diff --git a/configure.d/1_append_bio.conf b/configure.d/1_append_bio.conf
deleted file mode 100644
index 8c53463..0000000
--- a/configure.d/1_append_bio.conf
+++ /dev/null
@@ -1,34 +0,0 @@
-#!/bin/bash
-#
-# Copyright(c) 2012-2022 Intel Corporation
-# SPDX-License-Identifier: BSD-3-Clause
-#
-
-. $(dirname $3)/conf_framework.sh
-
-
-check() {
-	cur_name=$(basename $2)
-	config_file_path=$1
-    if compile_module $cur_name "struct bio *b;blk_rq_append_bio(NULL, &b);" "linux/blkdev.h"
-    then
-        echo $cur_name 1 >> $config_file_path
-    else
-        echo $cur_name 2 >> $config_file_path
-    fi
-}
-
-apply() {
-    case "$1" in
-    "1")
-        add_define "cas_blk_rq_append_bio(rq, bounce_bio) \\
-            blk_rq_append_bio(rq, &bounce_bio)" ;;
-    "2")
-        add_define "cas_blk_rq_append_bio(rq, bounce_bio) \\
-            blk_rq_append_bio(rq, bounce_bio)" ;;
-    *)
-        exit 1
-    esac
-}
-
-conf_run $@
diff --git a/configure.d/1_bio_iter.conf b/configure.d/1_bio_iter.conf
index be16c13..dfef373 100644
--- a/configure.d/1_bio_iter.conf
+++ b/configure.d/1_bio_iter.conf
@@ -1,6 +1,7 @@
 #!/bin/bash
 #
 # Copyright(c) 2012-2022 Intel Corporation
+# Copyright(c) 2024 Huawei Technologies
 # SPDX-License-Identifier: BSD-3-Clause
 #
 
@@ -25,15 +26,11 @@ apply() {
     "1")
 		add_define "CAS_BIO_BISIZE(bio) \\
 			bio->bi_iter.bi_size"
-		add_define "CAS_BIO_BIIDX(bio) \\
-			bio->bi_iter.bi_idx"
 		add_define "CAS_BIO_BISECTOR(bio) \\
 			bio->bi_iter.bi_sector" ;;
     "2")
 		add_define "CAS_BIO_BISIZE(bio) \\
 			bio->bi_size"
-		add_define "CAS_BIO_BIIDX(bio) \\
-			bio->bi_idx"
 		add_define "CAS_BIO_BISECTOR(bio) \\
 			bio->bi_sector" ;;
     *)
diff --git a/configure.d/1_blk_end_req.conf b/configure.d/1_blk_end_req.conf
deleted file mode 100644
index 91a22bc..0000000
--- a/configure.d/1_blk_end_req.conf
+++ /dev/null
@@ -1,34 +0,0 @@
-#!/bin/bash
-#
-# Copyright(c) 2012-2022 Intel Corporation
-# SPDX-License-Identifier: BSD-3-Clause
-#
-
-. $(dirname $3)/conf_framework.sh
-
-check() {
-	cur_name=$(basename $2)
-	config_file_path=$1
-	if compile_module $cur_name "blk_mq_end_request(NULL, 0);" "linux/blk-mq.h"
-	then
-		echo $cur_name "1" >> $config_file_path
-	elif compile_module $cur_name "blk_end_request_all(NULL, 0);" "linux/blkdev.h"
-	then
-		echo $cur_name "2" >> $config_file_path
-	else
-		echo $cur_name "X" >> $config_file_path
-	fi
-}
-
-apply() {
-    case "$1" in
-    "1")
-		add_define "CAS_END_REQUEST_ALL blk_mq_end_request" ;;
-    "2")
-		add_define "CAS_END_REQUEST_ALL blk_end_request_all" ;;
-    *)
-        exit 1
-    esac
-}
-
-conf_run $@
diff --git a/configure.d/1_kallsyms_on_each_symbol.conf b/configure.d/1_kallsyms_on_each_symbol.conf
deleted file mode 100644
index 419afd4..0000000
--- a/configure.d/1_kallsyms_on_each_symbol.conf
+++ /dev/null
@@ -1,31 +0,0 @@
-#!/bin/bash
-#
-# Copyright(c) 2012-2022 Intel Corporation
-# SPDX-License-Identifier: BSD-3-Clause
-#
-
-. $(dirname $3)/conf_framework.sh
-
-check() {
-    cur_name=$(basename $2)
-    config_file_path=$1
-    if compile_module $cur_name "kallsyms_on_each_symbol(NULL, NULL);" "linux/fs.h"
-    then
-        echo $cur_name "1" >> $config_file_path
-    else
-        echo $cur_name "2" >> $config_file_path
-    fi
-}
-
-apply() {
-    case "$1" in
-    "1")
-        add_define "SYMBOL_LOOKUP_SUPPORTED 1" ;;
-    "2")
-        ;;
-    *)
-        exit 1
-    esac
-}
-
-conf_run $@
diff --git a/configure.d/1_queue_bounce.conf b/configure.d/1_queue_bounce.conf
deleted file mode 100644
index dc737c1..0000000
--- a/configure.d/1_queue_bounce.conf
+++ /dev/null
@@ -1,33 +0,0 @@
-#!/bin/bash
-#
-# Copyright(c) 2012-2022 Intel Corporation
-# SPDX-License-Identifier: BSD-3-Clause
-#
-
-. $(dirname $3)/conf_framework.sh
-
-check() {
-	cur_name=$(basename $2)
-	config_file_path=$1
-	if compile_module $cur_name "blk_queue_bounce(NULL, NULL);" "linux/blkdev.h"
-	then
-		echo $cur_name "1" >> $config_file_path
-	else
-        echo $cur_name "2" >> $config_file_path
-    fi
-}
-
-apply() {
-    case "$1" in
-    "1")
-		add_define "cas_blk_queue_bounce(q, bounce_bio) \\
-			blk_queue_bounce(q, bounce_bio)" ;;
-    "2")
-		add_define "cas_blk_queue_bounce(q, bounce_bio) \\
-			({})" ;;
-    *)
-        exit 1
-    esac
-}
-
-conf_run $@
diff --git a/configure.d/1_queue_lock.conf b/configure.d/1_queue_lock.conf
deleted file mode 100644
index 37f7722..0000000
--- a/configure.d/1_queue_lock.conf
+++ /dev/null
@@ -1,38 +0,0 @@
-#!/bin/bash
-#
-# Copyright(c) 2012-2022 Intel Corporation
-# SPDX-License-Identifier: BSD-3-Clause
-#
-
-. $(dirname $3)/conf_framework.sh
-
-check() {
-	cur_name=$(basename $2)
-	config_file_path=$1
-	if compile_module $cur_name "struct request_queue *q; spin_lock_irq(q->queue_lock);"\
-	       "linux/blkdev.h"
-	then
-		echo $cur_name "1" >> $config_file_path
-	elif compile_module $cur_name "struct request_queue *q; spin_lock_irq(&q->queue_lock);"\
-	       "linux/blkdev.h"
-        then
-		echo $cur_name "2" >> $config_file_path
-	else
-		echo $cur_name "X" >> $config_file_path
-    fi
-}
-
-apply() {
-    case "$1" in
-    "1")
-	    add_define "CAS_QUEUE_SPIN_LOCK(q) spin_lock_irq(q->queue_lock)"
-	    add_define "CAS_QUEUE_SPIN_UNLOCK(q) spin_unlock_irq(q->queue_lock)" ;;
-    "2")
-	    add_define "CAS_QUEUE_SPIN_LOCK(q) spin_lock_irq(&q->queue_lock)"
-	    add_define "CAS_QUEUE_SPIN_UNLOCK(q) spin_unlock_irq(&q->queue_lock)" ;;
-    *)
-        exit 1
-    esac
-}
-
-conf_run $@
diff --git a/modules/cas_cache/context.h b/modules/cas_cache/context.h
index b502330..164ea14 100644
--- a/modules/cas_cache/context.h
+++ b/modules/cas_cache/context.h
@@ -38,7 +38,7 @@ struct blk_data {
 	/**
 	 * @brief CAS IO with which data is associated
 	 */
-	struct ocf_io *io;
+	ocf_io_t io;
 
 	/**
 	 * @brief Timestamp of start processing request
diff --git a/modules/cas_cache/exp_obj.c b/modules/cas_cache/exp_obj.c
index d4b329a..fce9714 100644
--- a/modules/cas_cache/exp_obj.c
+++ b/modules/cas_cache/exp_obj.c
@@ -37,12 +37,9 @@ static inline void bd_release_from_disk(struct block_device *bdev,
 #elif LINUX_VERSION_CODE >= KERNEL_VERSION(4, 3, 0)
        #define KRETURN(x)      ({ return (x); })
        #define MAKE_RQ_RET_TYPE blk_qc_t
-#elif LINUX_VERSION_CODE >= KERNEL_VERSION(3, 2, 0)
+#else
        #define KRETURN(x)      return
        #define MAKE_RQ_RET_TYPE void
-#else
-       #define KRETURN(x)      ({ return (x); })
-       #define MAKE_RQ_RET_TYPE int
 #endif
 
 /* For RHEL 9.x we assume backport from kernel 5.18+ */
diff --git a/modules/cas_cache/linux_kernel_version.h b/modules/cas_cache/linux_kernel_version.h
index a9682fd..41eda5b 100644
--- a/modules/cas_cache/linux_kernel_version.h
+++ b/modules/cas_cache/linux_kernel_version.h
@@ -1,5 +1,6 @@
 /*
 * Copyright(c) 2012-2022 Intel Corporation
+* Copyright(c) 2024 Huawei Technologies
 * SPDX-License-Identifier: BSD-3-Clause
 */
 
@@ -47,23 +48,6 @@
 #include <linux/slab_def.h>
 #endif
 
-#if LINUX_VERSION_CODE > KERNEL_VERSION(3, 0, 0)
-	#include <generated/utsrelease.h>
-	#ifdef UTS_UBUNTU_RELEASE_ABI
-		#define CAS_UBUNTU
-	#endif
-#endif
-
-/*
- * For 8KB process kernel stack check if request is not continous and
- * submit each bio as separate request. This prevent nvme driver from
- * splitting requests.
- * For large requests, nvme splitting causes stack overrun.
- */
-#if THREAD_SIZE <= 8192
-	#define RQ_CHECK_CONTINOUS
-#endif
-
 #ifndef SHRT_MIN
 	#define SHRT_MIN ((s16)-32768)
 #endif
@@ -74,16 +58,6 @@
 
 #define ENOTSUP ENOTSUPP
 
-#ifdef RHEL_RELEASE_VERSION
-	#if RHEL_RELEASE_CODE == RHEL_RELEASE_VERSION(7, 3)
-		#define CAS_RHEL_73
-	#endif
-#endif
-
-#if LINUX_VERSION_CODE >= KERNEL_VERSION(3,8,0)
-	#define CAS_GARBAGE_COLLECTOR
-#endif
-
 /* rate-limited printk */
 #define CAS_PRINT_RL(...) \
 	if (printk_ratelimit()) \
diff --git a/modules/cas_cache/volume/vol_blk_utils.c b/modules/cas_cache/volume/vol_blk_utils.c
index 6cb0736..314413b 100644
--- a/modules/cas_cache/volume/vol_blk_utils.c
+++ b/modules/cas_cache/volume/vol_blk_utils.c
@@ -192,42 +192,6 @@ uint32_t cas_io_iter_zero(struct bio_vec_iter *dst, uint32_t bytes)
 	return zeroed;
 }
 
-/*
- *
- */
-int cas_blk_io_set_data(struct ocf_io *io,
-		ctx_data_t *ctx_data, uint32_t offset)
-{
-	struct blkio *blkio = cas_io_to_blkio(io);
-	struct blk_data *data = ctx_data;
-
-	/* Set BIO vector (IO data) and initialize iterator */
-	blkio->data = data;
-	if (blkio->data) {
-		cas_io_iter_init(&blkio->iter, blkio->data->vec,
-				blkio->data->size);
-
-		/* Move into specified offset in BIO vector iterator */
-		if (offset != cas_io_iter_move(&blkio->iter, offset)) {
-			/* TODO Log message */
-			blkio->error = -ENOBUFS;
-			return -ENOBUFS;
-		}
-	}
-
-	return 0;
-}
-
-/*
- *
- */
-ctx_data_t *cas_blk_io_get_data(struct ocf_io *io)
-{
-	struct blkio *blkio = cas_io_to_blkio(io);
-
-	return blkio->data;
-}
-
 int cas_blk_open_volume_by_bdev(ocf_volume_t *vol, struct block_device *bdev)
 {
 	struct bd_object *bdobj;
diff --git a/modules/cas_cache/volume/vol_blk_utils.h b/modules/cas_cache/volume/vol_blk_utils.h
index ddbad01..9d8dea1 100644
--- a/modules/cas_cache/volume/vol_blk_utils.h
+++ b/modules/cas_cache/volume/vol_blk_utils.h
@@ -1,5 +1,6 @@
 /*
 * Copyright(c) 2012-2022 Intel Corporation
+* Copyright(c) 2024 Huawei Technologies
 * SPDX-License-Identifier: BSD-3-Clause
 */
 
@@ -9,27 +10,6 @@
 #include "obj_blk.h"
 #include "context.h"
 
-struct blkio {
-	int error;
-	atomic_t rq_remaning;
-	atomic_t ref_counter;
-	int32_t dir;
-
-	struct blk_data *data; /* IO data buffer */
-
-	/* BIO vector iterator for sending IO */
-	struct bio_vec_iter iter;
-};
-
-static inline struct blkio *cas_io_to_blkio(struct ocf_io *io)
-{
-	return ocf_io_get_priv(io);
-}
-
-int cas_blk_io_set_data(struct ocf_io *io, ctx_data_t *data,
-		uint32_t offset);
-ctx_data_t *cas_blk_io_get_data(struct ocf_io *io);
-
 int cas_blk_open_volume_by_bdev(ocf_volume_t *vol,
 		struct block_device *bdev);
 void cas_blk_close_volume(ocf_volume_t vol);
diff --git a/modules/cas_cache/volume/vol_block_dev_bottom.c b/modules/cas_cache/volume/vol_block_dev_bottom.c
index 6495e83..6b6b3cc 100644
--- a/modules/cas_cache/volume/vol_block_dev_bottom.c
+++ b/modules/cas_cache/volume/vol_block_dev_bottom.c
@@ -86,15 +86,15 @@ static uint64_t block_dev_get_byte_length(ocf_volume_t vol)
  *
  */
 static inline struct bio *cas_bd_io_alloc_bio(struct block_device *bdev, 
-					      struct blkio *bdio)
+					      struct bio_vec_iter *iter)
 {
 	struct bio *bio
-		= cas_bio_alloc(bdev, GFP_NOIO, cas_io_iter_size_left(&bdio->iter));
+		= cas_bio_alloc(bdev, GFP_NOIO, cas_io_iter_size_left(iter));
 
 	if (bio)
 		return bio;
 
-	if (cas_io_iter_size_left(&bdio->iter) < MAX_LINES_PER_IO) {
+	if (cas_io_iter_size_left(iter) < MAX_LINES_PER_IO) {
 		/* BIO vector was small, so it was memory
 		 * common problem - NO RAM!!!
 		 */
@@ -105,211 +105,6 @@ static inline struct bio *cas_bd_io_alloc_bio(struct block_device *bdev,
 	return cas_bio_alloc(bdev, GFP_NOIO, MAX_LINES_PER_IO);
 }
 
-/*
- *
- */
-static void cas_bd_io_end(struct ocf_io *io, int error)
-{
-	struct blkio *bdio = cas_io_to_blkio(io);
-
-	if (error)
-		bdio->error |= error;
-
-	if (atomic_dec_return(&bdio->rq_remaning))
-		return;
-
-	CAS_DEBUG_MSG("Completion");
-
-	/* Send completion to caller */
-	io->end(io, bdio->error);
-}
-
-/*
- *
- */
-CAS_DECLARE_BLOCK_CALLBACK(cas_bd_io_end, struct bio *bio,
-		unsigned int bytes_done, int error)
-{
-	struct ocf_io *io;
-	struct blkio *bdio;
-	struct bd_object *bdobj;
-	int err;
-
-	BUG_ON(!bio);
-	BUG_ON(!bio->bi_private);
-	CAS_BLOCK_CALLBACK_INIT(bio);
-	io = bio->bi_private;
-	bdobj = bd_object(ocf_io_get_volume(io));
-	BUG_ON(!bdobj);
-	err = CAS_BLOCK_CALLBACK_ERROR(bio, error);
-	bdio = cas_io_to_blkio(io);
-	BUG_ON(!bdio);
-
-	CAS_DEBUG_TRACE();
-
-	if (err == -EOPNOTSUPP && (CAS_BIO_OP_FLAGS(bio) & CAS_BIO_DISCARD))
-		err = 0;
-
-	cas_bd_io_end(io, err);
-
-	bio_put(bio);
-	CAS_BLOCK_CALLBACK_RETURN();
-}
-
-static void block_dev_submit_flush(struct ocf_io *io)
-{
-	struct blkio *blkio = cas_io_to_blkio(io);
-	struct bd_object *bdobj = bd_object(ocf_io_get_volume(io));
-	struct block_device *bdev = bdobj->btm_bd;
-	struct request_queue *q = bdev_get_queue(bdev);
-	struct bio *bio = NULL;
-
-	/* Prevent races of completing IO */
-	atomic_set(&blkio->rq_remaning, 1);
-
-	if (q == NULL) {
-		/* No queue, error */
-		blkio->error = -EINVAL;
-		goto out;
-	}
-
-	if (!CAS_CHECK_QUEUE_FLUSH(q)) {
-		/* This block device does not support flush, call back */
-		goto out;
-	}
-
-	bio = cas_bio_alloc(bdev, GFP_NOIO, 0);
-	if (bio == NULL) {
-		CAS_PRINT_RL(KERN_ERR "Couldn't allocate memory for BIO\n");
-		blkio->error = -ENOMEM;
-		goto out;
-	}
-
-	blkio->dir = io->dir;
-
-	bio->bi_end_io = CAS_REFER_BLOCK_CALLBACK(cas_bd_io_end);
-	CAS_BIO_SET_DEV(bio, bdev);
-	bio->bi_private = io;
-
-	atomic_inc(&blkio->rq_remaning);
-	cas_submit_bio(CAS_SET_FLUSH(io->dir), bio);
-
-out:
-	cas_bd_io_end(io, blkio->error);
-}
-
-static void block_dev_submit_discard(struct ocf_io *io)
-{
-	struct blkio *blkio = cas_io_to_blkio(io);
-	struct bd_object *bdobj = bd_object(ocf_io_get_volume(io));
-	struct block_device *bd = bdobj->btm_bd;
-	struct request_queue *q = bdev_get_queue(bd);
-	struct bio *bio = NULL;
-
-	unsigned int max_discard_sectors, granularity, bio_sects;
-	int alignment;
-	sector_t sects, start, end, tmp;
-
-	/* Prevent races of completing IO */
-	atomic_set(&blkio->rq_remaning, 1);
-
-	if (!q) {
-		/* No queue, error */
-		blkio->error = -ENXIO;
-		goto out;
-	}
-
-	if (!cas_has_discard_support(bd)) {
-		/* Discard is not supported by bottom device, send completion
-		 * to caller
-		 */
-		goto out;
-	}
-
-	granularity = max(q->limits.discard_granularity >> SECTOR_SHIFT, 1U);
-	alignment = (bdev_discard_alignment(bd) >> SECTOR_SHIFT) % granularity;
-	max_discard_sectors =
-		min(q->limits.max_discard_sectors, UINT_MAX >> SECTOR_SHIFT);
-	max_discard_sectors -= max_discard_sectors % granularity;
-	if (unlikely(!max_discard_sectors))
-		goto out;
-
-	sects = io->bytes >> SECTOR_SHIFT;
-	start = io->addr >> SECTOR_SHIFT;
-
-	while (sects) {
-		bio = cas_bio_alloc(bd, GFP_NOIO, 1);
-		if (!bio) {
-			CAS_PRINT_RL(CAS_KERN_ERR "Couldn't allocate memory for BIO\n");
-			blkio->error = -ENOMEM;
-			break;
-		}
-
-		bio_sects = min_t(sector_t, sects, max_discard_sectors);
-		end = start + bio_sects;
-		tmp = end;
-		if (bio_sects < sects &&
-		    sector_div(tmp, granularity) != alignment) {
-			end = end - alignment;
-			sector_div(end, granularity);
-			end = end * granularity + alignment;
-			bio_sects = end - start;
-		}
-
-		CAS_BIO_SET_DEV(bio, bd);
-		CAS_BIO_BISECTOR(bio) = start;
-		CAS_BIO_BISIZE(bio) = bio_sects << SECTOR_SHIFT;
-		bio->bi_next = NULL;
-		bio->bi_private = io;
-		bio->bi_end_io = CAS_REFER_BLOCK_CALLBACK(cas_bd_io_end);
-
-		atomic_inc(&blkio->rq_remaning);
-		cas_submit_bio(CAS_BIO_DISCARD, bio);
-
-		sects -= bio_sects;
-		start = end;
-
-		cond_resched();
-	}
-
-out:
-	cas_bd_io_end(io, blkio->error);
-}
-
-static inline bool cas_bd_io_prepare(int *dir, struct ocf_io *io)
-{
-	struct blkio *bdio = cas_io_to_blkio(io);
-
-	/* Setup DIR */
-	bdio->dir = *dir;
-
-	/* Convert CAS direction into kernel values */
-	switch (bdio->dir) {
-	case OCF_READ:
-		*dir = READ;
-		break;
-
-	case OCF_WRITE:
-		*dir = WRITE;
-		break;
-
-	default:
-		bdio->error = -EINVAL;
-		break;
-	}
-
-	if (!io->bytes) {
-		/* Don not accept empty request */
-		CAS_PRINT_RL(KERN_ERR "Invalid zero size IO\n");
-		bdio->error = -EINVAL;
-	}
-
-	if (bdio->error)
-		return false;
-
-	return true;
-}
-
 /*
  * Returns only flags that are relevant to request's direction.
  */
@@ -328,44 +123,57 @@ static inline uint64_t filter_req_flags(int dir, uint64_t flags)
 /*
  *
  */
-static void block_dev_submit_io(struct ocf_io *io)
+CAS_DECLARE_BLOCK_CALLBACK(cas_bd_forward_end, struct bio *bio,
+		unsigned int bytes_done, int error)
 {
-	struct blkio *bdio = cas_io_to_blkio(io);
-	struct bd_object *bdobj = bd_object(ocf_io_get_volume(io));
-	struct bio_vec_iter *iter = &bdio->iter;
-	uint64_t addr = io->addr;
-	uint32_t bytes = io->bytes;
-	int dir = io->dir;
-	struct blk_plug plug;
+	ocf_forward_token_t token;
+	int err;
 
-	if (CAS_IS_SET_FLUSH(io->flags)) {
-		CAS_DEBUG_MSG("Flush request");
-		/* It is flush requests handle it */
-		block_dev_submit_flush(io);
-		return;
-	}
+	CAS_BLOCK_CALLBACK_INIT(bio);
+	token = (ocf_forward_token_t)bio->bi_private;
+	err = CAS_BLOCK_CALLBACK_ERROR(bio, error);
+
+	CAS_DEBUG_TRACE();
+
+	if (err == -EOPNOTSUPP && (CAS_BIO_OP_FLAGS(bio) & CAS_BIO_DISCARD))
+		err = 0;
+
+	ocf_forward_end(token, err);
+
+	bio_put(bio);
+	CAS_BLOCK_CALLBACK_RETURN();
+}
+
+
+static void block_dev_forward_io(ocf_volume_t volume,
+		ocf_forward_token_t token, int dir, uint64_t addr,
+		uint64_t bytes, uint64_t offset)
+{
+	struct bd_object *bdobj = bd_object(volume);
+	struct blk_data *data = ocf_forward_get_data(token);
+	uint64_t flags = ocf_forward_get_flags(token);
+	int bio_dir = (dir == OCF_READ) ? READ : WRITE;
+	struct bio_vec_iter iter;
+	struct blk_plug plug;
+	int error = 0;
 
 	CAS_DEBUG_PARAM("Address = %llu, bytes = %u\n", addr, bytes);
 
-	/* Prevent races of completing IO */
-	atomic_set(&bdio->rq_remaning, 1);
-
-	if (!cas_bd_io_prepare(&dir, io)) {
-		CAS_DEBUG_MSG("Invalid request");
-		cas_bd_io_end(io, -EINVAL);
+	cas_io_iter_init(&iter, data->vec, data->size);
+	if (offset != cas_io_iter_move(&iter, offset)) {
+		ocf_forward_end(token, -OCF_ERR_INVAL);
 		return;
 	}
 
 	blk_start_plug(&plug);
-
-	while (cas_io_iter_is_next(iter) && bytes) {
+	while (cas_io_iter_is_next(&iter) && bytes) {
 		/* Still IO vectors to be sent */
 
 		/* Allocate BIO */
-		struct bio *bio = cas_bd_io_alloc_bio(bdobj->btm_bd, bdio);
+		struct bio *bio = cas_bd_io_alloc_bio(bdobj->btm_bd, &iter);
 
 		if (!bio) {
-			bdio->error = -ENOMEM;
+			error = -ENOMEM;
 			break;
 		}
 
@@ -373,15 +181,15 @@ static void block_dev_submit_io(struct ocf_io *io)
 		CAS_BIO_SET_DEV(bio, bdobj->btm_bd);
 		CAS_BIO_BISECTOR(bio) = addr / SECTOR_SIZE;
 		bio->bi_next = NULL;
-		bio->bi_private = io;
-		CAS_BIO_OP_FLAGS(bio) |= filter_req_flags(dir, io->flags);
-		bio->bi_end_io = CAS_REFER_BLOCK_CALLBACK(cas_bd_io_end);
+		bio->bi_private = (void *)token;
+		CAS_BIO_OP_FLAGS(bio) |= filter_req_flags(bio_dir, flags);
+		bio->bi_end_io = CAS_REFER_BLOCK_CALLBACK(cas_bd_forward_end);
 
 		/* Add pages */
-		while (cas_io_iter_is_next(iter) && bytes) {
-			struct page *page = cas_io_iter_current_page(iter);
-			uint32_t offset = cas_io_iter_current_offset(iter);
-			uint32_t length = cas_io_iter_current_length(iter);
+		while (cas_io_iter_is_next(&iter) && bytes) {
+			struct page *page = cas_io_iter_current_page(&iter);
+			uint32_t offset = cas_io_iter_current_offset(&iter);
+			uint32_t length = cas_io_iter_current_length(&iter);
 			int added;
 
 			if (length > bytes)
@@ -400,19 +208,19 @@ static void block_dev_submit_io(struct ocf_io *io)
 			addr += added;
 
 			/* Update BIO vector iterator */
-			if (added != cas_io_iter_move(iter, added)) {
-				bdio->error = -ENOBUFS;
+			if (added != cas_io_iter_move(&iter, added)) {
+				error = -ENOBUFS;
 				break;
 			}
 		}
 
-		if (bdio->error == 0) {
+		if (error == 0) {
 			/* Increase IO reference for sending this IO */
-			atomic_inc(&bdio->rq_remaning);
 
+			ocf_forward_get(token);
 			/* Send BIO */
 			CAS_DEBUG_MSG("Submit IO");
-			cas_submit_bio(dir, bio);
+			cas_submit_bio(bio_dir, bio);
 			bio = NULL;
 		} else {
 			if (bio) {
@@ -424,41 +232,146 @@ static void block_dev_submit_io(struct ocf_io *io)
 			break;
 		}
 	}
-
 	blk_finish_plug(&plug);
 
-	if (bytes && bdio->error == 0) {
+	if (bytes && error == 0) {
 		/* Not all bytes sent, mark error */
-		bdio->error = -ENOBUFS;
+		error = -ENOBUFS;
 	}
 
 	/* Prevent races of completing IO when
 	 * there are still child IOs not being send.
 	 */
-	cas_bd_io_end(io, 0);
+	ocf_forward_end(token, error);
+}
+
+static void block_dev_forward_flush(ocf_volume_t volume,
+		ocf_forward_token_t token)
+{
+	struct bd_object *bdobj = bd_object(volume);
+	struct request_queue *q = bdev_get_queue(bdobj->btm_bd);
+	struct bio *bio;
+
+	if (!q) {
+		/* No queue, error */
+		ocf_forward_end(token, -OCF_ERR_INVAL);
+		return;
+	}
+
+	if (!CAS_CHECK_QUEUE_FLUSH(q)) {
+		/* This block device does not support flush, call back */
+		ocf_forward_end(token, 0);
+		return;
+	}
+
+	bio = cas_bio_alloc(bdobj->btm_bd, GFP_NOIO, 0);
+	if (!bio) {
+		CAS_PRINT_RL(KERN_ERR "Couldn't allocate memory for BIO\n");
+		ocf_forward_end(token, -OCF_ERR_NO_MEM);
+		return;
+	}
+
+	CAS_BIO_SET_DEV(bio, bdobj->btm_bd);
+	bio->bi_private = (void *)token;
+	bio->bi_end_io = CAS_REFER_BLOCK_CALLBACK(cas_bd_forward_end);
+
+	cas_submit_bio(CAS_SET_FLUSH(0), bio);
+
+}
+
+static void block_dev_forward_discard(ocf_volume_t volume,
+		ocf_forward_token_t token, uint64_t addr, uint64_t bytes)
+{
+	struct bd_object *bdobj = bd_object(volume);
+	struct request_queue *q = bdev_get_queue(bdobj->btm_bd);
+	struct bio *bio;
+	int error = 0;
+
+	unsigned int max_discard_sectors, granularity, bio_sects;
+	int alignment;
+	sector_t sects, start, end, tmp;
+
+	if (!q) {
+		/* No queue, error */
+		ocf_forward_end(token, -OCF_ERR_INVAL);
+		return;
+	}
+
+	if (!cas_has_discard_support(bdobj->btm_bd)) {
+		/* Discard is not supported by bottom device, send completion
+		 * to caller
+		 */
+		ocf_forward_end(token, 0);
+		return;
+	}
+
+	granularity = max(q->limits.discard_granularity >> SECTOR_SHIFT, 1U);
+	alignment = (bdev_discard_alignment(bdobj->btm_bd) >> SECTOR_SHIFT)
+			% granularity;
+	max_discard_sectors =
+		min(q->limits.max_discard_sectors, UINT_MAX >> SECTOR_SHIFT);
+	max_discard_sectors -= max_discard_sectors % granularity;
+	if (unlikely(!max_discard_sectors)) {
+		ocf_forward_end(token, -OCF_ERR_INVAL);
+		return;
+	}
+
+	sects = bytes >> SECTOR_SHIFT;
+	start = addr >> SECTOR_SHIFT;
+
+	while (sects) {
+		bio = cas_bio_alloc(bdobj->btm_bd, GFP_NOIO, 1);
+		if (!bio) {
+			CAS_PRINT_RL(CAS_KERN_ERR "Couldn't allocate memory for BIO\n");
+			error = -OCF_ERR_NO_MEM;
+			break;
+		}
+
+		bio_sects = min_t(sector_t, sects, max_discard_sectors);
+		end = start + bio_sects;
+		tmp = end;
+		if (bio_sects < sects &&
+		    sector_div(tmp, granularity) != alignment) {
+			end = end - alignment;
+			sector_div(end, granularity);
+			end = end * granularity + alignment;
+			bio_sects = end - start;
+		}
+
+		CAS_BIO_SET_DEV(bio, bdobj->btm_bd);
+		CAS_BIO_BISECTOR(bio) = start;
+		CAS_BIO_BISIZE(bio) = bio_sects << SECTOR_SHIFT;
+		bio->bi_next = NULL;
+		bio->bi_private = (void *)token;
+		bio->bi_end_io = CAS_REFER_BLOCK_CALLBACK(cas_bd_forward_end);
+
+		ocf_forward_get(token);
+		cas_submit_bio(CAS_BIO_DISCARD, bio);
+
+		sects -= bio_sects;
+		start = end;
+
+		cond_resched();
+	}
+
+	ocf_forward_end(token, error);
 }
 
 const struct ocf_volume_properties cas_object_blk_properties = {
 	.name = "Block_Device",
-	.io_priv_size = sizeof(struct blkio),
 	.volume_priv_size = sizeof(struct bd_object),
 	.caps = {
 		.atomic_writes = 0, /* Atomic writes not supported */
 	},
 	.ops = {
-		.submit_io = block_dev_submit_io,
-		.submit_flush = block_dev_submit_flush,
-		.submit_metadata = NULL,
-		.submit_discard = block_dev_submit_discard,
+		.forward_io = block_dev_forward_io,
+		.forward_flush = block_dev_forward_flush,
+		.forward_discard = block_dev_forward_discard,
 		.open = block_dev_open_object,
 		.close = block_dev_close_object,
 		.get_max_io_size = block_dev_get_max_io_size,
 		.get_length = block_dev_get_byte_length,
 	},
-	.io_ops = {
-		.set_data = cas_blk_io_set_data,
-		.get_data = cas_blk_io_get_data,
-	},
 	.deinit = NULL,
 };
 
diff --git a/modules/cas_cache/volume/vol_block_dev_bottom.h b/modules/cas_cache/volume/vol_block_dev_bottom.h
index a727f86..581e474 100644
--- a/modules/cas_cache/volume/vol_block_dev_bottom.h
+++ b/modules/cas_cache/volume/vol_block_dev_bottom.h
@@ -6,7 +6,6 @@
 
 #ifndef __VOL_BLOCK_DEV_BOTTOM_H__
 #define __VOL_BLOCK_DEV_BOTTOM_H__
-
 int block_dev_init(void);
 
 #endif /* __VOL_BLOCK_DEV_BOTTOM_H__ */
diff --git a/modules/cas_cache/volume/vol_block_dev_top.c b/modules/cas_cache/volume/vol_block_dev_top.c
index 9a7c79e..be79dc0 100644
--- a/modules/cas_cache/volume/vol_block_dev_top.c
+++ b/modules/cas_cache/volume/vol_block_dev_top.c
@@ -189,10 +189,11 @@ static void blkdev_complete_data_master(struct blk_data *master, int error)
 	cas_free_blk_data(master);
 }
 
-static void blkdev_complete_data(struct ocf_io *io, int error)
+static void blkdev_complete_data(ocf_io_t io, void *priv1, void *priv2,
+		int error)
 {
-	struct bio *bio = io->priv1;
-	struct blk_data *master = io->priv2;
+	struct bio *bio = priv1;
+	struct blk_data *master = priv2;
 	struct blk_data *data = ocf_io_get_data(io);
 
 	ocf_io_put(io);
@@ -217,7 +218,7 @@ static int blkdev_handle_data_single(struct bd_object *bvol, struct bio *bio,
 	ocf_cache_t cache = ocf_volume_get_cache(bvol->front_volume);
 	struct cache_priv *cache_priv = ocf_cache_get_priv(cache);
 	ocf_queue_t queue = cache_priv->io_queues[smp_processor_id()];
-	struct ocf_io *io;
+	ocf_io_t io;
 	struct blk_data *data;
 	uint64_t flags = CAS_BIO_OP_FLAGS(bio);
 	int ret;
@@ -317,9 +318,10 @@ err:
 		CAS_BIO_ENDIO(bio, CAS_BIO_BISIZE(bio), CAS_ERRNO_TO_BLK_STS(error));
 }
 
-static void blkdev_complete_discard(struct ocf_io *io, int error)
+static void blkdev_complete_discard(ocf_io_t io, void *priv1, void *priv2,
+		int error)
 {
-	struct bio *bio = io->priv1;
+	struct bio *bio = priv1;
 	int result = map_cas_err_to_generic(error);
 
 	CAS_BIO_ENDIO(bio, CAS_BIO_BISIZE(bio), CAS_ERRNO_TO_BLK_STS(result));
@@ -331,7 +333,7 @@ static void blkdev_handle_discard(struct bd_object *bvol, struct bio *bio)
 	ocf_cache_t cache = ocf_volume_get_cache(bvol->front_volume);
 	struct cache_priv *cache_priv = ocf_cache_get_priv(cache);
 	ocf_queue_t queue = cache_priv->io_queues[smp_processor_id()];
-	struct ocf_io *io;
+	ocf_io_t io;
 
 	io = ocf_volume_new_io(bvol->front_volume, queue,
 			CAS_BIO_BISECTOR(bio) << SECTOR_SHIFT,
@@ -356,10 +358,11 @@ static void blkdev_handle_bio_noflush(struct bd_object *bvol, struct bio *bio)
 		blkdev_handle_data(bvol, bio);
 }
 
-static void blkdev_complete_flush(struct ocf_io *io, int error)
+static void blkdev_complete_flush(ocf_io_t io, void *priv1, void *priv2,
+		int error)
 {
-	struct bio *bio = io->priv1;
-	struct bd_object *bvol = io->priv2;
+	struct bio *bio = priv1;
+	struct bd_object *bvol = priv2;
 	int result = map_cas_err_to_generic(error);
 
 	ocf_io_put(io);
@@ -378,7 +381,7 @@ static void blkdev_handle_flush(struct bd_object *bvol, struct bio *bio)
 	ocf_cache_t cache = ocf_volume_get_cache(bvol->front_volume);
 	struct cache_priv *cache_priv = ocf_cache_get_priv(cache);
 	ocf_queue_t queue = cache_priv->io_queues[smp_processor_id()];
-	struct ocf_io *io;
+	ocf_io_t io;
 
 	io = ocf_volume_new_io(bvol->front_volume, queue, 0, 0, OCF_WRITE, 0,
 			CAS_SET_FLUSH(0));
diff --git a/ocf b/ocf
index 1fbb00d..6907abe 160000
--- a/ocf
+++ b/ocf
@@ -1 +1 @@
-Subproject commit 1fbb00de8f40bab42eb5625053a61536c49382cb
+Subproject commit 6907abeba2359fc02b75b0301bcc9112ef717cd5