p5551: add a script to test fetch pack-dir rescans
[gitweb.git] / submodule.c
index 9d8d91c207e7a2c09bdf997ba05c67b1262d8091..36f45f5a5a2adedd7e598371f98f6a54880668df 100644 (file)
@@ -1,4 +1,6 @@
 #include "cache.h"
+#include "repository.h"
+#include "config.h"
 #include "submodule-config.h"
 #include "submodule.h"
 #include "dir.h"
 #include "blob.h"
 #include "thread-utils.h"
 #include "quote.h"
+#include "remote.h"
+#include "worktree.h"
+#include "parse-options.h"
 
 static int config_fetch_recurse_submodules = RECURSE_SUBMODULES_ON_DEMAND;
+static int config_update_recurse_submodules = RECURSE_SUBMODULES_OFF;
 static int parallel_jobs = 1;
-static struct string_list changed_submodule_paths = STRING_LIST_INIT_NODUP;
+static struct string_list changed_submodule_paths = STRING_LIST_INIT_DUP;
 static int initialized_fetch_ref_tips;
-static struct sha1_array ref_tips_before_fetch;
-static struct sha1_array ref_tips_after_fetch;
+static struct oid_array ref_tips_before_fetch;
+static struct oid_array ref_tips_after_fetch;
 
 /*
  * The following flag is set if the .gitmodules file is unmerged. We then
@@ -123,35 +129,16 @@ void stage_updated_gitmodules(void)
 static int add_submodule_odb(const char *path)
 {
        struct strbuf objects_directory = STRBUF_INIT;
-       struct alternate_object_database *alt_odb;
        int ret = 0;
-       size_t alloc;
 
-       strbuf_git_path_submodule(&objects_directory, path, "objects/");
+       ret = strbuf_git_path_submodule(&objects_directory, path, "objects/");
+       if (ret)
+               goto done;
        if (!is_directory(objects_directory.buf)) {
                ret = -1;
                goto done;
        }
-       /* avoid adding it twice */
-       prepare_alt_odb();
-       for (alt_odb = alt_odb_list; alt_odb; alt_odb = alt_odb->next)
-               if (alt_odb->name - alt_odb->base == objects_directory.len &&
-                               !strncmp(alt_odb->base, objects_directory.buf,
-                                       objects_directory.len))
-                       goto done;
-
-       alloc = st_add(objects_directory.len, 42); /* for "12/345..." sha1 */
-       alt_odb = xmalloc(st_add(sizeof(*alt_odb), alloc));
-       alt_odb->next = alt_odb_list;
-       xsnprintf(alt_odb->base, alloc, "%s", objects_directory.buf);
-       alt_odb->name = alt_odb->base + objects_directory.len;
-       alt_odb->name[2] = '/';
-       alt_odb->name[40] = '\0';
-       alt_odb->name[41] = '\0';
-       alt_odb_list = alt_odb;
-
-       /* add possible alternates from the submodule */
-       read_info_alternates(objects_directory.buf, 0);
+       add_to_alternates_memory(objects_directory.buf);
 done:
        strbuf_release(&objects_directory);
        return ret;
@@ -169,7 +156,8 @@ void set_diffopt_flags_from_submodule_config(struct diff_options *diffopt,
        }
 }
 
-int submodule_config(const char *var, const char *value, void *cb)
+/* For loading from the .gitmodules file. */
+static int git_modules_config(const char *var, const char *value, void *cb)
 {
        if (!strcmp(var, "submodule.fetchjobs")) {
                parallel_jobs = git_config_int(var, value);
@@ -185,6 +173,56 @@ int submodule_config(const char *var, const char *value, void *cb)
        return 0;
 }
 
+/* Loads all submodule settings from the config. */
+int submodule_config(const char *var, const char *value, void *cb)
+{
+       if (!strcmp(var, "submodule.recurse")) {
+               int v = git_config_bool(var, value) ?
+                       RECURSE_SUBMODULES_ON : RECURSE_SUBMODULES_OFF;
+               config_update_recurse_submodules = v;
+               return 0;
+       } else {
+               return git_modules_config(var, value, cb);
+       }
+}
+
+/* Cheap function that only determines if we're interested in submodules at all */
+int git_default_submodule_config(const char *var, const char *value, void *cb)
+{
+       if (!strcmp(var, "submodule.recurse")) {
+               int v = git_config_bool(var, value) ?
+                       RECURSE_SUBMODULES_ON : RECURSE_SUBMODULES_OFF;
+               config_update_recurse_submodules = v;
+       }
+       return 0;
+}
+
+int option_parse_recurse_submodules_worktree_updater(const struct option *opt,
+                                                    const char *arg, int unset)
+{
+       if (unset) {
+               config_update_recurse_submodules = RECURSE_SUBMODULES_OFF;
+               return 0;
+       }
+       if (arg)
+               config_update_recurse_submodules =
+                       parse_update_recurse_submodules_arg(opt->long_name,
+                                                           arg);
+       else
+               config_update_recurse_submodules = RECURSE_SUBMODULES_ON;
+
+       return 0;
+}
+
+void load_submodule_cache(void)
+{
+       if (config_update_recurse_submodules == RECURSE_SUBMODULES_OFF)
+               return;
+
+       gitmodules_config();
+       git_config(submodule_config, NULL);
+}
+
 void gitmodules_config(void)
 {
        const char *work_tree = get_git_work_tree();
@@ -212,11 +250,166 @@ void gitmodules_config(void)
                }
 
                if (!gitmodules_is_unmerged)
-                       git_config_from_file(submodule_config, gitmodules_path.buf, NULL);
+                       git_config_from_file(git_modules_config,
+                               gitmodules_path.buf, NULL);
                strbuf_release(&gitmodules_path);
        }
 }
 
+static int gitmodules_cb(const char *var, const char *value, void *data)
+{
+       struct repository *repo = data;
+       return submodule_config_option(repo, var, value);
+}
+
+void repo_read_gitmodules(struct repository *repo)
+{
+       char *gitmodules_path = repo_worktree_path(repo, ".gitmodules");
+
+       git_config_from_file(gitmodules_cb, gitmodules_path, repo);
+       free(gitmodules_path);
+}
+
+void gitmodules_config_sha1(const unsigned char *commit_sha1)
+{
+       struct strbuf rev = STRBUF_INIT;
+       unsigned char sha1[20];
+
+       if (gitmodule_sha1_from_commit(commit_sha1, sha1, &rev)) {
+               git_config_from_blob_sha1(git_modules_config, rev.buf,
+                                         sha1, NULL);
+       }
+       strbuf_release(&rev);
+}
+
+/*
+ * Determine if a submodule has been initialized at a given 'path'
+ */
+int is_submodule_active(struct repository *repo, const char *path)
+{
+       int ret = 0;
+       char *key = NULL;
+       char *value = NULL;
+       const struct string_list *sl;
+       const struct submodule *module;
+
+       module = submodule_from_cache(repo, null_sha1, path);
+
+       /* early return if there isn't a path->module mapping */
+       if (!module)
+               return 0;
+
+       /* submodule.<name>.active is set */
+       key = xstrfmt("submodule.%s.active", module->name);
+       if (!repo_config_get_bool(repo, key, &ret)) {
+               free(key);
+               return ret;
+       }
+       free(key);
+
+       /* submodule.active is set */
+       sl = repo_config_get_value_multi(repo, "submodule.active");
+       if (sl) {
+               struct pathspec ps;
+               struct argv_array args = ARGV_ARRAY_INIT;
+               const struct string_list_item *item;
+
+               for_each_string_list_item(item, sl) {
+                       argv_array_push(&args, item->string);
+               }
+
+               parse_pathspec(&ps, 0, 0, NULL, args.argv);
+               ret = match_pathspec(&ps, path, strlen(path), 0, NULL, 1);
+
+               argv_array_clear(&args);
+               clear_pathspec(&ps);
+               return ret;
+       }
+
+       /* fallback to checking if the URL is set */
+       key = xstrfmt("submodule.%s.url", module->name);
+       ret = !repo_config_get_string(repo, key, &value);
+
+       free(value);
+       free(key);
+       return ret;
+}
+
+int is_submodule_populated_gently(const char *path, int *return_error_code)
+{
+       int ret = 0;
+       char *gitdir = xstrfmt("%s/.git", path);
+
+       if (resolve_gitdir_gently(gitdir, return_error_code))
+               ret = 1;
+
+       free(gitdir);
+       return ret;
+}
+
+/*
+ * Dies if the provided 'prefix' corresponds to an unpopulated submodule
+ */
+void die_in_unpopulated_submodule(const struct index_state *istate,
+                                 const char *prefix)
+{
+       int i, prefixlen;
+
+       if (!prefix)
+               return;
+
+       prefixlen = strlen(prefix);
+
+       for (i = 0; i < istate->cache_nr; i++) {
+               struct cache_entry *ce = istate->cache[i];
+               int ce_len = ce_namelen(ce);
+
+               if (!S_ISGITLINK(ce->ce_mode))
+                       continue;
+               if (prefixlen <= ce_len)
+                       continue;
+               if (strncmp(ce->name, prefix, ce_len))
+                       continue;
+               if (prefix[ce_len] != '/')
+                       continue;
+
+               die(_("in unpopulated submodule '%s'"), ce->name);
+       }
+}
+
+/*
+ * Dies if any paths in the provided pathspec descends into a submodule
+ */
+void die_path_inside_submodule(const struct index_state *istate,
+                              const struct pathspec *ps)
+{
+       int i, j;
+
+       for (i = 0; i < istate->cache_nr; i++) {
+               struct cache_entry *ce = istate->cache[i];
+               int ce_len = ce_namelen(ce);
+
+               if (!S_ISGITLINK(ce->ce_mode))
+                       continue;
+
+               for (j = 0; j < ps->nr ; j++) {
+                       const struct pathspec_item *item = &ps->items[j];
+
+                       if (item->len <= ce_len)
+                               continue;
+                       if (item->match[ce_len] != '/')
+                               continue;
+                       if (strncmp(ce->name, item->match, ce_len))
+                               continue;
+                       if (item->len == ce_len + 1)
+                               continue;
+
+                       die(_("Pathspec '%s' is in submodule '%.*s'"),
+                           item->original, ce_len, ce->name);
+               }
+       }
+}
+
 int parse_submodule_update_strategy(const char *value,
                struct submodule_update_strategy *dst)
 {
@@ -278,9 +471,9 @@ void handle_ignore_submodules_arg(struct diff_options *diffopt,
 
 static int prepare_submodule_summary(struct rev_info *rev, const char *path,
                struct commit *left, struct commit *right,
-               int *fast_forward, int *fast_backward)
+               struct commit_list *merge_bases)
 {
-       struct commit_list *merge_bases, *list;
+       struct commit_list *list;
 
        init_revisions(rev, NULL);
        setup_revisions(0, NULL, rev, NULL);
@@ -289,13 +482,6 @@ static int prepare_submodule_summary(struct rev_info *rev, const char *path,
        left->object.flags |= SYMMETRIC_LEFT;
        add_pending_object(rev, &left->object, path);
        add_pending_object(rev, &right->object, path);
-       merge_bases = get_merge_bases(left, right);
-       if (merge_bases) {
-               if (merge_bases->item == left)
-                       *fast_forward = 1;
-               else if (merge_bases->item == right)
-                       *fast_backward = 1;
-       }
        for (list = merge_bases; list; list = list->next) {
                list->item->object.flags |= UNINTERESTING;
                add_pending_object(rev, &list->item->object,
@@ -333,31 +519,40 @@ static void print_submodule_summary(struct rev_info *rev, FILE *f,
        strbuf_release(&sb);
 }
 
-void show_submodule_summary(FILE *f, const char *path,
+static void prepare_submodule_repo_env_no_git_dir(struct argv_array *out)
+{
+       const char * const *var;
+
+       for (var = local_repo_env; *var; var++) {
+               if (strcmp(*var, CONFIG_DATA_ENVIRONMENT))
+                       argv_array_push(out, *var);
+       }
+}
+
+void prepare_submodule_repo_env(struct argv_array *out)
+{
+       prepare_submodule_repo_env_no_git_dir(out);
+       argv_array_pushf(out, "%s=%s", GIT_DIR_ENVIRONMENT,
+                        DEFAULT_GIT_DIR_ENVIRONMENT);
+}
+
+/* Helper function to display the submodule header line prior to the full
+ * summary output. If it can locate the submodule objects directory it will
+ * attempt to lookup both the left and right commits and put them into the
+ * left and right pointers.
+ */
+static void show_submodule_header(FILE *f, const char *path,
                const char *line_prefix,
-               unsigned char one[20], unsigned char two[20],
+               struct object_id *one, struct object_id *two,
                unsigned dirty_submodule, const char *meta,
-               const char *del, const char *add, const char *reset)
+               const char *reset,
+               struct commit **left, struct commit **right,
+               struct commit_list **merge_bases)
 {
-       struct rev_info rev;
-       struct commit *left = NULL, *right = NULL;
        const char *message = NULL;
        struct strbuf sb = STRBUF_INIT;
        int fast_forward = 0, fast_backward = 0;
 
-       if (is_null_sha1(two))
-               message = "(submodule deleted)";
-       else if (add_submodule_odb(path))
-               message = "(not checked out)";
-       else if (is_null_sha1(one))
-               message = "(new submodule)";
-       else if (!(left = lookup_commit_reference(one)) ||
-                !(right = lookup_commit_reference(two)))
-               message = "(commits not present)";
-       else if (prepare_submodule_summary(&rev, path, left, right,
-                                          &fast_forward, &fast_backward))
-               message = "(revision walker failed)";
-
        if (dirty_submodule & DIRTY_SUBMODULE_UNTRACKED)
                fprintf(f, "%sSubmodule %s contains untracked content\n",
                        line_prefix, path);
@@ -365,29 +560,164 @@ void show_submodule_summary(FILE *f, const char *path,
                fprintf(f, "%sSubmodule %s contains modified content\n",
                        line_prefix, path);
 
-       if (!hashcmp(one, two)) {
+       if (is_null_oid(one))
+               message = "(new submodule)";
+       else if (is_null_oid(two))
+               message = "(submodule deleted)";
+
+       if (add_submodule_odb(path)) {
+               if (!message)
+                       message = "(not initialized)";
+               goto output_header;
+       }
+
+       /*
+        * Attempt to lookup the commit references, and determine if this is
+        * a fast forward or fast backwards update.
+        */
+       *left = lookup_commit_reference(one);
+       *right = lookup_commit_reference(two);
+
+       /*
+        * Warn about missing commits in the submodule project, but only if
+        * they aren't null.
+        */
+       if ((!is_null_oid(one) && !*left) ||
+            (!is_null_oid(two) && !*right))
+               message = "(commits not present)";
+
+       *merge_bases = get_merge_bases(*left, *right);
+       if (*merge_bases) {
+               if ((*merge_bases)->item == *left)
+                       fast_forward = 1;
+               else if ((*merge_bases)->item == *right)
+                       fast_backward = 1;
+       }
+
+       if (!oidcmp(one, two)) {
                strbuf_release(&sb);
                return;
        }
 
+output_header:
        strbuf_addf(&sb, "%s%sSubmodule %s ", line_prefix, meta, path);
-       strbuf_add_unique_abbrev(&sb, one, DEFAULT_ABBREV);
+       strbuf_add_unique_abbrev(&sb, one->hash, DEFAULT_ABBREV);
        strbuf_addstr(&sb, (fast_backward || fast_forward) ? ".." : "...");
-       strbuf_add_unique_abbrev(&sb, two, DEFAULT_ABBREV);
+       strbuf_add_unique_abbrev(&sb, two->hash, DEFAULT_ABBREV);
        if (message)
                strbuf_addf(&sb, " %s%s\n", message, reset);
        else
                strbuf_addf(&sb, "%s:%s\n", fast_backward ? " (rewind)" : "", reset);
        fwrite(sb.buf, sb.len, 1, f);
 
-       if (!message) /* only NULL if we succeeded in setting up the walk */
-               print_submodule_summary(&rev, f, line_prefix, del, add, reset);
+       strbuf_release(&sb);
+}
+
+void show_submodule_summary(FILE *f, const char *path,
+               const char *line_prefix,
+               struct object_id *one, struct object_id *two,
+               unsigned dirty_submodule, const char *meta,
+               const char *del, const char *add, const char *reset)
+{
+       struct rev_info rev;
+       struct commit *left = NULL, *right = NULL;
+       struct commit_list *merge_bases = NULL;
+
+       show_submodule_header(f, path, line_prefix, one, two, dirty_submodule,
+                             meta, reset, &left, &right, &merge_bases);
+
+       /*
+        * If we don't have both a left and a right pointer, there is no
+        * reason to try and display a summary. The header line should contain
+        * all the information the user needs.
+        */
+       if (!left || !right)
+               goto out;
+
+       /* Treat revision walker failure the same as missing commits */
+       if (prepare_submodule_summary(&rev, path, left, right, merge_bases)) {
+               fprintf(f, "%s(revision walker failed)\n", line_prefix);
+               goto out;
+       }
+
+       print_submodule_summary(&rev, f, line_prefix, del, add, reset);
+
+out:
+       if (merge_bases)
+               free_commit_list(merge_bases);
+       clear_commit_marks(left, ~0);
+       clear_commit_marks(right, ~0);
+}
+
+void show_submodule_inline_diff(FILE *f, const char *path,
+               const char *line_prefix,
+               struct object_id *one, struct object_id *two,
+               unsigned dirty_submodule, const char *meta,
+               const char *del, const char *add, const char *reset,
+               const struct diff_options *o)
+{
+       const struct object_id *old = &empty_tree_oid, *new = &empty_tree_oid;
+       struct commit *left = NULL, *right = NULL;
+       struct commit_list *merge_bases = NULL;
+       struct strbuf submodule_dir = STRBUF_INIT;
+       struct child_process cp = CHILD_PROCESS_INIT;
+
+       show_submodule_header(f, path, line_prefix, one, two, dirty_submodule,
+                             meta, reset, &left, &right, &merge_bases);
+
+       /* We need a valid left and right commit to display a difference */
+       if (!(left || is_null_oid(one)) ||
+           !(right || is_null_oid(two)))
+               goto done;
+
+       if (left)
+               old = one;
+       if (right)
+               new = two;
+
+       fflush(f);
+       cp.git_cmd = 1;
+       cp.dir = path;
+       cp.out = dup(fileno(f));
+       cp.no_stdin = 1;
+
+       /* TODO: other options may need to be passed here. */
+       argv_array_pushl(&cp.args, "diff", "--submodule=diff", NULL);
+
+       argv_array_pushf(&cp.args, "--line-prefix=%s", line_prefix);
+       if (DIFF_OPT_TST(o, REVERSE_DIFF)) {
+               argv_array_pushf(&cp.args, "--src-prefix=%s%s/",
+                                o->b_prefix, path);
+               argv_array_pushf(&cp.args, "--dst-prefix=%s%s/",
+                                o->a_prefix, path);
+       } else {
+               argv_array_pushf(&cp.args, "--src-prefix=%s%s/",
+                                o->a_prefix, path);
+               argv_array_pushf(&cp.args, "--dst-prefix=%s%s/",
+                                o->b_prefix, path);
+       }
+       argv_array_push(&cp.args, oid_to_hex(old));
+       /*
+        * If the submodule has modified content, we will diff against the
+        * work tree, under the assumption that the user has asked for the
+        * diff format and wishes to actually see all differences even if they
+        * haven't yet been committed to the submodule yet.
+        */
+       if (!(dirty_submodule & DIRTY_SUBMODULE_MODIFIED))
+               argv_array_push(&cp.args, oid_to_hex(new));
+
+       prepare_submodule_repo_env(&cp.env_array);
+       if (run_command(&cp))
+               fprintf(f, "(diff failed)\n");
+
+done:
+       strbuf_release(&submodule_dir);
+       if (merge_bases)
+               free_commit_list(merge_bases);
        if (left)
                clear_commit_marks(left, ~0);
        if (right)
                clear_commit_marks(right, ~0);
-
-       strbuf_release(&sb);
 }
 
 void set_config_fetch_recurse_submodules(int value)
@@ -395,33 +725,211 @@ void set_config_fetch_recurse_submodules(int value)
        config_fetch_recurse_submodules = value;
 }
 
+int should_update_submodules(void)
+{
+       return config_update_recurse_submodules == RECURSE_SUBMODULES_ON;
+}
+
+const struct submodule *submodule_from_ce(const struct cache_entry *ce)
+{
+       if (!S_ISGITLINK(ce->ce_mode))
+               return NULL;
+
+       if (!should_update_submodules())
+               return NULL;
+
+       return submodule_from_path(null_sha1, ce->name);
+}
+
+static struct oid_array *submodule_commits(struct string_list *submodules,
+                                          const char *path)
+{
+       struct string_list_item *item;
+
+       item = string_list_insert(submodules, path);
+       if (item->util)
+               return (struct oid_array *) item->util;
+
+       /* NEEDSWORK: should we have oid_array_init()? */
+       item->util = xcalloc(1, sizeof(struct oid_array));
+       return (struct oid_array *) item->util;
+}
+
+static void collect_changed_submodules_cb(struct diff_queue_struct *q,
+                                         struct diff_options *options,
+                                         void *data)
+{
+       int i;
+       struct string_list *changed = data;
+
+       for (i = 0; i < q->nr; i++) {
+               struct diff_filepair *p = q->queue[i];
+               struct oid_array *commits;
+               if (!S_ISGITLINK(p->two->mode))
+                       continue;
+
+               if (S_ISGITLINK(p->one->mode)) {
+                       /*
+                        * NEEDSWORK: We should honor the name configured in
+                        * the .gitmodules file of the commit we are examining
+                        * here to be able to correctly follow submodules
+                        * being moved around.
+                        */
+                       commits = submodule_commits(changed, p->two->path);
+                       oid_array_append(commits, &p->two->oid);
+               } else {
+                       /* Submodule is new or was moved here */
+                       /*
+                        * NEEDSWORK: When the .git directories of submodules
+                        * live inside the superprojects .git directory some
+                        * day we should fetch new submodules directly into
+                        * that location too when config or options request
+                        * that so they can be checked out from there.
+                        */
+                       continue;
+               }
+       }
+}
+
+/*
+ * Collect the paths of submodules in 'changed' which have changed based on
+ * the revisions as specified in 'argv'.  Each entry in 'changed' will also
+ * have a corresponding 'struct oid_array' (in the 'util' field) which lists
+ * what the submodule pointers were updated to during the change.
+ */
+static void collect_changed_submodules(struct string_list *changed,
+                                      struct argv_array *argv)
+{
+       struct rev_info rev;
+       const struct commit *commit;
+
+       init_revisions(&rev, NULL);
+       setup_revisions(argv->argc, argv->argv, &rev, NULL);
+       if (prepare_revision_walk(&rev))
+               die("revision walk setup failed");
+
+       while ((commit = get_revision(&rev))) {
+               struct rev_info diff_rev;
+
+               init_revisions(&diff_rev, NULL);
+               diff_rev.diffopt.output_format |= DIFF_FORMAT_CALLBACK;
+               diff_rev.diffopt.format_callback = collect_changed_submodules_cb;
+               diff_rev.diffopt.format_callback_data = changed;
+               diff_tree_combined_merge(commit, 1, &diff_rev);
+       }
+
+       reset_revision_walk();
+}
+
+static void free_submodules_oids(struct string_list *submodules)
+{
+       struct string_list_item *item;
+       for_each_string_list_item(item, submodules)
+               oid_array_clear((struct oid_array *) item->util);
+       string_list_clear(submodules, 1);
+}
+
 static int has_remote(const char *refname, const struct object_id *oid,
                      int flags, void *cb_data)
 {
        return 1;
 }
 
-static int submodule_needs_pushing(const char *path, const unsigned char sha1[20])
+static int append_oid_to_argv(const struct object_id *oid, void *data)
 {
-       if (add_submodule_odb(path) || !lookup_commit_reference(sha1))
+       struct argv_array *argv = data;
+       argv_array_push(argv, oid_to_hex(oid));
+       return 0;
+}
+
+static int check_has_commit(const struct object_id *oid, void *data)
+{
+       int *has_commit = data;
+
+       if (!lookup_commit_reference(oid))
+               *has_commit = 0;
+
+       return 0;
+}
+
+static int submodule_has_commits(const char *path, struct oid_array *commits)
+{
+       int has_commit = 1;
+
+       /*
+        * Perform a cheap, but incorrect check for the existence of 'commits'.
+        * This is done by adding the submodule's object store to the in-core
+        * object store, and then querying for each commit's existence.  If we
+        * do not have the commit object anywhere, there is no chance we have
+        * it in the object store of the correct submodule and have it
+        * reachable from a ref, so we can fail early without spawning rev-list
+        * which is expensive.
+        */
+       if (add_submodule_odb(path))
+               return 0;
+
+       oid_array_for_each_unique(commits, check_has_commit, &has_commit);
+
+       if (has_commit) {
+               /*
+                * Even if the submodule is checked out and the commit is
+                * present, make sure it exists in the submodule's object store
+                * and that it is reachable from a ref.
+                */
+               struct child_process cp = CHILD_PROCESS_INIT;
+               struct strbuf out = STRBUF_INIT;
+
+               argv_array_pushl(&cp.args, "rev-list", "-n", "1", NULL);
+               oid_array_for_each_unique(commits, append_oid_to_argv, &cp.args);
+               argv_array_pushl(&cp.args, "--not", "--all", NULL);
+
+               prepare_submodule_repo_env(&cp.env_array);
+               cp.git_cmd = 1;
+               cp.no_stdin = 1;
+               cp.dir = path;
+
+               if (capture_command(&cp, &out, GIT_MAX_HEXSZ + 1) || out.len)
+                       has_commit = 0;
+
+               strbuf_release(&out);
+       }
+
+       return has_commit;
+}
+
+static int submodule_needs_pushing(const char *path, struct oid_array *commits)
+{
+       if (!submodule_has_commits(path, commits))
+               /*
+                * NOTE: We do consider it safe to return "no" here. The
+                * correct answer would be "We do not know" instead of
+                * "No push needed", but it is quite hard to change
+                * the submodule pointer without having the submodule
+                * around. If a user did however change the submodules
+                * without having the submodule around, this indicates
+                * an expert who knows what they are doing or a
+                * maintainer integrating work from other people. In
+                * both cases it should be safe to skip this check.
+                */
                return 0;
 
        if (for_each_remote_ref_submodule(path, has_remote, NULL) > 0) {
                struct child_process cp = CHILD_PROCESS_INIT;
-               const char *argv[] = {"rev-list", NULL, "--not", "--remotes", "-n", "1" , NULL};
                struct strbuf buf = STRBUF_INIT;
                int needs_pushing = 0;
 
-               argv[1] = sha1_to_hex(sha1);
-               cp.argv = argv;
+               argv_array_push(&cp.args, "rev-list");
+               oid_array_for_each_unique(commits, append_oid_to_argv, &cp.args);
+               argv_array_pushl(&cp.args, "--not", "--remotes", "-n", "1" , NULL);
+
                prepare_submodule_repo_env(&cp.env_array);
                cp.git_cmd = 1;
                cp.no_stdin = 1;
                cp.out = -1;
                cp.dir = path;
                if (start_command(&cp))
-                       die("Could not run 'git rev-list %s --not --remotes -n 1' command in submodule %s",
-                               sha1_to_hex(sha1), path);
+                       die("Could not run 'git rev-list <commits> --not --remotes -n 1' command in submodule %s",
+                                       path);
                if (strbuf_read(&buf, cp.out, 41))
                        needs_pushing = 1;
                finish_command(&cp);
@@ -433,74 +941,64 @@ static int submodule_needs_pushing(const char *path, const unsigned char sha1[20
        return 0;
 }
 
-static void collect_submodules_from_diff(struct diff_queue_struct *q,
-                                        struct diff_options *options,
-                                        void *data)
-{
-       int i;
-       struct string_list *needs_pushing = data;
-
-       for (i = 0; i < q->nr; i++) {
-               struct diff_filepair *p = q->queue[i];
-               if (!S_ISGITLINK(p->two->mode))
-                       continue;
-               if (submodule_needs_pushing(p->two->path, p->two->oid.hash))
-                       string_list_insert(needs_pushing, p->two->path);
-       }
-}
-
-static void find_unpushed_submodule_commits(struct commit *commit,
-               struct string_list *needs_pushing)
-{
-       struct rev_info rev;
-
-       init_revisions(&rev, NULL);
-       rev.diffopt.output_format |= DIFF_FORMAT_CALLBACK;
-       rev.diffopt.format_callback = collect_submodules_from_diff;
-       rev.diffopt.format_callback_data = needs_pushing;
-       diff_tree_combined_merge(commit, 1, &rev);
-}
-
-int find_unpushed_submodules(unsigned char new_sha1[20],
+int find_unpushed_submodules(struct oid_array *commits,
                const char *remotes_name, struct string_list *needs_pushing)
 {
-       struct rev_info rev;
-       struct commit *commit;
-       const char *argv[] = {NULL, NULL, "--not", "NULL", NULL};
-       int argc = ARRAY_SIZE(argv) - 1;
-       char *sha1_copy;
+       struct string_list submodules = STRING_LIST_INIT_DUP;
+       struct string_list_item *submodule;
+       struct argv_array argv = ARGV_ARRAY_INIT;
 
-       struct strbuf remotes_arg = STRBUF_INIT;
+       /* argv.argv[0] will be ignored by setup_revisions */
+       argv_array_push(&argv, "find_unpushed_submodules");
+       oid_array_for_each_unique(commits, append_oid_to_argv, &argv);
+       argv_array_push(&argv, "--not");
+       argv_array_pushf(&argv, "--remotes=%s", remotes_name);
 
-       strbuf_addf(&remotes_arg, "--remotes=%s", remotes_name);
-       init_revisions(&rev, NULL);
-       sha1_copy = xstrdup(sha1_to_hex(new_sha1));
-       argv[1] = sha1_copy;
-       argv[3] = remotes_arg.buf;
-       setup_revisions(argc, argv, &rev, NULL);
-       if (prepare_revision_walk(&rev))
-               die("revision walk setup failed");
+       collect_changed_submodules(&submodules, &argv);
 
-       while ((commit = get_revision(&rev)) != NULL)
-               find_unpushed_submodule_commits(commit, needs_pushing);
+       for_each_string_list_item(submodule, &submodules) {
+               struct oid_array *commits = submodule->util;
+               const char *path = submodule->string;
 
-       reset_revision_walk();
-       free(sha1_copy);
-       strbuf_release(&remotes_arg);
+               if (submodule_needs_pushing(path, commits))
+                       string_list_insert(needs_pushing, path);
+       }
+
+       free_submodules_oids(&submodules);
+       argv_array_clear(&argv);
 
        return needs_pushing->nr;
 }
 
-static int push_submodule(const char *path)
+static int push_submodule(const char *path,
+                         const struct remote *remote,
+                         const char **refspec, int refspec_nr,
+                         const struct string_list *push_options,
+                         int dry_run)
 {
        if (add_submodule_odb(path))
                return 1;
 
        if (for_each_remote_ref_submodule(path, has_remote, NULL) > 0) {
                struct child_process cp = CHILD_PROCESS_INIT;
-               const char *argv[] = {"push", NULL};
+               argv_array_push(&cp.args, "push");
+               if (dry_run)
+                       argv_array_push(&cp.args, "--dry-run");
+
+               if (push_options && push_options->nr) {
+                       const struct string_list_item *item;
+                       for_each_string_list_item(item, push_options)
+                               argv_array_pushf(&cp.args, "--push-option=%s",
+                                                item->string);
+               }
+
+               if (remote->origin != REMOTE_UNCONFIGURED) {
+                       int i;
+                       argv_array_push(&cp.args, remote->name);
+                       for (i = 0; i < refspec_nr; i++)
+                               argv_array_push(&cp.args, refspec[i]);
+               }
 
-               cp.argv = argv;
                prepare_submodule_repo_env(&cp.env_array);
                cp.git_cmd = 1;
                cp.no_stdin = 1;
@@ -513,18 +1011,79 @@ static int push_submodule(const char *path)
        return 1;
 }
 
-int push_unpushed_submodules(unsigned char new_sha1[20], const char *remotes_name)
+/*
+ * Perform a check in the submodule to see if the remote and refspec work.
+ * Die if the submodule can't be pushed.
+ */
+static void submodule_push_check(const char *path, const char *head,
+                                const struct remote *remote,
+                                const char **refspec, int refspec_nr)
+{
+       struct child_process cp = CHILD_PROCESS_INIT;
+       int i;
+
+       argv_array_push(&cp.args, "submodule--helper");
+       argv_array_push(&cp.args, "push-check");
+       argv_array_push(&cp.args, head);
+       argv_array_push(&cp.args, remote->name);
+
+       for (i = 0; i < refspec_nr; i++)
+               argv_array_push(&cp.args, refspec[i]);
+
+       prepare_submodule_repo_env(&cp.env_array);
+       cp.git_cmd = 1;
+       cp.no_stdin = 1;
+       cp.no_stdout = 1;
+       cp.dir = path;
+
+       /*
+        * Simply indicate if 'submodule--helper push-check' failed.
+        * More detailed error information will be provided by the
+        * child process.
+        */
+       if (run_command(&cp))
+               die("process for submodule '%s' failed", path);
+}
+
+int push_unpushed_submodules(struct oid_array *commits,
+                            const struct remote *remote,
+                            const char **refspec, int refspec_nr,
+                            const struct string_list *push_options,
+                            int dry_run)
 {
        int i, ret = 1;
        struct string_list needs_pushing = STRING_LIST_INIT_DUP;
 
-       if (!find_unpushed_submodules(new_sha1, remotes_name, &needs_pushing))
+       if (!find_unpushed_submodules(commits, remote->name, &needs_pushing))
                return 1;
 
+       /*
+        * Verify that the remote and refspec can be propagated to all
+        * submodules.  This check can be skipped if the remote and refspec
+        * won't be propagated due to the remote being unconfigured (e.g. a URL
+        * instead of a remote name).
+        */
+       if (remote->origin != REMOTE_UNCONFIGURED) {
+               char *head;
+               struct object_id head_oid;
+
+               head = resolve_refdup("HEAD", 0, head_oid.hash, NULL);
+               if (!head)
+                       die(_("Failed to resolve HEAD as a valid ref."));
+
+               for (i = 0; i < needs_pushing.nr; i++)
+                       submodule_push_check(needs_pushing.items[i].string,
+                                            head, remote,
+                                            refspec, refspec_nr);
+               free(head);
+       }
+
+       /* Actually push the submodules */
        for (i = 0; i < needs_pushing.nr; i++) {
                const char *path = needs_pushing.items[i].string;
                fprintf(stderr, "Pushing submodule '%s'\n", path);
-               if (!push_submodule(path)) {
+               if (!push_submodule(path, remote, refspec, refspec_nr,
+                                   push_options, dry_run)) {
                        fprintf(stderr, "Unable to push submodule '%s'\n", path);
                        ret = 0;
                }
@@ -535,130 +1094,88 @@ int push_unpushed_submodules(unsigned char new_sha1[20], const char *remotes_nam
        return ret;
 }
 
-static int is_submodule_commit_present(const char *path, unsigned char sha1[20])
-{
-       int is_present = 0;
-       if (!add_submodule_odb(path) && lookup_commit_reference(sha1)) {
-               /* Even if the submodule is checked out and the commit is
-                * present, make sure it is reachable from a ref. */
-               struct child_process cp = CHILD_PROCESS_INIT;
-               const char *argv[] = {"rev-list", "-n", "1", NULL, "--not", "--all", NULL};
-               struct strbuf buf = STRBUF_INIT;
-
-               argv[3] = sha1_to_hex(sha1);
-               cp.argv = argv;
-               prepare_submodule_repo_env(&cp.env_array);
-               cp.git_cmd = 1;
-               cp.no_stdin = 1;
-               cp.dir = path;
-               if (!capture_command(&cp, &buf, 1024) && !buf.len)
-                       is_present = 1;
-
-               strbuf_release(&buf);
-       }
-       return is_present;
-}
-
-static void submodule_collect_changed_cb(struct diff_queue_struct *q,
-                                        struct diff_options *options,
-                                        void *data)
-{
-       int i;
-       for (i = 0; i < q->nr; i++) {
-               struct diff_filepair *p = q->queue[i];
-               if (!S_ISGITLINK(p->two->mode))
-                       continue;
-
-               if (S_ISGITLINK(p->one->mode)) {
-                       /* NEEDSWORK: We should honor the name configured in
-                        * the .gitmodules file of the commit we are examining
-                        * here to be able to correctly follow submodules
-                        * being moved around. */
-                       struct string_list_item *path;
-                       path = unsorted_string_list_lookup(&changed_submodule_paths, p->two->path);
-                       if (!path && !is_submodule_commit_present(p->two->path, p->two->oid.hash))
-                               string_list_append(&changed_submodule_paths, xstrdup(p->two->path));
-               } else {
-                       /* Submodule is new or was moved here */
-                       /* NEEDSWORK: When the .git directories of submodules
-                        * live inside the superprojects .git directory some
-                        * day we should fetch new submodules directly into
-                        * that location too when config or options request
-                        * that so they can be checked out from there. */
-                       continue;
-               }
-       }
-}
-
-static int add_sha1_to_array(const char *ref, const struct object_id *oid,
-                            int flags, void *data)
+static int append_oid_to_array(const char *ref, const struct object_id *oid,
+                              int flags, void *data)
 {
-       sha1_array_append(data, oid->hash);
+       struct oid_array *array = data;
+       oid_array_append(array, oid);
        return 0;
 }
 
-void check_for_new_submodule_commits(unsigned char new_sha1[20])
+void check_for_new_submodule_commits(struct object_id *oid)
 {
        if (!initialized_fetch_ref_tips) {
-               for_each_ref(add_sha1_to_array, &ref_tips_before_fetch);
+               for_each_ref(append_oid_to_array, &ref_tips_before_fetch);
                initialized_fetch_ref_tips = 1;
        }
 
-       sha1_array_append(&ref_tips_after_fetch, new_sha1);
-}
-
-static void add_sha1_to_argv(const unsigned char sha1[20], void *data)
-{
-       argv_array_push(data, sha1_to_hex(sha1));
+       oid_array_append(&ref_tips_after_fetch, oid);
 }
 
 static void calculate_changed_submodule_paths(void)
 {
-       struct rev_info rev;
-       struct commit *commit;
        struct argv_array argv = ARGV_ARRAY_INIT;
+       struct string_list changed_submodules = STRING_LIST_INIT_DUP;
+       const struct string_list_item *item;
 
        /* No need to check if there are no submodules configured */
        if (!submodule_from_path(NULL, NULL))
                return;
 
-       init_revisions(&rev, NULL);
        argv_array_push(&argv, "--"); /* argv[0] program name */
-       sha1_array_for_each_unique(&ref_tips_after_fetch,
-                                  add_sha1_to_argv, &argv);
+       oid_array_for_each_unique(&ref_tips_after_fetch,
+                                  append_oid_to_argv, &argv);
        argv_array_push(&argv, "--not");
-       sha1_array_for_each_unique(&ref_tips_before_fetch,
-                                  add_sha1_to_argv, &argv);
-       setup_revisions(argv.argc, argv.argv, &rev, NULL);
-       if (prepare_revision_walk(&rev))
-               die("revision walk setup failed");
+       oid_array_for_each_unique(&ref_tips_before_fetch,
+                                  append_oid_to_argv, &argv);
 
        /*
         * Collect all submodules (whether checked out or not) for which new
         * commits have been recorded upstream in "changed_submodule_paths".
         */
-       while ((commit = get_revision(&rev))) {
-               struct commit_list *parent = commit->parents;
-               while (parent) {
-                       struct diff_options diff_opts;
-                       diff_setup(&diff_opts);
-                       DIFF_OPT_SET(&diff_opts, RECURSIVE);
-                       diff_opts.output_format |= DIFF_FORMAT_CALLBACK;
-                       diff_opts.format_callback = submodule_collect_changed_cb;
-                       diff_setup_done(&diff_opts);
-                       diff_tree_sha1(parent->item->object.oid.hash, commit->object.oid.hash, "", &diff_opts);
-                       diffcore_std(&diff_opts);
-                       diff_flush(&diff_opts);
-                       parent = parent->next;
-               }
+       collect_changed_submodules(&changed_submodules, &argv);
+
+       for_each_string_list_item(item, &changed_submodules) {
+               struct oid_array *commits = item->util;
+               const char *path = item->string;
+
+               if (!submodule_has_commits(path, commits))
+                       string_list_append(&changed_submodule_paths, path);
        }
 
+       free_submodules_oids(&changed_submodules);
        argv_array_clear(&argv);
-       sha1_array_clear(&ref_tips_before_fetch);
-       sha1_array_clear(&ref_tips_after_fetch);
+       oid_array_clear(&ref_tips_before_fetch);
+       oid_array_clear(&ref_tips_after_fetch);
        initialized_fetch_ref_tips = 0;
 }
 
+int submodule_touches_in_range(struct object_id *excl_oid,
+                              struct object_id *incl_oid)
+{
+       struct string_list subs = STRING_LIST_INIT_DUP;
+       struct argv_array args = ARGV_ARRAY_INIT;
+       int ret;
+
+       gitmodules_config();
+       /* No need to check if there are no submodules configured */
+       if (!submodule_from_path(NULL, NULL))
+               return 0;
+
+       argv_array_push(&args, "--"); /* args[0] program name */
+       argv_array_push(&args, oid_to_hex(incl_oid));
+       argv_array_push(&args, "--not");
+       argv_array_push(&args, oid_to_hex(excl_oid));
+
+       collect_changed_submodules(&subs, &args);
+       ret = subs.nr;
+
+       argv_array_clear(&args);
+
+       free_submodules_oids(&subs);
+       return ret;
+}
+
 struct submodule_parallel_fetch {
        int count;
        struct argv_array args;
@@ -816,67 +1333,78 @@ int fetch_populated_submodules(const struct argv_array *options,
 
 unsigned is_submodule_modified(const char *path, int ignore_untracked)
 {
-       ssize_t len;
        struct child_process cp = CHILD_PROCESS_INIT;
-       const char *argv[] = {
-               "status",
-               "--porcelain",
-               NULL,
-               NULL,
-       };
        struct strbuf buf = STRBUF_INIT;
+       FILE *fp;
        unsigned dirty_submodule = 0;
-       const char *line, *next_line;
        const char *git_dir;
+       int ignore_cp_exit_code = 0;
 
        strbuf_addf(&buf, "%s/.git", path);
        git_dir = read_gitfile(buf.buf);
        if (!git_dir)
                git_dir = buf.buf;
-       if (!is_directory(git_dir)) {
+       if (!is_git_directory(git_dir)) {
+               if (is_directory(git_dir))
+                       die(_("'%s' not recognized as a git repository"), git_dir);
                strbuf_release(&buf);
                /* The submodule is not checked out, so it is not modified */
                return 0;
-
        }
        strbuf_reset(&buf);
 
+       argv_array_pushl(&cp.args, "status", "--porcelain=2", NULL);
        if (ignore_untracked)
-               argv[2] = "-uno";
+               argv_array_push(&cp.args, "-uno");
 
-       cp.argv = argv;
        prepare_submodule_repo_env(&cp.env_array);
        cp.git_cmd = 1;
        cp.no_stdin = 1;
        cp.out = -1;
        cp.dir = path;
        if (start_command(&cp))
-               die("Could not run 'git status --porcelain' in submodule %s", path);
+               die("Could not run 'git status --porcelain=2' in submodule %s", path);
 
-       len = strbuf_read(&buf, cp.out, 1024);
-       line = buf.buf;
-       while (len > 2) {
-               if ((line[0] == '?') && (line[1] == '?')) {
+       fp = xfdopen(cp.out, "r");
+       while (strbuf_getwholeline(&buf, fp, '\n') != EOF) {
+               /* regular untracked files */
+               if (buf.buf[0] == '?')
                        dirty_submodule |= DIRTY_SUBMODULE_UNTRACKED;
-                       if (dirty_submodule & DIRTY_SUBMODULE_MODIFIED)
-                               break;
-               } else {
-                       dirty_submodule |= DIRTY_SUBMODULE_MODIFIED;
-                       if (ignore_untracked ||
-                           (dirty_submodule & DIRTY_SUBMODULE_UNTRACKED))
-                               break;
+
+               if (buf.buf[0] == 'u' ||
+                   buf.buf[0] == '1' ||
+                   buf.buf[0] == '2') {
+                       /* T = line type, XY = status, SSSS = submodule state */
+                       if (buf.len < strlen("T XY SSSS"))
+                               die("BUG: invalid status --porcelain=2 line %s",
+                                   buf.buf);
+
+                       if (buf.buf[5] == 'S' && buf.buf[8] == 'U')
+                               /* nested untracked file */
+                               dirty_submodule |= DIRTY_SUBMODULE_UNTRACKED;
+
+                       if (buf.buf[0] == 'u' ||
+                           buf.buf[0] == '2' ||
+                           memcmp(buf.buf + 5, "S..U", 4))
+                               /* other change */
+                               dirty_submodule |= DIRTY_SUBMODULE_MODIFIED;
                }
-               next_line = strchr(line, '\n');
-               if (!next_line)
+
+               if ((dirty_submodule & DIRTY_SUBMODULE_MODIFIED) &&
+                   ((dirty_submodule & DIRTY_SUBMODULE_UNTRACKED) ||
+                    ignore_untracked)) {
+                       /*
+                        * We're not interested in any further information from
+                        * the child any more, neither output nor its exit code.
+                        */
+                       ignore_cp_exit_code = 1;
                        break;
-               next_line++;
-               len -= (next_line - line);
-               line = next_line;
+               }
        }
-       close(cp.out);
+       fclose(fp);
 
-       if (finish_command(&cp))
-               die("'git status --porcelain' failed in submodule %s", path);
+       if (finish_command(&cp) && !ignore_cp_exit_code)
+               die("'git status --porcelain=2' failed in submodule %s", path);
 
        strbuf_release(&buf);
        return dirty_submodule;
@@ -918,45 +1446,232 @@ int submodule_uses_gitfile(const char *path)
        return 1;
 }
 
-int ok_to_remove_submodule(const char *path)
+/*
+ * Check if it is a bad idea to remove a submodule, i.e. if we'd lose data
+ * when doing so.
+ *
+ * Return 1 if we'd lose data, return 0 if the removal is fine,
+ * and negative values for errors.
+ */
+int bad_to_remove_submodule(const char *path, unsigned flags)
 {
        ssize_t len;
        struct child_process cp = CHILD_PROCESS_INIT;
-       const char *argv[] = {
-               "status",
-               "--porcelain",
-               "-u",
-               "--ignore-submodules=none",
-               NULL,
-       };
        struct strbuf buf = STRBUF_INIT;
-       int ok_to_remove = 1;
+       int ret = 0;
 
        if (!file_exists(path) || is_empty_dir(path))
-               return 1;
+               return 0;
 
        if (!submodule_uses_gitfile(path))
-               return 0;
+               return 1;
+
+       argv_array_pushl(&cp.args, "status", "--porcelain",
+                                  "--ignore-submodules=none", NULL);
+
+       if (flags & SUBMODULE_REMOVAL_IGNORE_UNTRACKED)
+               argv_array_push(&cp.args, "-uno");
+       else
+               argv_array_push(&cp.args, "-uall");
+
+       if (!(flags & SUBMODULE_REMOVAL_IGNORE_IGNORED_UNTRACKED))
+               argv_array_push(&cp.args, "--ignored");
 
-       cp.argv = argv;
        prepare_submodule_repo_env(&cp.env_array);
        cp.git_cmd = 1;
        cp.no_stdin = 1;
        cp.out = -1;
        cp.dir = path;
-       if (start_command(&cp))
-               die("Could not run 'git status --porcelain -uall --ignore-submodules=none' in submodule %s", path);
+       if (start_command(&cp)) {
+               if (flags & SUBMODULE_REMOVAL_DIE_ON_ERROR)
+                       die(_("could not start 'git status' in submodule '%s'"),
+                               path);
+               ret = -1;
+               goto out;
+       }
 
        len = strbuf_read(&buf, cp.out, 1024);
        if (len > 2)
-               ok_to_remove = 0;
+               ret = 1;
        close(cp.out);
 
-       if (finish_command(&cp))
-               die("'git status --porcelain -uall --ignore-submodules=none' failed in submodule %s", path);
-
+       if (finish_command(&cp)) {
+               if (flags & SUBMODULE_REMOVAL_DIE_ON_ERROR)
+                       die(_("could not run 'git status' in submodule '%s'"),
+                               path);
+               ret = -1;
+       }
+out:
        strbuf_release(&buf);
-       return ok_to_remove;
+       return ret;
+}
+
+static const char *get_super_prefix_or_empty(void)
+{
+       const char *s = get_super_prefix();
+       if (!s)
+               s = "";
+       return s;
+}
+
+static int submodule_has_dirty_index(const struct submodule *sub)
+{
+       struct child_process cp = CHILD_PROCESS_INIT;
+
+       prepare_submodule_repo_env(&cp.env_array);
+
+       cp.git_cmd = 1;
+       argv_array_pushl(&cp.args, "diff-index", "--quiet",
+                                  "--cached", "HEAD", NULL);
+       cp.no_stdin = 1;
+       cp.no_stdout = 1;
+       cp.dir = sub->path;
+       if (start_command(&cp))
+               die("could not recurse into submodule '%s'", sub->path);
+
+       return finish_command(&cp);
+}
+
+static void submodule_reset_index(const char *path)
+{
+       struct child_process cp = CHILD_PROCESS_INIT;
+       prepare_submodule_repo_env(&cp.env_array);
+
+       cp.git_cmd = 1;
+       cp.no_stdin = 1;
+       cp.dir = path;
+
+       argv_array_pushf(&cp.args, "--super-prefix=%s%s/",
+                                  get_super_prefix_or_empty(), path);
+       argv_array_pushl(&cp.args, "read-tree", "-u", "--reset", NULL);
+
+       argv_array_push(&cp.args, EMPTY_TREE_SHA1_HEX);
+
+       if (run_command(&cp))
+               die("could not reset submodule index");
+}
+
+/**
+ * Moves a submodule at a given path from a given head to another new head.
+ * For edge cases (a submodule coming into existence or removing a submodule)
+ * pass NULL for old or new respectively.
+ */
+int submodule_move_head(const char *path,
+                        const char *old,
+                        const char *new,
+                        unsigned flags)
+{
+       int ret = 0;
+       struct child_process cp = CHILD_PROCESS_INIT;
+       const struct submodule *sub;
+       int *error_code_ptr, error_code;
+
+       if (!is_submodule_active(the_repository, path))
+               return 0;
+
+       if (flags & SUBMODULE_MOVE_HEAD_FORCE)
+               /*
+                * Pass non NULL pointer to is_submodule_populated_gently
+                * to prevent die()-ing. We'll use connect_work_tree_and_git_dir
+                * to fixup the submodule in the force case later.
+                */
+               error_code_ptr = &error_code;
+       else
+               error_code_ptr = NULL;
+
+       if (old && !is_submodule_populated_gently(path, error_code_ptr))
+               return 0;
+
+       sub = submodule_from_path(null_sha1, path);
+
+       if (!sub)
+               die("BUG: could not get submodule information for '%s'", path);
+
+       if (old && !(flags & SUBMODULE_MOVE_HEAD_FORCE)) {
+               /* Check if the submodule has a dirty index. */
+               if (submodule_has_dirty_index(sub))
+                       return error(_("submodule '%s' has dirty index"), path);
+       }
+
+       if (!(flags & SUBMODULE_MOVE_HEAD_DRY_RUN)) {
+               if (old) {
+                       if (!submodule_uses_gitfile(path))
+                               absorb_git_dir_into_superproject("", path,
+                                       ABSORB_GITDIR_RECURSE_SUBMODULES);
+               } else {
+                       char *gitdir = xstrfmt("%s/modules/%s",
+                                   get_git_common_dir(), sub->name);
+                       connect_work_tree_and_git_dir(path, gitdir);
+                       free(gitdir);
+
+                       /* make sure the index is clean as well */
+                       submodule_reset_index(path);
+               }
+
+               if (old && (flags & SUBMODULE_MOVE_HEAD_FORCE)) {
+                       char *gitdir = xstrfmt("%s/modules/%s",
+                                   get_git_common_dir(), sub->name);
+                       connect_work_tree_and_git_dir(path, gitdir);
+                       free(gitdir);
+               }
+       }
+
+       prepare_submodule_repo_env(&cp.env_array);
+
+       cp.git_cmd = 1;
+       cp.no_stdin = 1;
+       cp.dir = path;
+
+       argv_array_pushf(&cp.args, "--super-prefix=%s%s/",
+                       get_super_prefix_or_empty(), path);
+       argv_array_pushl(&cp.args, "read-tree", "--recurse-submodules", NULL);
+
+       if (flags & SUBMODULE_MOVE_HEAD_DRY_RUN)
+               argv_array_push(&cp.args, "-n");
+       else
+               argv_array_push(&cp.args, "-u");
+
+       if (flags & SUBMODULE_MOVE_HEAD_FORCE)
+               argv_array_push(&cp.args, "--reset");
+       else
+               argv_array_push(&cp.args, "-m");
+
+       argv_array_push(&cp.args, old ? old : EMPTY_TREE_SHA1_HEX);
+       argv_array_push(&cp.args, new ? new : EMPTY_TREE_SHA1_HEX);
+
+       if (run_command(&cp)) {
+               ret = -1;
+               goto out;
+       }
+
+       if (!(flags & SUBMODULE_MOVE_HEAD_DRY_RUN)) {
+               if (new) {
+                       child_process_init(&cp);
+                       /* also set the HEAD accordingly */
+                       cp.git_cmd = 1;
+                       cp.no_stdin = 1;
+                       cp.dir = path;
+
+                       prepare_submodule_repo_env(&cp.env_array);
+                       argv_array_pushl(&cp.args, "update-ref", "HEAD", new, NULL);
+
+                       if (run_command(&cp)) {
+                               ret = -1;
+                               goto out;
+                       }
+               } else {
+                       struct strbuf sb = STRBUF_INIT;
+
+                       strbuf_addf(&sb, "%s/.git", path);
+                       unlink_or_warn(sb.buf);
+                       strbuf_release(&sb);
+
+                       if (is_empty_dir(path))
+                               rmdir_or_warn(path);
+               }
+       }
+out:
+       return ret;
 }
 
 static int find_first_merges(struct object_array *result, const char *path,
@@ -977,7 +1692,7 @@ static int find_first_merges(struct object_array *result, const char *path,
        memset(&rev_opts, 0, sizeof(rev_opts));
 
        /* get all revisions that merge commit a */
-       snprintf(merged_revision, sizeof(merged_revision), "^%s",
+       xsnprintf(merged_revision, sizeof(merged_revision), "^%s",
                        oid_to_hex(&a->object.oid));
        init_revisions(&revs, NULL);
        rev_opts.submodule = path;
@@ -1030,9 +1745,9 @@ static void print_commit(struct commit *commit)
 #define MERGE_WARNING(path, msg) \
        warning("Failed to merge submodule %s (%s)", path, msg);
 
-int merge_submodule(unsigned char result[20], const char *path,
-                   const unsigned char base[20], const unsigned char a[20],
-                   const unsigned char b[20], int search)
+int merge_submodule(struct object_id *result, const char *path,
+                   const struct object_id *base, const struct object_id *a,
+                   const struct object_id *b, int search)
 {
        struct commit *commit_base, *commit_a, *commit_b;
        int parent_count;
@@ -1041,14 +1756,14 @@ int merge_submodule(unsigned char result[20], const char *path,
        int i;
 
        /* store a in result in case we fail */
-       hashcpy(result, a);
+       oidcpy(result, a);
 
        /* we can not handle deletion conflicts */
-       if (is_null_sha1(base))
+       if (is_null_oid(base))
                return 0;
-       if (is_null_sha1(a))
+       if (is_null_oid(a))
                return 0;
-       if (is_null_sha1(b))
+       if (is_null_oid(b))
                return 0;
 
        if (add_submodule_odb(path)) {
@@ -1072,11 +1787,11 @@ int merge_submodule(unsigned char result[20], const char *path,
 
        /* Case #1: a is contained in b or vice versa */
        if (in_merge_bases(commit_a, commit_b)) {
-               hashcpy(result, b);
+               oidcpy(result, b);
                return 1;
        }
        if (in_merge_bases(commit_b, commit_a)) {
-               hashcpy(result, a);
+               oidcpy(result, a);
                return 1;
        }
 
@@ -1122,42 +1837,241 @@ int merge_submodule(unsigned char result[20], const char *path,
        return 0;
 }
 
-/* Update gitfile and core.worktree setting to connect work tree and git dir */
-void connect_work_tree_and_git_dir(const char *work_tree, const char *git_dir)
+int parallel_submodules(void)
+{
+       return parallel_jobs;
+}
+
+/*
+ * Embeds a single submodules git directory into the superprojects git dir,
+ * non recursively.
+ */
+static void relocate_single_git_dir_into_superproject(const char *prefix,
+                                                     const char *path)
 {
-       struct strbuf file_name = STRBUF_INIT;
-       struct strbuf rel_path = STRBUF_INIT;
-       const char *real_work_tree = xstrdup(real_path(work_tree));
+       char *old_git_dir = NULL, *real_old_git_dir = NULL, *real_new_git_dir = NULL;
+       const char *new_git_dir;
+       const struct submodule *sub;
+
+       if (submodule_uses_worktrees(path))
+               die(_("relocate_gitdir for submodule '%s' with "
+                     "more than one worktree not supported"), path);
+
+       old_git_dir = xstrfmt("%s/.git", path);
+       if (read_gitfile(old_git_dir))
+               /* If it is an actual gitfile, it doesn't need migration. */
+               return;
+
+       real_old_git_dir = real_pathdup(old_git_dir, 1);
+
+       sub = submodule_from_path(null_sha1, path);
+       if (!sub)
+               die(_("could not lookup name for submodule '%s'"), path);
+
+       new_git_dir = git_path("modules/%s", sub->name);
+       if (safe_create_leading_directories_const(new_git_dir) < 0)
+               die(_("could not create directory '%s'"), new_git_dir);
+       real_new_git_dir = real_pathdup(new_git_dir, 1);
 
-       /* Update gitfile */
-       strbuf_addf(&file_name, "%s/.git", work_tree);
-       write_file(file_name.buf, "gitdir: %s",
-                  relative_path(git_dir, real_work_tree, &rel_path));
+       fprintf(stderr, _("Migrating git directory of '%s%s' from\n'%s' to\n'%s'\n"),
+               get_super_prefix_or_empty(), path,
+               real_old_git_dir, real_new_git_dir);
 
-       /* Update core.worktree setting */
-       strbuf_reset(&file_name);
-       strbuf_addf(&file_name, "%s/config", git_dir);
-       git_config_set_in_file(file_name.buf, "core.worktree",
-                              relative_path(real_work_tree, git_dir,
-                                            &rel_path));
+       relocate_gitdir(path, real_old_git_dir, real_new_git_dir);
 
-       strbuf_release(&file_name);
-       strbuf_release(&rel_path);
-       free((void *)real_work_tree);
+       free(old_git_dir);
+       free(real_old_git_dir);
+       free(real_new_git_dir);
 }
 
-int parallel_submodules(void)
+/*
+ * Migrate the git directory of the submodule given by path from
+ * having its git directory within the working tree to the git dir nested
+ * in its superprojects git dir under modules/.
+ */
+void absorb_git_dir_into_superproject(const char *prefix,
+                                     const char *path,
+                                     unsigned flags)
 {
-       return parallel_jobs;
+       int err_code;
+       const char *sub_git_dir;
+       struct strbuf gitdir = STRBUF_INIT;
+       strbuf_addf(&gitdir, "%s/.git", path);
+       sub_git_dir = resolve_gitdir_gently(gitdir.buf, &err_code);
+
+       /* Not populated? */
+       if (!sub_git_dir) {
+               const struct submodule *sub;
+
+               if (err_code == READ_GITFILE_ERR_STAT_FAILED) {
+                       /* unpopulated as expected */
+                       strbuf_release(&gitdir);
+                       return;
+               }
+
+               if (err_code != READ_GITFILE_ERR_NOT_A_REPO)
+                       /* We don't know what broke here. */
+                       read_gitfile_error_die(err_code, path, NULL);
+
+               /*
+               * Maybe populated, but no git directory was found?
+               * This can happen if the superproject is a submodule
+               * itself and was just absorbed. The absorption of the
+               * superproject did not rewrite the git file links yet,
+               * fix it now.
+               */
+               sub = submodule_from_path(null_sha1, path);
+               if (!sub)
+                       die(_("could not lookup name for submodule '%s'"), path);
+               connect_work_tree_and_git_dir(path,
+                       git_path("modules/%s", sub->name));
+       } else {
+               /* Is it already absorbed into the superprojects git dir? */
+               char *real_sub_git_dir = real_pathdup(sub_git_dir, 1);
+               char *real_common_git_dir = real_pathdup(get_git_common_dir(), 1);
+
+               if (!starts_with(real_sub_git_dir, real_common_git_dir))
+                       relocate_single_git_dir_into_superproject(prefix, path);
+
+               free(real_sub_git_dir);
+               free(real_common_git_dir);
+       }
+       strbuf_release(&gitdir);
+
+       if (flags & ABSORB_GITDIR_RECURSE_SUBMODULES) {
+               struct child_process cp = CHILD_PROCESS_INIT;
+               struct strbuf sb = STRBUF_INIT;
+
+               if (flags & ~ABSORB_GITDIR_RECURSE_SUBMODULES)
+                       die("BUG: we don't know how to pass the flags down?");
+
+               strbuf_addstr(&sb, get_super_prefix_or_empty());
+               strbuf_addstr(&sb, path);
+               strbuf_addch(&sb, '/');
+
+               cp.dir = path;
+               cp.git_cmd = 1;
+               cp.no_stdin = 1;
+               argv_array_pushl(&cp.args, "--super-prefix", sb.buf,
+                                          "submodule--helper",
+                                          "absorb-git-dirs", NULL);
+               prepare_submodule_repo_env(&cp.env_array);
+               if (run_command(&cp))
+                       die(_("could not recurse into submodule '%s'"), path);
+
+               strbuf_release(&sb);
+       }
 }
 
-void prepare_submodule_repo_env(struct argv_array *out)
+const char *get_superproject_working_tree(void)
 {
-       const char * const *var;
+       struct child_process cp = CHILD_PROCESS_INIT;
+       struct strbuf sb = STRBUF_INIT;
+       const char *one_up = real_path_if_valid("../");
+       const char *cwd = xgetcwd();
+       const char *ret = NULL;
+       const char *subpath;
+       int code;
+       ssize_t len;
 
-       for (var = local_repo_env; *var; var++) {
-               if (strcmp(*var, CONFIG_DATA_ENVIRONMENT))
-                       argv_array_push(out, *var);
+       if (!is_inside_work_tree())
+               /*
+                * FIXME:
+                * We might have a superproject, but it is harder
+                * to determine.
+                */
+               return NULL;
+
+       if (!one_up)
+               return NULL;
+
+       subpath = relative_path(cwd, one_up, &sb);
+
+       prepare_submodule_repo_env(&cp.env_array);
+       argv_array_pop(&cp.env_array);
+
+       argv_array_pushl(&cp.args, "--literal-pathspecs", "-C", "..",
+                       "ls-files", "-z", "--stage", "--full-name", "--",
+                       subpath, NULL);
+       strbuf_reset(&sb);
+
+       cp.no_stdin = 1;
+       cp.no_stderr = 1;
+       cp.out = -1;
+       cp.git_cmd = 1;
+
+       if (start_command(&cp))
+               die(_("could not start ls-files in .."));
+
+       len = strbuf_read(&sb, cp.out, PATH_MAX);
+       close(cp.out);
+
+       if (starts_with(sb.buf, "160000")) {
+               int super_sub_len;
+               int cwd_len = strlen(cwd);
+               char *super_sub, *super_wt;
+
+               /*
+                * There is a superproject having this repo as a submodule.
+                * The format is <mode> SP <hash> SP <stage> TAB <full name> \0,
+                * We're only interested in the name after the tab.
+                */
+               super_sub = strchr(sb.buf, '\t') + 1;
+               super_sub_len = sb.buf + sb.len - super_sub - 1;
+
+               if (super_sub_len > cwd_len ||
+                   strcmp(&cwd[cwd_len - super_sub_len], super_sub))
+                       die (_("BUG: returned path string doesn't match cwd?"));
+
+               super_wt = xstrdup(cwd);
+               super_wt[cwd_len - super_sub_len] = '\0';
+
+               ret = real_path(super_wt);
+               free(super_wt);
+       }
+       strbuf_release(&sb);
+
+       code = finish_command(&cp);
+
+       if (code == 128)
+               /* '../' is not a git repository */
+               return NULL;
+       if (code == 0 && len == 0)
+               /* There is an unrelated git repository at '../' */
+               return NULL;
+       if (code)
+               die(_("ls-tree returned unexpected return code %d"), code);
+
+       return ret;
+}
+
+int submodule_to_gitdir(struct strbuf *buf, const char *submodule)
+{
+       const struct submodule *sub;
+       const char *git_dir;
+       int ret = 0;
+
+       strbuf_reset(buf);
+       strbuf_addstr(buf, submodule);
+       strbuf_complete(buf, '/');
+       strbuf_addstr(buf, ".git");
+
+       git_dir = read_gitfile(buf->buf);
+       if (git_dir) {
+               strbuf_reset(buf);
+               strbuf_addstr(buf, git_dir);
        }
-       argv_array_push(out, "GIT_DIR=.git");
+       if (!is_git_directory(buf->buf)) {
+               gitmodules_config();
+               sub = submodule_from_path(null_sha1, submodule);
+               if (!sub) {
+                       ret = -1;
+                       goto cleanup;
+               }
+               strbuf_reset(buf);
+               strbuf_git_path(buf, "%s/%s", "modules", sub->name);
+       }
+
+cleanup:
+       return ret;
 }