Index: uspace/srv/fs/fat/fat.h
===================================================================
--- uspace/srv/fs/fat/fat.h	(revision 4979403852d819a9253efa1bff79a2dc52611a3f)
+++ uspace/srv/fs/fat/fat.h	(revision 32f623d91ff1aedbf20c62ea4be540fa7c29bf6d)
@@ -230,4 +230,8 @@
 } fat_node_t;
 
+typedef struct {
+	bool lfn_enabled;
+} fat_instance_t;
+
 extern vfs_out_ops_t fat_ops;
 extern libfs_ops_t fat_libfs_ops;
Index: uspace/srv/fs/fat/fat_directory.c
===================================================================
--- uspace/srv/fs/fat/fat_directory.c	(revision 4979403852d819a9253efa1bff79a2dc52611a3f)
+++ uspace/srv/fs/fat/fat_directory.c	(revision 32f623d91ff1aedbf20c62ea4be540fa7c29bf6d)
@@ -262,5 +262,10 @@
 {
 	int rc;
-	bool enable_lfn = true; /* TODO: make this a mount option */
+	void *data;
+	fat_instance_t *instance;
+
+	rc = fs_instance_get(di->nodep->idx->service_id, &data);
+	assert(rc == EOK);
+	instance = (fat_instance_t *) data;
 	
 	if (fat_valid_short_name(name)) {
@@ -277,5 +282,5 @@
 		rc = fat_directory_write_dentry(di, de);
 		return rc;
-	} else if (enable_lfn && fat_valid_name(name)) {
+	} else if (instance->lfn_enabled && fat_valid_name(name)) {
 		/* We should create long entries to store name */
 		int long_entry_count;
@@ -292,5 +297,5 @@
 		if (lfn_size % FAT_LFN_ENTRY_SIZE)
 			long_entry_count++;
-		rc = fat_directory_lookup_free(di, long_entry_count+1);
+		rc = fat_directory_lookup_free(di, long_entry_count + 1);
 		if (rc != EOK)
 			return rc;
@@ -328,5 +333,5 @@
 		FAT_LFN_ORDER(d) |= FAT_LFN_LAST;
 
-		rc = fat_directory_seek(di, start_pos+long_entry_count);
+		rc = fat_directory_seek(di, start_pos + long_entry_count);
 		return rc;
 	}
Index: uspace/srv/fs/fat/fat_ops.c
===================================================================
--- uspace/srv/fs/fat/fat_ops.c	(revision 4979403852d819a9253efa1bff79a2dc52611a3f)
+++ uspace/srv/fs/fat/fat_ops.c	(revision 32f623d91ff1aedbf20c62ea4be540fa7c29bf6d)
@@ -873,5 +873,12 @@
 	enum cache_mode cmode;
 	fat_bs_t *bs;
-	int rc;
+	fat_instance_t *instance;
+	int rc;
+
+	instance = malloc(sizeof(fat_instance_t));
+	if (!instance)
+		return ENOMEM;
+
+	instance->lfn_enabled = true;
 
 	/* Check for option enabling write through. */
@@ -883,10 +890,13 @@
 	/* initialize libblock */
 	rc = block_init(EXCHANGE_SERIALIZE, service_id, BS_SIZE);
-	if (rc != EOK)
-		return rc;
+	if (rc != EOK) {
+		free(instance);
+		return rc;
+	}
 
 	/* prepare the boot block */
 	rc = block_bb_read(service_id, BS_BLOCK);
 	if (rc != EOK) {
+		free(instance);
 		block_fini(service_id);
 		return rc;
@@ -897,4 +907,5 @@
 	
 	if (BPS(bs) != BS_SIZE) {
+		free(instance);
 		block_fini(service_id);
 		return ENOTSUP;
@@ -904,4 +915,5 @@
 	rc = block_cache_init(service_id, BPS(bs), 0 /* XXX */, cmode);
 	if (rc != EOK) {
+		free(instance);
 		block_fini(service_id);
 		return rc;
@@ -911,4 +923,5 @@
 	rc = fat_sanity_check(bs, service_id);
 	if (rc != EOK) {
+		free(instance);
 		(void) block_cache_fini(service_id);
 		block_fini(service_id);
@@ -918,4 +931,5 @@
 	rc = fat_idx_init_by_service_id(service_id);
 	if (rc != EOK) {
+		free(instance);
 		(void) block_cache_fini(service_id);
 		block_fini(service_id);
@@ -926,4 +940,5 @@
 	fs_node_t *rfn = (fs_node_t *)malloc(sizeof(fs_node_t));
 	if (!rfn) {
+		free(instance);
 		(void) block_cache_fini(service_id);
 		block_fini(service_id);
@@ -935,4 +950,5 @@
 	fat_node_t *rootp = (fat_node_t *)malloc(sizeof(fat_node_t));
 	if (!rootp) {
+		free(instance);
 		free(rfn);
 		(void) block_cache_fini(service_id);
@@ -945,4 +961,5 @@
 	fat_idx_t *ridxp = fat_idx_get_by_pos(service_id, FAT_CLST_ROOTPAR, 0);
 	if (!ridxp) {
+		free(instance);
 		free(rfn);
 		free(rootp);
@@ -964,4 +981,6 @@
 		rc = fat_clusters_get(&clusters, bs, service_id, rootp->firstc);
 		if (rc != EOK) {
+			fibril_mutex_unlock(&ridxp->lock);
+			free(instance);
 			free(rfn);
 			free(rootp);
@@ -974,4 +993,16 @@
 	} else
 		rootp->size = RDE(bs) * sizeof(fat_dentry_t);
+
+	rc = fs_instance_create(service_id, instance);
+	if (rc != EOK) {
+		fibril_mutex_unlock(&ridxp->lock);
+		free(instance);
+		free(rfn);
+		free(rootp);
+		(void) block_cache_fini(service_id);
+		block_fini(service_id);
+		fat_idx_fini_by_service_id(service_id);
+		return rc;
+	}
 
 	rootp->idx = ridxp;
@@ -1024,4 +1055,10 @@
 	(void) block_cache_fini(service_id);
 	block_fini(service_id);
+
+	void *data;
+	if (fs_instance_get(service_id, &data) == EOK) {
+		fs_instance_destroy(service_id);
+		free(data);
+	}
 
 	return EOK;