Merge pull request #852 from arrbee/submodule-extensions

Submodule extensions

12 files changed, 1485 insertions, 241 deletions

diff --git a/src/buffer.c b/src/buffer.c
index b57998e1..61cfaf9e 100644
--- a/src/buffer.c
+++ b/src/buffer.c
@@ -144,31 +144,40 @@ int git_buf_puts(git_buf *buf, const char *string)
 int git_buf_puts_escaped(
 	git_buf *buf, const char *string, const char *esc_chars, const char *esc_with)
 {
-	const char *scan = string;
-	size_t total = 0, esc_with_len = strlen(esc_with);
+	const char *scan;
+	size_t total = 0, esc_len = strlen(esc_with), count;
 
-	while (*scan) {
-		size_t count = strcspn(scan, esc_chars);
-		total += count + 1 + esc_with_len;
-		scan += count + 1;
+	if (!string)
+		return 0;
+
+	for (scan = string; *scan; ) {
+		/* count run of non-escaped characters */
+		count = strcspn(scan, esc_chars);
+		total += count;
+		scan += count;
+		/* count run of escaped characters */
+		count = strspn(scan, esc_chars);
+		total += count * (esc_len + 1);
+		scan += count;
 	}
 
 	ENSURE_SIZE(buf, buf->size + total + 1);
 
 	for (scan = string; *scan; ) {
-		size_t count = strcspn(scan, esc_chars);
+		count = strcspn(scan, esc_chars);
 
 		memmove(buf->ptr + buf->size, scan, count);
 		scan += count;
 		buf->size += count;
 
-		if (*scan) {
-			memmove(buf->ptr + buf->size, esc_with, esc_with_len);
-			buf->size += esc_with_len;
-
-			memmove(buf->ptr + buf->size, scan, 1);
-			scan += 1;
-			buf->size += 1;
+		for (count = strspn(scan, esc_chars); count > 0; --count) {
+			/* copy escape sequence */
+			memmove(buf->ptr + buf->size, esc_with, esc_len);
+			buf->size += esc_len;
+			/* copy character to be escaped */
+			buf->ptr[buf->size] = *scan;
+			buf->size++;
+			scan++;
 		}
 	}
 
diff --git a/src/checkout.c b/src/checkout.c
index ac540391..88df2128 100644
--- a/src/checkout.c
+++ b/src/checkout.c
@@ -31,7 +31,7 @@ typedef struct tree_walk_data
 	git_checkout_opts *opts;
 	git_repository *repo;
 	git_odb *odb;
-	bool do_symlinks;
+	bool no_symlinks;
 } tree_walk_data;
 
 
@@ -48,9 +48,9 @@ static int blob_contents_to_link(tree_walk_data *data, git_buf *fnbuf,
 			/* Create the link */
 			const char *new = git_buf_cstr(&linktarget),
 						  *old = git_buf_cstr(fnbuf);
-			retcode = data->do_symlinks
-				? p_symlink(new, old)
-				: git_futils_fake_symlink(new, old);
+			retcode = data->no_symlinks
+				? git_futils_fake_symlink(new, old)
+				: p_symlink(new, old);
 		}
 		git_buf_free(&linktarget);
 		git_blob_free(blob);
@@ -176,13 +176,14 @@ int git_checkout_head(git_repository *repo, git_checkout_opts *opts, git_indexer
 		return GIT_ERROR;
 	}
 
+	memset(&payload, 0, sizeof(payload));
+
 	/* Determine if symlinks should be handled */
-	if (!git_repository_config(&cfg, repo)) {
+	if (!git_repository_config__weakptr(&cfg, repo)) {
 		int temp = true;
 		if (!git_config_get_bool(&temp, cfg, "core.symlinks")) {
-			payload.do_symlinks = !!temp;
+			payload.no_symlinks = !temp;
 		}
-		git_config_free(cfg);
 	}
 
 	stats->total = stats->processed = 0;
diff --git a/src/config_file.c b/src/config_file.c
index 547509b9..d3fb56aa 100644
--- a/src/config_file.c
+++ b/src/config_file.c
@@ -195,7 +195,7 @@ static int file_foreach(
 	void *data)
 {
 	diskfile_backend *b = (diskfile_backend *)backend;
-	cvar_t *var;
+	cvar_t *var, *next_var;
 	const char *key;
 	regex_t regex;
 	int result = 0;
@@ -212,7 +212,9 @@ static int file_foreach(
 	}
 
 	git_strmap_foreach(b->values, key, var,
-		for (; var != NULL; var = CVAR_LIST_NEXT(var)) {
+		for (; var != NULL; var = next_var) {
+			next_var = CVAR_LIST_NEXT(var);
+
 			/* skip non-matching keys if regexp was provided */
 			if (regexp && regexec(&regex, key, 0, NULL, 0) != 0)
 				continue;
@@ -253,11 +255,17 @@ static int config_set(git_config_file *cfg, const char *name, const char *value)
 		char *tmp = NULL;
 
 		git__free(key);
+
 		if (existing->next != NULL) {
 			giterr_set(GITERR_CONFIG, "Multivar incompatible with simple set");
 			return -1;
 		}
 
+		/* don't update if old and new values already match */
+		if ((!existing->value && !value) ||
+			(existing->value && value && !strcmp(existing->value, value)))
+			return 0;
+
 		if (value) {
 			tmp = git__strdup(value);
 			GITERR_CHECK_ALLOC(tmp);
diff --git a/src/config_file.h b/src/config_file.h
index c3129288..bf687b51 100644
--- a/src/config_file.h
+++ b/src/config_file.h
@@ -19,12 +19,24 @@ GIT_INLINE(void) git_config_file_free(git_config_file *cfg)
 	cfg->free(cfg);
 }
 
+GIT_INLINE(int) git_config_file_get_string(
+	const char **out, git_config_file *cfg, const char *name)
+{
+	return cfg->get(cfg, name, out);
+}
+
 GIT_INLINE(int) git_config_file_set_string(
 	git_config_file *cfg, const char *name, const char *value)
 {
 	return cfg->set(cfg, name, value);
 }
 
+GIT_INLINE(int) git_config_file_delete(
+	git_config_file *cfg, const char *name)
+{
+	return cfg->del(cfg, name);
+}
+
 GIT_INLINE(int) git_config_file_foreach(
 	git_config_file *cfg,
 	int (*fn)(const char *key, const char *value, void *data),
diff --git a/src/diff.c b/src/diff.c
index 9abf8b9f..430f52e0 100644
--- a/src/diff.c
+++ b/src/diff.c
@@ -530,7 +530,7 @@ static int maybe_modified(
 				status = GIT_DELTA_UNMODIFIED;
 			else if (git_submodule_lookup(&sub, diff->repo, nitem->path) < 0)
 				return -1;
-			else if (sub->ignore == GIT_SUBMODULE_IGNORE_ALL)
+			else if (git_submodule_ignore(sub) == GIT_SUBMODULE_IGNORE_ALL)
 				status = GIT_DELTA_UNMODIFIED;
 			else {
 				/* TODO: support other GIT_SUBMODULE_IGNORE values */
diff --git a/src/diff_output.c b/src/diff_output.c
index d269a4ce..2bf939f3 100644
--- a/src/diff_output.c
+++ b/src/diff_output.c
@@ -718,6 +718,25 @@ int git_diff_print_patch(
 	return error;
 }
 
+int git_diff_entrycount(git_diff_list *diff, int delta_t)
+{
+	int count = 0;
+	unsigned int i;
+	git_diff_delta *delta;
+
+	assert(diff);
+
+	if (delta_t < 0)
+		return diff->deltas.length;
+
+	git_vector_foreach(&diff->deltas, i, delta) {
+		if (delta->status == (git_delta_t)delta_t)
+			count++;
+	}
+
+	return count;
+}
+
 int git_diff_blobs(
 	git_blob *old_blob,
 	git_blob *new_blob,
diff --git a/src/errors.c b/src/errors.c
index d43d7d9b..802ad364 100644
--- a/src/errors.c
+++ b/src/errors.c
@@ -110,6 +110,11 @@ void giterr_set_regex(const regex_t *regex, int error_code)
 void giterr_clear(void)
 {
 	GIT_GLOBAL->last_error = NULL;
+
+	errno = 0;
+#ifdef GIT_WIN32
+	SetLastError(0);
+#endif
 }
 
 const git_error *giterr_last(void)
diff --git a/src/filebuf.c b/src/filebuf.c
index 8b3ebb3e..cfc8528e 100644
--- a/src/filebuf.c
+++ b/src/filebuf.c
@@ -50,6 +50,7 @@ static int lock_file(git_filebuf *file, int flags)
 		if (flags & GIT_FILEBUF_FORCE)
 			p_unlink(file->path_lock);
 		else {
+			giterr_clear(); /* actual OS error code just confuses */
 			giterr_set(GITERR_OS,
 				"Failed to lock file '%s' for writing", file->path_lock);
 			return -1;
diff --git a/src/iterator.c b/src/iterator.c
index 819b0e22..92fe6713 100644
--- a/src/iterator.c
+++ b/src/iterator.c
@@ -525,7 +525,9 @@ static int workdir_iterator__advance(
 	while ((wf = wi->stack) != NULL) {
 		next = git_vector_get(&wf->entries, ++wf->index);
 		if (next != NULL) {
-			if (strcmp(next->path, DOT_GIT "/") == 0)
+			/* match git's behavior of ignoring anything named ".git" */
+			if (strcmp(next->path, DOT_GIT "/") == 0 ||
+				strcmp(next->path, DOT_GIT) == 0)
 				continue;
 			/* else found a good entry */
 			break;
@@ -607,8 +609,8 @@ static int workdir_iterator__update_entry(workdir_iterator *wi)
 
 	wi->entry.path = ps->path;
 
-	/* skip over .git directory */
-	if (strcmp(ps->path, DOT_GIT "/") == 0)
+	/* skip over .git entry */
+	if (strcmp(ps->path, DOT_GIT "/") == 0 || strcmp(ps->path, DOT_GIT) == 0)
 		return workdir_iterator__advance((git_iterator *)wi, NULL);
 
 	/* if there is an error processing the entry, treat as ignored */
@@ -629,15 +631,10 @@ static int workdir_iterator__update_entry(workdir_iterator *wi)
 
 	/* detect submodules */
 	if (S_ISDIR(wi->entry.mode)) {
-		bool is_submodule = git_path_contains(&wi->path, DOT_GIT);
-
-		/* if there is no .git, still check submodules data */
-		if (!is_submodule) {
-			int res = git_submodule_lookup(NULL, wi->repo, wi->entry.path);
-			is_submodule = (res == 0);
-			if (res == GIT_ENOTFOUND)
-				giterr_clear();
-		}
+		int res = git_submodule_lookup(NULL, wi->repo, wi->entry.path);
+		bool is_submodule = (res == 0);
+		if (res == GIT_ENOTFOUND)
+			giterr_clear();
 
 		/* if submodule, mark as GITLINK and remove trailing slash */
 		if (is_submodule) {
diff --git a/src/repository.c b/src/repository.c
index 18788d18..c12df25c 100644
--- a/src/repository.c
+++ b/src/repository.c
@@ -146,8 +146,13 @@ static int load_workdir(git_repository *repo, git_buf *parent_path)
 		return -1;
 
 	error = git_config_get_string(&worktree, config, "core.worktree");
-	if (!error && worktree != NULL)
-		repo->workdir = git__strdup(worktree);
+	if (!error && worktree != NULL) {
+		error = git_path_prettify_dir(
+			&worktree_buf, worktree, repo->path_repository);
+		if (error < 0)
+			return error;
+		repo->workdir = git_buf_detach(&worktree_buf);
+	}
 	else if (error != GIT_ENOTFOUND)
 		return error;
 	else {
diff --git a/src/submodule.c b/src/submodule.c
index b8537cb8..a9de9ee6 100644
--- a/src/submodule.c
+++ b/src/submodule.c
@@ -17,18 +17,24 @@
 #include "config_file.h"
 #include "config.h"
 #include "repository.h"
+#include "submodule.h"
+#include "tree.h"
+#include "iterator.h"
+
+#define GIT_MODULES_FILE ".gitmodules"
 
 static git_cvar_map _sm_update_map[] = {
 	{GIT_CVAR_STRING, "checkout", GIT_SUBMODULE_UPDATE_CHECKOUT},
 	{GIT_CVAR_STRING, "rebase", GIT_SUBMODULE_UPDATE_REBASE},
-	{GIT_CVAR_STRING, "merge", GIT_SUBMODULE_UPDATE_MERGE}
+	{GIT_CVAR_STRING, "merge", GIT_SUBMODULE_UPDATE_MERGE},
+	{GIT_CVAR_STRING, "none", GIT_SUBMODULE_UPDATE_NONE},
 };
 
 static git_cvar_map _sm_ignore_map[] = {
-	{GIT_CVAR_STRING, "all", GIT_SUBMODULE_IGNORE_ALL},
-	{GIT_CVAR_STRING, "dirty", GIT_SUBMODULE_IGNORE_DIRTY},
+	{GIT_CVAR_STRING, "none", GIT_SUBMODULE_IGNORE_NONE},
 	{GIT_CVAR_STRING, "untracked", GIT_SUBMODULE_IGNORE_UNTRACKED},
-	{GIT_CVAR_STRING, "none", GIT_SUBMODULE_IGNORE_NONE}
+	{GIT_CVAR_STRING, "dirty", GIT_SUBMODULE_IGNORE_DIRTY},
+	{GIT_CVAR_STRING, "all", GIT_SUBMODULE_IGNORE_ALL},
 };
 
 static kh_inline khint_t str_hash_no_trailing_slash(const char *s)
@@ -36,7 +42,7 @@ static kh_inline khint_t str_hash_no_trailing_slash(const char *s)
 	khint_t h;
 
 	for (h = 0; *s; ++s)
-		if (s[1] || *s != '/')
+		if (s[1] != '\0' || *s != '/')
 			h = (h << 5) - h + *s;
 
 	return h;
@@ -47,29 +53,790 @@ static kh_inline int str_equal_no_trailing_slash(const char *a, const char *b)
 	size_t alen = a ? strlen(a) : 0;
 	size_t blen = b ? strlen(b) : 0;
 
-	if (alen && a[alen] == '/')
+	if (alen > 0 && a[alen - 1] == '/')
 		alen--;
-	if (blen && b[blen] == '/')
+	if (blen > 0 && b[blen - 1] == '/')
 		blen--;
 
 	return (alen == blen && strncmp(a, b, alen) == 0);
 }
 
-__KHASH_IMPL(str, static kh_inline, const char *, void *, 1, str_hash_no_trailing_slash, str_equal_no_trailing_slash);
+__KHASH_IMPL(
+	str, static kh_inline, const char *, void *, 1,
+	str_hash_no_trailing_slash, str_equal_no_trailing_slash);
+
+static int load_submodule_config(git_repository *repo, bool force);
+static git_config_file *open_gitmodules(git_repository *, bool, const git_oid *);
+static int lookup_head_remote(git_buf *url, git_repository *repo);
+static int submodule_get(git_submodule **, git_repository *, const char *, const char *);
+static void submodule_release(git_submodule *sm, int decr);
+static int submodule_load_from_index(git_repository *, const git_index_entry *);
+static int submodule_load_from_head(git_repository*, const char*, const git_oid*);
+static int submodule_load_from_config(const char *, const char *, void *);
+static int submodule_load_from_wd_lite(git_submodule *, const char *, void *);
+static int submodule_update_config(git_submodule *, const char *, const char *, bool, bool);
+static void submodule_mode_mismatch(git_repository *, const char *, unsigned int);
+static int submodule_index_status(unsigned int *status, git_submodule *sm);
+static int submodule_wd_status(unsigned int *status, git_submodule *sm);
 
-static git_submodule *submodule_alloc(const char *name)
+static int submodule_cmp(const void *a, const void *b)
 {
-	git_submodule *sm = git__calloc(1, sizeof(git_submodule));
-	if (sm == NULL)
-		return sm;
+	return strcmp(((git_submodule *)a)->name, ((git_submodule *)b)->name);
+}
 
-	sm->path = sm->name = git__strdup(name);
-	if (!sm->name) {
-		git__free(sm);
+static int submodule_config_key_trunc_puts(git_buf *key, const char *suffix)
+{
+	ssize_t idx = git_buf_rfind(key, '.');
+	git_buf_truncate(key, (size_t)(idx + 1));
+	return git_buf_puts(key, suffix);
+}
+
+/*
+ * PUBLIC APIS
+ */
+
+int git_submodule_lookup(
+	git_submodule **sm_ptr, /* NULL if user only wants to test existence */
+	git_repository *repo,
+	const char *name)       /* trailing slash is allowed */
+{
+	int error;
+	khiter_t pos;
+
+	assert(repo && name);
+
+	if ((error = load_submodule_config(repo, false)) < 0)
+		return error;
+
+	pos = git_strmap_lookup_index(repo->submodules, name);
+
+	if (!git_strmap_valid_index(repo->submodules, pos)) {
+		error = GIT_ENOTFOUND;
+
+		/* check if a plausible submodule exists at path */
+		if (git_repository_workdir(repo)) {
+			git_buf path = GIT_BUF_INIT;
+
+			if (git_buf_joinpath(&path, git_repository_workdir(repo), name) < 0)
+				return -1;
+
+			if (git_path_contains_dir(&path, DOT_GIT))
+				error = GIT_EEXISTS;
+
+			git_buf_free(&path);
+		}
+
+		return error;
+	}
+
+	if (sm_ptr)
+		*sm_ptr = git_strmap_value_at(repo->submodules, pos);
+
+	return 0;
+}
+
+int git_submodule_foreach(
+	git_repository *repo,
+	int (*callback)(git_submodule *sm, const char *name, void *payload),
+	void *payload)
+{
+	int error;
+	git_submodule *sm;
+	git_vector seen = GIT_VECTOR_INIT;
+	seen._cmp = submodule_cmp;
+
+	assert(repo && callback);
+
+	if ((error = load_submodule_config(repo, false)) < 0)
+		return error;
+
+	git_strmap_foreach_value(repo->submodules, sm, {
+		/* Usually the following will not come into play - it just prevents
+		 * us from issuing a callback twice for a submodule where the name
+		 * and path are not the same.
+		 */
+		if (sm->refcount > 1) {
+			if (git_vector_bsearch(&seen, sm) != GIT_ENOTFOUND)
+				continue;
+			if ((error = git_vector_insert(&seen, sm)) < 0)
+				break;
+		}
+
+		if (callback(sm, sm->name, payload)) {
+			error = GIT_EUSER;
+			break;
+		}
+	});
+
+	git_vector_free(&seen);
+
+	return error;
+}
+
+void git_submodule_config_free(git_repository *repo)
+{
+	git_strmap *smcfg;
+	git_submodule *sm;
+
+	assert(repo);
+
+	smcfg = repo->submodules;
+	repo->submodules = NULL;
+
+	if (smcfg == NULL)
+		return;
+
+	git_strmap_foreach_value(smcfg, sm, {
+		submodule_release(sm,1);
+	});
+	git_strmap_free(smcfg);
+}
+
+int git_submodule_add_setup(
+	git_submodule **submodule,
+	git_repository *repo,
+	const char *url,
+	const char *path,
+	int use_gitlink)
+{
+	int error = 0;
+	git_config_file *mods = NULL;
+	git_submodule *sm;
+	git_buf name = GIT_BUF_INIT, real_url = GIT_BUF_INIT;
+	git_repository_init_options initopt;
+	git_repository *subrepo = NULL;
+
+	assert(repo && url && path);
+
+	/* see if there is already an entry for this submodule */
+
+	if (git_submodule_lookup(&sm, repo, path) < 0)
+		giterr_clear();
+	else {
+		giterr_set(GITERR_SUBMODULE,
+			"Attempt to add a submodule that already exists");
+		return GIT_EEXISTS;
+	}
+
+	/* resolve parameters */
+
+	if (url[0] == '.' && (url[1] == '/' || (url[1] == '.' && url[2] == '/'))) {
+		if (!(error = lookup_head_remote(&real_url, repo)))
+			error = git_path_apply_relative(&real_url, url);
+	} else if (strchr(url, ':') != NULL || url[0] == '/') {
+		error = git_buf_sets(&real_url, url);
+	} else {
+		giterr_set(GITERR_SUBMODULE, "Invalid format for submodule URL");
+		error = -1;
+	}
+	if (error)
+		goto cleanup;
+
+	/* validate and normalize path */
+
+	if (git__prefixcmp(path, git_repository_workdir(repo)) == 0)
+		path += strlen(git_repository_workdir(repo));
+
+	if (git_path_root(path) >= 0) {
+		giterr_set(GITERR_SUBMODULE, "Submodule path must be a relative path");
+		error = -1;
+		goto cleanup;
+	}
+
+	/* update .gitmodules */
+
+	if ((mods = open_gitmodules(repo, true, NULL)) == NULL) {
+		giterr_set(GITERR_SUBMODULE,
+			"Adding submodules to a bare repository is not supported (for now)");
+		return -1;
+	}
+
+	if ((error = git_buf_printf(&name, "submodule.%s.path", path)) < 0 ||
+		(error = git_config_file_set_string(mods, name.ptr, path)) < 0)
+		goto cleanup;
+
+	if ((error = submodule_config_key_trunc_puts(&name, "url")) < 0 ||
+		(error = git_config_file_set_string(mods, name.ptr, real_url.ptr)) < 0)
+		goto cleanup;
+
+	git_buf_clear(&name);
+
+	/* init submodule repository and add origin remote as needed */
+
+	error = git_buf_joinpath(&name, git_repository_workdir(repo), path);
+	if (error < 0)
+		goto cleanup;
+
+	/* New style: sub-repo goes in <repo-dir>/modules/<name>/ with a
+	 * gitlink in the sub-repo workdir directory to that repository
+	 *
+	 * Old style: sub-repo goes directly into repo/<name>/.git/
+	 */
+
+	memset(&initopt, 0, sizeof(initopt));
+	initopt.flags = GIT_REPOSITORY_INIT_MKPATH |
+		GIT_REPOSITORY_INIT_NO_REINIT;
+	initopt.origin_url = real_url.ptr;
+
+	if (git_path_exists(name.ptr) &&
+		git_path_contains(&name, DOT_GIT))
+	{
+		/* repo appears to already exist - reinit? */
+	}
+	else if (use_gitlink) {
+		git_buf repodir = GIT_BUF_INIT;
+
+		error = git_buf_join_n(
+			&repodir, '/', 3, git_repository_path(repo), "modules", path);
+		if (error < 0)
+			goto cleanup;
+
+		initopt.workdir_path = name.ptr;
+		initopt.flags |= GIT_REPOSITORY_INIT_NO_DOTGIT_DIR;
+
+		error = git_repository_init_ext(&subrepo, repodir.ptr, &initopt);
+
+		git_buf_free(&repodir);
+	}
+	else {
+		error = git_repository_init_ext(&subrepo, name.ptr, &initopt);
+	}
+	if (error < 0)
+		goto cleanup;
+
+	/* add submodule to hash and "reload" it */
+
+	if (!(error = submodule_get(&sm, repo, path, NULL)) &&
+		!(error = git_submodule_reload(sm)))
+		error = git_submodule_init(sm, false);
+
+cleanup:
+	if (submodule != NULL)
+		*submodule = !error ? sm : NULL;
+
+	if (mods != NULL)
+		git_config_file_free(mods);
+	git_repository_free(subrepo);
+	git_buf_free(&real_url);
+	git_buf_free(&name);
+
+	return error;
+}
+
+int git_submodule_add_finalize(git_submodule *sm)
+{
+	int error;
+	git_index *index;
+
+	assert(sm);
+
+	if ((error = git_repository_index__weakptr(&index, sm->owner)) < 0 ||
+		(error = git_index_add(index, GIT_MODULES_FILE, 0)) < 0)
+		return error;
+
+	return git_submodule_add_to_index(sm, true);
+}
+
+int git_submodule_add_to_index(git_submodule *sm, int write_index)
+{
+	int error;
+	git_repository *repo, *sm_repo;
+	git_index *index;
+	git_buf path = GIT_BUF_INIT;
+	git_commit *head;
+	git_index_entry entry;
+	struct stat st;
+
+	assert(sm);
+
+	repo = sm->owner;
+
+	/* force reload of wd OID by git_submodule_open */
+	sm->flags = sm->flags & ~GIT_SUBMODULE_STATUS__WD_OID_VALID;
+
+	if ((error = git_repository_index__weakptr(&index, repo)) < 0 ||
+		(error = git_buf_joinpath(
+			&path, git_repository_workdir(repo), sm->path)) < 0 ||
+		(error = git_submodule_open(&sm_repo, sm)) < 0)
+		goto cleanup;
+
+	/* read stat information for submodule working directory */
+	if (p_stat(path.ptr, &st) < 0) {
+		giterr_set(GITERR_SUBMODULE,
+			"Cannot add submodule without working directory");
+		error = -1;
+		goto cleanup;
+	}
+
+	memset(&entry, 0, sizeof(entry));
+	entry.path = sm->path;
+	git_index__init_entry_from_stat(&st, &entry);
+
+	/* calling git_submodule_open will have set sm->wd_oid if possible */
+	if ((sm->flags & GIT_SUBMODULE_STATUS__WD_OID_VALID) == 0) {
+		giterr_set(GITERR_SUBMODULE,
+			"Cannot add submodule without HEAD to index");
+		error = -1;
+		goto cleanup;
+	}
+	git_oid_cpy(&entry.oid, &sm->wd_oid);
+
+	if ((error = git_commit_lookup(&head, sm_repo, &sm->wd_oid)) < 0)
+		goto cleanup;
+
+	entry.ctime.seconds = git_commit_time(head);
+	entry.ctime.nanoseconds = 0;
+	entry.mtime.seconds = git_commit_time(head);
+	entry.mtime.nanoseconds = 0;
+
+	git_commit_free(head);
+
+	/* add it */
+	error = git_index_add2(index, &entry);
+
+	/* write it, if requested */
+	if (!error && write_index) {
+		error = git_index_write(index);
+
+		if (!error)
+			git_oid_cpy(&sm->index_oid, &sm->wd_oid);
+	}
+
+cleanup:
+	git_repository_free(sm_repo);
+	git_buf_free(&path);
+	return error;
+}
+
+int git_submodule_save(git_submodule *submodule)
+{
+	int error = 0;
+	git_config_file *mods;
+	git_buf key = GIT_BUF_INIT;
+
+	assert(submodule);
+
+	mods = open_gitmodules(submodule->owner, true, NULL);
+	if (!mods) {
+		giterr_set(GITERR_SUBMODULE,
+			"Adding submodules to a bare repository is not supported (for now)");
+		return -1;
+	}
+
+	if ((error = git_buf_printf(&key, "submodule.%s.", submodule->name)) < 0)
+		goto cleanup;
+
+	/* save values for path, url, update, ignore, fetchRecurseSubmodules */
+
+	if ((error = submodule_config_key_trunc_puts(&key, "path")) < 0 ||
+		(error = git_config_file_set_string(mods, key.ptr, submodule->path)) < 0)
+		goto cleanup;
+
+	if ((error = submodule_config_key_trunc_puts(&key, "url")) < 0 ||
+		(error = git_config_file_set_string(mods, key.ptr, submodule->url)) < 0)
+		goto cleanup;
+
+	if (!(error = submodule_config_key_trunc_puts(&key, "update")) &&
+		submodule->update != GIT_SUBMODULE_UPDATE_DEFAULT)
+	{
+		const char *val = (submodule->update == GIT_SUBMODULE_UPDATE_CHECKOUT) ?
+			NULL : _sm_update_map[submodule->update].str_match;
+		error = git_config_file_set_string(mods, key.ptr, val);
+	}
+	if (error < 0)
+		goto cleanup;
+
+	if (!(error = submodule_config_key_trunc_puts(&key, "ignore")) &&
+		submodule->ignore != GIT_SUBMODULE_IGNORE_DEFAULT)
+	{
+		const char *val = (submodule->ignore == GIT_SUBMODULE_IGNORE_NONE) ?
+			NULL : _sm_ignore_map[submodule->ignore].str_match;
+		error = git_config_file_set_string(mods, key.ptr, val);
+	}
+	if (error < 0)
+		goto cleanup;
+
+	if ((error = submodule_config_key_trunc_puts(
+			&key, "fetchRecurseSubmodules")) < 0 ||
+		(error = git_config_file_set_string(
+			mods, key.ptr, submodule->fetch_recurse ? "true" : "false")) < 0)
+		goto cleanup;
+
+	/* update internal defaults */
+
+	submodule->ignore_default = submodule->ignore;
+	submodule->update_default = submodule->update;
+	submodule->flags |= GIT_SUBMODULE_STATUS_IN_CONFIG;
+
+cleanup:
+	if (mods != NULL)
+		git_config_file_free(mods);
+	git_buf_free(&key);
+
+	return error;
+}
+
+git_repository *git_submodule_owner(git_submodule *submodule)
+{
+	assert(submodule);
+	return submodule->owner;
+}
+
+const char *git_submodule_name(git_submodule *submodule)
+{
+	assert(submodule);
+	return submodule->name;
+}
+
+const char *git_submodule_path(git_submodule *submodule)
+{
+	assert(submodule);
+	return submodule->path;
+}
+
+const char *git_submodule_url(git_submodule *submodule)
+{
+	assert(submodule);
+	return submodule->url;
+}
+
+int git_submodule_set_url(git_submodule *submodule, const char *url)
+{
+	assert(submodule && url);
+
+	git__free(submodule->url);
+
+	submodule->url = git__strdup(url);
+	GITERR_CHECK_ALLOC(submodule->url);
+
+	return 0;
+}
+
+const git_oid *git_submodule_index_oid(git_submodule *submodule)
+{
+	assert(submodule);
+
+	if (submodule->flags & GIT_SUBMODULE_STATUS__INDEX_OID_VALID)
+		return &submodule->index_oid;
+	else
+		return NULL;
+}
+
+const git_oid *git_submodule_head_oid(git_submodule *submodule)
+{
+	assert(submodule);
+
+	if (submodule->flags & GIT_SUBMODULE_STATUS__HEAD_OID_VALID)
+		return &submodule->head_oid;
+	else
 		return NULL;
+}
+
+const git_oid *git_submodule_wd_oid(git_submodule *submodule)
+{
+	assert(submodule);
+
+	if (!(submodule->flags & GIT_SUBMODULE_STATUS__WD_OID_VALID)) {
+		git_repository *subrepo;
+
+		/* calling submodule open grabs the HEAD OID if possible */
+		if (!git_submodule_open(&subrepo, submodule))
+			git_repository_free(subrepo);
+		else
+			giterr_clear();
 	}
 
+	if (submodule->flags & GIT_SUBMODULE_STATUS__WD_OID_VALID)
+		return &submodule->wd_oid;
+	else
+		return NULL;
+}
+
+git_submodule_ignore_t git_submodule_ignore(git_submodule *submodule)
+{
+	assert(submodule);
+	return submodule->ignore;
+}
+
+git_submodule_ignore_t git_submodule_set_ignore(
+	git_submodule *submodule, git_submodule_ignore_t ignore)
+{
+	git_submodule_ignore_t old;
+
+	assert(submodule);
+
+	if (ignore == GIT_SUBMODULE_IGNORE_DEFAULT)
+		ignore = submodule->ignore_default;
+
+	old = submodule->ignore;
+	submodule->ignore = ignore;
+	return old;
+}
+
+git_submodule_update_t git_submodule_update(git_submodule *submodule)
+{
+	assert(submodule);
+	return submodule->update;
+}
+
+git_submodule_update_t git_submodule_set_update(
+	git_submodule *submodule, git_submodule_update_t update)
+{
+	git_submodule_update_t old;
+
+	assert(submodule);
+
+	if (update == GIT_SUBMODULE_UPDATE_DEFAULT)
+		update = submodule->update_default;
+
+	old = submodule->update;
+	submodule->update = update;
+	return old;
+}
+
+int git_submodule_init(git_submodule *submodule, int overwrite)
+{
+	int error;
+
+	/* write "submodule.NAME.url" */
+
+	if (!submodule->url) {
+		giterr_set(GITERR_SUBMODULE,
+			"No URL configured for submodule '%s'", submodule->name);
+		return -1;
+	}
+
+	error = submodule_update_config(
+		submodule, "url", submodule->url, overwrite != 0, false);
+	if (error < 0)
+		return error;
+
+	/* write "submodule.NAME.update" if not default */
+
+	if (submodule->update == GIT_SUBMODULE_UPDATE_CHECKOUT)
+		error = submodule_update_config(
+			submodule, "update", NULL, (overwrite != 0), false);
+	else if (submodule->update != GIT_SUBMODULE_UPDATE_DEFAULT)
+		error = submodule_update_config(
+			submodule, "update",
+			_sm_update_map[submodule->update].str_match,
+			(overwrite != 0), false);
+
+	return error;
+}
+
+int git_submodule_sync(git_submodule *submodule)
+{
+	if (!submodule->url) {
+		giterr_set(GITERR_SUBMODULE,
+			"No URL configured for submodule '%s'", submodule->name);
+		return -1;
+	}
+
+	/* copy URL over to config only if it already exists */
+
+	return submodule_update_config(
+		submodule, "url", submodule->url, true, true);
+}
+
+int git_submodule_open(
+	git_repository **subrepo,
+	git_submodule *submodule)
+{
+	int error;
+	git_buf path = GIT_BUF_INIT;
+	git_repository *repo;
+	const char *workdir;
+
+	assert(submodule && subrepo);
+
+	repo = submodule->owner;
+	workdir = git_repository_workdir(repo);
+
+	if (!workdir) {
+		giterr_set(GITERR_REPOSITORY,
+			"Cannot open submodule repository in a bare repo");
+		return GIT_ENOTFOUND;
+	}
+
+	if ((submodule->flags & GIT_SUBMODULE_STATUS_IN_WD) == 0) {
+		giterr_set(GITERR_REPOSITORY,
+			"Cannot open submodule repository that is not checked out");
+		return GIT_ENOTFOUND;
+	}
+
+	if (git_buf_joinpath(&path, workdir, submodule->path) < 0)
+		return -1;
+
+	error = git_repository_open(subrepo, path.ptr);
+
+	git_buf_free(&path);
+
+	/* if we have opened the submodule successfully, let's grab the HEAD OID */
+	if (!error && !(submodule->flags & GIT_SUBMODULE_STATUS__WD_OID_VALID)) {
+		if (!git_reference_name_to_oid(
+				&submodule->wd_oid, *subrepo, GIT_HEAD_FILE))
+			submodule->flags |= GIT_SUBMODULE_STATUS__WD_OID_VALID;
+		else
+			giterr_clear();
+	}
+
+	return error;
+}
+
+int git_submodule_reload_all(git_repository *repo)
+{
+	assert(repo);
+	return load_submodule_config(repo, true);
+}
+
+int git_submodule_reload(git_submodule *submodule)
+{
+	git_repository *repo;
+	git_index *index;
+	int pos, error;
+	git_tree *head;
+	git_config_file *mods;
+
+	assert(submodule);
+
+	/* refresh index data */
+
+	repo = submodule->owner;
+	if (git_repository_index__weakptr(&index, repo) < 0)
+		return -1;
+
+	submodule->flags = submodule->flags &
+		~(GIT_SUBMODULE_STATUS_IN_INDEX |
+		  GIT_SUBMODULE_STATUS__INDEX_OID_VALID);
+
+	pos = git_index_find(index, submodule->path);
+	if (pos >= 0) {
+		git_index_entry *entry = git_index_get(index, pos);
+
+		if (S_ISGITLINK(entry->mode)) {
+			if ((error = submodule_load_from_index(repo, entry)) < 0)
+				return error;
+		} else {
+			submodule_mode_mismatch(
+				repo, entry->path, GIT_SUBMODULE_STATUS__INDEX_NOT_SUBMODULE);
+		}
+	}
+
+	/* refresh HEAD tree data */
+
+	if (!(error = git_repository_head_tree(&head, repo))) {
+		git_tree_entry *te;
+
+		submodule->flags = submodule->flags &
+			~(GIT_SUBMODULE_STATUS_IN_HEAD |
+			  GIT_SUBMODULE_STATUS__HEAD_OID_VALID);
+
+		if (!(error = git_tree_entry_bypath(&te, head, submodule->path))) {
+
+			if (S_ISGITLINK(te->attr)) {
+				error = submodule_load_from_head(repo, submodule->path, &te->oid);
+			} else {
+				submodule_mode_mismatch(
+					repo, submodule->path,
+					GIT_SUBMODULE_STATUS__HEAD_NOT_SUBMODULE);
+			}
+
+			git_tree_entry_free(te);
+		}
+		else if (error == GIT_ENOTFOUND) {
+			giterr_clear();
+			error = 0;
+		}
+
+		git_tree_free(head);
+	}
+
+	if (error < 0)
+		return error;
+
+	/* refresh config data */
+
+	if ((mods = open_gitmodules(repo, false, NULL)) != NULL) {
+		git_buf path = GIT_BUF_INIT;
+
+		git_buf_sets(&path, "submodule\\.");
+		git_buf_puts_escape_regex(&path, submodule->name);
+		git_buf_puts(&path, ".*");
+
+		if (git_buf_oom(&path))
+			error = -1;
+		else
+			error = git_config_file_foreach_match(
+				mods, path.ptr, submodule_load_from_config, repo);
+
+		git_buf_free(&path);
+		git_config_file_free(mods);
+	}
+
+	if (error < 0)
+		return error;
+
+	/* refresh wd data */
+
+	submodule->flags = submodule->flags &
+		~(GIT_SUBMODULE_STATUS_IN_WD | GIT_SUBMODULE_STATUS__WD_OID_VALID);
+
+	error = submodule_load_from_wd_lite(submodule, submodule->path, NULL);
+
+	return error;
+}
+
+int git_submodule_status(
+	unsigned int *status,
+	git_submodule *submodule)
+{
+	int error = 0;
+	unsigned int status_val;
+
+	assert(status && submodule);
+
+	status_val = GIT_SUBMODULE_STATUS__CLEAR_INTERNAL(submodule->flags);
+
+	if (submodule->ignore != GIT_SUBMODULE_IGNORE_ALL) {
+		if (!(error = submodule_index_status(&status_val, submodule)))
+			error = submodule_wd_status(&status_val, submodule);
+	}
+
+	*status = status_val;
+
+	return error;
+}
+
+/*
+ * INTERNAL FUNCTIONS
+ */
+
+static git_submodule *submodule_alloc(git_repository *repo, const char *name)
+{
+	git_submodule *sm;
+
+	if (!name || !strlen(name)) {
+		giterr_set(GITERR_SUBMODULE, "Invalid submodule name");
+		return NULL;
+	}
+
+	sm = git__calloc(1, sizeof(git_submodule));
+	if (sm == NULL)
+		goto fail;
+
+	sm->path = sm->name = git__strdup(name);
+	if (!sm->name)
+		goto fail;
+
+	sm->owner = repo;
+	sm->refcount = 1;
+
 	return sm;
+
+fail:
+	submodule_release(sm, 0);
+	return NULL;
 }
 
 static void submodule_release(git_submodule *sm, int decr)
@@ -80,70 +847,121 @@ static void submodule_release(git_submodule *sm, int decr)
 	sm->refcount -= decr;
 
 	if (sm->refcount == 0) {
-		if (sm->name != sm->path)
+		if (sm->name != sm->path) {
 			git__free(sm->path);
+			sm->path = NULL;
+		}
+
 		git__free(sm->name);
+		sm->name = NULL;
+
 		git__free(sm->url);
+		sm->url = NULL;
+
+		sm->owner = NULL;
+
 		git__free(sm);
 	}
 }
 
-static int submodule_from_entry(
-	git_strmap *smcfg, git_index_entry *entry)
+static int submodule_get(
+	git_submodule **sm_ptr,
+	git_repository *repo,
+	const char *name,
+	const char *alternate)
 {
-	git_submodule *sm;
-	void *old_sm;
+	git_strmap *smcfg = repo->submodules;
 	khiter_t pos;
+	git_submodule *sm;
 	int error;
 
-	pos = git_strmap_lookup_index(smcfg, entry->path);
+	assert(repo && name);
 
-	if (git_strmap_valid_index(smcfg, pos))
-		sm = git_strmap_value_at(smcfg, pos);
-	else
-		sm = submodule_alloc(entry->path);
+	pos = git_strmap_lookup_index(smcfg, name);
 
-	git_oid_cpy(&sm->oid, &entry->oid);
+	if (!git_strmap_valid_index(smcfg, pos) && alternate)
+		pos = git_strmap_lookup_index(smcfg, alternate);
 
-	if (strcmp(sm->path, entry->path) != 0) {
-		if (sm->path != sm->name) {
-			git__free(sm->path);
-			sm->path = sm->name;
+	if (!git_strmap_valid_index(smcfg, pos)) {
+		sm = submodule_alloc(repo, name);
+
+		/* insert value at name - if another thread beats us to it, then use
+		 * their record and release our own.
+		 */
+		pos = kh_put(str, smcfg, sm->name, &error);
+
+		if (error < 0) {
+			submodule_release(sm, 1);
+			sm = NULL;
+		} else if (error == 0) {
+			submodule_release(sm, 1);
+			sm = git_strmap_value_at(smcfg, pos);
+		} else {
+			git_strmap_set_value_at(smcfg, pos, sm);
 		}
-		sm->path = git__strdup(entry->path);
-		if (!sm->path)
-			goto fail;
+	} else {
+		sm = git_strmap_value_at(smcfg, pos);
 	}
 
-	git_strmap_insert2(smcfg, sm->path, sm, old_sm, error);
-	if (error < 0)
-		goto fail;
-	sm->refcount++;
+	*sm_ptr = sm;
 
-	if (old_sm && ((git_submodule *)old_sm) != sm) {
-		/* TODO: log warning about multiple entrys for same submodule path */
-		submodule_release(old_sm, 1);
+	return (sm != NULL) ? 0 : -1;
+}
+
+static int submodule_load_from_index(
+	git_repository *repo, const git_index_entry *entry)
+{
+	git_submodule *sm;
+
+	if (submodule_get(&sm, repo, entry->path, NULL) < 0)
+		return -1;
+
+	if (sm->flags & GIT_SUBMODULE_STATUS_IN_INDEX) {
+		sm->flags |= GIT_SUBMODULE_STATUS__INDEX_MULTIPLE_ENTRIES;
+		return 0;
 	}
 
+	sm->flags |= GIT_SUBMODULE_STATUS_IN_INDEX;
+
+	git_oid_cpy(&sm->index_oid, &entry->oid);
+	sm->flags |= GIT_SUBMODULE_STATUS__INDEX_OID_VALID;
+
 	return 0;
+}
 
-fail:
-	submodule_release(sm, 0);
+static int submodule_load_from_head(
+	git_repository *repo, const char *path, const git_oid *oid)
+{
+	git_submodule *sm;
+
+	if (submodule_get(&sm, repo, path, NULL) < 0)
+		return -1;
+
+	sm->flags |= GIT_SUBMODULE_STATUS_IN_HEAD;
+
+	git_oid_cpy(&sm->head_oid, oid);
+	sm->flags |= GIT_SUBMODULE_STATUS__HEAD_OID_VALID;
+
+	return 0;
+}
+
+static int submodule_config_error(const char *property, const char *value)
+{
+	giterr_set(GITERR_INVALID,
+		"Invalid value for submodule '%s' property: '%s'", property, value);
 	return -1;
 }
 
-static int submodule_from_config(
+static int submodule_load_from_config(
 	const char *key, const char *value, void *data)
 {
-	git_strmap *smcfg = data;
-	const char *namestart;
-	const char *property;
+	git_repository *repo = data;
+	git_strmap *smcfg = repo->submodules;
+	const char *namestart, *property, *alternate = NULL;
 	git_buf name = GIT_BUF_INIT;
 	git_submodule *sm;
-	void *old_sm = NULL;
 	bool is_path;
-	khiter_t pos;
-	int error;
+	int error = 0;
 
 	if (git__prefixcmp(key, "submodule.") != 0)
 		return 0;
@@ -153,235 +971,502 @@ static int submodule_from_config(
 	if (property == NULL)
 		return 0;
 	property++;
-	is_path = (strcmp(property, "path") == 0);
+	is_path = (strcasecmp(property, "path") == 0);
 
 	if (git_buf_set(&name, namestart, property - namestart - 1) < 0)
 		return -1;
 
-	pos = git_strmap_lookup_index(smcfg, name.ptr);
-	if (!git_strmap_valid_index(smcfg, pos) && is_path)
-		pos = git_strmap_lookup_index(smcfg, value);
-	if (!git_strmap_valid_index(smcfg, pos))
-		sm = submodule_alloc(name.ptr);
-	else
-		sm = git_strmap_value_at(smcfg, pos);
-	if (!sm)
-		goto fail;
+	if (submodule_get(&sm, repo, name.ptr, is_path ? value : NULL) < 0) {
+		git_buf_free(&name);
+		return -1;
+	}
 
-	if (strcmp(sm->name, name.ptr) != 0) {
-		assert(sm->path == sm->name);
-		sm->name = git_buf_detach(&name);
+	sm->flags |= GIT_SUBMODULE_STATUS_IN_CONFIG;
 
-		git_strmap_insert2(smcfg, sm->name, sm, old_sm, error);
-		if (error < 0)
-			goto fail;
-		sm->refcount++;
+	/* Only from config might we get differing names & paths.  If so, then
+	 * update the submodule and insert under the alternative key.
+	 */
+
+	/* TODO: if case insensitive filesystem, then the following strcmps
+	 * should be strcasecmp
+	 */
+
+	if (strcmp(sm->name, name.ptr) != 0) {
+		alternate = sm->name = git_buf_detach(&name);
+	} else if (is_path && value && strcmp(sm->path, value) != 0) {
+		alternate = sm->path = git__strdup(value);
+		if (!sm->path)
+			error = -1;
 	}
-	else if (is_path && strcmp(sm->path, value) != 0) {
-		assert(sm->path == sm->name);
-		sm->path = git__strdup(value);
-		if (sm->path == NULL)
-			goto fail;
+	if (alternate) {
+		void *old_sm = NULL;
+		git_strmap_insert2(smcfg, alternate, sm, old_sm, error);
 
-		git_strmap_insert2(smcfg, sm->path, sm, old_sm, error);
-		if (error < 0)
-			goto fail;
-		sm->refcount++;
+		if (error >= 0)
+			sm->refcount++; /* inserted under a new key */
+
+		/* if we replaced an old module under this key, release the old one */
+		if (old_sm && ((git_submodule *)old_sm) != sm) {
+			submodule_release(old_sm, 1);
+			/* TODO: log warning about multiple submodules with same path */
+		}
 	}
+
 	git_buf_free(&name);
+	if (error < 0)
+		return error;
 
-	if (old_sm && ((git_submodule *)old_sm) != sm) {
-		/* TODO: log warning about multiple submodules with same path */
-		submodule_release(old_sm, 1);
-	}
+	/* TODO: Look up path in index and if it is present but not a GITLINK
+	 * then this should be deleted (at least to match git's behavior)
+	 */
 
 	if (is_path)
 		return 0;
 
 	/* copy other properties into submodule entry */
-	if (strcmp(property, "url") == 0) {
-		if (sm->url) {
-			git__free(sm->url);
-			sm->url = NULL;
-		}
-		if ((sm->url = git__strdup(value)) == NULL)
-			goto fail;
+	if (strcasecmp(property, "url") == 0) {
+		git__free(sm->url);
+		sm->url = NULL;
+
+		if (value != NULL && (sm->url = git__strdup(value)) == NULL)
+			return -1;
 	}
-	else if (strcmp(property, "update") == 0) {
+	else if (strcasecmp(property, "update") == 0) {
 		int val;
 		if (git_config_lookup_map_value(
-			_sm_update_map, ARRAY_SIZE(_sm_update_map), value, &val) < 0) {
-			giterr_set(GITERR_INVALID,
-				"Invalid value for submodule update property: '%s'", value);
-			goto fail;
-		}
-		sm->update = (git_submodule_update_t)val;
+			_sm_update_map, ARRAY_SIZE(_sm_update_map), value, &val) < 0)
+			return submodule_config_error("update", value);
+		sm->update_default = sm->update = (git_submodule_update_t)val;
 	}
-	else if (strcmp(property, "fetchRecurseSubmodules") == 0) {
-		if (git__parse_bool(&sm->fetch_recurse, value) < 0) {
-			giterr_set(GITERR_INVALID,
-				"Invalid value for submodule 'fetchRecurseSubmodules' property: '%s'", value);
-			goto fail;
-		}
+	else if (strcasecmp(property, "fetchRecurseSubmodules") == 0) {
+		if (git__parse_bool(&sm->fetch_recurse, value) < 0)
+			return submodule_config_error("fetchRecurseSubmodules", value);
 	}
-	else if (strcmp(property, "ignore") == 0) {
+	else if (strcasecmp(property, "ignore") == 0) {
 		int val;
 		if (git_config_lookup_map_value(
-			_sm_ignore_map, ARRAY_SIZE(_sm_ignore_map), value, &val) < 0) {
-			giterr_set(GITERR_INVALID,
-				"Invalid value for submodule ignore property: '%s'", value);
-			goto fail;
-		}
-		sm->ignore = (git_submodule_ignore_t)val;
+			_sm_ignore_map, ARRAY_SIZE(_sm_ignore_map), value, &val) < 0)
+			return submodule_config_error("ignore", value);
+		sm->ignore_default = sm->ignore = (git_submodule_ignore_t)val;
 	}
 	/* ignore other unknown submodule properties */
 
 	return 0;
+}
 
-fail:
-	submodule_release(sm, 0);
-	git_buf_free(&name);
-	return -1;
+static int submodule_load_from_wd_lite(
+	git_submodule *sm, const char *name, void *payload)
+{
+	git_repository *repo = git_submodule_owner(sm);
+	git_buf path = GIT_BUF_INIT;
+
+	GIT_UNUSED(name);
+	GIT_UNUSED(payload);
+
+	if (git_buf_joinpath(&path, git_repository_workdir(repo), sm->path) < 0)
+		return -1;
+
+	if (git_path_isdir(path.ptr))
+		sm->flags |= GIT_SUBMODULE_STATUS__WD_SCANNED;
+
+	if (git_path_contains(&path, DOT_GIT))
+		sm->flags |= GIT_SUBMODULE_STATUS_IN_WD;
+
+	git_buf_free(&path);
+
+	return 0;
+}
+
+static void submodule_mode_mismatch(
+	git_repository *repo, const char *path, unsigned int flag)
+{
+	khiter_t pos = git_strmap_lookup_index(repo->submodules, path);
+
+	if (git_strmap_valid_index(repo->submodules, pos)) {
+		git_submodule *sm = git_strmap_value_at(repo->submodules, pos);
+
+		sm->flags |= flag;
+	}
 }
 
-static int load_submodule_config(git_repository *repo)
+static int load_submodule_config_from_index(
+	git_repository *repo, git_oid *gitmodules_oid)
 {
 	int error;
-	git_index *index;
-	unsigned int i, max_i;
-	git_oid gitmodules_oid;
-	git_strmap *smcfg;
-	struct git_config_file *mods = NULL;
+	git_iterator *i;
+	const git_index_entry *entry;
 
-	if (repo->submodules)
-		return 0;
+	if ((error = git_iterator_for_index(&i, repo)) < 0)
+		return error;
 
-	/* submodule data is kept in a hashtable with each submodule stored
-	 * under both its name and its path.  These are usually the same, but
-	 * that is not guaranteed.
-	 */
-	smcfg = git_strmap_alloc();
-	GITERR_CHECK_ALLOC(smcfg);
+	error = git_iterator_current(i, &entry);
 
-	/* scan index for gitmodules (and .gitmodules entry) */
-	if ((error = git_repository_index__weakptr(&index, repo)) < 0)
-		goto cleanup;
-	memset(&gitmodules_oid, 0, sizeof(gitmodules_oid));
-	max_i = git_index_entrycount(index);
+	while (!error && entry != NULL) {
+
+		if (S_ISGITLINK(entry->mode)) {
+			error = submodule_load_from_index(repo, entry);
+			if (error < 0)
+				break;
+		} else {
+			submodule_mode_mismatch(
+				repo, entry->path, GIT_SUBMODULE_STATUS__INDEX_NOT_SUBMODULE);
+
+			if (strcmp(entry->path, GIT_MODULES_FILE) == 0)
+				git_oid_cpy(gitmodules_oid, &entry->oid);
+		}
+
+		error = git_iterator_advance(i, &entry);
+	}
+
+	git_iterator_free(i);
+
+	return error;
+}
+
+static int load_submodule_config_from_head(
+	git_repository *repo, git_oid *gitmodules_oid)
+{
+	int error;
+	git_tree *head;
+	git_iterator *i;
+	const git_index_entry *entry;
+
+	if ((error = git_repository_head_tree(&head, repo)) < 0)
+		return error;
+
+	if ((error = git_iterator_for_tree(&i, repo, head)) < 0) {
+		git_tree_free(head);
+		return error;
+	}
+
+	error = git_iterator_current(i, &entry);
+
+	while (!error && entry != NULL) {
 
-	for (i = 0; i < max_i; i++) {
-		git_index_entry *entry = git_index_get(index, i);
 		if (S_ISGITLINK(entry->mode)) {
-			if ((error = submodule_from_entry(smcfg, entry)) < 0)
-				goto cleanup;
+			error = submodule_load_from_head(repo, entry->path, &entry->oid);
+			if (error < 0)
+				break;
+		} else {
+			submodule_mode_mismatch(
+				repo, entry->path, GIT_SUBMODULE_STATUS__HEAD_NOT_SUBMODULE);
+
+			if (strcmp(entry->path, GIT_MODULES_FILE) == 0 &&
+				git_oid_iszero(gitmodules_oid))
+				git_oid_cpy(gitmodules_oid, &entry->oid);
 		}
-		else if (strcmp(entry->path, ".gitmodules") == 0)
-			git_oid_cpy(&gitmodules_oid, &entry->oid);
+
+		error = git_iterator_advance(i, &entry);
 	}
 
-	/* load .gitmodules from workdir if it exists */
-	if (git_repository_workdir(repo) != NULL) {
-		/* look in workdir for .gitmodules */
-		git_buf path = GIT_BUF_INIT;
-		if (!git_buf_joinpath(
-				&path, git_repository_workdir(repo), ".gitmodules") &&
-			git_path_isfile(path.ptr))
-		{
-			if (!(error = git_config_file__ondisk(&mods, path.ptr)))
-				error = git_config_file_open(mods);
+	git_iterator_free(i);
+	git_tree_free(head);
+
+	return error;
+}
+
+static git_config_file *open_gitmodules(
+	git_repository *repo,
+	bool okay_to_create,
+	const git_oid *gitmodules_oid)
+{
+	const char *workdir = git_repository_workdir(repo);
+	git_buf path = GIT_BUF_INIT;
+	git_config_file *mods = NULL;
+
+	if (workdir != NULL) {
+		if (git_buf_joinpath(&path, workdir, GIT_MODULES_FILE) != 0)
+			return NULL;
+
+		if (okay_to_create || git_path_isfile(path.ptr)) {
+			/* git_config_file__ondisk should only fail if OOM */
+			if (git_config_file__ondisk(&mods, path.ptr) < 0)
+				mods = NULL;
+			/* open should only fail here if the file is malformed */
+			else if (git_config_file_open(mods) < 0) {
+				git_config_file_free(mods);
+				mods = NULL;
+			}
 		}
-		git_buf_free(&path);
 	}
 
-	/* load .gitmodules from object cache if not in workdir */
-	if (!error && mods == NULL && !git_oid_iszero(&gitmodules_oid)) {
-		/* TODO: is it worth loading gitmodules from object cache? */
+	if (!mods && gitmodules_oid && !git_oid_iszero(gitmodules_oid)) {
+		/* TODO: Retrieve .gitmodules content from ODB */
+
+		/* Should we actually do this?  Core git does not, but it means you
+		 * can't really get much information about submodules on bare repos.
+		 */
+	}
+
+	git_buf_free(&path);
+
+	return mods;
+}
+
+static int load_submodule_config(git_repository *repo, bool force)
+{
+	int error;
+	git_oid gitmodules_oid;
+	git_buf path = GIT_BUF_INIT;
+	git_config_file *mods = NULL;
+
+	if (repo->submodules && !force)
+		return 0;
+
+	memset(&gitmodules_oid, 0, sizeof(gitmodules_oid));
+
+	/* Submodule data is kept in a hashtable keyed by both name and path.
+	 * These are usually the same, but that is not guaranteed.
+	 */
+	if (!repo->submodules) {
+		repo->submodules = git_strmap_alloc();
+		GITERR_CHECK_ALLOC(repo->submodules);
 	}
 
-	/* process .gitmodules info */
-	if (!error && mods != NULL)
-		error = git_config_file_foreach(mods, submodule_from_config, smcfg);
+	/* add submodule information from index */
+
+	if ((error = load_submodule_config_from_index(repo, &gitmodules_oid)) < 0)
+		goto cleanup;
+
+	/* add submodule information from HEAD */
+
+	if ((error = load_submodule_config_from_head(repo, &gitmodules_oid)) < 0)
+		goto cleanup;
+
+	/* add submodule information from .gitmodules */
 
-	/* store submodule config in repo */
-	if (!error)
-		repo->submodules = smcfg;
+	if ((mods = open_gitmodules(repo, false, &gitmodules_oid)) != NULL)
+		error = git_config_file_foreach(mods, submodule_load_from_config, repo);
+
+	if (error != 0)
+		goto cleanup;
+
+	/* shallow scan submodules in work tree */
+
+	if (!git_repository_is_bare(repo))
+		error = git_submodule_foreach(repo, submodule_load_from_wd_lite, NULL);
 
 cleanup:
+	git_buf_free(&path);
+
 	if (mods != NULL)
 		git_config_file_free(mods);
+
 	if (error)
-		git_strmap_free(smcfg);
+		git_submodule_config_free(repo);
+
 	return error;
 }
 
-void git_submodule_config_free(git_repository *repo)
+static int lookup_head_remote(git_buf *url, git_repository *repo)
 {
-	git_strmap *smcfg = repo->submodules;
-	git_submodule *sm;
+	int error;
+	git_config *cfg;
+	git_reference *head = NULL, *remote = NULL;
+	const char *tgt, *scan;
+	git_buf key = GIT_BUF_INIT;
+
+	/* 1. resolve HEAD -> refs/heads/BRANCH
+	 * 2. lookup config branch.BRANCH.remote -> ORIGIN
+	 * 3. lookup remote.ORIGIN.url
+	 */
 
-	repo->submodules = NULL;
+	if ((error = git_repository_config__weakptr(&cfg, repo)) < 0)
+		return error;
 
-	if (smcfg == NULL)
-		return;
+	if (git_reference_lookup(&head, repo, GIT_HEAD_FILE) < 0) {
+		giterr_set(GITERR_SUBMODULE,
+			"Cannot resolve relative URL when HEAD cannot be resolved");
+		error = GIT_ENOTFOUND;
+		goto cleanup;
+	}
 
-	git_strmap_foreach_value(smcfg, sm, {
-		submodule_release(sm,1);
-	});
-	git_strmap_free(smcfg);
-}
+	if (git_reference_type(head) != GIT_REF_SYMBOLIC) {
+		giterr_set(GITERR_SUBMODULE,
+			"Cannot resolve relative URL when HEAD is not symbolic");
+		error = GIT_ENOTFOUND;
+		goto cleanup;
+	}
 
-static int submodule_cmp(const void *a, const void *b)
-{
-	return strcmp(((git_submodule *)a)->name, ((git_submodule *)b)->name);
+	if ((error = git_branch_tracking(&remote, head)) < 0)
+		goto cleanup;
+
+	/* remote should refer to something like refs/remotes/ORIGIN/BRANCH */
+
+	if (git_reference_type(remote) != GIT_REF_SYMBOLIC ||
+		git__prefixcmp(git_reference_target(remote), "refs/remotes/") != 0)
+	{
+		giterr_set(GITERR_SUBMODULE,
+			"Cannot resolve relative URL when HEAD is not symbolic");
+		error = GIT_ENOTFOUND;
+		goto cleanup;
+	}
+
+	scan = tgt = git_reference_target(remote) + strlen("refs/remotes/");
+	while (*scan && (*scan != '/' || (scan > tgt && scan[-1] != '\\')))
+		scan++; /* find non-escaped slash to end ORIGIN name */
+
+	error = git_buf_printf(&key, "remote.%.*s.url", (int)(scan - tgt), tgt);
+	if (error < 0)
+		goto cleanup;
+
+	if ((error = git_config_get_string(&tgt, cfg, key.ptr)) < 0)
+		goto cleanup;
+
+	error = git_buf_sets(url, tgt);
+
+cleanup:
+	git_buf_free(&key);
+	git_reference_free(head);
+	git_reference_free(remote);
+
+	return error;
 }
 
-int git_submodule_foreach(
-	git_repository *repo,
-	int (*callback)(const char *name, void *payload),
-	void *payload)
+static int submodule_update_config(
+	git_submodule *submodule,
+	const char *attr,
+	const char *value,
+	bool overwrite,
+	bool only_existing)
 {
 	int error;
-	git_submodule *sm;
-	git_vector seen = GIT_VECTOR_INIT;
-	seen._cmp = submodule_cmp;
+	git_config *config;
+	git_buf key = GIT_BUF_INIT;
+	const char *old = NULL;
+
+	assert(submodule);
 
-	if ((error = load_submodule_config(repo)) < 0)
+	error = git_repository_config__weakptr(&config, submodule->owner);
+	if (error < 0)
 		return error;
 
-	git_strmap_foreach_value(repo->submodules, sm, {
-		/* usually the following will not come into play */
-		if (sm->refcount > 1) {
-			if (git_vector_bsearch(&seen, sm) != GIT_ENOTFOUND)
-				continue;
-			if ((error = git_vector_insert(&seen, sm)) < 0)
-				break;
-		}
+	error = git_buf_printf(&key, "submodule.%s.%s", submodule->name, attr);
+	if (error < 0)
+		goto cleanup;
 
-		if ((error = callback(sm->name, payload)) < 0)
-			break;
-	});
+	if (git_config_get_string(&old, config, key.ptr) < 0)
+		giterr_clear();
 
-	git_vector_free(&seen);
+	if (!old && only_existing)
+		goto cleanup;
+	if (old && !overwrite)
+		goto cleanup;
+	if ((!old && !value) || (old && value && strcmp(old, value) == 0))
+		goto cleanup;
+
+	if (!value)
+		error = git_config_delete(config, key.ptr);
+	else
+		error = git_config_set_string(config, key.ptr, value);
 
+cleanup:
+	git_buf_free(&key);
 	return error;
 }
 
-int git_submodule_lookup(
-	git_submodule **sm_ptr, /* NULL allowed if user only wants to test */
-	git_repository *repo,
-	const char *name)       /* trailing slash is allowed */
+static int submodule_index_status(unsigned int *status, git_submodule *sm)
 {
-	khiter_t pos;
+	const git_oid *head_oid  = git_submodule_head_oid(sm);
+	const git_oid *index_oid = git_submodule_index_oid(sm);
 
-	if (load_submodule_config(repo) < 0)
-		return -1;
+	if (!head_oid) {
+		if (index_oid)
+			*status |= GIT_SUBMODULE_STATUS_INDEX_ADDED;
+	}
+	else if (!index_oid)
+		*status |= GIT_SUBMODULE_STATUS_INDEX_DELETED;
+	else if (!git_oid_equal(head_oid, index_oid))
+		*status |= GIT_SUBMODULE_STATUS_INDEX_MODIFIED;
 
-	pos = git_strmap_lookup_index(repo->submodules, name);
-	if (!git_strmap_valid_index(repo->submodules, pos))
-		return GIT_ENOTFOUND;
+	return 0;
+}
 
-	if (sm_ptr)
-		*sm_ptr = git_strmap_value_at(repo->submodules, pos);
+static int submodule_wd_status(unsigned int *status, git_submodule *sm)
+{
+	int error = 0;
+	const git_oid *wd_oid, *index_oid;
+	git_repository *sm_repo = NULL;
+
+	/* open repo now if we need it (so wd_oid() call won't reopen) */
+	if ((sm->ignore == GIT_SUBMODULE_IGNORE_NONE ||
+		 sm->ignore == GIT_SUBMODULE_IGNORE_UNTRACKED) &&
+		(sm->flags & GIT_SUBMODULE_STATUS_IN_WD) != 0)
+	{
+		if ((error = git_submodule_open(&sm_repo, sm)) < 0)
+			return error;
+	}
 
-	return 0;
+	index_oid = git_submodule_index_oid(sm);
+	wd_oid    = git_submodule_wd_oid(sm);
+
+	if (!index_oid) {
+		if (wd_oid)
+			*status |= GIT_SUBMODULE_STATUS_WD_ADDED;
+	}
+	else if (!wd_oid) {
+		if ((sm->flags & GIT_SUBMODULE_STATUS__WD_SCANNED) != 0 &&
+			(sm->flags & GIT_SUBMODULE_STATUS_IN_WD) == 0)
+			*status |= GIT_SUBMODULE_STATUS_WD_UNINITIALIZED;
+		else
+			*status |= GIT_SUBMODULE_STATUS_WD_DELETED;
+	}
+	else if (!git_oid_equal(index_oid, wd_oid))
+		*status |= GIT_SUBMODULE_STATUS_WD_MODIFIED;
+
+	if (sm_repo != NULL) {
+		git_tree *sm_head;
+		git_diff_options opt;
+		git_diff_list *diff;
+
+		/* the diffs below could be optimized with an early termination
+		 * option to the git_diff functions, but for now this is sufficient
+		 * (and certainly no worse that what core git does).
+		 */
+
+		/* perform head-to-index diff on submodule */
+
+		if ((error = git_repository_head_tree(&sm_head, sm_repo)) < 0)
+			return error;
+
+		memset(&opt, 0, sizeof(opt));
+		if (sm->ignore == GIT_SUBMODULE_IGNORE_NONE)
+			opt.flags |= GIT_DIFF_INCLUDE_UNTRACKED;
+
+		error = git_diff_index_to_tree(sm_repo, &opt, sm_head, &diff);
+
+		if (!error) {
+			if (git_diff_entrycount(diff, -1) > 0)
+				*status |= GIT_SUBMODULE_STATUS_WD_INDEX_MODIFIED;
+
+			git_diff_list_free(diff);
+			diff = NULL;
+		}
+
+		git_tree_free(sm_head);
+
+		if (error < 0)
+			return error;
+
+		/* perform index-to-workdir diff on submodule */
+
+		error = git_diff_workdir_to_index(sm_repo, &opt, &diff);
+
+		if (!error) {
+			int untracked = git_diff_entrycount(diff, GIT_DELTA_UNTRACKED);
+
+			if (untracked > 0)
+				*status |= GIT_SUBMODULE_STATUS_WD_UNTRACKED;
+
+			if (git_diff_entrycount(diff, -1) - untracked > 0)
+				*status |= GIT_SUBMODULE_STATUS_WD_WD_MODIFIED;
+
+			git_diff_list_free(diff);
+			diff = NULL;
+		}
+
+		git_repository_free(sm_repo);
+	}
+
+	return error;
 }
diff --git a/src/submodule.h b/src/submodule.h
new file mode 100644
index 00000000..c7a6aaf7
--- /dev/null
+++ b/src/submodule.h
@@ -0,0 +1,102 @@
+/*
+ * Copyright (C) 2012 the libgit2 contributors
+ *
+ * This file is part of libgit2, distributed under the GNU GPL v2 with
+ * a Linking Exception. For full terms see the included COPYING file.
+ */
+#ifndef INCLUDE_submodule_h__
+#define INCLUDE_submodule_h__
+
+/* Notes:
+ *
+ * Submodule information can be in four places: the index, the config files
+ * (both .git/config and .gitmodules), the HEAD tree, and the working
+ * directory.
+ *
+ * In the index:
+ * - submodule is found by path
+ * - may be missing, present, or of the wrong type
+ * - will have an oid if present
+ *
+ * In the HEAD tree:
+ * - submodule is found by path
+ * - may be missing, present, or of the wrong type
+ * - will have an oid if present
+ *
+ * In the config files:
+ * - submodule is found by submodule "name" which is usually the path
+ * - may be missing or present
+ * - will have a name, path, url, and other properties
+ *
+ * In the working directory:
+ * - submodule is found by path
+ * - may be missing, an empty directory, a checked out directory,
+ *   or of the wrong type
+ * - if checked out, will have a HEAD oid
+ * - if checked out, will have git history that can be used to compare oids
+ * - if checked out, may have modified files and/or untracked files
+ */
+
+/**
+ * Description of submodule
+ *
+ * This record describes a submodule found in a repository.  There should be
+ * an entry for every submodule found in the HEAD and index, and for every
+ * submodule described in .gitmodules.  The fields are as follows:
+ *
+ * - `owner` is the git_repository containing this submodule
+ * - `name` is the name of the submodule from .gitmodules.
+ * - `path` is the path to the submodule from the repo root.  It is almost
+ *    always the same as `name`.
+ * - `url` is the url for the submodule.
+ * - `tree_oid` is the SHA1 for the submodule path in the repo HEAD.
+ * - `index_oid` is the SHA1 for the submodule recorded in the index.
+ * - `workdir_oid` is the SHA1 for the HEAD of the checked out submodule.
+ * - `update` is a git_submodule_update_t value - see gitmodules(5) update.
+ * - `ignore` is a git_submodule_ignore_t value - see gitmodules(5) ignore.
+ * - `fetch_recurse` is 0 or 1 - see gitmodules(5) fetchRecurseSubmodules.
+ * - `refcount` tracks how many hashmap entries there are for this submodule.
+ *   It only comes into play if the name and path of the submodule differ.
+ * - `flags` is for internal use, tracking where this submodule has been
+ *   found (head, index, config, workdir) and other misc info about it.
+ *
+ * If the submodule has been added to .gitmodules but not yet git added,
+ * then the `index_oid` will be valid and zero.  If the submodule has been
+ * deleted, but the delete has not been committed yet, then the `index_oid`
+ * will be set, but the `url` will be NULL.
+ */
+struct git_submodule {
+	git_repository *owner;
+	char *name;
+	char *path; /* important: may point to same string data as "name" */
+	char *url;
+	uint32_t flags;
+	git_oid head_oid;
+	git_oid index_oid;
+	git_oid wd_oid;
+	/* information from config */
+	git_submodule_update_t update;
+	git_submodule_update_t update_default;
+	git_submodule_ignore_t ignore;
+	git_submodule_ignore_t ignore_default;
+	int fetch_recurse;
+	/* internal information */
+	int refcount;
+};
+
+/* Additional flags on top of public GIT_SUBMODULE_STATUS values */
+enum {
+	GIT_SUBMODULE_STATUS__WD_SCANNED          = (1u << 20),
+	GIT_SUBMODULE_STATUS__HEAD_OID_VALID      = (1u << 21),
+	GIT_SUBMODULE_STATUS__INDEX_OID_VALID     = (1u << 22),
+	GIT_SUBMODULE_STATUS__WD_OID_VALID        = (1u << 23),
+	GIT_SUBMODULE_STATUS__HEAD_NOT_SUBMODULE  = (1u << 24),
+	GIT_SUBMODULE_STATUS__INDEX_NOT_SUBMODULE = (1u << 25),
+	GIT_SUBMODULE_STATUS__WD_NOT_SUBMODULE    = (1u << 26),
+	GIT_SUBMODULE_STATUS__INDEX_MULTIPLE_ENTRIES = (1u << 27),
+};
+
+#define GIT_SUBMODULE_STATUS__CLEAR_INTERNAL(S) \
+	((S) & ~(0xFFFFFFFFu << 20))
+
+#endif