mingw: enable atomic O_APPEND
[gitweb.git] / merge-recursive.c
index 7ea70c4a9d11dea43ebdad1ba8718410793c263a..f110e1c5ecbd6254b25785ecb49f29cb7bb75ed2 100644 (file)
@@ -23,6 +23,7 @@
 #include "merge-recursive.h"
 #include "dir.h"
 #include "submodule.h"
+#include "revision.h"
 
 struct path_hashmap_entry {
        struct hashmap_entry e;
@@ -162,7 +163,7 @@ static struct commit *make_virtual_commit(struct tree *tree, const char *comment
        struct commit *commit = alloc_commit_node();
 
        set_merge_remote_desc(commit, comment, (struct object *)commit);
-       commit->tree = tree;
+       commit->maybe_tree = tree;
        commit->object.parsed = 1;
        return commit;
 }
@@ -290,7 +291,7 @@ static void output_commit_title(struct merge_options *o, struct commit *commit)
                strbuf_addf(&o->obuf, "virtual %s\n",
                        merge_remote_util(commit)->name);
        else {
-               strbuf_add_unique_abbrev(&o->obuf, commit->object.oid.hash,
+               strbuf_add_unique_abbrev(&o->obuf, &commit->object.oid,
                                         DEFAULT_ABBREV);
                strbuf_addch(&o->obuf, ' ');
                if (parse_commit(commit) != 0)
@@ -316,7 +317,7 @@ static int add_cacheinfo(struct merge_options *o,
 
        ce = make_cache_entry(mode, oid ? oid->hash : null_sha1, path, stage, 0);
        if (!ce)
-               return err(o, _("addinfo_cache failed for path '%s'"), path);
+               return err(o, _("add_cacheinfo failed for path '%s'; merge aborting."), path);
 
        ret = add_cache_entry(ce, options);
        if (refresh) {
@@ -324,7 +325,7 @@ static int add_cacheinfo(struct merge_options *o,
 
                nce = refresh_cache_entry(ce, CE_MATCH_REFRESH | CE_MATCH_IGNORE_MISSING);
                if (!nce)
-                       return err(o, _("addinfo_cache failed for path '%s'"), path);
+                       return err(o, _("add_cacheinfo failed to refresh for path '%s'; merge aborting."), path);
                if (nce != ce)
                        ret = add_cache_entry(nce, options);
        }
@@ -337,13 +338,14 @@ static void init_tree_desc_from_tree(struct tree_desc *desc, struct tree *tree)
        init_tree_desc(desc, tree->buffer, tree->size);
 }
 
-static int git_merge_trees(struct merge_options *o,
-                          struct tree *common,
-                          struct tree *head,
-                          struct tree *merge)
+static int unpack_trees_start(struct merge_options *o,
+                             struct tree *common,
+                             struct tree *head,
+                             struct tree *merge)
 {
        int rc;
        struct tree_desc t[3];
+       struct index_state tmp_index = { NULL };
 
        memset(&o->unpack_opts, 0, sizeof(o->unpack_opts));
        if (o->call_depth)
@@ -354,7 +356,8 @@ static int git_merge_trees(struct merge_options *o,
        o->unpack_opts.head_idx = 2;
        o->unpack_opts.fn = threeway_merge;
        o->unpack_opts.src_index = &the_index;
-       o->unpack_opts.dst_index = &the_index;
+       o->unpack_opts.dst_index = &tmp_index;
+       o->unpack_opts.aggressive = !merge_detect_rename(o);
        setup_unpack_trees_porcelain(&o->unpack_opts, "merge");
 
        init_tree_desc_from_tree(t+0, common);
@@ -362,16 +365,27 @@ static int git_merge_trees(struct merge_options *o,
        init_tree_desc_from_tree(t+2, merge);
 
        rc = unpack_trees(3, t, &o->unpack_opts);
+       cache_tree_free(&active_cache_tree);
+
        /*
-        * unpack_trees NULLifies src_index, but it's used in verify_uptodate,
-        * so set to the new index which will usually have modification
-        * timestamp info copied over.
+        * Update the_index to match the new results, AFTER saving a copy
+        * in o->orig_index.  Update src_index to point to the saved copy.
+        * (verify_uptodate() checks src_index, and the original index is
+        * the one that had the necessary modification timestamps.)
         */
-       o->unpack_opts.src_index = &the_index;
-       cache_tree_free(&active_cache_tree);
+       o->orig_index = the_index;
+       the_index = tmp_index;
+       o->unpack_opts.src_index = &o->orig_index;
+
        return rc;
 }
 
+static void unpack_trees_finish(struct merge_options *o)
+{
+       discard_index(&o->orig_index);
+       clear_unpack_trees_porcelain(&o->unpack_opts);
+}
+
 struct tree *write_tree_from_memory(struct merge_options *o)
 {
        struct tree *result = NULL;
@@ -385,7 +399,7 @@ struct tree *write_tree_from_memory(struct merge_options *o)
                                fprintf(stderr, "BUG: %d %.*s\n", ce_stage(ce),
                                        (int)ce_namelen(ce), ce->name);
                }
-               die("BUG: unmerged index entries in merge-recursive.c");
+               BUG("unmerged index entries in merge-recursive.c");
        }
 
        if (!active_cache_tree)
@@ -402,7 +416,7 @@ struct tree *write_tree_from_memory(struct merge_options *o)
        return result;
 }
 
-static int save_files_dirs(const unsigned char *sha1,
+static int save_files_dirs(const struct object_id *oid,
                struct strbuf *base, const char *path,
                unsigned int mode, int stage, void *context)
 {
@@ -427,16 +441,16 @@ static void get_files_dirs(struct merge_options *o, struct tree *tree)
        read_tree_recursive(tree, "", 0, 0, &match_all, save_files_dirs, o);
 }
 
-static int get_tree_entry_if_blob(const unsigned char *tree,
+static int get_tree_entry_if_blob(const struct object_id *tree,
                                  const char *path,
-                                 unsigned char *hashy,
+                                 struct object_id *hashy,
                                  unsigned int *mode_o)
 {
        int ret;
 
        ret = get_tree_entry(tree, path, hashy, mode_o);
        if (S_ISDIR(*mode_o)) {
-               hashcpy(hashy, null_sha1);
+               oidcpy(hashy, &null_oid);
                *mode_o = 0;
        }
        return ret;
@@ -452,12 +466,12 @@ static struct stage_data *insert_stage_data(const char *path,
 {
        struct string_list_item *item;
        struct stage_data *e = xcalloc(1, sizeof(struct stage_data));
-       get_tree_entry_if_blob(o->object.oid.hash, path,
-                              e->stages[1].oid.hash, &e->stages[1].mode);
-       get_tree_entry_if_blob(a->object.oid.hash, path,
-                              e->stages[2].oid.hash, &e->stages[2].mode);
-       get_tree_entry_if_blob(b->object.oid.hash, path,
-                              e->stages[3].oid.hash, &e->stages[3].mode);
+       get_tree_entry_if_blob(&o->object.oid, path,
+                              &e->stages[1].oid, &e->stages[1].mode);
+       get_tree_entry_if_blob(&a->object.oid, path,
+                              &e->stages[2].oid, &e->stages[2].mode);
+       get_tree_entry_if_blob(&b->object.oid, path,
+                              &e->stages[3].oid, &e->stages[3].mode);
        item = string_list_insert(entries, path);
        item->util = e;
        return e;
@@ -773,31 +787,78 @@ static int dir_in_way(const char *path, int check_working_copy, int empty_ok)
                !(empty_ok && is_empty_dir(path));
 }
 
-static int was_tracked(const char *path)
+/*
+ * Returns whether path was tracked in the index before the merge started,
+ * and its oid and mode match the specified values
+ */
+static int was_tracked_and_matches(struct merge_options *o, const char *path,
+                                  const struct object_id *oid, unsigned mode)
 {
-       int pos = cache_name_pos(path, strlen(path));
+       int pos = index_name_pos(&o->orig_index, path, strlen(path));
+       struct cache_entry *ce;
+
+       if (0 > pos)
+               /* we were not tracking this path before the merge */
+               return 0;
+
+       /* See if the file we were tracking before matches */
+       ce = o->orig_index.cache[pos];
+       return (oid_eq(&ce->oid, oid) && ce->ce_mode == mode);
+}
+
+/*
+ * Returns whether path was tracked in the index before the merge started
+ */
+static int was_tracked(struct merge_options *o, const char *path)
+{
+       int pos = index_name_pos(&o->orig_index, path, strlen(path));
 
        if (0 <= pos)
-               /* we have been tracking this path */
+               /* we were tracking this path before the merge */
                return 1;
 
-       /*
-        * Look for an unmerged entry for the path,
-        * specifically stage #2, which would indicate
-        * that "our" side before the merge started
-        * had the path tracked (and resulted in a conflict).
-        */
-       for (pos = -1 - pos;
-            pos < active_nr && !strcmp(path, active_cache[pos]->name);
-            pos++)
-               if (ce_stage(active_cache[pos]) == 2)
-                       return 1;
        return 0;
 }
 
 static int would_lose_untracked(const char *path)
 {
-       return !was_tracked(path) && file_exists(path);
+       /*
+        * This may look like it can be simplified to:
+        *   return !was_tracked(o, path) && file_exists(path)
+        * but it can't.  This function needs to know whether path was in
+        * the working tree due to EITHER having been tracked in the index
+        * before the merge OR having been put into the working copy and
+        * index by unpack_trees().  Due to that either-or requirement, we
+        * check the current index instead of the original one.
+        *
+        * Note that we do not need to worry about merge-recursive itself
+        * updating the index after unpack_trees() and before calling this
+        * function, because we strictly require all code paths in
+        * merge-recursive to update the working tree first and the index
+        * second.  Doing otherwise would break
+        * update_file()/would_lose_untracked(); see every comment in this
+        * file which mentions "update_stages".
+        */
+       int pos = cache_name_pos(path, strlen(path));
+
+       if (pos < 0)
+               pos = -1 - pos;
+       while (pos < active_nr &&
+              !strcmp(path, active_cache[pos]->name)) {
+               /*
+                * If stage #0, it is definitely tracked.
+                * If it has stage #2 then it was tracked
+                * before this merge started.  All other
+                * cases the path was not tracked.
+                */
+               switch (ce_stage(active_cache[pos])) {
+               case 0:
+               case 2:
+                       return 0;
+               }
+               pos++;
+       }
+       return file_exists(path);
 }
 
 static int was_dirty(struct merge_options *o, const char *path)
@@ -805,12 +866,12 @@ static int was_dirty(struct merge_options *o, const char *path)
        struct cache_entry *ce;
        int dirty = 1;
 
-       if (o->call_depth || !was_tracked(path))
+       if (o->call_depth || !was_tracked(o, path))
                return !dirty;
 
-       ce = cache_file_exists(path, strlen(path), ignore_case);
-       dirty = (ce->ce_stat_data.sd_mtime.sec > 0 &&
-                verify_uptodate(ce, &o->unpack_opts) != 0);
+       ce = index_file_exists(o->unpack_opts.src_index,
+                              path, strlen(path), ignore_case);
+       dirty = verify_uptodate(ce, &o->unpack_opts) != 0;
        return dirty;
 }
 
@@ -891,7 +952,7 @@ static int update_file_flags(struct merge_options *o,
                        goto update_index;
                }
 
-               buf = read_sha1_file(oid->hash, &type, &size);
+               buf = read_object_file(oid, &type, &size);
                if (!buf)
                        return err(o, _("cannot read object %s '%s'"), oid_to_hex(oid), path);
                if (type != OBJ_BLOB) {
@@ -942,7 +1003,9 @@ static int update_file_flags(struct merge_options *o,
        }
  update_index:
        if (!ret && update_cache)
-               add_cacheinfo(o, mode, oid, path, 0, update_wd, ADD_CACHE_OK_TO_ADD);
+               if (add_cacheinfo(o, mode, oid, path, 0, update_wd,
+                                 ADD_CACHE_OK_TO_ADD))
+                       return -1;
        return ret;
 }
 
@@ -1026,13 +1089,193 @@ static int merge_3way(struct merge_options *o,
        return merge_status;
 }
 
+static int find_first_merges(struct object_array *result, const char *path,
+               struct commit *a, struct commit *b)
+{
+       int i, j;
+       struct object_array merges = OBJECT_ARRAY_INIT;
+       struct commit *commit;
+       int contains_another;
+
+       char merged_revision[42];
+       const char *rev_args[] = { "rev-list", "--merges", "--ancestry-path",
+                                  "--all", merged_revision, NULL };
+       struct rev_info revs;
+       struct setup_revision_opt rev_opts;
+
+       memset(result, 0, sizeof(struct object_array));
+       memset(&rev_opts, 0, sizeof(rev_opts));
+
+       /* get all revisions that merge commit a */
+       xsnprintf(merged_revision, sizeof(merged_revision), "^%s",
+                       oid_to_hex(&a->object.oid));
+       init_revisions(&revs, NULL);
+       rev_opts.submodule = path;
+       /* FIXME: can't handle linked worktrees in submodules yet */
+       revs.single_worktree = path != NULL;
+       setup_revisions(ARRAY_SIZE(rev_args)-1, rev_args, &revs, &rev_opts);
+
+       /* save all revisions from the above list that contain b */
+       if (prepare_revision_walk(&revs))
+               die("revision walk setup failed");
+       while ((commit = get_revision(&revs)) != NULL) {
+               struct object *o = &(commit->object);
+               if (in_merge_bases(b, commit))
+                       add_object_array(o, NULL, &merges);
+       }
+       reset_revision_walk();
+
+       /* Now we've got all merges that contain a and b. Prune all
+        * merges that contain another found merge and save them in
+        * result.
+        */
+       for (i = 0; i < merges.nr; i++) {
+               struct commit *m1 = (struct commit *) merges.objects[i].item;
+
+               contains_another = 0;
+               for (j = 0; j < merges.nr; j++) {
+                       struct commit *m2 = (struct commit *) merges.objects[j].item;
+                       if (i != j && in_merge_bases(m2, m1)) {
+                               contains_another = 1;
+                               break;
+                       }
+               }
+
+               if (!contains_another)
+                       add_object_array(merges.objects[i].item, NULL, result);
+       }
+
+       object_array_clear(&merges);
+       return result->nr;
+}
+
+static void print_commit(struct commit *commit)
+{
+       struct strbuf sb = STRBUF_INIT;
+       struct pretty_print_context ctx = {0};
+       ctx.date_mode.type = DATE_NORMAL;
+       format_commit_message(commit, " %h: %m %s", &sb, &ctx);
+       fprintf(stderr, "%s\n", sb.buf);
+       strbuf_release(&sb);
+}
+
+static int merge_submodule(struct merge_options *o,
+                          struct object_id *result, const char *path,
+                          const struct object_id *base, const struct object_id *a,
+                          const struct object_id *b)
+{
+       struct commit *commit_base, *commit_a, *commit_b;
+       int parent_count;
+       struct object_array merges;
+
+       int i;
+       int search = !o->call_depth;
+
+       /* store a in result in case we fail */
+       oidcpy(result, a);
+
+       /* we can not handle deletion conflicts */
+       if (is_null_oid(base))
+               return 0;
+       if (is_null_oid(a))
+               return 0;
+       if (is_null_oid(b))
+               return 0;
+
+       if (add_submodule_odb(path)) {
+               output(o, 1, _("Failed to merge submodule %s (not checked out)"), path);
+               return 0;
+       }
+
+       if (!(commit_base = lookup_commit_reference(base)) ||
+           !(commit_a = lookup_commit_reference(a)) ||
+           !(commit_b = lookup_commit_reference(b))) {
+               output(o, 1, _("Failed to merge submodule %s (commits not present)"), path);
+               return 0;
+       }
+
+       /* check whether both changes are forward */
+       if (!in_merge_bases(commit_base, commit_a) ||
+           !in_merge_bases(commit_base, commit_b)) {
+               output(o, 1, _("Failed to merge submodule %s (commits don't follow merge-base)"), path);
+               return 0;
+       }
+
+       /* Case #1: a is contained in b or vice versa */
+       if (in_merge_bases(commit_a, commit_b)) {
+               oidcpy(result, b);
+               if (show(o, 3)) {
+                       output(o, 3, _("Fast-forwarding submodule %s to the following commit:"), path);
+                       output_commit_title(o, commit_b);
+               } else if (show(o, 2))
+                       output(o, 2, _("Fast-forwarding submodule %s"), path);
+               else
+                       ; /* no output */
+
+               return 1;
+       }
+       if (in_merge_bases(commit_b, commit_a)) {
+               oidcpy(result, a);
+               if (show(o, 3)) {
+                       output(o, 3, _("Fast-forwarding submodule %s to the following commit:"), path);
+                       output_commit_title(o, commit_a);
+               } else if (show(o, 2))
+                       output(o, 2, _("Fast-forwarding submodule %s"), path);
+               else
+                       ; /* no output */
+
+               return 1;
+       }
+
+       /*
+        * Case #2: There are one or more merges that contain a and b in
+        * the submodule. If there is only one, then present it as a
+        * suggestion to the user, but leave it marked unmerged so the
+        * user needs to confirm the resolution.
+        */
+
+       /* Skip the search if makes no sense to the calling context.  */
+       if (!search)
+               return 0;
+
+       /* find commit which merges them */
+       parent_count = find_first_merges(&merges, path, commit_a, commit_b);
+       switch (parent_count) {
+       case 0:
+               output(o, 1, _("Failed to merge submodule %s (merge following commits not found)"), path);
+               break;
+
+       case 1:
+               output(o, 1, _("Failed to merge submodule %s (not fast-forward)"), path);
+               output(o, 2, _("Found a possible merge resolution for the submodule:\n"));
+               print_commit((struct commit *) merges.objects[0].item);
+               output(o, 2, _(
+                       "If this is correct simply add it to the index "
+                       "for example\n"
+                       "by using:\n\n"
+                       "  git update-index --cacheinfo 160000 %s \"%s\"\n\n"
+                       "which will accept this suggestion.\n"),
+                       oid_to_hex(&merges.objects[0].item->oid), path);
+               break;
+
+       default:
+               output(o, 1, _("Failed to merge submodule %s (multiple merges found)"), path);
+               for (i = 0; i < merges.nr; i++)
+                       print_commit((struct commit *) merges.objects[i].item);
+       }
+
+       object_array_clear(&merges);
+       return 0;
+}
+
 static int merge_file_1(struct merge_options *o,
-                                          const struct diff_filespec *one,
-                                          const struct diff_filespec *a,
-                                          const struct diff_filespec *b,
-                                          const char *branch1,
-                                          const char *branch2,
-                                          struct merge_file_info *result)
+                       const struct diff_filespec *one,
+                       const struct diff_filespec *a,
+                       const struct diff_filespec *b,
+                       const char *filename,
+                       const char *branch1,
+                       const char *branch2,
+                       struct merge_file_info *result)
 {
        result->merge = 0;
        result->clean = 1;
@@ -1077,8 +1320,9 @@ static int merge_file_1(struct merge_options *o,
                        if ((merge_status < 0) || !result_buf.ptr)
                                ret = err(o, _("Failed to execute internal merge"));
 
-                       if (!ret && write_sha1_file(result_buf.ptr, result_buf.size,
-                                                   blob_type, result->oid.hash))
+                       if (!ret &&
+                           write_object_file(result_buf.ptr, result_buf.size,
+                                             blob_type, &result->oid))
                                ret = err(o, _("Unable to add %s to database"),
                                          a->path);
 
@@ -1087,33 +1331,45 @@ static int merge_file_1(struct merge_options *o,
                                return ret;
                        result->clean = (merge_status == 0);
                } else if (S_ISGITLINK(a->mode)) {
-                       result->clean = merge_submodule(&result->oid,
+                       result->clean = merge_submodule(o, &result->oid,
                                                       one->path,
                                                       &one->oid,
                                                       &a->oid,
-                                                      &b->oid,
-                                                      !o->call_depth);
+                                                      &b->oid);
                } else if (S_ISLNK(a->mode)) {
-                       oidcpy(&result->oid, &a->oid);
-
-                       if (!oid_eq(&a->oid, &b->oid))
-                               result->clean = 0;
+                       switch (o->recursive_variant) {
+                       case MERGE_RECURSIVE_NORMAL:
+                               oidcpy(&result->oid, &a->oid);
+                               if (!oid_eq(&a->oid, &b->oid))
+                                       result->clean = 0;
+                               break;
+                       case MERGE_RECURSIVE_OURS:
+                               oidcpy(&result->oid, &a->oid);
+                               break;
+                       case MERGE_RECURSIVE_THEIRS:
+                               oidcpy(&result->oid, &b->oid);
+                               break;
+                       }
                } else
-                       die("BUG: unsupported object type in the tree");
+                       BUG("unsupported object type in the tree");
        }
 
+       if (result->merge)
+               output(o, 2, _("Auto-merging %s"), filename);
+
        return 0;
 }
 
 static int merge_file_special_markers(struct merge_options *o,
-                          const struct diff_filespec *one,
-                          const struct diff_filespec *a,
-                          const struct diff_filespec *b,
-                          const char *branch1,
-                          const char *filename1,
-                          const char *branch2,
-                          const char *filename2,
-                          struct merge_file_info *mfi)
+                                     const struct diff_filespec *one,
+                                     const struct diff_filespec *a,
+                                     const struct diff_filespec *b,
+                                     const char *target_filename,
+                                     const char *branch1,
+                                     const char *filename1,
+                                     const char *branch2,
+                                     const char *filename2,
+                                     struct merge_file_info *mfi)
 {
        char *side1 = NULL;
        char *side2 = NULL;
@@ -1124,22 +1380,23 @@ static int merge_file_special_markers(struct merge_options *o,
        if (filename2)
                side2 = xstrfmt("%s:%s", branch2, filename2);
 
-       ret = merge_file_1(o, one, a, b,
+       ret = merge_file_1(o, one, a, b, target_filename,
                           side1 ? side1 : branch1,
                           side2 ? side2 : branch2, mfi);
+
        free(side1);
        free(side2);
        return ret;
 }
 
 static int merge_file_one(struct merge_options *o,
-                                        const char *path,
-                                        const struct object_id *o_oid, int o_mode,
-                                        const struct object_id *a_oid, int a_mode,
-                                        const struct object_id *b_oid, int b_mode,
-                                        const char *branch1,
-                                        const char *branch2,
-                                        struct merge_file_info *mfi)
+                         const char *path,
+                         const struct object_id *o_oid, int o_mode,
+                         const struct object_id *a_oid, int a_mode,
+                         const struct object_id *b_oid, int b_mode,
+                         const char *branch1,
+                         const char *branch2,
+                         struct merge_file_info *mfi)
 {
        struct diff_filespec one, a, b;
 
@@ -1150,7 +1407,7 @@ static int merge_file_one(struct merge_options *o,
        a.mode = a_mode;
        oidcpy(&b.oid, b_oid);
        b.mode = b_mode;
-       return merge_file_1(o, &one, &a, &b, branch1, branch2, mfi);
+       return merge_file_1(o, &one, &a, &b, path, branch1, branch2, mfi);
 }
 
 static int conflict_rename_dir(struct merge_options *o,
@@ -1428,6 +1685,8 @@ static int conflict_rename_rename_2to1(struct merge_options *o,
        struct diff_filespec *c1 = ci->pair1->two;
        struct diff_filespec *c2 = ci->pair2->two;
        char *path = c1->path; /* == c2->path */
+       char *path_side_1_desc;
+       char *path_side_2_desc;
        struct merge_file_info mfi_c1;
        struct merge_file_info mfi_c2;
        int ret;
@@ -1441,13 +1700,19 @@ static int conflict_rename_rename_2to1(struct merge_options *o,
        remove_file(o, 1, a->path, o->call_depth || would_lose_untracked(a->path));
        remove_file(o, 1, b->path, o->call_depth || would_lose_untracked(b->path));
 
+       path_side_1_desc = xstrfmt("%s (was %s)", path, a->path);
+       path_side_2_desc = xstrfmt("%s (was %s)", path, b->path);
        if (merge_file_special_markers(o, a, c1, &ci->ren1_other,
+                                      path_side_1_desc,
                                       o->branch1, c1->path,
                                       o->branch2, ci->ren1_other.path, &mfi_c1) ||
            merge_file_special_markers(o, b, &ci->ren2_other, c2,
+                                      path_side_2_desc,
                                       o->branch1, ci->ren2_other.path,
                                       o->branch2, c2->path, &mfi_c2))
                return -1;
+       free(path_side_1_desc);
+       free(path_side_2_desc);
 
        if (o->call_depth) {
                /*
@@ -1524,7 +1789,15 @@ static struct diff_queue_struct *get_diffpairs(struct merge_options *o,
        diff_setup(&opts);
        opts.flags.recursive = 1;
        opts.flags.rename_empty = 0;
-       opts.detect_rename = DIFF_DETECT_RENAME;
+       opts.detect_rename = merge_detect_rename(o);
+       /*
+        * We do not have logic to handle the detection of copies.  In
+        * fact, it may not even make sense to add such logic: would we
+        * really want a change to a base file to be propagated through
+        * multiple other files by a merge?
+        */
+       if (opts.detect_rename > DIFF_DETECT_RENAME)
+               opts.detect_rename = DIFF_DETECT_RENAME;
        opts.rename_limit = o->merge_rename_limit >= 0 ? o->merge_rename_limit :
                            o->diff_rename_limit >= 0 ? o->diff_rename_limit :
                            1000;
@@ -1549,11 +1822,11 @@ static struct diff_queue_struct *get_diffpairs(struct merge_options *o,
 
 static int tree_has_path(struct tree *tree, const char *path)
 {
-       unsigned char hashy[GIT_MAX_RAWSZ];
+       struct object_id hashy;
        unsigned int mode_o;
 
-       return !get_tree_entry(tree->object.oid.hash, path,
-                              hashy, &mode_o);
+       return !get_tree_entry(&tree->object.oid, path,
+                              &hashy, &mode_o);
 }
 
 /*
@@ -1851,7 +2124,7 @@ static struct hashmap *get_directory_renames(struct diff_queue_struct *pairs,
         * renames, finding out how often each directory rename pair
         * possibility occurs.
         */
-       dir_renames = xmalloc(sizeof(struct hashmap));
+       dir_renames = xmalloc(sizeof(*dir_renames));
        dir_rename_init(dir_renames);
        for (i = 0; i < pairs->nr; ++i) {
                struct string_list_item *item;
@@ -1871,7 +2144,7 @@ static struct hashmap *get_directory_renames(struct diff_queue_struct *pairs,
 
                entry = dir_rename_find_entry(dir_renames, old_dir);
                if (!entry) {
-                       entry = xmalloc(sizeof(struct dir_rename_entry));
+                       entry = xmalloc(sizeof(*entry));
                        dir_rename_entry_init(entry, old_dir);
                        hashmap_put(dir_renames, entry);
                } else {
@@ -2149,9 +2422,9 @@ static void apply_directory_rename_modifications(struct merge_options *o,
         * the various conflict_rename_*() functions update the index
         * explicitly rather than relying on unpack_trees() to have done it.
         */
-       get_tree_entry(tree->object.oid.hash,
+       get_tree_entry(&tree->object.oid,
                       pair->two->path,
-                      re->dst_entry->stages[stage].oid.hash,
+                      &re->dst_entry->stages[stage].oid,
                       &re->dst_entry->stages[stage].mode);
 
        /* Update pair status */
@@ -2331,7 +2604,7 @@ static int process_renames(struct merge_options *o,
                        const char *ren2_dst = ren2->pair->two->path;
                        enum rename_type rename_type;
                        if (strcmp(ren1_src, ren2_src) != 0)
-                               die("BUG: ren1_src != ren2_src");
+                               BUG("ren1_src != ren2_src");
                        ren2->dst_entry->processed = 1;
                        ren2->processed = 1;
                        if (strcmp(ren1_dst, ren2_dst) != 0) {
@@ -2365,7 +2638,7 @@ static int process_renames(struct merge_options *o,
                        ren2 = lookup->util;
                        ren2_dst = ren2->pair->two->path;
                        if (strcmp(ren1_dst, ren2_dst) != 0)
-                               die("BUG: ren1_dst != ren2_dst");
+                               BUG("ren1_dst != ren2_dst");
 
                        clean_merge = 0;
                        ren2->processed = 1;
@@ -2407,7 +2680,7 @@ static int process_renames(struct merge_options *o,
                         * add-source case).
                         */
                        remove_file(o, 1, ren1_src,
-                                   renamed_stage == 2 || !was_tracked(ren1_src));
+                                   renamed_stage == 2 || !was_tracked(o, ren1_src));
 
                        oidcpy(&src_other.oid,
                               &ren1->src_entry->stages[other_stage].oid);
@@ -2564,7 +2837,7 @@ static int handle_renames(struct merge_options *o,
        ri->head_renames = NULL;
        ri->merge_renames = NULL;
 
-       if (!o->detect_rename)
+       if (!merge_detect_rename(o))
                return 1;
 
        head_pairs = get_diffpairs(o, common, head);
@@ -2636,7 +2909,7 @@ static int read_oid_strbuf(struct merge_options *o,
        void *buf;
        enum object_type type;
        unsigned long size;
-       buf = read_sha1_file(oid->hash, &type, &size);
+       buf = read_object_file(oid, &type, &size);
        if (!buf)
                return err(o, _("cannot read object %s"), oid_to_hex(oid));
        if (type != OBJ_BLOB) {
@@ -2715,7 +2988,7 @@ static int handle_modify_delete(struct merge_options *o,
 
 static int merge_content(struct merge_options *o,
                         const char *path,
-                        int file_in_way,
+                        int is_dirty,
                         struct object_id *o_oid, int o_mode,
                         struct object_id *a_oid, int a_mode,
                         struct object_id *b_oid, int b_mode,
@@ -2756,27 +3029,26 @@ static int merge_content(struct merge_options *o,
                               S_ISGITLINK(pair1->two->mode)))
                        df_conflict_remains = 1;
        }
-       if (merge_file_special_markers(o, &one, &a, &b,
+       if (merge_file_special_markers(o, &one, &a, &b, path,
                                       o->branch1, path1,
                                       o->branch2, path2, &mfi))
                return -1;
 
-       if (mfi.clean && !df_conflict_remains &&
-           oid_eq(&mfi.oid, a_oid) && mfi.mode == a_mode) {
+       /*
+        * We can skip updating the working tree file iff:
+        *   a) The merge is clean
+        *   b) The merge matches what was in HEAD (content, mode, pathname)
+        *   c) The target path is usable (i.e. not involved in D/F conflict)
+        */
+       if (mfi.clean &&
+           was_tracked_and_matches(o, path, &mfi.oid, mfi.mode) &&
+           !df_conflict_remains) {
                output(o, 3, _("Skipped %s (merged same as existing)"), path);
-               /*
-                * The content merge resulted in the same file contents we
-                * already had.  We can return early if those file contents
-                * are recorded at the correct path (which may not be true
-                * if the merge involves a rename).
-                */
-               if (was_tracked(path)) {
-                       add_cacheinfo(o, mfi.mode, &mfi.oid, path,
-                                     0, (!o->call_depth), 0);
-                       return mfi.clean;
-               }
-       } else
-               output(o, 2, _("Auto-merging %s"), path);
+               if (add_cacheinfo(o, mfi.mode, &mfi.oid, path,
+                                 0, (!o->call_depth && !is_dirty), 0))
+                       return -1;
+               return mfi.clean;
+       }
 
        if (!mfi.clean) {
                if (S_ISGITLINK(mfi.mode))
@@ -2788,7 +3060,7 @@ static int merge_content(struct merge_options *o,
                                return -1;
        }
 
-       if (df_conflict_remains || file_in_way) {
+       if (df_conflict_remains || is_dirty) {
                char *new_path;
                if (o->call_depth) {
                        remove_file_from_cache(path);
@@ -2797,7 +3069,7 @@ static int merge_content(struct merge_options *o,
                                if (update_stages(o, path, &one, &a, &b))
                                        return -1;
                        } else {
-                               int file_from_stage2 = was_tracked(path);
+                               int file_from_stage2 = was_tracked(o, path);
                                struct diff_filespec merged;
                                oidcpy(&merged.oid, &mfi.oid);
                                merged.mode = mfi.mode;
@@ -2810,6 +3082,10 @@ static int merge_content(struct merge_options *o,
 
                }
                new_path = unique_path(o, path, rename_conflict_info->branch1);
+               if (is_dirty) {
+                       output(o, 1, _("Refusing to lose dirty file at %s"),
+                              path);
+               }
                output(o, 1, _("Adding as %s instead"), new_path);
                if (update_file(o, 0, &mfi.oid, mfi.mode, new_path)) {
                        free(new_path);
@@ -2819,7 +3095,7 @@ static int merge_content(struct merge_options *o,
                mfi.clean = 0;
        } else if (update_file(o, mfi.clean, &mfi.oid, mfi.mode, path))
                return -1;
-       return mfi.clean;
+       return !is_dirty && mfi.clean;
 }
 
 static int conflict_rename_normal(struct merge_options *o,
@@ -2829,21 +3105,10 @@ static int conflict_rename_normal(struct merge_options *o,
                                  struct object_id *b_oid, unsigned int b_mode,
                                  struct rename_conflict_info *ci)
 {
-       int clean_merge;
-       int file_in_the_way = 0;
-
-       if (was_dirty(o, path)) {
-               file_in_the_way = 1;
-               output(o, 1, _("Refusing to lose dirty file at %s"), path);
-       }
-
        /* Merge the content and write it out */
-       clean_merge = merge_content(o, path, file_in_the_way,
-                                   o_oid, o_mode, a_oid, a_mode, b_oid, b_mode,
-                                   ci);
-       if (clean_merge > 0 && file_in_the_way)
-               clean_merge = 0;
-       return clean_merge;
+       return merge_content(o, path, was_dirty(o, path),
+                            o_oid, o_mode, a_oid, a_mode, b_oid, b_mode,
+                            ci);
 }
 
 /* Per entry merge function */
@@ -2966,7 +3231,8 @@ static int process_entry(struct merge_options *o,
        } else if (a_oid && b_oid) {
                /* Case C: Added in both (check for same permissions) and */
                /* case D: Modified in both, but differently. */
-               clean_merge = merge_content(o, path, 0 /* file_in_way */,
+               int is_dirty = 0; /* unpack_trees would have bailed if dirty */
+               clean_merge = merge_content(o, path, is_dirty,
                                            o_oid, o_mode, a_oid, a_mode, b_oid, b_mode,
                                            NULL);
        } else if (!o_oid && !a_oid && !b_oid) {
@@ -2976,7 +3242,7 @@ static int process_entry(struct merge_options *o,
                 */
                remove_file(o, 1, path, !a_mode);
        } else
-               die("BUG: fatal merge failure, shouldn't happen.");
+               BUG("fatal merge failure, shouldn't happen.");
 
        return clean_merge;
 }
@@ -3007,13 +3273,14 @@ int merge_trees(struct merge_options *o,
                return 1;
        }
 
-       code = git_merge_trees(o, common, head, merge);
+       code = unpack_trees_start(o, common, head, merge);
 
        if (code != 0) {
                if (show(o, 4) || o->call_depth)
                        err(o, _("merging of trees %s and %s failed"),
                            oid_to_hex(&head->object.oid),
                            oid_to_hex(&merge->object.oid));
+               unpack_trees_finish(o);
                return -1;
        }
 
@@ -3054,7 +3321,7 @@ int merge_trees(struct merge_options *o,
                for (i = 0; i < entries->nr; i++) {
                        struct stage_data *e = entries->items[i].util;
                        if (!e->processed)
-                               die("BUG: unprocessed path??? %s",
+                               BUG("unprocessed path??? %s",
                                    entries->items[i].string);
                }
 
@@ -3066,12 +3333,16 @@ int merge_trees(struct merge_options *o,
 
                hashmap_free(&o->current_file_dir_set, 1);
 
-               if (clean < 0)
+               if (clean < 0) {
+                       unpack_trees_finish(o);
                        return clean;
+               }
        }
        else
                clean = 1;
 
+       unpack_trees_finish(o);
+
        if (o->call_depth && !(*result = write_tree_from_memory(o)))
                return -1;
 
@@ -3101,7 +3372,7 @@ int merge_recursive(struct merge_options *o,
 {
        struct commit_list *iter;
        struct commit *merged_common_ancestors;
-       struct tree *mrtree = mrtree;
+       struct tree *mrtree;
        int clean;
 
        if (show(o, 4)) {
@@ -3165,7 +3436,8 @@ int merge_recursive(struct merge_options *o,
                read_cache();
 
        o->ancestor = "merged common ancestors";
-       clean = merge_trees(o, h1->tree, h2->tree, merged_common_ancestors->tree,
+       clean = merge_trees(o, get_commit_tree(h1), get_commit_tree(h2),
+                           get_commit_tree(merged_common_ancestors),
                            &mrtree);
        if (clean < 0) {
                flush_output(o);
@@ -3229,11 +3501,13 @@ int merge_recursive_generic(struct merge_options *o,
        hold_locked_index(&lock, LOCK_DIE_ON_ERROR);
        clean = merge_recursive(o, head_commit, next_commit, ca,
                        result);
-       if (clean < 0)
+       if (clean < 0) {
+               rollback_lock_file(&lock);
                return clean;
+       }
 
-       if (active_cache_changed &&
-           write_locked_index(&the_index, &lock, COMMIT_LOCK))
+       if (write_locked_index(&the_index, &lock,
+                              COMMIT_LOCK | SKIP_IF_UNCHANGED))
                return err(o, _("Unable to write index."));
 
        return clean ? 0 : 1;
@@ -3241,9 +3515,18 @@ int merge_recursive_generic(struct merge_options *o,
 
 static void merge_recursive_config(struct merge_options *o)
 {
+       char *value = NULL;
        git_config_get_int("merge.verbosity", &o->verbosity);
        git_config_get_int("diff.renamelimit", &o->diff_rename_limit);
        git_config_get_int("merge.renamelimit", &o->merge_rename_limit);
+       if (!git_config_get_string("diff.renames", &value)) {
+               o->diff_detect_rename = git_config_rename("diff.renames", value);
+               free(value);
+       }
+       if (!git_config_get_string("merge.renames", &value)) {
+               o->merge_detect_rename = git_config_rename("merge.renames", value);
+               free(value);
+       }
        git_config(git_xmerge_config, NULL);
 }
 
@@ -3256,7 +3539,8 @@ void init_merge_options(struct merge_options *o)
        o->diff_rename_limit = -1;
        o->merge_rename_limit = -1;
        o->renormalize = 0;
-       o->detect_rename = 1;
+       o->diff_detect_rename = -1;
+       o->merge_detect_rename = -1;
        merge_recursive_config(o);
        merge_verbosity = getenv("GIT_MERGE_VERBOSITY");
        if (merge_verbosity)
@@ -3307,16 +3591,16 @@ int parse_merge_opt(struct merge_options *o, const char *s)
        else if (!strcmp(s, "no-renormalize"))
                o->renormalize = 0;
        else if (!strcmp(s, "no-renames"))
-               o->detect_rename = 0;
+               o->merge_detect_rename = 0;
        else if (!strcmp(s, "find-renames")) {
-               o->detect_rename = 1;
+               o->merge_detect_rename = 1;
                o->rename_score = 0;
        }
        else if (skip_prefix(s, "find-renames=", &arg) ||
                 skip_prefix(s, "rename-threshold=", &arg)) {
                if ((o->rename_score = parse_rename_score(&arg)) == -1 || *arg != 0)
                        return -1;
-               o->detect_rename = 1;
+               o->merge_detect_rename = 1;
        }
        else
                return -1;