IFMT & b->mode)) {
 859                 result.clean = 0;
 860                 if (S_ISREG(a->mode)) {
 861                         result.mode = a->mode;
 862                         hashcpy(result.sha, a->sha1);
 863                 } else {
 864                         result.mode = b->mode;
 865                         hashcpy(result.sha, b->sha1);
 866                 }
 867         } else {
 868                 if (!sha_eq(a->sha1, one->sha1) && !sha_eq(b->sha1, one->sha1))
 869                         result.merge = 1;
 870
 871                 /*
 872                  * Merge modes
 873                  */
 874                 if (a->mode == b->mode || a->mode == one->mode)
 875                         result.mode = b->mode;
 876                 else {
 877                         result.mode = a->mode;
 878                         if (b->mode != one->mode) {
 879                                 result.clean = 0;
 880                                 result.merge = 1;
 881                         }
 882                 }
 883
 884                 if (sha_eq(a->sha1, b->sha1) || sha_eq(a->sha1, one->sha1))
 885                         hashcpy(result.sha, b->sha1);
 886                 else if (sha_eq(b->sha1, one->sha1))
 887                         hashcpy(result.sha, a->sha1);
 888                 else if (S_ISREG(a->mode)) {
 889                         mmbuffer_t result_buf;
 890                         int merge_status;
 891
 892                         merge_status = merge_3way(o, &result_buf, one, a, b,
 893                                                   branch1, branch2);
 894
 895                         if ((merge_status < 0) || !result_buf.ptr)
 896                                 die("Failed to execute internal merge");
 897
 898                         if (write_sha1_file(result_buf.ptr, result_buf.size,
 899                                             blob_type, result.sha))
 900                                 die("Unable to add %s to database",
 901                                     a->path);
 902
 903                         free(result_buf.ptr);
 904                         result.clean = (merge_status == 0);
 905                 } else if (S_ISGITLINK(a->mode)) {
 906                         result.clean = merge_submodule(result.sha, one->path, one->sha1,
 907                                                        a->sha1, b->sha1);
 908                 } else if (S_ISLNK(a->mode)) {
 909                         hashcpy(result.sha, a->sha1);
 910
 911                         if (!sha_eq(a->sha1, b->sha1))
 912                                 result.clean = 0;
 913                 } else {
 914                         die("unsupported object type in the tree");
 915                 }
 916         }
 917
 918         return result;
 919 }
 920
 921 static void conflict_rename_delete(struct merge_options *o,
 922                                    struct diff_filepair *pair,
 923                                    const char *rename_branch,
 924                                    const char *other_branch)
 925 {
 926         char *dest_name = pair->two->path;
 927         int df_conflict = 0;
 928
 929         output(o, 1, "CONFLICT (rename/delete): Rename %s->%s in %s "
 930                "and deleted in %s",
 931                pair->one->path, pair->two->path, rename_branch,
 932                other_branch);
 933         if (!o->call_depth)
 934                 update_stages(dest_name, NULL,
 935                               rename_branch == o->branch1 ? pair->two : NULL,
 936                               rename_branch == o->branch1 ? NULL : pair->two);
 937         if (dir_in_way(dest_name, !o->call_depth)) {
 938                 dest_name = unique_path(o, dest_name, rename_branch);
 939                 df_conflict = 1;
 940         }
 941         update_file(o, 0, pair->two->sha1, pair->two->mode, dest_name);
 942         if (df_conflict)
 943                 free(dest_name);
 944 }
 945
 946 static void conflict_rename_rename_1to2(struct merge_options *o,
 947                                         struct diff_filepair *pair1,
 948                                         const char *branch1,
 949                                         struct diff_filepair *pair2,
 950                                         const char *branch2)
 951 {
 952         /* One file was renamed in both branches, but to different names. */
 953         char *del[2];
 954         int delp = 0;
 955         const char *src      = pair1->one->path;
 956         const char *ren1_dst = pair1->two->path;
 957         const char *ren2_dst = pair2->two->path;
 958         const char *dst_name1 = ren1_dst;
 959         const char *dst_name2 = ren2_dst;
 960
 961         output(o, 1, "CONFLICT (rename/rename): "
 962                "Rename \"%s\"->\"%s\" in branch \"%s\" "
 963                "rename \"%s\"->\"%s\" in \"%s\"%s",
 964                src, pair1->two->path, branch1,
 965                src, pair2->two->path, branch2,
 966                o->call_depth ? " (left unresolved)" : "");
 967         if (o->call_depth) {
 968                 /*
 969                  * FIXME: Why remove file from cache, and then
 970                  * immediately readd it?  Why not just overwrite using
 971                  * update_file only?  Also...this is buggy for
 972                  * rename/add-source situations...
 973                  */
 974                 remove_file_from_cache(src);
 975                 update_file(o, 0, pair1->one->sha1, pair1->one->mode, src);
 976         }
 977
 978         if (dir_in_way(ren1_dst, !o->call_depth)) {
 979                 dst_name1 = del[delp++] = unique_path(o, ren1_dst, branch1);
 980                 output(o, 1, "%s is a directory in %s adding as %s instead",
 981                        ren1_dst, branch2, dst_name1);
 982         }
 983         if (dir_in_way(ren2_dst, !o->call_depth)) {
 984                 dst_name2 = del[delp++] = unique_path(o, ren2_dst, branch2);
 985                 output(o, 1, "%s is a directory in %s adding as %s instead",
 986                        ren2_dst, branch1, dst_name2);
 987         }
 988         if (o->call_depth) {
 989                 remove_file_from_cache(dst_name1);
 990                 remove_file_from_cache(dst_name2);
 991                 /*
 992                  * Uncomment to leave the conflicting names in the resulting tree
 993                  *
 994                  * update_file(o, 0, pair1->two->sha1, pair1->two->mode, dst_name1);
 995                  * update_file(o, 0, pair2->two->sha1, pair2->two->mode, dst_name2);
 996                  */
 997         } else {
 998                 update_stages(ren1_dst, NULL, pair1->two, NULL);
 999                 update_stages(ren2_dst, NULL, NULL, pair2->two);
1000
1001                 update_file(o, 0, pair1->two->sha1, pair1->two->mode, dst_name1);
1002                 update_file(o, 0, pair2->two->sha1, pair2->two->mode, dst_name2);
1003         }
1004         while (delp--)
1005                 free(del[delp]);
1006 }
1007
1008 static void conflict_rename_rename_2to1(struct merge_options *o,
1009                                         struct rename *ren1,
1010                                         const char *branch1,
1011                                         struct rename *ren2,
1012                                         const char *branch2)
1013 {
1014         char *path = ren1->pair->two->path; /* same as ren2->pair->two->path */
1015         /* Two files were renamed to the same thing. */
1016         if (o->call_depth) {
1017                 struct merge_file_info mfi;
1018                 struct diff_filespec one, a, b;
1019
1020                 one.path = a.path = b.path = path;
1021                 hashcpy(one.sha1, null_sha1);
1022                 one.mode = 0;
1023                 hashcpy(a.sha1, ren1->pair->two->sha1);
1024                 a.mode = ren1->pair->two->mode;
1025                 hashcpy(b.sha1, ren2->pair->two->sha1);
1026                 b.mode = ren2->pair->two->mode;
1027                 mfi = merge_file(o, &one, &a, &b, branch1, branch2);
1028                 output(o, 1, "Adding merged %s", path);
1029                 update_file(o, 0, mfi.sha, mfi.mode, path);
1030         } else {
1031                 char *new_path1 = unique_path(o, path, branch1);
1032                 char *new_path2 = unique_path(o, path, branch2);
1033                 output(o, 1, "Renaming %s to %s and %s to %s instead",
1034                        ren1->pair->one->path, new_path1,
1035                        ren2->pair->one->path, new_path2);
1036                 remove_file(o, 0, path, 0);
1037                 update_file(o, 0, ren1->pair->two->sha1, ren1->pair->two->mode,
1038                             new_path1);
1039                 update_file(o, 0, ren2->pair->two->sha1, ren2->pair->two->mode,
1040                             new_path2);
1041                 free(new_path2);
1042                 free(new_path1);
1043         }
1044 }
1045
1046 static int process_renames(struct merge_options *o,
1047                            struct string_list *a_renames,
1048                            struct string_list *b_renames)
1049 {
1050         int clean_merge = 1, i, j;
1051         struct string_list a_by_dst = STRING_LIST_INIT_NODUP;
1052         struct string_list b_by_dst = STRING_LIST_INIT_NODUP;
1053         const struct rename *sre;
1054
1055         for (i = 0; i < a_renames->nr; i++) {
1056                 sre = a_renames->items[i].util;
1057                 string_list_insert(&a_by_dst, sre->pair->two->path)->util
1058                         = (void *)sre;
1059         }
1060         for (i = 0; i < b_renames->nr; i++) {
1061                 sre = b_renames->items[i].util;
1062                 string_list_insert(&b_by_dst, sre->pair->two->path)->util
1063                         = (void *)sre;
1064         }
1065
1066         for (i = 0, j = 0; i < a_renames->nr || j < b_renames->nr;) {
1067                 struct string_list *renames1, *renames2Dst;
1068                 struct rename *ren1 = NULL, *ren2 = NULL;
1069                 const char *branch1, *branch2;
1070                 const char *ren1_src, *ren1_dst;
1071
1072                 if (i >= a_renames->nr) {
1073                         ren2 = b_renames->items[j++].util;
1074                 } else if (j >= b_renames->nr) {
1075                         ren1 = a_renames->items[i++].util;
1076                 } else {
1077                         int compare = strcmp(a_renames->items[i].string,
1078                                              b_renames->items[j].string);
1079                         if (compare <= 0)
1080                                 ren1 = a_renames->items[i++].util;
1081                         if (compare >= 0)
1082                                 ren2 = b_renames->items[j++].util;
1083                 }
1084
1085                 /* TODO: refactor, so that 1/2 are not needed */
1086                 if (ren1) {
1087                         renames1 = a_renames;
1088                         renames2Dst = &b_by_dst;
1089                         branch1 = o->branch1;
1090                         branch2 = o->branch2;
1091                 } else {
1092                         struct rename *tmp;
1093                         renames1 = b_renames;
1094                         renames2Dst = &a_by_dst;
1095                         branch1 = o->branch2;
1096                         branch2 = o->branch1;
1097                         tmp = ren2;
1098                         ren2 = ren1;
1099                         ren1 = tmp;
1100                 }
1101
1102                 ren1->dst_entry->processed = 1;
1103                 /* BUG: We should only mark src_entry as processed if we
1104                  * are not dealing with a rename + add-source case.
1105                  */
1106                 ren1->src_entry->processed = 1;
1107
1108                 if (ren1->processed)
1109                         continue;
1110                 ren1->processed = 1;
1111
1112                 ren1_src = ren1->pair->one->path;
1113                 ren1_dst = ren1->pair->two->path;
1114
1115                 if (ren2) {
1116                         const char *ren2_src = ren2->pair->one->path;
1117                         const char *ren2_dst = ren2->pair->two->path;
1118                         enum rename_type rename_type;
1119                         /* Renamed in 1 and renamed in 2 */
1120                         if (strcmp(ren1_src, ren2_src) != 0)
1121                                 die("ren1.src != ren2.src");
1122                         ren2->dst_entry->processed = 1;
1123                         ren2->processed = 1;
1124                         if (strcmp(ren1_dst, ren2_dst) != 0) {
1125                                 rename_type = RENAME_ONE_FILE_TO_TWO;
1126                         } else {
1127                                 rename_type = RENAME_ONE_FILE_TO_ONE;
1128                                 /* BUG: We should only remove ren1_src in
1129                                  * the base stage (think of rename +
1130                                  * add-source cases).
1131                                  */
1132                                 remove_file(o, 1, ren1_src, 1);
1133                                 update_entry(ren1->dst_entry,
1134                                              ren1->pair->one,
1135                                              ren1->pair->two,
1136                                              ren2->pair->two);
1137                         }
1138                         setup_rename_conflict_info(rename_type,
1139                                                    ren1->pair,
1140                                                    ren2->pair,
1141                                                    branch1,
1142                                                    branch2,
1143                                                    ren1->dst_entry,
1144                                                    ren2->dst_entry);
1145                 } else {
1146                         /* Renamed in 1, maybe changed in 2 */
1147                         struct string_list_item *item;
1148                         /* we only use sha1 and mode of these */
1149                         struct diff_filespec src_other, dst_other;
1150                         int try_merge;
1151
1152                         /*
1153                          * unpack_trees loads entries from common-commit
1154                          * into stage 1, from head-commit into stage 2, and
1155                          * from merge-commit into stage 3.  We keep track
1156                          * of which side corresponds to the rename.
1157                          */
1158                         int renamed_stage = a_renames == renames1 ? 2 : 3;
1159                         int other_stage =   a_renames == renames1 ? 3 : 2;
1160
1161                         /* BUG: We should only remove ren1_src in the base
1162                          * stage and in other_stage (think of rename +
1163                          * add-source case).
1164                          */
1165                         remove_file(o, 1, ren1_src,
1166                                     renamed_stage == 2 || !was_tracked(ren1_src));
1167
1168                         hashcpy(src_other.sha1, ren1->src_entry->stages[other_stage].sha);
1169                         src_other.mode = ren1->src_entry->stages[other_stage].mode;
1170                         hashcpy(dst_other.sha1, ren1->dst_entry->stages[other_stage].sha);
1171                         dst_other.mode = ren1->dst_entry->stages[other_stage].mode;
1172                         try_merge = 0;
1173
1174                         if (sha_eq(src_other.sha1, null_sha1)) {
1175                                 setup_rename_conflict_info(RENAME_DELETE,
1176                                                            ren1->pair,
1177                                                            NULL,
1178                                                            branch1,
1179                                                            branch2,
1180                                                            ren1->dst_entry,
1181                                                            NULL);
1182                         } else if ((item = string_list_lookup(renames2Dst, ren1_dst))) {
1183                                 char *ren2_src, *ren2_dst;
1184                                 ren2 = item->util;
1185                                 ren2_src = ren2->pair->one->path;
1186                                 ren2_dst = ren2->pair->two->path;
1187
1188                                 clean_merge = 0;
1189                                 ren2->processed = 1;
1190                                 remove_file(o, 1, ren2_src,
1191                                             renamed_stage == 3 || would_lose_untracked(ren1_src));
1192
1193                                 output(o, 1, "CONFLICT (rename/rename): "
1194                                        "Rename %s->%s in %s. "
1195                                        "Rename %s->%s in %s",
1196                                        ren1_src, ren1_dst, branch1,
1197                                        ren2_src, ren2_dst, branch2);
1198                                 conflict_rename_rename_2to1(o, ren1, branch1, ren2, branch2);
1199                         } else if ((dst_other.mode == ren1->pair->two->mode) &&
1200                                    sha_eq(dst_other.sha1, ren1->pair->two->sha1)) {
1201                                 /* Added file on the other side
1202                                    identical to the file being
1203                                    renamed: clean merge */
1204                                 update_file(o, 1, ren1->pair->two->sha1, ren1->pair->two->mode, ren1_dst);
1205                         } else if (!sha_eq(dst_other.sha1, null_sha1)) {
1206                                 clean_merge = 0;
1207                                 try_merge = 1;
1208                                 output(o, 1, "CONFLICT (rename/add): Rename %s->%s in %s. "
1209                                        "%s added in %s",
1210                                        ren1_src, ren1_dst, branch1,
1211                                        ren1_dst, branch2);
1212                                 if (o->call_depth) {
1213                                         struct merge_file_info mfi;
1214                                         struct diff_filespec one, a, b;
1215
1216                                         one.path = a.path = b.path =
1217                                                 (char *)ren1_dst;
1218                                         hashcpy(one.sha1, null_sha1);
1219                                         one.mode = 0;
1220                                         hashcpy(a.sha1, ren1->pair->two->sha1);
1221                                         a.mode = ren1->pair->two->mode;
1222                                         hashcpy(b.sha1, dst_other.sha1);
1223                                         b.mode = dst_other.mode;
1224                                         mfi = merge_file(o, &one, &a, &b,
1225                                                          branch1,
1226                                                          branch2);
1227                                         output(o, 1, "Adding merged %s", ren1_dst);
1228                                         update_file(o, 0,
1229                                                     mfi.sha,
1230                                                     mfi.mode,
1231                                                     ren1_dst);
1232                                         try_merge = 0;
1233                                 } else {
1234                                         char *new_path = unique_path(o, ren1_dst, branch2);
1235                                         output(o, 1, "Adding as %s instead", new_path);
1236                                         update_file(o, 0, dst_other.sha1, dst_other.mode, new_path);
1237                                         free(new_path);
1238                                 }
1239                         } else
1240                                 try_merge = 1;
1241
1242                         if (try_merge) {
1243                                 struct diff_filespec *one, *a, *b;
1244                                 src_other.path = (char *)ren1_src;
1245
1246                                 one = ren1->pair->one;
1247                                 if (a_renames == renames1) {
1248                                         a = ren1->pair->two;
1249                                         b = &src_other;
1250                                 } else {
1251                                         b = ren1->pair->two;
1252                                         a = &src_other;
1253                                 }
1254                                 update_entry(ren1->dst_entry, one, a, b);
1255                                 setup_rename_conflict_info(RENAME_NORMAL,
1256                                                            ren1->pair,
1257                                                            NULL,
1258                                                            branch1,
1259                                                            NULL,
1260                                                            ren1->dst_entry,
1261                                                            NULL);
1262                         }
1263                 }
1264         }
1265         string_list_clear(&a_by_dst, 0);
1266         string_list_clear(&b_by_dst, 0);
1267
1268         return clean_merge;
1269 }
1270
1271 static unsigned char *stage_sha(const unsigned char *sha, unsigned mode)
1272 {
1273         return (is_null_sha1(sha) || mode == 0) ? NULL: (unsigned char *)sha;
1274 }
1275
1276 static int read_sha1_strbuf(const unsigned char *sha1, struct strbuf *dst)
1277 {
1278         void *buf;
1279         enum object_type type;
1280         unsigned long size;
1281         buf = read_sha1_file(sha1, &type, &size);
1282         if (!buf)
1283                 return error("cannot read object %s", sha1_to_hex(sha1));
1284         if (type != OBJ_BLOB) {
1285                 free(buf);
1286                 return error("object %s is not a blob", sha1_to_hex(sha1));
1287         }
1288         strbuf_attach(dst, buf, size, size + 1);
1289         return 0;
1290 }
1291
1292 static int blob_unchanged(const unsigned char *o_sha,
1293                           const unsigned char *a_sha,
1294                           int renormalize, const char *path)
1295 {
1296         struct strbuf o = STRBUF_INIT;
1297         struct strbuf a = STRBUF_INIT;
1298         int ret = 0; /* assume changed for safety */
1299
1300         if (sha_eq(o_sha, a_sha))
1301                 return 1;
1302         if (!renormalize)
1303                 return 0;
1304
1305         assert(o_sha && a_sha);
1306         if (read_sha1_strbuf(o_sha, &o) || read_sha1_strbuf(a_sha, &a))
1307                 goto error_return;
1308         /*
1309          * Note: binary | is used so that both renormalizations are
1310          * performed.  Comparison can be skipped if both files are
1311          * unchanged since their sha1s have already been compared.
1312          */
1313         if (renormalize_buffer(path, o.buf, o.len, &o) |
1314             renormalize_buffer(path, a.buf, o.len, &a))
1315                 ret = (o.len == a.len && !memcmp(o.buf, a.buf, o.len));
1316
1317 error_return:
1318         strbuf_release(&o);
1319         strbuf_release(&a);
1320         return ret;
1321 }
1322
1323 static void handle_delete_modify(struct merge_options *o,
1324                                  const char *path,
1325                                  const char *new_path,
1326                                  unsigned char *a_sha, int a_mode,
1327                                  unsigned char *b_sha, int b_mode)
1328 {
1329         if (!a_sha) {
1330                 output(o, 1, "CONFLICT (delete/modify): %s deleted in %s "
1331                        "and modified in %s. Version %s of %s left in tree%s%s.",
1332                        path, o->branch1,
1333                        o->branch2, o->branch2, path,
1334                        NULL == new_path ? "" : " at ",
1335                        NULL == new_path ? "" : new_path);
1336                 update_file(o, 0, b_sha, b_mode, new_path ? new_path : path);
1337         } else {
1338                 output(o, 1, "CONFLICT (delete/modify): %s deleted in %s "
1339                        "and modified in %s. Version %s of %s left in tree%s%s.",
1340                        path, o->branch2,
1341                        o->branch1, o->branch1, path,
1342                        NULL == new_path ? "" : " at ",
1343                        NULL == new_path ? "" : new_path);
1344                 update_file(o, 0, a_sha, a_mode, new_path ? new_path : path);
1345         }
1346 }
1347
1348 static int merge_content(struct merge_options *o,
1349                          const char *path,
1350                          unsigned char *o_sha, int o_mode,
1351                          unsigned char *a_sha, int a_mode,
1352                          unsigned char *b_sha, int b_mode,
1353                          struct rename_conflict_info *rename_conflict_info)
1354 {
1355         const char *reason = "content";
1356         char *side1 = NULL, *side2 = NULL;
1357         const char *path1 = NULL, *path2 = NULL;
1358         struct merge_file_info mfi;
1359         struct diff_filespec one, a, b;
1360         unsigned df_conflict_remains = 0;
1361
1362         if (!o_sha) {
1363                 reason = "add/add";
1364                 o_sha = (unsigned char *)null_sha1;
1365         }
1366         one.path = a.path = b.path = (char *)path;
1367         hashcpy(one.sha1, o_sha);
1368         one.mode = o_mode;
1369         hashcpy(a.sha1, a_sha);
1370         a.mode = a_mode;
1371         hashcpy(b.sha1, b_sha);
1372         b.mode = b_mode;
1373
1374         if (rename_conflict_info) {
1375                 struct diff_filepair *pair1 = rename_conflict_info->pair1;
1376
1377                 path1 = (o->branch1 == rename_conflict_info->branch1) ?
1378                         pair1->two->path : pair1->one->path;
1379                 /* If rename_conflict_info->pair2 != NULL, we are in
1380                  * RENAME_ONE_FILE_TO_ONE case.  Otherwise, we have a
1381                  * normal rename.
1382                  */
1383                 path2 = (rename_conflict_info->pair2 ||
1384                          o->branch2 == rename_conflict_info->branch1) ?
1385                         pair1->two->path : pair1->one->path;
1386                 side1 = xmalloc(strlen(o->branch1) + strlen(path1) + 2);
1387                 side2 = xmalloc(strlen(o->branch2) + strlen(path2) + 2);
1388                 sprintf(side1, "%s:%s", o->branch1, path1);
1389                 sprintf(side2, "%s:%s", o->branch2, path2);
1390
1391                 if (dir_in_way(path, !o->call_depth))
1392                         df_conflict_remains = 1;
1393         }
1394         mfi = merge_file(o, &one, &a, &b,
1395                          side1 ? side1 : o->branch1, side2 ? side2 : o->branch2);
1396         free(side1);
1397         free(side2);
1398
1399         if (mfi.clean && !df_conflict_remains &&
1400             sha_eq(mfi.sha, a_sha) && mfi.mode == a_mode) {
1401                 int path_renamed_outside_HEAD;
1402                 output(o, 3, "Skipped %s (merged same as existing)", path);
1403                 /*
1404                  * The content merge resulted in the same file contents we
1405                  * already had.  We can return early if those file contents
1406                  * are recorded at the correct path (which may not be true
1407                  * if the merge involves a rename).
1408                  */
1409                 path_renamed_outside_HEAD = !path2 || !strcmp(path, path2);
1410                 if (!path_renamed_outside_HEAD) {
1411                         add_cacheinfo(mfi.mode, mfi.sha, path,
1412                                       0 /*stage*/, 1 /*refresh*/, 0 /*options*/);
1413                         return mfi.clean;
1414                 }
1415         } else
1416                 output(o, 2, "Auto-merging %s", path);
1417
1418         if (!mfi.clean) {
1419                 if (S_ISGITLINK(mfi.mode))
1420                         reason = "submodule";
1421                 output(o, 1, "CONFLICT (%s): Merge conflict in %s",
1422                                 reason, path);
1423                 if (rename_conflict_info && !df_conflict_remains)
1424                         update_stages(path, &one, &a, &b);
1425         }
1426
1427         if (df_conflict_remains) {
1428                 char *new_path;
1429                 if (o->call_depth) {
1430                         remove_file_from_cache(path);
1431                 } else {
1432                         if (!mfi.clean)
1433                                 update_stages(path, &one, &a, &b);
1434                         else {
1435                                 int file_from_stage2 = was_tracked(path);
1436                                 struct diff_filespec merged;
1437                                 hashcpy(merged.sha1, mfi.sha);
1438                                 merged.mode = mfi.mode;
1439
1440                                 update_stages(path, NULL,
1441                                               file_from_stage2 ? &merged : NULL,
1442                                               file_from_stage2 ? NULL : &merged);
1443                         }
1444
1445                 }
1446                 new_path = unique_path(o, path, rename_conflict_info->branch1);
1447                 output(o, 1, "Adding as %s instead", new_path);
1448                 update_file(o, 0, mfi.sha, mfi.mode, new_path);
1449                 free(new_path);
1450                 mfi.clean = 0;
1451         } else {
1452                 update_file(o, mfi.clean, mfi.sha, mfi.mode, path);
1453         }
1454         return mfi.clean;
1455
1456 }
1457
1458 /* Per entry merge function */
1459 static int process_entry(struct merge_options *o,
1460                          const char *path, struct stage_data *entry)
1461 {
1462         /*
1463         printf("processing entry, clean cache: %s\n", index_only ? "yes": "no");
1464         print_index_entry("\tpath: ", entry);
1465         */
1466         int clean_merge = 1;
1467         int normalize = o->renormalize;
1468         unsigned o_mode = entry->stages[1].mode;
1469         unsigned a_mode = entry->stages[2].mode;
1470         unsigned b_mode = entry->stages[3].mode;
1471         unsigned char *o_sha = stage_sha(entry->stages[1].sha, o_mode);
1472         unsigned char *a_sha = stage_sha(entry->stages[2].sha, a_mode);
1473         unsigned char *b_sha = stage_sha(entry->stages[3].sha, b_mode);
1474
1475         entry->processed = 1;
1476         if (entry->rename_conflict_info) {
1477                 struct rename_conflict_info *conflict_info = entry->rename_conflict_info;
1478                 switch (conflict_info->rename_type) {
1479                 case RENAME_NORMAL:
1480                 case RENAME_ONE_FILE_TO_ONE:
1481                         clean_merge = merge_content(o, path,
1482                                                     o_sha, o_mode, a_sha, a_mode, b_sha, b_mode,
1483                                                     conflict_info);
1484                         break;
1485                 case RENAME_DELETE:
1486                         clean_merge = 0;
1487                         conflict_rename_delete(o, conflict_info->pair1,
1488                                                conflict_info->branch1,
1489                                                conflict_info->branch2);
1490                         break;
1491                 case RENAME_ONE_FILE_TO_TWO:
1492                         clean_merge = 0;
1493                         conflict_rename_rename_1to2(o, conflict_info->pair1,
1494                                                     conflict_info->branch1,
1495                                                     conflict_info->pair2,
1496                                                     conflict_info->branch2);
1497                         break;
1498                 default:
1499                         entry->processed = 0;
1500                         break;
1501                 }
1502         } else if (o_sha && (!a_sha || !b_sha)) {
1503                 /* Case A: Deleted in one */
1504                 if ((!a_sha && !b_sha) ||
1505                     (!b_sha && blob_unchanged(o_sha, a_sha, normalize, path)) ||
1506                     (!a_sha && blob_unchanged(o_sha, b_sha, normalize, path))) {
1507                         /* Deleted in both or deleted in one and
1508                          * unchanged in the other */
1509                         if (a_sha)
1510                                 output(o, 2, "Removing %s", path);
1511                         /* do not touch working file if it did not exist */
1512                         remove_file(o, 1, path, !a_sha);
1513                 } else {
1514                         /* Modify/delete; deleted side may have put a directory in the way */
1515                         char *renamed = NULL;
1516                         clean_merge = 0;
1517                         if (dir_in_way(path, !o->call_depth)) {
1518                                 renamed = unique_path(o, path, a_sha ? o->branch1 : o->branch2);
1519                         }
1520                         handle_delete_modify(o, path, renamed,
1521                                              a_sha, a_mode, b_sha, b_mode);
1522                         free(renamed);
1523                 }
1524         } else if ((!o_sha && a_sha && !b_sha) ||
1525                    (!o_sha && !a_sha && b_sha)) {
1526                 /* Case B: Added in one. */
1527                 /* [nothing|directory] -> ([nothing|directory], file) */
1528
1529                 const char *add_branch;
1530                 const char *other_branch;
1531                 unsigned mode;
1532                 const unsigned char *sha;
1533                 const char *conf;
1534
1535                 if (a_sha) {
1536                         add_branch = o->branch1;
1537                         other_branch = o->branch2;
1538                         mode = a_mode;
1539                         sha = a_sha;
1540                         conf = "file/directory";
1541                 } else {
1542                         add_branch = o->branch2;
1543                         other_branch = o->branch1;
1544                         mode = b_mode;
1545                         sha = b_sha;
1546                         conf = "directory/file";
1547                 }
1548                 if (dir_in_way(path, !o->call_depth)) {
1549                         char *new_path = unique_path(o, path, add_branch);
1550                         clean_merge = 0;
1551                         output(o, 1, "CONFLICT (%s): There is a directory with name %s in %s. "
1552                                "Adding %s as %s",
1553                                conf, path, other_branch, path, new_path);
1554                         if (o->call_depth)
1555                                 remove_file_from_cache(path);
1556                         update_file(o, 0, sha, mode, new_path);
1557                         if (o->call_depth)
1558                                 remove_file_from_cache(path);
1559                         free(new_path);
1560                 } else {
1561                         output(o, 2, "Adding %s", path);
1562                         update_file(o, 1, sha, mode, path);
1563                 }
1564         } else if (a_sha && b_sha) {
1565                 /* Case C: Added in both (check for same permissions) and */
1566                 /* case D: Modified in both, but differently. */
1567                 clean_merge = merge_content(o, path,
1568                                             o_sha, o_mode, a_sha, a_mode, b_sha, b_mode,
1569                                             NULL);
1570         } else if (!o_sha && !a_sha && !b_sha) {
1571                 /*
1572                  * this entry was deleted altogether. a_mode == 0 means
1573                  * we had that path and want to actively remove it.
1574                  */
1575                 remove_file(o, 1, path, !a_mode);
1576         } else
1577                 die("Fatal merge failure, shouldn't happen.");
1578
1579         return clean_merge;
1580 }
1581
1582 int merge_trees(struct merge_options *o,
1583                 struct tree *head,
1584                 struct tree *merge,
1585                 struct tree *common,
1586                 struct tree **result)
1587 {
1588         int code, clean;
1589
1590         if (o->subtree_shift) {
1591                 merge = shift_tree_object(head, merge, o->subtree_shift);
1592                 common = shift_tree_object(head, common, o->subtree_shift);
1593         }
1594
1595         if (sha_eq(common->object.sha1, merge->object.sha1)) {
1596                 output(o, 0, "Already up-to-date!");
1597                 *result = head;
1598                 return 1;
1599         }
1600
1601         code = git_merge_trees(o->call_depth, common, head, merge);
1602
1603         if (code != 0) {
1604                 if (show(o, 4) || o->call_depth)
1605                         die("merging of trees %s and %s failed",
1606                             sha1_to_hex(head->object.sha1),
1607                             sha1_to_hex(merge->object.sha1));
1608                 else
1609                         exit(128);
1610         }
1611
1612         if (unmerged_cache()) {
1613                 struct string_list *entries, *re_head, *re_merge;
1614                 int i;
1615                 string_list_clear(&o->current_file_set, 1);
1616                 string_list_clear(&o->current_directory_set, 1);
1617                 get_files_dirs(o, head);
1618                 get_files_dirs(o, merge);
1619
1620                 entries = get_unmerged();
1621                 record_df_conflict_files(o, entries);
1622                 re_head  = get_renames(o, head, common, head, merge, entries);
1623                 re_merge = get_renames(o, merge, common, head, merge, entries);
1624                 clean = process_renames(o, re_head, re_merge);
1625                 for (i = entries->nr-1; 0 <= i; i--) {
1626                         const char *path = entries->items[i].string;
1627                         struct stage_data *e = entries->items[i].util;
1628                         if (!e->processed
1629                                 && !process_entry(o, path, e))
1630                                 clean = 0;
1631                 }
1632                 for (i = 0; i < entries->nr; i++) {
1633                         struct stage_data *e = entries->items[i].util;
1634                         if (!e->processed)
1635                                 die("Unprocessed path??? %s",
1636                                     entries->items[i].string);
1637                 }
1638
1639                 string_list_clear(re_merge, 0);
1640                 string_list_clear(re_head, 0);
1641                 string_list_clear(entries, 1);
1642
1643         }
1644         else
1645                 clean = 1;
1646
1647         if (o->call_depth)
1648                 *result = write_tree_from_memory(o);
1649
1650         return clean;
1651 }
1652
1653 static struct commit_list *reverse_commit_list(struct commit_list *list)
1654 {
1655         struct commit_list *next = NULL, *current, *backup;
1656         for (current = list; current; current = backup) {
1657                 backup = current->next;
1658                 current->next = next;
1659                 next = current;
1660         }
1661         return next;
1662 }
1663
1664 /*
1665  * Merge the commits h1 and h2, return the resulting virtual
1666  * commit object and a flag indicating the cleanness of the merge.
1667  */
1668 int merge_recursive(struct merge_options *o,
1669                     struct commit *h1,
1670                     struct commit *h2,
1671                     struct commit_list *ca,
1672                     struct commit **result)
1673 {
1674         struct commit_list *iter;
1675         struct commit *merged_common_ancestors;
1676         struct tree *mrtree = mrtree;
1677         int clean;
1678
1679         if (show(o, 4)) {
1680                 output(o, 4, "Merging:");
1681                 output_commit_title(o, h1);
1682                 output_commit_title(o, h2);
1683         }
1684
1685         if (!ca) {
1686                 ca = get_merge_bases(h1, h2, 1);
1687                 ca = reverse_commit_list(ca);
1688         }
1689
1690         if (show(o, 5)) {
1691                 output(o, 5, "found %u common ancestor(s):", commit_list_count(ca));
1692                 for (iter = ca; iter; iter = iter->next)
1693                         output_commit_title(o, iter->item);
1694         }
1695
1696         merged_common_ancestors = pop_commit(&ca);
1697         if (merged_common_ancestors == NULL) {
1698                 /* if there is no common ancestor, make an empty tree */
1699                 struct tree *tree = xcalloc(1, sizeof(struct tree));
1700
1701                 tree->object.parsed = 1;
1702                 tree->object.type = OBJ_TREE;
1703                 pretend_sha1_file(NULL, 0, OBJ_TREE, tree->object.sha1);
1704                 merged_common_ancestors = make_virtual_commit(tree, "ancestor");
1705         }
1706
1707         for (iter = ca; iter; iter = iter->next) {
1708                 const char *saved_b1, *saved_b2;
1709                 o->call_depth++;
1710                 /*
1711                  * When the merge fails, the result contains files
1712                  * with conflict markers. The cleanness flag is
1713                  * ignored, it was never actually used, as result of
1714                  * merge_trees has always overwritten it: the committed
1715                  * "conflicts" were already resolved.
1716                  */
1717                 discard_cache();
1718                 saved_b1 = o->branch1;
1719                 saved_b2 = o->branch2;
1720                 o->branch1 = "Temporary merge branch 1";
1721                 o->branch2 = "Temporary merge branch 2";
1722                 merge_recursive(o, merged_common_ancestors, iter->item,
1723                                 NULL, &merged_common_ancestors);
1724                 o->branch1 = saved_b1;
1725                 o->branch2 = saved_b2;
1726                 o->call_depth--;
1727
1728                 if (!merged_common_ancestors)
1729                         die("merge returned no commit");
1730         }
1731
1732         discard_cache();
1733         if (!o->call_depth)
1734                 read_cache();
1735
1736         o->ancestor = "merged common ancestors";
1737         clean = merge_trees(o, h1->tree, h2->tree, merged_common_ancestors->tree,
1738                             &mrtree);
1739
1740         if (o->call_depth) {
1741                 *result = make_virtual_commit(mrtree, "merged tree");
1742                 commit_list_insert(h1, &(*result)->parents);
1743                 commit_list_insert(h2, &(*result)->parents->next);
1744         }
1745         flush_output(o);
1746         if (show(o, 2))
1747                 diff_warn_rename_limit("merge.renamelimit",
1748                                        o->needed_rename_limit, 0);
1749         return clean;
1750 }
1751
1752 static struct commit *get_ref(const unsigned char *sha1, const char *name)
1753 {
1754         struct object *object;
1755
1756         object = deref_tag(parse_object(sha1), name, strlen(name));
1757         if (!object)
1758                 return NULL;
1759         if (object->type == OBJ_TREE)
1760                 return make_virtual_commit((struct tree*)object, name);
1761         if (object->type != OBJ_COMMIT)
1762                 return NULL;
1763         if (parse_commit((struct commit *)object))
1764                 return NULL;
1765         return (struct commit *)object;
1766 }
1767
1768 int merge_recursive_generic(struct merge_options *o,
1769                             const unsigned char *head,
1770                             const unsigned char *merge,
1771                             int num_base_list,
1772                             const unsigned char **base_list,
1773                             struct commit **result)
1774 {
1775         int clean, index_fd;
1776         struct lock_file *lock = xcalloc(1, sizeof(struct lock_file));
1777         struct commit *head_commit = get_ref(head, o->branch1);
1778         struct commit *next_commit = get_ref(merge, o->branch2);
1779         struct commit_list *ca = NULL;
1780
1781         if (base_list) {
1782                 int i;
1783                 for (i = 0; i < num_base_list; ++i) {
1784                         struct commit *base;
1785                         if (!(base = get_ref(base_list[i], sha1_to_hex(base_list[i]))))
1786                                 return error("Could not parse object '%s'",
1787                                         sha1_to_hex(base_list[i]));
1788                         commit_list_insert(base, &ca);
1789                 }
1790         }
1791
1792         index_fd = hold_locked_index(lock, 1);
1793         clean = merge_recursive(o, head_commit, next_commit, ca,
1794                         result);
1795         if (active_cache_changed &&
1796                         (write_cache(index_fd, active_cache, active_nr) ||
1797                          commit_locked_index(lock)))
1798                 return error("Unable to write index.");
1799
1800         return clean ? 0 : 1;
1801 }
1802
1803 static int merge_recursive_config(const char *var, const char *value, void *cb)
1804 {
1805         struct merge_options *o = cb;
1806         if (!strcmp(var, "merge.verbosity")) {
1807                 o->verbosity = git_config_int(var, value);
1808                 return 0;
1809         }
1810         if (!strcmp(var, "diff.renamelimit")) {
1811                 o->diff_rename_limit = git_config_int(var, value);
1812                 return 0;
1813         }
1814         if (!strcmp(var, "merge.renamelimit")) {
1815                 o->merge_rename_limit = git_config_int(var, value);
1816                 return 0;
1817         }
1818         return git_xmerge_config(var, value, cb);
1819 }
1820
1821 void init_merge_options(struct merge_options *o)
1822 {
1823         memset(o, 0, sizeof(struct merge_options));
1824         o->verbosity = 2;
1825         o->buffer_output = 1;
1826         o->diff_rename_limit = -1;
1827         o->merge_rename_limit = -1;
1828         o->renormalize = 0;
1829         git_config(merge_recursive_config, o);
1830         if (getenv("GIT_MERGE_VERBOSITY"))
1831                 o->verbosity =
1832                         strtol(getenv("GIT_MERGE_VERBOSITY"), NULL, 10);
1833         if (o->verbosity >= 5)
1834                 o->buffer_output = 0;
1835         strbuf_init(&o->obuf, 0);
1836         memset(&o->current_file_set, 0, sizeof(struct string_list));
1837         o->current_file_set.strdup_strings = 1;
1838         memset(&o->current_directory_set, 0, sizeof(struct string_list));
1839         o->current_directory_set.strdup_strings = 1;
1840         memset(&o->df_conflict_file_set, 0, sizeof(struct string_list));
1841         o->df_conflict_file_set.strdup_strings = 1;
1842 }
1843
1844 int parse_merge_opt(struct merge_options *o, const char *s)
1845 {
1846         if (!s || !*s)
1847                 return -1;
1848         if (!strcmp(s, "ours"))
1849                 o->recursive_variant = MERGE_RECURSIVE_OURS;
1850         else if (!strcmp(s, "theirs"))
1851                 o->recursive_variant = MERGE_RECURSIVE_THEIRS;
1852         else if (!strcmp(s, "subtree"))
1853                 o->subtree_shift = "";
1854         else if (!prefixcmp(s, "subtree="))
1855                 o->subtree_shift = s + strlen("subtree=");
1856         else if (!strcmp(s, "patience"))
1857                 o->xdl_opts |= XDF_PATIENCE_DIFF;
1858         else if (!strcmp(s, "ignore-space-change"))
1859                 o->xdl_opts |= XDF_IGNORE_WHITESPACE_CHANGE;
1860         else if (!strcmp(s, "ignore-all-space"))
1861                 o->xdl_opts |= XDF_IGNORE_WHITESPACE;
1862         else if (!strcmp(s, "ignore-space-at-eol"))
1863                 o->xdl_opts |= XDF_IGNORE_WHITESPACE_AT_EOL;
1864         else if (!strcmp(s, "renormalize"))
1865                 o->renormalize = 1;
1866         else if (!strcmp(s, "no-renormalize"))
1867                 o->renormalize = 0;
1868         else if (!prefixcmp(s, "rename-threshold=")) {
1869                 const char *score = s + strlen("rename-threshold=");
1870                 if ((o->rename_score = parse_rename_score(&score)) == -1 || *score != 0)
1871                         return -1;
1872         }
1873         else
1874                 return -1;
1875         return 0;
1876 }