Initial commit of tldk code.

Change-Id: Ib96fdd2c57bae0a51ed420137c35eb8e2ee58473
Signed-off-by: Konstantin Ananyev <konstantin.ananyev@intel.com>
Signed-off-by: Ed Warnicke <eaw@cisco.com>

12 files changed, 3211 insertions, 0 deletions

diff --git a/lib/libtle_udp/Makefile b/lib/libtle_udp/Makefile
new file mode 100644
index 0000000..100755c
--- /dev/null
+++ b/lib/libtle_udp/Makefile
@@ -0,0 +1,50 @@
+# Copyright (c) 2016 Intel Corporation.
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at:
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+ifeq ($(RTE_SDK),)
+$(error "Please define RTE_SDK environment variable")
+endif
+
+# Default target, can be overwritten by command line or environment
+RTE_TARGET ?= x86_64-native-linuxapp-gcc
+
+include $(RTE_SDK)/mk/rte.vars.mk
+
+# library name
+LIB = libtle_udp.a
+
+CFLAGS += -O3
+CFLAGS += $(WERROR_FLAGS) -I$(SRCDIR)
+
+EXPORT_MAP := tle_udp_version.map
+
+LIBABIVER := 1
+
+#source files
+SRCS-y += buf_cage.c
+SRCS-y += event.c
+SRCS-y += udp_ctl.c
+SRCS-y += udp_rxtx.c
+
+# install this header file
+SYMLINK-y-include += tle_udp_impl.h
+SYMLINK-y-include += tle_event.h
+
+# this library depends on
+DEPDIRS-y += $(RTE_SDK)/lib/librte_eal
+DEPDIRS-y += $(RTE_SDK)/lib/librte_ether
+DEPDIRS-y += $(RTE_SDK)/lib/librte_mbuf
+DEPDIRS-y += $(RTE_SDK)lib/librte_net
+DEPDIRS-y += $(RTE_SDK)lib/librte_ip_frag
+
+include $(RTE_SDK)/mk/rte.extlib.mk
diff --git a/lib/libtle_udp/buf_cage.c b/lib/libtle_udp/buf_cage.c
new file mode 100644
index 0000000..0ae21b0
--- /dev/null
+++ b/lib/libtle_udp/buf_cage.c
@@ -0,0 +1,81 @@
+/*
+ * Copyright (c) 2016  Intel Corporation.
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at:
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include <rte_errno.h>
+#include <rte_malloc.h>
+#include <rte_log.h>
+
+#include "buf_cage.h"
+#include "osdep.h"
+
+struct bcg_store *
+bcg_create(const struct bcg_store_prm *prm)
+{
+	struct buf_cage *bc;
+	struct bcg_store *st;
+	uintptr_t end, p;
+	size_t sz, tsz;
+	uint32_t n;
+
+	if (prm == NULL || (prm->cage_align != 0 &&
+			rte_is_power_of_2(prm->cage_align) == 0)) {
+		rte_errno = EINVAL;
+		return NULL;
+	}
+
+	/* number of cages required. */
+	n = (prm->max_bufs + prm->cage_bufs - 1) / prm->cage_bufs;
+	n = RTE_MAX(n, prm->min_cages);
+
+	/* size of each cage. */
+	sz = prm->cage_bufs * sizeof(bc->bufs[0]) + sizeof(*bc);
+	sz = RTE_ALIGN_CEIL(sz, prm->cage_align);
+
+	/* total number of bytes required. */
+	tsz = n * sz + RTE_ALIGN_CEIL(sizeof(*st), prm->cage_align);
+
+	st = rte_zmalloc_socket(NULL, tsz, RTE_CACHE_LINE_SIZE, prm->socket_id);
+	if (st == NULL) {
+		UDP_LOG(ERR, "%s: allocation of %zu bytes on "
+			"socket %d failed\n",
+			__func__, tsz, prm->socket_id);
+		return NULL;
+	}
+
+	st->prm = prm[0];
+	bcg_queue_reset(&st->free);
+
+	p = (uintptr_t)RTE_PTR_ALIGN_CEIL((st + 1), prm->cage_align);
+	end = p + n * sz;
+
+	for (; p != end; p += sz) {
+		bc = (struct buf_cage *)p;
+		bc->st = st;
+		bc->num = prm->cage_bufs;
+		STAILQ_INSERT_TAIL(&st->free.queue, bc, ql);
+	}
+
+	st->free.num = n;
+	st->nb_cages = n;
+	st->cage_sz = sz;
+	st->total_sz = tsz;
+	return st;
+}
+
+void
+bcg_destroy(struct bcg_store *st)
+{
+	rte_free(st);
+}
diff --git a/lib/libtle_udp/buf_cage.h b/lib/libtle_udp/buf_cage.h
new file mode 100644
index 0000000..3b3c429
--- /dev/null
+++ b/lib/libtle_udp/buf_cage.h
@@ -0,0 +1,231 @@
+/*
+ * Copyright (c) 2016  Intel Corporation.
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at:
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#ifndef _BUF_CAGE_H_
+#define _BUF_CAGE_H_
+
+#include <rte_common.h>
+#include <rte_atomic.h>
+#include <rte_spinlock.h>
+#include <sys/queue.h>
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+struct bcg_store;
+
+struct buf_cage {
+	struct bcg_store *st;
+	STAILQ_ENTRY(buf_cage) ql;
+	uint32_t num;
+	uint32_t rp;
+	uint32_t wp;
+	const void *bufs[0];
+};
+
+struct bcg_queue {
+	rte_spinlock_t lock;
+	uint32_t num;
+	STAILQ_HEAD(, buf_cage) queue;
+};
+
+struct bcg_store_prm {
+	void *user_data;
+	int32_t socket_id;     /* NUMA socket to allocate memory from. */
+	uint32_t max_bufs;     /* total number of bufs to cage. */
+	uint32_t min_cages;    /* min number of cages per store. */
+	uint32_t cage_bufs;    /* min number of bufs per cage. */
+	uint32_t cage_align;   /* each cage to be aligned (power of 2). */
+};
+
+struct bcg_store {
+	struct bcg_queue free;
+	uint32_t nb_cages;
+	size_t cage_sz;
+	size_t total_sz;
+	struct bcg_store_prm prm;
+} __rte_cache_aligned;
+
+struct bcg_store *bcg_create(const struct bcg_store_prm *prm);
+void bcg_destroy(struct bcg_store *st);
+
+static inline int
+bcg_store_full(const struct bcg_store *st)
+{
+	return st->nb_cages == st->free.num;
+}
+
+static inline void
+bcg_queue_reset(struct bcg_queue *bq)
+{
+	STAILQ_INIT(&bq->queue);
+	bq->num = 0;
+	rte_spinlock_init(&bq->lock);
+}
+
+static inline void
+bcg_reset(struct buf_cage *bc)
+{
+	bc->rp = 0;
+	bc->wp = 0;
+}
+
+static inline void *
+bcg_get_udata(struct buf_cage *bc)
+{
+	return bc->st->prm.user_data;
+}
+
+static inline struct buf_cage *
+__bcg_dequeue_head(struct bcg_queue *bq)
+{
+	struct buf_cage *bc;
+
+	bc = STAILQ_FIRST(&bq->queue);
+	if (bc != NULL) {
+		STAILQ_REMOVE_HEAD(&bq->queue, ql);
+		bq->num--;
+	}
+	return bc;
+}
+
+static inline struct buf_cage *
+bcg_dequeue_head(struct bcg_queue *bq)
+{
+	struct buf_cage *bc;
+
+	if (bq->num == 0)
+		return NULL;
+
+	rte_compiler_barrier();
+
+	rte_spinlock_lock(&bq->lock);
+	bc = __bcg_dequeue_head(bq);
+	rte_spinlock_unlock(&bq->lock);
+	return bc;
+}
+
+static inline uint32_t
+__bcg_enqueue_head(struct bcg_queue *bq, struct buf_cage *bc)
+{
+	STAILQ_INSERT_HEAD(&bq->queue, bc, ql);
+	return ++bq->num;
+}
+
+static inline uint32_t
+bcg_enqueue_head(struct bcg_queue *bq, struct buf_cage *bc)
+{
+	uint32_t n;
+
+	rte_spinlock_lock(&bq->lock);
+	n = __bcg_enqueue_head(bq, bc);
+	rte_spinlock_unlock(&bq->lock);
+	return n;
+}
+
+static inline uint32_t
+__bcg_enqueue_tail(struct bcg_queue *bq, struct buf_cage *bc)
+{
+	STAILQ_INSERT_TAIL(&bq->queue, bc, ql);
+	return ++bq->num;
+}
+
+static inline uint32_t
+bcg_enqueue_tail(struct bcg_queue *bq, struct buf_cage *bc)
+{
+	uint32_t n;
+
+	rte_spinlock_lock(&bq->lock);
+	n = __bcg_enqueue_tail(bq, bc);
+	rte_spinlock_unlock(&bq->lock);
+	return n;
+}
+
+static inline uint32_t
+bcg_queue_append(struct bcg_queue *dst, struct bcg_queue *src)
+{
+	rte_spinlock_lock(&src->lock);
+	STAILQ_CONCAT(&dst->queue, &src->queue);
+	dst->num += src->num;
+	src->num = 0;
+	rte_spinlock_unlock(&src->lock);
+	return dst->num;
+}
+
+static inline uint32_t
+bcg_free_count(const struct buf_cage *bc)
+{
+	return bc->num - bc->wp;
+}
+
+
+static inline uint32_t
+bcg_fill_count(const struct buf_cage *bc)
+{
+	return bc->wp - bc->rp;
+}
+
+/* !!! if going to keep it - try to unroll copying stuff. !!! */
+static inline uint32_t
+bcg_get(struct buf_cage *bc, const void *bufs[], uint32_t num)
+{
+	uint32_t i, n, r;
+
+	r = bc->rp;
+	n = RTE_MIN(num, bc->wp - r);
+	for (i = 0; i != n; i++)
+		bufs[i] = bc->bufs[r + i];
+
+	bc->rp = r + n;
+	return n;
+}
+
+static inline uint32_t
+bcg_put(struct buf_cage *bc, const void *bufs[], uint32_t num)
+{
+	uint32_t i, n, w;
+
+	w = bc->wp;
+	n = RTE_MIN(num, bc->num - w);
+	for (i = 0; i != n; i++)
+		bc->bufs[w + i] = bufs[i];
+
+	bc->wp = w + n;
+	return n;
+}
+
+
+static inline struct buf_cage *
+bcg_alloc(struct bcg_store *st)
+{
+	return bcg_dequeue_head(&st->free);
+}
+
+static inline uint32_t
+bcg_free(struct buf_cage *bc)
+{
+	struct bcg_store *st;
+
+	st = bc->st;
+	bcg_reset(bc);
+	return bcg_enqueue_head(&st->free, bc);
+}
+
+#ifdef __cplusplus
+}
+#endif
+
+#endif /* _BUF_CAGE_H_ */
diff --git a/lib/libtle_udp/event.c b/lib/libtle_udp/event.c
new file mode 100644
index 0000000..7e340e8
--- /dev/null
+++ b/lib/libtle_udp/event.c
@@ -0,0 +1,104 @@
+/*
+ * Copyright (c) 2016  Intel Corporation.
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at:
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include <rte_errno.h>
+#include <rte_malloc.h>
+#include <rte_log.h>
+#include <tle_event.h>
+
+#include "osdep.h"
+
+struct tle_evq *
+tle_evq_create(const struct tle_evq_param *prm)
+{
+	struct tle_evq *evq;
+	size_t sz;
+	uint32_t i;
+
+	if (prm == NULL) {
+		rte_errno = EINVAL;
+		return NULL;
+	}
+
+	sz = sizeof(*evq) + sizeof(evq->events[0]) * prm->max_events;
+	evq =  rte_zmalloc_socket(NULL, sz, RTE_CACHE_LINE_SIZE,
+		prm->socket_id);
+	if (evq == NULL) {
+		UDP_LOG(ERR, "allocation of %zu bytes for "
+			"new tle_evq(%u) on socket %d failed\n",
+			sz, prm->max_events, prm->socket_id);
+		return NULL;
+	}
+
+	TAILQ_INIT(&evq->armed);
+	TAILQ_INIT(&evq->free);
+
+	for (i = 0; i != prm->max_events; i++) {
+		evq->events[i].head = evq;
+		TAILQ_INSERT_TAIL(&evq->free, evq->events + i, ql);
+	}
+
+	evq->nb_events = i;
+	evq->nb_free = i;
+
+	return evq;
+}
+
+void
+tle_evq_destroy(struct tle_evq *evq)
+{
+	rte_free(evq);
+}
+
+struct tle_event *
+tle_event_alloc(struct tle_evq *evq, const void *data)
+{
+	struct tle_event *h;
+
+	if (evq == NULL) {
+		rte_errno = EINVAL;
+		return NULL;
+	}
+
+	rte_spinlock_lock(&evq->lock);
+	h = TAILQ_FIRST(&evq->free);
+	if (h != NULL) {
+		TAILQ_REMOVE(&evq->free, h, ql);
+		evq->nb_free--;
+		h->data = data;
+	} else
+		rte_errno = -ENOMEM;
+	rte_spinlock_unlock(&evq->lock);
+	return h;
+}
+
+void
+tle_event_free(struct tle_event *ev)
+{
+	struct tle_evq *q;
+
+	if (ev == NULL) {
+		rte_errno = EINVAL;
+		return;
+	}
+
+	q = ev->head;
+	rte_spinlock_lock(&q->lock);
+	ev->data = NULL;
+	ev->state = TLE_SEV_IDLE;
+	TAILQ_INSERT_HEAD(&q->free, ev, ql);
+	q->nb_free++;
+	rte_spinlock_unlock(&q->lock);
+}
diff --git a/lib/libtle_udp/misc.h b/lib/libtle_udp/misc.h
new file mode 100644
index 0000000..3874647
--- /dev/null
+++ b/lib/libtle_udp/misc.h
@@ -0,0 +1,296 @@
+/*
+ * Copyright (c) 2016  Intel Corporation.
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at:
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#ifndef _MISC_H_
+#define _MISC_H_
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+static inline int
+ymm_mask_cmp(const _ymm_t *da, const _ymm_t *sa, const _ymm_t *sm)
+{
+	uint64_t ret;
+
+	ret = ((sa->u64[0] & sm->u64[0]) ^ da->u64[0]) |
+		((sa->u64[1] & sm->u64[1]) ^ da->u64[1]) |
+		((sa->u64[2] & sm->u64[2]) ^ da->u64[2]) |
+		((sa->u64[3] & sm->u64[3]) ^ da->u64[3]);
+
+	return (ret == 0);
+}
+
+/*
+ * Setup tx_offload field inside mbuf using raw 64-bit field.
+ * Consider to move it into DPDK librte_mbuf.
+ */
+static inline uint64_t
+_mbuf_tx_offload(uint64_t il2, uint64_t il3, uint64_t il4, uint64_t tso,
+	uint64_t ol3, uint64_t ol2)
+{
+	return il2 | il3 << 7 | il4 << 16 | tso << 24 | ol3 << 40 | ol2 << 49;
+}
+
+/*
+ * Routines to calculate L3/L4 checksums in SW.
+ * Pretty similar to ones from DPDK librte_net/rte_ip.h,
+ * but provide better performance (at least for tested configurations),
+ * and extended functionality.
+ * Consider to move them into DPDK librte_net/rte_ip.h.
+ */
+
+/* make compiler to generate: add %r1, %r2; adc $0, %r1. */
+#define CKSUM_ADD_CARRY(s, v)	do {       \
+	(s) += (v);                        \
+	(s) = ((s) < (v)) ? (s) + 1 : (s); \
+} while (0)
+
+/**
+ * Process the non-complemented checksum of a buffer.
+ * Similar  to rte_raw_cksum(), but provide better perfomance
+ * (at least on IA platforms).
+ * @param buf
+ *   Pointer to the buffer.
+ * @param len
+ *   Length of the buffer.
+ * @return
+ *   The non-complemented checksum.
+ */
+static inline uint16_t
+__raw_cksum(const uint8_t *buf, uint32_t size)
+{
+	uint64_t s, sum;
+	uint32_t i, n;
+	uint32_t  dw1, dw2;
+	uint16_t w1, w2;
+	const uint64_t *b;
+
+	b = (const uint64_t *)buf;
+	n = size / sizeof(*b);
+	sum = 0;
+
+	/* main loop, consume 8 bytes per iteration. */
+	for (i = 0; i != n; i++) {
+		s = b[i];
+		CKSUM_ADD_CARRY(sum, s);
+	}
+
+	/* consume the remainder. */
+	n = size % sizeof(*b);
+	if (n != 0) {
+		/* position of the of last 8 bytes of data. */
+		b = (const uint64_t *)((uintptr_t)(b + i) + n - sizeof(*b));
+		/* calculate shift amount. */
+		n = (sizeof(*b) - n) * CHAR_BIT;
+		s = b[0] >> n;
+		CKSUM_ADD_CARRY(sum, s);
+	}
+
+	/* reduce to 16 bits */
+	dw1 = sum;
+	dw2 = sum >> 32;
+	CKSUM_ADD_CARRY(dw1, dw2);
+	w1 = dw1;
+	w2 = dw1 >> 16;
+	CKSUM_ADD_CARRY(w1, w2);
+	return w1;
+}
+
+
+/**
+ * Process UDP or TCP checksum over possibly multi-segmented packet.
+ * @param mb
+ *   The pointer to the mbuf with the packet.
+ * @param l4_ofs
+ *   Offset to the beginning of the L4 header (should be in first segment).
+ * @param cksum
+ *   Already pre-calculated pseudo-header checksum value.
+ * @return
+ *   The complemented checksum.
+ */
+static inline uint32_t
+__udptcp_mbuf_cksum(const struct rte_mbuf *mb, uint16_t l4_ofs,
+	uint32_t cksum)
+{
+	uint32_t dlen, i, plen;
+	const struct rte_mbuf *ms;
+	const void *data;
+
+	plen = rte_pktmbuf_pkt_len(mb);
+	ms = mb;
+
+	for (i = l4_ofs; i < plen && ms != NULL; i += dlen) {
+		data = rte_pktmbuf_mtod_offset(ms, const void *, l4_ofs);
+		dlen = rte_pktmbuf_data_len(ms) - l4_ofs;
+		cksum += __raw_cksum(data, dlen);
+		ms = ms->next;
+		l4_ofs = 0;
+	}
+
+	cksum = ((cksum & 0xffff0000) >> 16) + (cksum & 0xffff);
+	cksum = (~cksum) & 0xffff;
+	if (cksum == 0)
+		cksum = 0xffff;
+
+	return cksum;
+}
+
+/**
+ * Process the pseudo-header checksum of an IPv4 header.
+ *
+ * Depending on the ol_flags, the pseudo-header checksum expected by the
+ * drivers is not the same. For instance, when TSO is enabled, the IP
+ * payload length must not be included in the packet.
+ *
+ * When ol_flags is 0, it computes the standard pseudo-header checksum.
+ *
+ * @param ipv4_hdr
+ *   The pointer to the contiguous IPv4 header.
+ * @param ipv4_len
+ *   Length of the IPv4 header.
+ * @param ol_flags
+ *   The ol_flags of the associated mbuf.
+ * @return
+ *   The non-complemented checksum to set in the L4 header.
+ */
+static inline uint16_t
+_ipv4x_phdr_cksum(const struct ipv4_hdr *ipv4_hdr, size_t ipv4h_len,
+	uint64_t ol_flags)
+{
+	uint32_t s0, s1;
+
+	s0 = ipv4_hdr->src_addr;
+	s1 = ipv4_hdr->dst_addr;
+	CKSUM_ADD_CARRY(s0, s1);
+
+	if (ol_flags & PKT_TX_TCP_SEG)
+		s1 = 0;
+	else
+		s1 = rte_cpu_to_be_16(
+			(uint16_t)(rte_be_to_cpu_16(ipv4_hdr->total_length) -
+			ipv4h_len));
+
+	s1 += rte_cpu_to_be_16(ipv4_hdr->next_proto_id);
+	CKSUM_ADD_CARRY(s0, s1);
+
+	return __rte_raw_cksum_reduce(s0);
+}
+
+/**
+ * Process the IPv4 UDP or TCP checksum.
+ *
+ * @param mb
+ *   The pointer to the IPv4 packet.
+ * @param l4_ofs
+ *   Offset to the beginning of the L4 header (should be in first segment).
+ * @param ipv4_hdr
+ *   The pointer to the contiguous IPv4 header.
+ * @return
+ *   The complemented checksum to set in the IP packet.
+ */
+static inline int
+_ipv4_udptcp_mbuf_cksum(const struct rte_mbuf *mb, uint16_t l4_ofs,
+	const struct ipv4_hdr *ipv4_hdr)
+{
+	uint32_t cksum;
+
+	cksum = _ipv4x_phdr_cksum(ipv4_hdr, mb->l3_len, 0);
+	cksum  = __udptcp_mbuf_cksum(mb, l4_ofs, cksum);
+
+	return cksum;
+}
+
+/**
+ * Process the IPv6 UDP or TCP checksum.
+ *
+ * @param mb
+ *   The pointer to the IPv6 packet.
+ * @param l4_ofs
+ *   Offset to the beginning of the L4 header (should be in first segment).
+ * @param ipv6_hdr
+ *   The pointer to the contiguous IPv6 header.
+ * @return
+ *   The complemented checksum to set in the IP packet.
+ */
+static inline int
+_ipv6_udptcp_mbuf_cksum(const struct rte_mbuf *mb, uint16_t l4_ofs,
+	const struct ipv6_hdr *ipv6_hdr)
+{
+	uint32_t cksum;
+
+	cksum = rte_ipv6_phdr_cksum(ipv6_hdr, 0);
+	cksum  = __udptcp_mbuf_cksum(mb, l4_ofs, cksum);
+
+	return cksum;
+}
+
+static inline uint16_t
+_ipv4x_cksum(const void *iph, size_t len)
+{
+	uint16_t cksum;
+
+	cksum = __raw_cksum(iph, len);
+	return (cksum == 0xffff) ? cksum : ~cksum;
+}
+
+
+/*
+ * Analog of read-write locks, very much in favour of read side.
+ * Assumes, that there are no more then INT32_MAX concurrent readers.
+ * Consider to move into DPDK librte_eal.
+ */
+
+static inline int
+rwl_try_acquire(rte_atomic32_t *p)
+{
+	return rte_atomic32_add_return(p, 1);
+}
+
+static inline void
+rwl_release(rte_atomic32_t *p)
+{
+	rte_atomic32_sub(p, 1);
+}
+
+static inline int
+rwl_acquire(rte_atomic32_t *p)
+{
+	int32_t rc;
+
+	rc = rwl_try_acquire(p);
+	if (rc < 0)
+		rwl_release(p);
+	return rc;
+}
+
+static inline void
+rwl_down(rte_atomic32_t *p)
+{
+	 while (rte_atomic32_cmpset((volatile uint32_t *)p, 0, INT32_MIN) == 0)
+		rte_pause();
+}
+
+static inline void
+rwl_up(rte_atomic32_t *p)
+{
+	rte_atomic32_sub(p, INT32_MIN);
+}
+
+#ifdef __cplusplus
+}
+#endif
+
+#endif /* _MISC_H_ */
diff --git a/lib/libtle_udp/osdep.h b/lib/libtle_udp/osdep.h
new file mode 100644
index 0000000..6161242
--- /dev/null
+++ b/lib/libtle_udp/osdep.h
@@ -0,0 +1,56 @@
+/*
+ * Copyright (c) 2016  Intel Corporation.
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at:
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#ifndef _OSDEP_H_
+#define _OSDEP_H_
+
+#include <rte_vect.h>
+#include <rte_log.h>
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+#define UDP_LOG(lvl, fmt, args...)      RTE_LOG(lvl, USER1, fmt, ##args)
+
+/*
+ * if no AVX support, define _ymm_t here.
+ */
+
+#ifdef __AVX__
+
+#define _ymm_t	rte_ymm_t
+
+#else
+
+#define YMM_SIZE        (2 * sizeof(rte_xmm_t))
+#define YMM_MASK        (YMM_SIZE - 1)
+
+typedef union _ymm {
+	xmm_t    x[YMM_SIZE / sizeof(xmm_t)];
+	uint8_t  u8[YMM_SIZE / sizeof(uint8_t)];
+	uint16_t u16[YMM_SIZE / sizeof(uint16_t)];
+	uint32_t u32[YMM_SIZE / sizeof(uint32_t)];
+	uint64_t u64[YMM_SIZE / sizeof(uint64_t)];
+	double   pd[YMM_SIZE / sizeof(double)];
+} _ymm_t;
+
+#endif /* __AVX__ */
+
+#ifdef __cplusplus
+}
+#endif
+
+#endif /* _OSDEP_H_ */
diff --git a/lib/libtle_udp/port_bitmap.h b/lib/libtle_udp/port_bitmap.h
new file mode 100644
index 0000000..6aff4e6
--- /dev/null
+++ b/lib/libtle_udp/port_bitmap.h
@@ -0,0 +1,112 @@
+/*
+ * Copyright (c) 2016  Intel Corporation.
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at:
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#ifndef _PORT_BITMAP_H_
+#define _PORT_BITMAP_H_
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+/*
+ * Simple implementation of bitmap for all possible UDP ports [0-UINT16_MAX].
+ */
+
+#define MAX_PORT_NUM	(UINT16_MAX + 1)
+
+#define PORT_BLK(p)	((p) / (sizeof(uint32_t) * CHAR_BIT))
+#define PORT_IDX(p)	((p) % (sizeof(uint32_t) * CHAR_BIT))
+
+#define	MAX_PORT_BLK	PORT_BLK(MAX_PORT_NUM)
+
+struct udp_pbm {
+	uint32_t nb_set; /* number of bits set. */
+	uint32_t blk;    /* last block with free entry. */
+	uint32_t bm[MAX_PORT_BLK];
+};
+
+static inline void
+udp_pbm_init(struct udp_pbm *pbm, uint32_t blk)
+{
+	pbm->bm[0] = 1;
+	pbm->nb_set = 1;
+	pbm->blk = blk;
+}
+
+static inline void
+udp_pbm_set(struct udp_pbm *pbm, uint16_t port)
+{
+	uint32_t i, b, v;
+
+	i = PORT_BLK(port);
+	b = 1 << PORT_IDX(port);
+	v = pbm->bm[i];
+	pbm->bm[i] = v | b;
+	pbm->nb_set += (v & b) == 0;
+}
+
+static inline void
+udp_pbm_clear(struct udp_pbm *pbm, uint16_t port)
+{
+	uint32_t i, b, v;
+
+	i = PORT_BLK(port);
+	b = 1 << PORT_IDX(port);
+	v = pbm->bm[i];
+	pbm->bm[i] = v & ~b;
+	pbm->nb_set -= (v & b) != 0;
+}
+
+
+static inline uint32_t
+udp_pbm_check(const struct udp_pbm *pbm, uint16_t port)
+{
+	uint32_t i, v;
+
+	i = PORT_BLK(port);
+	v = pbm->bm[i] >> PORT_IDX(port);
+	return v & 1;
+}
+
+static inline uint16_t
+udp_pbm_find_range(struct udp_pbm *pbm, uint32_t start_blk, uint32_t end_blk)
+{
+	uint32_t i, v;
+	uint16_t p;
+
+	if (pbm->nb_set == MAX_PORT_NUM)
+		return 0;
+
+	p = 0;
+	for (i = start_blk; i != end_blk; i++) {
+		i %= RTE_DIM(pbm->bm);
+		v = pbm->bm[i];
+		if (v != UINT32_MAX) {
+			for (p = i * (sizeof(pbm->bm[0]) * CHAR_BIT);
+					(v & 1) != 0; v >>= 1, p++)
+				;
+
+			pbm->blk = i;
+			break;
+		}
+	}
+	return p;
+}
+
+#ifdef __cplusplus
+}
+#endif
+
+#endif /* _PORT_BITMAP_H_ */
diff --git a/lib/libtle_udp/tle_event.h b/lib/libtle_udp/tle_event.h
new file mode 100644
index 0000000..1a5c436
--- /dev/null
+++ b/lib/libtle_udp/tle_event.h
@@ -0,0 +1,257 @@
+/*
+ * Copyright (c) 2016  Intel Corporation.
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at:
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#ifndef _SEV_IMPL_H_
+#define _SEV_IMPL_H_
+
+#include <rte_common.h>
+#include <rte_spinlock.h>
+#include <rte_atomic.h>
+#include <sys/queue.h>
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+struct tle_evq;
+
+/**
+ * Possible states of the event.
+ */
+enum tle_ev_state {
+	TLE_SEV_IDLE,
+	TLE_SEV_DOWN,
+	TLE_SEV_UP,
+	TLE_SEV_NUM
+};
+
+struct tle_event {
+	TAILQ_ENTRY(tle_event) ql;
+	struct tle_evq *head;
+	const void *data;
+	enum tle_ev_state state;
+} __rte_cache_aligned;
+
+struct tle_evq {
+	rte_spinlock_t lock;
+	uint32_t nb_events;
+	uint32_t nb_armed;
+	uint32_t nb_free;
+	TAILQ_HEAD(, tle_event) armed;
+	TAILQ_HEAD(, tle_event) free;
+	struct tle_event events[0];
+};
+
+/**
+ * event queue creation parameters.
+ */
+struct tle_evq_param {
+	int32_t socket_id;    /**< socket ID to allocate memory from. */
+	uint32_t max_events;  /**< max number of events in queue. */
+};
+
+/**
+ * create event queue.
+ * @param prm
+ *   Parameters used to create and initialise the queue.
+ * @return
+ *   Pointer to new event queue structure,
+ *   or NULL on error, with error code set in rte_errno.
+ *   Possible rte_errno errors include:
+ *   - EINVAL - invalid parameter passed to function
+ *   - ENOMEM - out of memory
+ */
+struct tle_evq *tle_evq_create(const struct tle_evq_param *prm);
+
+/**
+ * Destroy given event queue.
+ *
+ * @param evq
+ *   event queue to destroy
+ */
+void tle_evq_destroy(struct tle_evq *evq);
+
+/**
+ * allocate a new event within given event queue.
+ * @param evq
+ *    event queue to allocate a new stream within.
+ * @param data
+ *   User data to be associated with that event.
+ * @return
+ *   Pointer to event structure that can be used in future tle_event API calls,
+ *   or NULL on error, with error code set in rte_errno.
+ *   Possible rte_errno errors include:
+ *   - EINVAL - invalid parameter passed to function
+ *   - ENOMEM - max limit of allocated events reached for that context
+ */
+struct tle_event *tle_event_alloc(struct tle_evq *evq, const void *data);
+
+/**
+ * free an allocated event.
+ * @param ev
+ *   Pointer to the event to free.
+ */
+void tle_event_free(struct tle_event *ev);
+
+
+/**
+ * move event from DOWN to UP state.
+ * @param ev
+ *   Pointer to the event.
+ */
+static inline void
+tle_event_raise(struct tle_event *ev)
+{
+	struct tle_evq *q;
+
+	if (ev->state != TLE_SEV_DOWN)
+		return;
+
+	q = ev->head;
+	rte_compiler_barrier();
+
+	rte_spinlock_lock(&q->lock);
+	if (ev->state == TLE_SEV_DOWN) {
+		ev->state = TLE_SEV_UP;
+		TAILQ_INSERT_TAIL(&q->armed, ev, ql);
+		q->nb_armed++;
+	}
+	rte_spinlock_unlock(&q->lock);
+}
+
+/**
+ * move event from UP to DOWN state.
+ * @param ev
+ *   Pointer to the event.
+ */
+static inline void
+tle_event_down(struct tle_event *ev)
+{
+	struct tle_evq *q;
+
+	if (ev->state != TLE_SEV_UP)
+		return;
+
+	q = ev->head;
+	rte_compiler_barrier();
+
+	rte_spinlock_lock(&q->lock);
+	if (ev->state == TLE_SEV_UP) {
+		ev->state = TLE_SEV_DOWN;
+		TAILQ_REMOVE(&q->armed, ev, ql);
+		q->nb_armed--;
+	}
+	rte_spinlock_unlock(&q->lock);
+}
+
+/**
+ * move from IDLE to DOWN/UP state.
+ * @param ev
+ *   Pointer to the event.
+ * @param st
+ *   new state for the event.
+ */
+static inline void
+tle_event_active(struct tle_event *ev, enum tle_ev_state st)
+{
+	struct tle_evq *q;
+
+	if (st == ev->state)
+		return;
+
+	q = ev->head;
+	rte_compiler_barrier();
+
+	rte_spinlock_lock(&q->lock);
+	if (st > ev->state) {
+		if (st == TLE_SEV_UP) {
+			TAILQ_INSERT_TAIL(&q->armed, ev, ql);
+			q->nb_armed++;
+		}
+		ev->state = st;
+	}
+	rte_spinlock_unlock(&q->lock);
+}
+
+/**
+ * move event IDLE state.
+ * @param ev
+ *   Pointer to the event.
+ */
+static inline void
+tle_event_idle(struct tle_event *ev)
+{
+	struct tle_evq *q;
+
+	if (ev->state == TLE_SEV_IDLE)
+		return;
+
+	q = ev->head;
+	rte_compiler_barrier();
+
+	rte_spinlock_lock(&q->lock);
+	if (ev->state == TLE_SEV_UP) {
+		TAILQ_REMOVE(&q->armed, ev, ql);
+		q->nb_armed--;
+	}
+	ev->state = TLE_SEV_IDLE;
+	rte_spinlock_unlock(&q->lock);
+}
+
+
+/*
+ * return up to *num* user data pointers associated with
+ * the events that were in the UP state.
+ * Each retrieved event is automatically moved into the DOWN state.
+ * @param evq
+ *   event queue to retrieve events from.
+ * @param evd
+ *   An array of user data pointers associated with the events retrieved.
+ *   It must be large enough to store up to *num* pointers in it.
+ * @param num
+ *   Number of elements in the *evd* array.
+ * @return
+ *   number of of entries filled inside *evd* array.
+ */
+static inline int32_t
+tle_evq_get(struct tle_evq *evq, const void *evd[], uint32_t num)
+{
+	uint32_t i, n;
+	struct tle_event *ev;
+
+	if (evq->nb_armed == 0)
+		return 0;
+
+	rte_compiler_barrier();
+
+	rte_spinlock_lock(&evq->lock);
+	n = RTE_MIN(num, evq->nb_armed);
+	for (i = 0; i != n; i++) {
+		ev = TAILQ_FIRST(&evq->armed);
+		ev->state = TLE_SEV_DOWN;
+		TAILQ_REMOVE(&evq->armed, ev, ql);
+		evd[i] = ev->data;
+	}
+	evq->nb_armed -= n;
+	rte_spinlock_unlock(&evq->lock);
+	return n;
+}
+
+
+#ifdef __cplusplus
+}
+#endif
+
+#endif /* _SEV_IMPL_H_ */
diff --git a/lib/libtle_udp/tle_udp_impl.h b/lib/libtle_udp/tle_udp_impl.h
new file mode 100644
index 0000000..a5d17e1
--- /dev/null
+++ b/lib/libtle_udp/tle_udp_impl.h
@@ -0,0 +1,373 @@
+/*
+ * Copyright (c) 2016  Intel Corporation.
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at:
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#ifndef _TLE_UDP_IMPL_H_
+#define _TLE_UDP_IMPL_H_
+
+#include <stdint.h>
+#include <sys/socket.h>
+#include <netinet/in.h>
+#include <rte_common.h>
+#include <rte_mbuf.h>
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+/**
+ * <udp_ctx>  - each such ctx represents an 'independent copy of the stack'.
+ * It owns set of <udp_stream>s and <udp_dev>s entities and provides
+ * (de)multiplexing input/output packets from/into UDP devices into/from
+ * UDP streams.
+ * <udp_dev> is an abstraction for the underlying device, that is able
+ * to RX/TX packets and may provide some HW offload capabilities.
+ * It is a user responsibility to add to the <udp_ctx> all <udp_dev>s,
+ * that context has to manage, before starting to do stream operations
+ * (open/send/recv,close) over that context.
+ * Right now adding/deleting <udp_dev>s to the context with open
+ * streams is not supported.
+ * <udp_stream> represents an UDP endpoint <addr, port> and is an analogy to
+ * socket entity.
+ * As with a socket, there are ability to do recv/send over it.
+ * <udp_stream> belongs to particular <udp_ctx> but is visible globally across
+ * the process, i.e. any thread within the process can do recv/send over it
+ * without any further synchronisation.
+ * While 'upper' layer API is thread safe, lower layer API (rx_bulk/tx_bulk)
+ * is not thread safe and is not supposed to be run on multiple threads
+ * in parallel.
+ * So single thread can drive multiple <udp_ctx>s and do IO for them,
+ * but multiple threads can't drive same <udp_ctx> without some
+ * explicit synchronization.
+ */
+
+struct tle_udp_ctx;
+struct tle_udp_dev;
+
+/**
+ * UDP device parameters.
+ */
+struct tle_udp_dev_param {
+	uint32_t rx_offload; /**< DEV_RX_OFFLOAD_* supported. */
+	uint32_t tx_offload; /**< DEV_TX_OFFLOAD_* supported. */
+	struct in_addr local_addr4;  /**< local IPv4 address assigned. */
+	struct in6_addr local_addr6; /**< local IPv6 address assigned. */
+};
+
+#define TLE_UDP_MAX_HDR	0x60
+
+struct tle_udp_dest {
+	struct rte_mempool *head_mp; /**< MP for fragment feaders. */
+	struct tle_udp_dev *dev;     /**< device to send packets through. */
+	uint16_t mtu;                /**< MTU for given destination. */
+	uint8_t l2_len;  /**< L2 header lenght. */
+	uint8_t l3_len;  /**< L3 header lenght. */
+	uint8_t hdr[TLE_UDP_MAX_HDR]; /**< L2/L3 headers. */
+};
+
+/**
+ * UDP context creation parameters.
+ */
+struct tle_udp_ctx_param {
+	int32_t socket_id;         /**< socket ID to allocate memory for. */
+	uint32_t max_streams;      /**< max number of streams in context. */
+	uint32_t max_stream_rbufs; /**< max recv mbufs per stream. */
+	uint32_t max_stream_sbufs; /**< max send mbufs per stream. */
+
+	int (*lookup4)(void *opaque, const struct in_addr *addr,
+		struct tle_udp_dest *res);
+	/**< will be called by send() to get IPv4 packet destination info. */
+	void *lookup4_data;
+	/**< opaque data pointer for lookup4() callback. */
+
+	int (*lookup6)(void *opaque, const struct in6_addr *addr,
+		struct tle_udp_dest *res);
+	/**< will be called by send() to get IPv6 packet destination info. */
+	void *lookup6_data;
+	/**< opaque data pointer for lookup6() callback. */
+};
+
+/**
+ * create UDP context.
+ * @param ctx_prm
+ *   Parameters used to create and initialise the UDP context.
+ * @return
+ *   Pointer to UDP context structure that can be used in future UDP
+ *   operations, or NULL on error, with error code set in rte_errno.
+ *   Possible rte_errno errors include:
+ *   - EINVAL - invalid parameter passed to function
+ *   - ENOMEM - out of memory
+ */
+struct tle_udp_ctx *
+tle_udp_create(const struct tle_udp_ctx_param *ctx_prm);
+
+/**
+ * Destroy given UDP context.
+ *
+ * @param ctx
+ *   UDP context to destroy
+ */
+void tle_udp_destroy(struct tle_udp_ctx *ctx);
+
+/**
+ * Add new device into the given UDP context.
+ * This function is not multi-thread safe.
+ *
+ * @param ctx
+ *   UDP context to add new device into.
+ * @param dev_prm
+ *   Parameters used to create and initialise new device inside the
+ *   UDP context.
+ * @return
+ *   Pointer to UDP device structure that can be used in future UDP
+ *   operations, or NULL on error, with error code set in rte_errno.
+ *   Possible rte_errno errors include:
+ *   - EINVAL - invalid parameter passed to function
+ *   - ENODEV - max possible value of open devices is reached
+ *   - ENOMEM - out of memory
+ */
+struct tle_udp_dev *
+tle_udp_add_dev(struct tle_udp_ctx *ctx,
+		const struct tle_udp_dev_param *dev_prm);
+
+/**
+ * Remove and destroy previously added device from the given UDP context.
+ * This function is not multi-thread safe.
+ *
+ * @param dev
+ *   UDP device to remove and destroy.
+ * @return
+ *   zero on successful completion.
+ *   - -EINVAL - invalid parameter passed to function
+ */
+int tle_udp_del_dev(struct tle_udp_dev *dev);
+
+/**
+ * Flags to the UDP context that destinations info might be changed,
+ * so if it has any destinations data cached, then
+ * it has to be invalidated.
+ * @param ctx
+ *   UDP context to invalidate.
+ */
+void tle_udp_ctx_invalidate(struct tle_udp_ctx *ctx);
+
+struct tle_udp_stream;
+
+/**
+ * Stream asynchronous notification mechanisms:
+ * a) recv/send callback.
+ * Stream recv/send notification callbacks behaviour is edge-triggered (ET).
+ * recv callback will be invoked if stream receive buffer was empty and
+ * new packet(s) have arrived.
+ * send callback will be invoked when stream send buffer was full,
+ * and some packets belonging to that stream were sent
+ * (part of send buffer became free again).
+ * Note that both recv and send callbacks are called with sort of read lock
+ * held on that stream. So it is not permitted to call stream_close()
+ * within the callback function. Doing that would cause a deadlock.
+ * While it is allowed to call stream send/recv functions within the
+ * callback, it is not recommended: callback function will be invoked
+ * within tle_udp_rx_bulk/tle_udp_tx_bulk context and some heavy processing
+ * within the callback functions might cause performance degradation
+ * or even loss of packets for further streams.
+ * b) recv/send event.
+ * Stream recv/send events behavour is level-triggered (LT).
+ * receive event will be raised by either
+ * tle_udp_rx_burst() or tle_udp_stream_recv() as long as there are any
+ * remaining packets inside stream receive buffer.
+ * send event will be raised by either
+ * tle_udp_tx_burst() or tle_udp_stream_send() as long as there are any
+ * free space inside stream send buffer.
+ * Note that callback and event are mutually exclusive on <stream, op> basis.
+ * It is not possible to  open a stream with both recv event and callback
+ * specified.
+ * Though it is possible to open a stream with recv callback and send event,
+ * or visa-versa.
+ * If the user doesn't need any notification mechanism for that stream,
+ * both event and callback could be set to zero.
+ */
+
+/**
+ * Stream recv/send callback function and data.
+ */
+struct tle_udp_stream_cb {
+	void (*func)(void *, struct tle_udp_stream *);
+	void *data;
+};
+
+struct tle_event;
+
+/**
+ * UDP stream creation parameters.
+ */
+struct tle_udp_stream_param {
+	struct sockaddr_storage local_addr;  /**< stream local address. */
+	struct sockaddr_storage remote_addr; /**< stream remote address. */
+
+	/* _cb and _ev are mutually exclusive */
+	struct tle_event *recv_ev;          /**< recv event to use.  */
+	struct tle_udp_stream_cb recv_cb;   /**< recv callback to use. */
+
+	struct tle_event *send_ev;          /**< send event to use. */
+	struct tle_udp_stream_cb send_cb;   /**< send callback to use. */
+};
+
+/**
+ * create a new stream within given UDP context.
+ * @param ctx
+ *   UDP context to create new stream within.
+ * @param prm
+ *   Parameters used to create and initialise the new stream.
+ * @return
+ *   Pointer to UDP stream structure that can be used in future UDP API calls,
+ *   or NULL on error, with error code set in rte_errno.
+ *   Possible rte_errno errors include:
+ *   - EINVAL - invalid parameter passed to function
+ *   - ENOFILE - max limit of open streams reached for that context
+ */
+struct tle_udp_stream *
+tle_udp_stream_open(struct tle_udp_ctx *ctx,
+	const struct tle_udp_stream_param *prm);
+
+/**
+ * close an open stream.
+ * All packets still remaining in stream receive buffer will be freed.
+ * All packets still remaining in stream transmit buffer will be kept
+ * for father transmission.
+ * @param s
+ *   Pointer to the stream to close.
+ * @return
+ *   zero on successful completion.
+ *   - -EINVAL - invalid parameter passed to function
+ */
+int tle_udp_stream_close(struct tle_udp_stream *s);
+
+/**
+ * get open stream parameters.
+ * @param s
+ *   Pointer to the stream.
+ * @return
+ *   zero on successful completion.
+ *   - EINVAL - invalid parameter passed to function
+ */
+int
+tle_udp_stream_get_param(const struct tle_udp_stream *s,
+	struct tle_udp_stream_param *prm);
+
+/**
+ * Take input mbufs and distribute them to open UDP streams.
+ * expects that for each input packet:
+ *	- l2_len, l3_len, l4_len are setup correctly
+ *	- (packet_type & (RTE_PTYPE_L3_IPV4 | RTE_PTYPE_L3_IPV6)) != 0,
+ *	- (packet_type & RTE_PTYPE_L4_UDP) != 0,
+ * During delivery L3/L4 checksums will be verified
+ * (either relies on HW offload or in SW).
+ * This function is not multi-thread safe.
+ * @param dev
+ *   UDP device the packets were received from.
+ * @param pkt
+ *   The burst of input packets that need to be processed.
+ * @param rp
+ *   The array that will contain pointers of unprocessed packets at return.
+ *   Should contain at least *num* elements.
+ * @param rc
+ *   The array that will contain error code for corresponding rp[] entry:
+ *   - ENOENT - no open stream matching this packet.
+ *   - ENOBUFS - receive buffer of the destination stream is full.
+ *   Should contain at least *num* elements.
+ * @param num
+ *   Number of elements in the *pkt* input array.
+ * @return
+ *   number of packets delivered to the UDP streams.
+ */
+uint16_t tle_udp_rx_bulk(struct tle_udp_dev *dev, struct rte_mbuf *pkt[],
+	struct rte_mbuf *rp[], int32_t rc[], uint16_t num);
+
+/**
+ * Fill *pkt* with pointers to the packets that have to be transmitted
+ * over given UDP device.
+ * Output packets have to be ready to be passed straight to rte_eth_tx_burst()
+ * without any extra processing.
+ * UDP/IPv4 checksum either already calculated or appropriate mbuf fields set
+ * properly for HW offload.
+ * This function is not multi-thread safe.
+ * @param dev
+ *   UDP device the output packets will be transmitted over.
+ * @param pkt
+ *   An array of pointers to *rte_mbuf* structures that
+ *   must be large enough to store up to *num* pointers in it.
+ * @param num
+ *   Number of elements in the *pkt* array.
+ * @return
+ *   number of of entries filled inside *pkt* array.
+ */
+uint16_t tle_udp_tx_bulk(struct tle_udp_dev *dev, struct rte_mbuf *pkt[],
+	uint16_t num);
+
+/*
+ * return up to *num* mbufs that was received for given UDP stream.
+ * For each returned mbuf:
+ * data_off set to the start of the packet's UDP data
+ * l2_len, l3_len, l4_len are setup properly
+ * (so user can still extract L2/L3 address info if needed)
+ * packet_type RTE_PTYPE_L2/L3/L4 bits are setup properly.
+ * L3/L4 checksum is verified.
+ * Packets with invalid L3/L4 checksum will be silently dropped.
+ * @param s
+ *   UDP stream to receive packets from.
+ * @param pkt
+ *   An array of pointers to *rte_mbuf* structures that
+ *   must be large enough to store up to *num* pointers in it.
+ * @param num
+ *   Number of elements in the *pkt* array.
+ * @return
+ *   number of of entries filled inside *pkt* array.
+ */
+uint16_t tle_udp_stream_recv(struct tle_udp_stream *s, struct rte_mbuf *pkt[],
+	uint16_t num);
+
+/**
+ * Consume and queue up to *num* packets, that will be sent eventually
+ * by tle_udp_tx_bulk().
+ * If *dst_addr* is NULL, then default remote address associated with that
+ * stream (if any) will be used.
+ * The main purpose of that function is to determine over which UDP dev
+ * given packets have to be sent out and do necessary preparations for that.
+ * Based on the *dst_addr* it does route lookup, fills L2/L3/L4 headers,
+ * and, if necessary, fragments packets.
+ * Depending on the underlying device information, it either does
+ * IP/UDP checksum calculations in SW or sets mbuf TX checksum
+ * offload fields properly.
+ * For each input mbuf the following conditions have to be met:
+ *	- data_off point to the start of packet's UDP data.
+ *	- there is enough header space to prepend L2/L3/L4 headers.
+ * @param s
+ *   UDP stream to send packets over.
+ * @param pkt
+ *   The burst of output packets that need to be send.
+ * @param num
+ *   Number of elements in the *pkt* array.
+ * @param dst_addr
+ *   Destination address to send packets to.
+ * @return
+ *   number of packets successfully queued in the stream send buffer.
+ */
+uint16_t tle_udp_stream_send(struct tle_udp_stream *s, struct rte_mbuf *pkt[],
+	uint16_t num, const struct sockaddr *dst_addr);
+
+#ifdef __cplusplus
+}
+#endif
+
+#endif /* _TLE_UDP_IMPL_H_ */
diff --git a/lib/libtle_udp/udp_ctl.c b/lib/libtle_udp/udp_ctl.c
new file mode 100644
index 0000000..36ec8a6
--- /dev/null
+++ b/lib/libtle_udp/udp_ctl.c
@@ -0,0 +1,723 @@
+/*
+ * Copyright (c) 2016  Intel Corporation.
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at:
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include <string.h>
+#include <rte_malloc.h>
+#include <rte_errno.h>
+#include <rte_ethdev.h>
+#include <rte_ip.h>
+#include <rte_udp.h>
+
+#include "udp_impl.h"
+#include "misc.h"
+
+#define	LPORT_START	0x8000
+#define	LPORT_END	MAX_PORT_NUM
+
+#define	LPORT_START_BLK	PORT_BLK(LPORT_START)
+#define	LPORT_END_BLK	PORT_BLK(LPORT_END)
+
+#define	MAX_BURST	0x20
+
+static const struct in6_addr tle_udp6_any = IN6ADDR_ANY_INIT;
+static const struct in6_addr tle_udp6_none = {
+	{
+		.__u6_addr32 = {
+			UINT32_MAX, UINT32_MAX, UINT32_MAX, UINT32_MAX
+		},
+	},
+};
+
+static int
+check_dev_prm(const struct tle_udp_dev_param *dev_prm)
+{
+	/* no valid IPv4/IPv6 addresses provided. */
+	if (dev_prm->local_addr4.s_addr == INADDR_ANY &&
+			memcmp(&dev_prm->local_addr6, &tle_udp6_any,
+			sizeof(tle_udp6_any)) == 0)
+		return -EINVAL;
+
+	return 0;
+}
+
+static void
+unuse_stream(struct tle_udp_stream *s)
+{
+	s->type = TLE_UDP_VNUM;
+	rte_atomic32_set(&s->rx.use, INT32_MIN);
+	rte_atomic32_set(&s->tx.use, INT32_MIN);
+}
+
+static int
+init_stream(struct tle_udp_ctx *ctx, struct tle_udp_stream *s)
+{
+	size_t sz;
+	uint32_t n;
+	struct bcg_store_prm sp;
+	char name[RTE_RING_NAMESIZE];
+
+	/* init RX part. */
+
+	n = RTE_MAX(ctx->prm.max_stream_rbufs, 1U);
+	n = rte_align32pow2(n);
+	sz = sizeof(*s->rx.q) + n * sizeof(s->rx.q->ring[0]);
+
+	s->rx.q = rte_zmalloc_socket(NULL, sz, RTE_CACHE_LINE_SIZE,
+		ctx->prm.socket_id);
+	if (s->rx.q == NULL) {
+		UDP_LOG(ERR, "%s(%p): allocation of %zu bytes on socket %d "
+			"failed with error code: %d\n",
+			__func__, s, sz, ctx->prm.socket_id, rte_errno);
+		return ENOMEM;
+	}
+
+	snprintf(name, sizeof(name), "%p@%zu", s, sz);
+	rte_ring_init(s->rx.q, name, n, RING_F_SP_ENQ);
+
+	/* init TX part. */
+
+	sp.socket_id = ctx->prm.socket_id;
+	sp.max_bufs = ctx->prm.max_stream_sbufs;
+	sp.min_cages = RTE_DIM(ctx->dev) + 1;
+	sp.cage_bufs = MAX_BURST;
+	sp.cage_align = RTE_CACHE_LINE_SIZE;
+	sp.user_data = s;
+
+	s->tx.st = bcg_create(&sp);
+	if (s->tx.st == NULL) {
+		UDP_LOG(ERR,
+			"%s(%p): bcg_create() failed with error code: %d\n",
+			__func__, s, rte_errno);
+		return ENOMEM;
+	}
+
+	s->ctx = ctx;
+	unuse_stream(s);
+	STAILQ_INSERT_TAIL(&ctx->streams.free, s, link);
+
+	return 0;
+}
+
+static void
+fini_stream(struct tle_udp_stream *s)
+{
+	bcg_destroy(s->tx.st);
+	rte_free(s->rx.q);
+}
+
+struct tle_udp_ctx *
+tle_udp_create(const struct tle_udp_ctx_param *ctx_prm)
+{
+	struct tle_udp_ctx *ctx;
+	size_t sz;
+	uint32_t i;
+
+	if (ctx_prm == NULL) {
+		rte_errno = EINVAL;
+		return NULL;
+	}
+
+	sz = sizeof(*ctx);
+	ctx = rte_zmalloc_socket(NULL, sz, RTE_CACHE_LINE_SIZE,
+		ctx_prm->socket_id);
+	if (ctx == NULL) {
+		UDP_LOG(ERR, "allocation of %zu bytes for new udp_ctx "
+			"on socket %d failed\n",
+			sz, ctx_prm->socket_id);
+		return NULL;
+	}
+
+	ctx->prm = *ctx_prm;
+
+	sz = sizeof(*ctx->streams.buf) * ctx_prm->max_streams;
+	ctx->streams.buf = rte_zmalloc_socket(NULL, sz, RTE_CACHE_LINE_SIZE,
+		ctx_prm->socket_id);
+	if (ctx->streams.buf == NULL) {
+		UDP_LOG(ERR, "allocation of %zu bytes on socket %d "
+			"for %u udp_streams failed\n",
+			sz, ctx_prm->socket_id, ctx_prm->max_streams);
+		tle_udp_destroy(ctx);
+		return NULL;
+	}
+
+	STAILQ_INIT(&ctx->streams.free);
+	for (i = 0; i != ctx_prm->max_streams &&
+			init_stream(ctx, &ctx->streams.buf[i]) == 0;
+			i++)
+		;
+
+	if (i != ctx_prm->max_streams) {
+		UDP_LOG(ERR, "initalisation of %u-th stream failed", i);
+		tle_udp_destroy(ctx);
+		return NULL;
+	}
+
+	for (i = 0; i != RTE_DIM(ctx->use); i++)
+		udp_pbm_init(ctx->use + i, LPORT_START_BLK);
+
+	ctx->streams.nb_free = ctx->prm.max_streams;
+	return ctx;
+}
+
+void
+tle_udp_destroy(struct tle_udp_ctx *ctx)
+{
+	uint32_t i;
+
+	if (ctx == NULL) {
+		rte_errno = EINVAL;
+		return;
+	}
+
+	if (ctx->streams.buf != 0) {
+		for (i = 0; i != ctx->prm.max_streams; i++)
+			fini_stream(&ctx->streams.buf[i]);
+		rte_free(ctx->streams.buf);
+	}
+
+	for (i = 0; i != RTE_DIM(ctx->dev); i++)
+		tle_udp_del_dev(ctx->dev + i);
+
+	rte_free(ctx);
+}
+
+void
+tle_udp_ctx_invalidate(struct tle_udp_ctx *ctx)
+{
+	RTE_SET_USED(ctx);
+}
+
+static int
+init_dev_proto(struct tle_udp_dev *dev, uint32_t idx, int32_t socket_id)
+{
+	size_t sz;
+
+	sz = sizeof(*dev->dp[idx]);
+	dev->dp[idx] = rte_zmalloc_socket(NULL, sz, RTE_CACHE_LINE_SIZE,
+		socket_id);
+
+	if (dev->dp[idx] == NULL) {
+		UDP_LOG(ERR, "allocation of %zu bytes on "
+			"socket %d for %u-th device failed\n",
+			sz, socket_id, idx);
+		return ENOMEM;
+	}
+
+	udp_pbm_init(&dev->dp[idx]->use, LPORT_START_BLK);
+	return 0;
+}
+
+static struct tle_udp_dev *
+find_free_dev(struct tle_udp_ctx *ctx)
+{
+	uint32_t i;
+
+	if (ctx->nb_dev < RTE_DIM(ctx->dev)) {
+		for (i = 0; i != RTE_DIM(ctx->dev); i++) {
+			if (ctx->dev[i].ctx != ctx)
+				return ctx->dev + i;
+		}
+	}
+
+	rte_errno = ENODEV;
+	return NULL;
+}
+
+struct tle_udp_dev *
+tle_udp_add_dev(struct tle_udp_ctx *ctx,
+	const struct tle_udp_dev_param *dev_prm)
+{
+	int32_t rc;
+	struct tle_udp_dev *dev;
+
+	if (ctx == NULL || dev_prm == NULL || check_dev_prm(dev_prm) != 0) {
+		rte_errno = EINVAL;
+		return NULL;
+	}
+
+	dev = find_free_dev(ctx);
+	if (dev == NULL)
+		return NULL;
+	rc = 0;
+
+	/* device can handle IPv4 traffic */
+	if (dev_prm->local_addr4.s_addr != INADDR_ANY)
+		rc = init_dev_proto(dev, TLE_UDP_V4, ctx->prm.socket_id);
+
+	/* device can handle IPv6 traffic */
+	if (rc == 0 && memcmp(&dev_prm->local_addr6, &tle_udp6_any,
+			sizeof(tle_udp6_any)) != 0)
+		rc = init_dev_proto(dev, TLE_UDP_V6, ctx->prm.socket_id);
+
+	if (rc != 0) {
+		/* cleanup and return an error. */
+		rte_free(dev->dp[TLE_UDP_V4]);
+		rte_free(dev->dp[TLE_UDP_V6]);
+		rte_errno = rc;
+		return NULL;
+	}
+
+	/* setup RX data. */
+	if (dev_prm->local_addr4.s_addr != INADDR_ANY &&
+			(dev_prm->rx_offload & DEV_RX_OFFLOAD_IPV4_CKSUM) == 0)
+		dev->rx.ol_flags[TLE_UDP_V4] |= PKT_RX_IP_CKSUM_BAD;
+	if ((dev_prm->rx_offload & DEV_RX_OFFLOAD_UDP_CKSUM) == 0) {
+		dev->rx.ol_flags[TLE_UDP_V4] |= PKT_RX_L4_CKSUM_BAD;
+		dev->rx.ol_flags[TLE_UDP_V6] |= PKT_RX_L4_CKSUM_BAD;
+	}
+
+	/* setup TX data. */
+	bcg_queue_reset(&dev->tx.beq);
+	bcg_queue_reset(&dev->tx.feq);
+
+	if ((dev_prm->tx_offload & DEV_TX_OFFLOAD_UDP_CKSUM) != 0) {
+		dev->tx.ol_flags[TLE_UDP_V4] |= PKT_TX_IPV4 | PKT_TX_UDP_CKSUM;
+		dev->tx.ol_flags[TLE_UDP_V6] |= PKT_TX_IPV6 | PKT_TX_UDP_CKSUM;
+	}
+	if ((dev_prm->tx_offload & DEV_TX_OFFLOAD_IPV4_CKSUM) != 0)
+		dev->tx.ol_flags[TLE_UDP_V4] |= PKT_TX_IPV4 | PKT_TX_IP_CKSUM;
+
+	dev->prm = *dev_prm;
+	dev->ctx = ctx;
+	ctx->nb_dev++;
+
+	return dev;
+}
+
+static void
+empty_cage(struct buf_cage *bc)
+{
+	uint32_t i, n;
+	struct rte_mbuf *pkt[MAX_BURST];
+
+	do {
+		n = bcg_get(bc, (const void **)(uintptr_t)pkt, RTE_DIM(pkt));
+		for (i = 0; i != n; i++)
+			rte_pktmbuf_free(pkt[i]);
+	} while (n != 0);
+
+	bcg_free(bc);
+}
+
+int
+tle_udp_del_dev(struct tle_udp_dev *dev)
+{
+	uint32_t p;
+	struct buf_cage *bc;
+	struct tle_udp_ctx *ctx;
+
+	ctx = dev->ctx;
+
+	if (dev == NULL || dev->ctx == NULL)
+		return -EINVAL;
+
+	p = dev - ctx->dev;
+
+	if (p >= RTE_DIM(ctx->dev) ||
+			(dev->dp[TLE_UDP_V4] == NULL &&
+			dev->dp[TLE_UDP_V6] == NULL))
+		return -EINVAL;
+
+	/* emtpy TX queues. */
+	if (dev->tx.bc != NULL)
+		empty_cage(dev->tx.bc);
+
+	bcg_queue_append(&dev->tx.beq, &dev->tx.feq);
+
+	while ((bc = __bcg_dequeue_head(&dev->tx.beq)) != NULL)
+		empty_cage(bc);
+
+	rte_free(dev->dp[TLE_UDP_V4]);
+	rte_free(dev->dp[TLE_UDP_V6]);
+	memset(dev, 0, sizeof(*dev));
+	ctx->nb_dev--;
+	return 0;
+}
+
+static inline void
+stream_down(struct tle_udp_stream *s)
+{
+	rwl_down(&s->rx.use);
+	rwl_down(&s->tx.use);
+}
+
+static inline void
+stream_up(struct tle_udp_stream *s)
+{
+	rwl_up(&s->rx.use);
+	rwl_up(&s->tx.use);
+}
+
+static struct tle_udp_dev *
+find_ipv4_dev(struct tle_udp_ctx *ctx, const struct in_addr *addr)
+{
+	uint32_t i;
+
+	for (i = 0; i != RTE_DIM(ctx->dev); i++) {
+		if (ctx->dev[i].prm.local_addr4.s_addr == addr->s_addr &&
+				ctx->dev[i].dp[TLE_UDP_V4] != NULL)
+			return ctx->dev + i;
+	}
+
+	return NULL;
+}
+
+static struct tle_udp_dev *
+find_ipv6_dev(struct tle_udp_ctx *ctx, const struct in6_addr *addr)
+{
+	uint32_t i;
+
+	for (i = 0; i != RTE_DIM(ctx->dev); i++) {
+		if (memcmp(&ctx->dev[i].prm.local_addr6, addr,
+				sizeof(*addr)) == 0 &&
+				ctx->dev[i].dp[TLE_UDP_V6] != NULL)
+			return ctx->dev + i;
+	}
+
+	return NULL;
+}
+
+static int
+stream_fill_dev(struct tle_udp_ctx *ctx, struct tle_udp_stream *s)
+{
+	struct tle_udp_dev *dev;
+	struct udp_pbm *pbm;
+	struct sockaddr_in *lin4;
+	struct sockaddr_in6 *lin6;
+	uint32_t i, p, sp, t;
+
+	if (s->prm.local_addr.ss_family == AF_INET) {
+		lin4 = (struct sockaddr_in *)&s->prm.local_addr;
+		t = TLE_UDP_V4;
+		p = lin4->sin_port;
+	} else if (s->prm.local_addr.ss_family == AF_INET6) {
+		lin6 = (struct sockaddr_in6 *)&s->prm.local_addr;
+		t = TLE_UDP_V6;
+		p = lin6->sin6_port;
+	} else
+		return EINVAL;
+
+	p = ntohs(p);
+
+	/* if local address is not wildcard, find device it belongs to. */
+	if (t == TLE_UDP_V4 && lin4->sin_addr.s_addr != INADDR_ANY) {
+		dev = find_ipv4_dev(ctx, &lin4->sin_addr);
+		if (dev == NULL)
+			return ENODEV;
+	} else if (t == TLE_UDP_V6 && memcmp(&tle_udp6_any, &lin6->sin6_addr,
+			sizeof(tle_udp6_any)) != 0) {
+		dev = find_ipv6_dev(ctx, &lin6->sin6_addr);
+		if (dev == NULL)
+			return ENODEV;
+	} else
+		dev = NULL;
+
+	if (dev != NULL)
+		pbm = &dev->dp[t]->use;
+	else
+		pbm = &ctx->use[t];
+
+	/* try to acquire local port number. */
+	if (p == 0) {
+		p = udp_pbm_find_range(pbm, pbm->blk, LPORT_END_BLK);
+		if (p == 0 && pbm->blk > LPORT_START_BLK)
+			p = udp_pbm_find_range(pbm, LPORT_START_BLK, pbm->blk);
+	} else if (udp_pbm_check(pbm, p) != 0)
+		return EEXIST;
+
+	if (p == 0)
+		return ENFILE;
+
+	/* fill socket's dst port and type */
+
+	sp = htons(p);
+	s->type = t;
+	s->port.dst = sp;
+
+	/* mark port as in-use */
+
+	udp_pbm_set(&ctx->use[t], p);
+	if (dev != NULL) {
+		udp_pbm_set(pbm, p);
+		dev->dp[t]->streams[sp] = s;
+	} else {
+		for (i = 0; i != RTE_DIM(ctx->dev); i++) {
+			if (ctx->dev[i].dp[t] != NULL) {
+				udp_pbm_set(&ctx->dev[i].dp[t]->use, p);
+				ctx->dev[i].dp[t]->streams[sp] = s;
+			}
+		}
+	}
+
+	return 0;
+}
+
+static int
+stream_clear_dev(struct tle_udp_ctx *ctx, struct tle_udp_stream *s)
+{
+	struct tle_udp_dev *dev;
+	uint32_t i, p, sp, t;
+
+	t = s->type;
+	sp = s->port.dst;
+	p = ntohs(sp);
+
+	/* if local address is not wildcard, find device it belongs to. */
+	if (t == TLE_UDP_V4 && s->ipv4.addr.dst != INADDR_ANY) {
+		dev = find_ipv4_dev(ctx, (struct in_addr *)&s->ipv4.addr.dst);
+		if (dev == NULL)
+			return ENODEV;
+	} else if (t == TLE_UDP_V6 && memcmp(&tle_udp6_any, &s->ipv6.addr.dst,
+			sizeof(tle_udp6_any)) != 0) {
+		dev = find_ipv6_dev(ctx, (struct in6_addr *)&s->ipv6.addr.dst);
+		if (dev == NULL)
+			return ENODEV;
+	} else
+		dev = NULL;
+
+	udp_pbm_clear(&ctx->use[t], p);
+	if (dev != NULL) {
+		udp_pbm_clear(&dev->dp[t]->use, p);
+		dev->dp[t]->streams[sp] = NULL;
+	} else {
+		for (i = 0; i != RTE_DIM(ctx->dev); i++) {
+			if (ctx->dev[i].dp[t] != NULL) {
+				udp_pbm_clear(&ctx->dev[i].dp[t]->use, p);
+				ctx->dev[i].dp[t]->streams[sp] = NULL;
+			}
+		}
+	}
+
+	return 0;
+}
+
+static struct tle_udp_stream *
+get_stream(struct tle_udp_ctx *ctx)
+{
+	struct tle_udp_stream *s;
+
+	s = NULL;
+	if (ctx->streams.nb_free == 0)
+		return s;
+
+	rte_spinlock_lock(&ctx->streams.lock);
+	if (ctx->streams.nb_free != 0) {
+		s = STAILQ_FIRST(&ctx->streams.free);
+		STAILQ_REMOVE_HEAD(&ctx->streams.free, link);
+		ctx->streams.nb_free--;
+	}
+	rte_spinlock_unlock(&ctx->streams.lock);
+	return s;
+}
+
+static void
+put_stream(struct tle_udp_ctx *ctx, struct tle_udp_stream *s, int32_t head)
+{
+	s->type = TLE_UDP_VNUM;
+	rte_spinlock_lock(&ctx->streams.lock);
+	if (head != 0)
+		STAILQ_INSERT_HEAD(&ctx->streams.free, s, link);
+	else
+		STAILQ_INSERT_TAIL(&ctx->streams.free, s, link);
+	ctx->streams.nb_free++;
+	rte_spinlock_unlock(&ctx->streams.lock);
+}
+
+static void
+fill_ipv4_am(const struct sockaddr_in *in, uint32_t *addr, uint32_t *mask)
+{
+	*addr = in->sin_addr.s_addr;
+	*mask = (*addr == INADDR_ANY) ? INADDR_ANY : INADDR_NONE;
+}
+
+static void
+fill_ipv6_am(const struct sockaddr_in6 *in, rte_xmm_t *addr, rte_xmm_t *mask)
+{
+	const struct in6_addr *pm;
+
+	memcpy(addr, &in->sin6_addr, sizeof(*addr));
+	if (memcmp(&tle_udp6_any, addr, sizeof(*addr)) == 0)
+		pm = &tle_udp6_any;
+	else
+		pm = &tle_udp6_none;
+
+	memcpy(mask, pm, sizeof(*mask));
+}
+
+static int
+check_stream_prm(const struct tle_udp_stream_param *prm)
+{
+	if ((prm->local_addr.ss_family != AF_INET &&
+			prm->local_addr.ss_family != AF_INET6) ||
+			prm->local_addr.ss_family != prm->remote_addr.ss_family)
+		return EINVAL;
+
+	/* callback and event notifications mechanisms are mutually exclusive */
+	if ((prm->recv_ev != NULL && prm->recv_cb.func != NULL) ||
+			(prm->send_ev != NULL && prm->send_cb.func != NULL))
+		return EINVAL;
+
+	return 0;
+}
+
+struct tle_udp_stream *
+tle_udp_stream_open(struct tle_udp_ctx *ctx,
+	const struct tle_udp_stream_param *prm)
+{
+	struct tle_udp_stream *s;
+	const struct sockaddr_in *rin;
+	int32_t rc;
+
+	if (ctx == NULL || prm == NULL || check_stream_prm(prm) != 0) {
+		rte_errno = EINVAL;
+		return NULL;
+	}
+
+	s = get_stream(ctx);
+	if (s == NULL)	{
+		rte_errno = ENFILE;
+		return NULL;
+
+	/* some TX still pending for that stream. */
+	} else if (bcg_store_full(s->tx.st) == 0) {
+		put_stream(ctx, s, 0);
+		rte_errno = EAGAIN;
+		return NULL;
+	}
+
+	/* copy input parameters. */
+	s->prm = *prm;
+
+	/* setup ports and port mask fields (except dst port). */
+	rin = (const struct sockaddr_in *)&prm->remote_addr;
+	s->port.src = rin->sin_port;
+	s->pmsk.src = (s->port.src == 0) ? 0 : UINT16_MAX;
+	s->pmsk.dst = UINT16_MAX;
+
+	/* setup src and dst addresses. */
+	if (prm->local_addr.ss_family == AF_INET) {
+		fill_ipv4_am((const struct sockaddr_in *)&prm->local_addr,
+			&s->ipv4.addr.dst, &s->ipv4.mask.dst);
+		fill_ipv4_am((const struct sockaddr_in *)&prm->remote_addr,
+			&s->ipv4.addr.src, &s->ipv4.mask.src);
+	} else if (prm->local_addr.ss_family == AF_INET6) {
+		fill_ipv6_am((const struct sockaddr_in6 *)&prm->local_addr,
+			&s->ipv6.addr.dst, &s->ipv6.mask.dst);
+		fill_ipv6_am((const struct sockaddr_in6 *)&prm->remote_addr,
+			&s->ipv6.addr.src, &s->ipv6.mask.src);
+	}
+
+	rte_spinlock_lock(&ctx->dev_lock);
+	rc = stream_fill_dev(ctx, s);
+	rte_spinlock_unlock(&ctx->dev_lock);
+
+	if (rc != 0) {
+		put_stream(ctx, s, 1);
+		s = NULL;
+		rte_errno = rc;
+	} else {
+		/* setup stream notification menchanism */
+		s->rx.ev = prm->recv_ev;
+		s->rx.cb = prm->recv_cb;
+		s->tx.ev = prm->send_ev;
+		s->tx.cb = prm->send_cb;
+
+		/* mark stream as avaialbe for RX/TX */
+		if (s->tx.ev != NULL)
+			tle_event_raise(s->tx.ev);
+		stream_up(s);
+	}
+
+	return s;
+}
+
+int
+tle_udp_stream_close(struct tle_udp_stream *s)
+{
+	uint32_t i, n;
+	int32_t rc;
+	struct tle_udp_ctx *ctx;
+	struct rte_mbuf *m[MAX_BURST];
+
+	static const struct tle_udp_stream_cb zcb;
+
+	if (s == NULL || s->type >= TLE_UDP_VNUM)
+		return EINVAL;
+
+	ctx = s->ctx;
+
+	/* mark stream as unavaialbe for RX/TX. */
+	stream_down(s);
+
+	/* reset TX cages. */
+	rte_spinlock_lock(&s->tx.lock);
+	memset(s->tx.cg, 0, sizeof(s->tx.cg));
+	rte_spinlock_unlock(&s->tx.lock);
+
+	/* reset stream events if any. */
+	if (s->rx.ev != NULL) {
+		tle_event_idle(s->rx.ev);
+		s->rx.ev = NULL;
+	}
+	if (s->tx.ev != NULL) {
+		tle_event_idle(s->tx.ev);
+		s->tx.ev = NULL;
+	}
+
+	s->rx.cb = zcb;
+	s->tx.cb = zcb;
+
+	/* free stream's destination port */
+	rte_spinlock_lock(&ctx->dev_lock);
+	rc = stream_clear_dev(ctx, s);
+	rte_spinlock_unlock(&ctx->dev_lock);
+
+	/* empty stream's RX queue */
+	do {
+		n = rte_ring_dequeue_burst(s->rx.q, (void **)m, RTE_DIM(m));
+		for (i = 0; i != n; i++)
+			rte_pktmbuf_free(m[i]);
+	} while (n != 0);
+
+	/*
+	 * mark the stream as free again.
+	 * if there still are pkts queued for TX,
+	 * then put this stream to the tail of free list.
+	 */
+	put_stream(ctx, s, bcg_store_full(s->tx.st));
+	return rc;
+}
+
+int
+tle_udp_stream_get_param(const struct tle_udp_stream *s,
+	struct tle_udp_stream_param *prm)
+{
+	struct sockaddr_in *lin4;
+	struct sockaddr_in6 *lin6;
+
+	if (prm == NULL || s == NULL || s->type >= TLE_UDP_VNUM)
+		return EINVAL;
+
+	prm[0] = s->prm;
+	if (prm->local_addr.ss_family == AF_INET) {
+		lin4 = (struct sockaddr_in *)&prm->local_addr;
+		lin4->sin_port = s->port.dst;
+	} else if (s->prm.local_addr.ss_family == AF_INET6) {
+		lin6 = (struct sockaddr_in6 *)&prm->local_addr;
+		lin6->sin6_port = s->port.dst;
+	}
+
+	return 0;
+}
diff --git a/lib/libtle_udp/udp_impl.h b/lib/libtle_udp/udp_impl.h
new file mode 100644
index 0000000..fbdb743
--- /dev/null
+++ b/lib/libtle_udp/udp_impl.h
@@ -0,0 +1,161 @@
+/*
+ * Copyright (c) 2016  Intel Corporation.
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at:
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#ifndef _UDP_IMPL_H_
+#define _UDP_IMPL_H_
+
+#include <rte_spinlock.h>
+#include <rte_vect.h>
+#include <tle_udp_impl.h>
+#include <tle_event.h>
+
+#include "buf_cage.h"
+#include "port_bitmap.h"
+#include "osdep.h"
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+enum {
+	TLE_UDP_V4,
+	TLE_UDP_V6,
+	TLE_UDP_VNUM
+};
+
+union udp_ports {
+	uint32_t raw;
+	struct {
+		uint16_t src;
+		uint16_t dst;
+	};
+};
+
+union udph {
+	uint64_t raw;
+	struct {
+		union udp_ports ports;
+		uint16_t len;
+		uint16_t cksum;
+	};
+};
+
+union ipv4_addrs {
+	uint64_t raw;
+	struct {
+		uint32_t src;
+		uint32_t dst;
+	};
+};
+
+union ipv6_addrs {
+	_ymm_t raw;
+	struct {
+		rte_xmm_t src;
+		rte_xmm_t dst;
+	};
+};
+
+union ip_addrs {
+	union ipv4_addrs v4;
+	union ipv6_addrs v6;
+};
+
+
+struct tle_udp_stream {
+
+	STAILQ_ENTRY(tle_udp_stream) link;
+	struct tle_udp_ctx *ctx;
+
+	uint8_t type;	       /* TLE_UDP_V4 or TLE_UDP_V6 */
+
+	struct {
+		struct rte_ring *q;
+		struct tle_event *ev;
+		struct tle_udp_stream_cb cb;
+		rte_atomic32_t use;
+	} rx;
+
+	union udp_ports port;
+	union udp_ports pmsk;
+
+	union {
+		struct {
+			union ipv4_addrs addr;
+			union ipv4_addrs mask;
+		} ipv4;
+		struct {
+			union ipv6_addrs addr;
+			union ipv6_addrs mask;
+		} ipv6;
+	};
+
+	struct {
+		rte_atomic32_t use;
+		rte_spinlock_t lock;
+		struct tle_event *ev;
+		struct tle_udp_stream_cb cb;
+		struct bcg_store *st;
+		struct buf_cage *cg[RTE_MAX_ETHPORTS];
+	} tx __rte_cache_aligned;
+
+	struct tle_udp_stream_param prm;
+} __rte_cache_aligned;
+
+struct tle_udp_dport {
+	struct udp_pbm use; /* ports in use. */
+	struct tle_udp_stream *streams[MAX_PORT_NUM]; /* port to stream. */
+};
+
+struct tle_udp_dev {
+	struct tle_udp_ctx *ctx;
+	struct {
+		uint64_t ol_flags[TLE_UDP_VNUM];
+	} rx;
+	struct {
+		/* used by FE. */
+		uint64_t ol_flags[TLE_UDP_VNUM];
+		rte_atomic32_t packet_id[TLE_UDP_VNUM];
+		struct bcg_queue feq;
+
+		/* used by BE only. */
+		struct bcg_queue beq __rte_cache_min_aligned;
+		struct buf_cage *bc;
+	} tx;
+	struct tle_udp_dev_param prm; /* copy of device paramaters. */
+	struct tle_udp_dport *dp[TLE_UDP_VNUM]; /* device udp ports */
+};
+
+struct tle_udp_ctx {
+	struct tle_udp_ctx_param prm;
+
+	struct {
+		rte_spinlock_t lock;
+		uint32_t nb_free; /* number of free streams. */
+		STAILQ_HEAD(, tle_udp_stream) free;
+		struct tle_udp_stream *buf; /* array of streams */
+	} streams;
+
+	rte_spinlock_t dev_lock;
+	uint32_t nb_dev;
+	struct udp_pbm use[TLE_UDP_VNUM];          /* all ports in use. */
+	struct tle_udp_dev dev[RTE_MAX_ETHPORTS];
+};
+
+#ifdef __cplusplus
+}
+#endif
+
+#endif /* _UDP_IMPL_H_ */
diff --git a/lib/libtle_udp/udp_rxtx.c b/lib/libtle_udp/udp_rxtx.c
new file mode 100644
index 0000000..d5d248e
--- /dev/null
+++ b/lib/libtle_udp/udp_rxtx.c
@@ -0,0 +1,767 @@
+
+#include <rte_malloc.h>
+#include <rte_errno.h>
+#include <rte_ethdev.h>
+#include <rte_ip.h>
+#include <rte_ip_frag.h>
+#include <rte_udp.h>
+
+#include "udp_impl.h"
+#include "misc.h"
+
+static inline struct tle_udp_stream *
+rx_stream_obtain(struct tle_udp_dev *dev, uint32_t type, uint32_t port)
+{
+	struct tle_udp_stream *s;
+
+	if (type >= TLE_UDP_VNUM || dev->dp[type] == NULL)
+		return NULL;
+
+	s = dev->dp[type]->streams[port];
+	if (s == NULL)
+		return NULL;
+
+	if (rwl_acquire(&s->rx.use) < 0)
+		return NULL;
+
+	return s;
+}
+
+static inline uint16_t
+get_pkt_type(const struct rte_mbuf *m)
+{
+	uint32_t v;
+
+	v = m->packet_type &
+		(RTE_PTYPE_L3_IPV4 | RTE_PTYPE_L3_IPV6 | RTE_PTYPE_L4_MASK);
+	if (v == (RTE_PTYPE_L3_IPV4 | RTE_PTYPE_L4_UDP))
+		return TLE_UDP_V4;
+	else if (v == (RTE_PTYPE_L3_IPV6 | RTE_PTYPE_L4_UDP))
+		return TLE_UDP_V6;
+	else
+		return TLE_UDP_VNUM;
+}
+
+static inline union udp_ports
+pkt_info(const struct tle_udp_dev *dev, struct rte_mbuf *m,
+	union udp_ports *ports, union ipv4_addrs *addr4,
+	union ipv6_addrs **addr6)
+{
+	uint32_t len;
+	union udp_ports ret, *up;
+	union ipv4_addrs *pa4;
+
+	ret.src = get_pkt_type(m);
+
+	len = m->l2_len;
+	if (ret.src == TLE_UDP_V4) {
+		pa4 = rte_pktmbuf_mtod_offset(m, union ipv4_addrs *,
+			len + offsetof(struct ipv4_hdr, src_addr));
+		addr4->raw = pa4->raw;
+		m->ol_flags |= dev->rx.ol_flags[TLE_UDP_V4];
+	} else if (ret.src == TLE_UDP_V6) {
+		*addr6 = rte_pktmbuf_mtod_offset(m, union ipv6_addrs *,
+			len + offsetof(struct ipv6_hdr, src_addr));
+		m->ol_flags |= dev->rx.ol_flags[TLE_UDP_V6];
+	}
+
+	len += m->l3_len;
+	up = rte_pktmbuf_mtod_offset(m, union udp_ports *,
+		len + offsetof(struct udp_hdr, src_port));
+	ports->raw = up->raw;
+	ret.dst = ports->dst;
+	return ret;
+}
+
+/*
+ * Helper routine, enqueues packets to the stream and calls RX
+ * notification callback, if needed.
+ */
+static inline uint16_t
+rx_stream(struct tle_udp_stream *s, void *mb[], struct rte_mbuf *rp[],
+	int32_t rc[], uint32_t num)
+{
+	uint32_t i, k, r;
+
+	r = rte_ring_enqueue_burst(s->rx.q, mb, num);
+
+	/* if RX queue was empty invoke user RX notification callback. */
+	if (s->rx.cb.func != NULL && r != 0 && rte_ring_count(s->rx.q) == r)
+		s->rx.cb.func(s->rx.cb.data, s);
+
+	for (i = r, k = 0; i != num; i++, k++) {
+		rc[k] = ENOBUFS;
+		rp[k] = mb[i];
+	}
+
+	return r;
+}
+
+static inline uint16_t
+rx_stream6(struct tle_udp_stream *s, struct rte_mbuf *pkt[],
+	union ipv6_addrs *addr[], union udp_ports port[],
+	struct rte_mbuf *rp[], int32_t rc[], uint16_t num)
+{
+	uint32_t i, k, n;
+	void *mb[num];
+
+	k = 0;
+	n = 0;
+
+	for (i = 0; i != num; i++) {
+
+		if ((port[i].raw & s->pmsk.raw) != s->port.raw ||
+				ymm_mask_cmp(&addr[i]->raw, &s->ipv6.addr.raw,
+				&s->ipv6.mask.raw) != 0) {
+			rc[k] = ENOENT;
+			rp[k] = pkt[i];
+			k++;
+		} else {
+			mb[n] = pkt[i];
+			n++;
+		}
+	}
+
+	return rx_stream(s, mb, rp + k, rc + k, n);
+}
+
+static inline uint16_t
+rx_stream4(struct tle_udp_stream *s, struct rte_mbuf *pkt[],
+	union ipv4_addrs addr[], union udp_ports port[],
+	struct rte_mbuf *rp[], int32_t rc[], uint16_t num)
+{
+	uint32_t i, k, n;
+	void *mb[num];
+
+	k = 0;
+	n = 0;
+
+	for (i = 0; i != num; i++) {
+
+		if ((addr[i].raw & s->ipv4.mask.raw) != s->ipv4.addr.raw ||
+				(port[i].raw & s->pmsk.raw) !=
+				s->port.raw) {
+			rc[k] = ENOENT;
+			rp[k] = pkt[i];
+			k++;
+		} else {
+			mb[n] = pkt[i];
+			n++;
+		}
+	}
+
+	return rx_stream(s, mb, rp + k, rc + k, n);
+}
+
+uint16_t
+tle_udp_rx_bulk(struct tle_udp_dev *dev, struct rte_mbuf *pkt[],
+	struct rte_mbuf *rp[], int32_t rc[], uint16_t num)
+{
+	struct tle_udp_stream *s;
+	uint32_t i, j, k, n, p, t;
+	union udp_ports tp[num], port[num];
+	union ipv4_addrs a4[num];
+	union ipv6_addrs *pa6[num];
+
+	for (i = 0; i != num; i++)
+		tp[i] = pkt_info(dev, pkt[i], &port[i], &a4[i], &pa6[i]);
+
+	k = 0;
+	for (i = 0; i != num; i = j) {
+
+		for (j = i + 1; j != num && tp[j].raw == tp[i].raw; j++)
+			;
+
+		t = tp[i].src;
+		p = tp[i].dst;
+		s = rx_stream_obtain(dev, t, p);
+		if (s != NULL) {
+
+			if (t == TLE_UDP_V4)
+				n = rx_stream4(s, pkt + i, a4 + i,
+					port + i, rp + k, rc + k, j - i);
+			else
+				n = rx_stream6(s, pkt + i, pa6 + i, port + i,
+					rp + k, rc + k, j - i);
+
+			k += j - i - n;
+
+			if (s->rx.ev != NULL)
+				tle_event_raise(s->rx.ev);
+			rwl_release(&s->rx.use);
+
+		} else {
+			for (; i != j; i++) {
+				rc[k] = ENOENT;
+				rp[k] = pkt[i];
+				k++;
+			}
+		}
+	}
+
+	return num - k;
+}
+
+static inline void
+tx_cage_release(struct buf_cage *bc)
+{
+	struct tle_udp_stream *s;
+	uint32_t n;
+
+	s = bcg_get_udata(bc);
+	n = bcg_free(bc);
+
+	/* If stream is still open, then mark it as avaialble for writing. */
+	if (rwl_try_acquire(&s->tx.use) > 0) {
+
+		if (s->tx.ev != NULL)
+			tle_event_raise(s->tx.ev);
+
+		/* if stream send buffer was full invoke TX callback */
+		else if (s->tx.cb.func != NULL && n == 1)
+			s->tx.cb.func(s->tx.cb.data, s);
+
+	}
+
+	rwl_release(&s->tx.use);
+}
+
+static inline void
+tx_cage_update(struct tle_udp_dev *dev, struct buf_cage *bc)
+{
+	struct tle_udp_stream *s;
+	struct tle_udp_ctx *ctx;
+	uint32_t idx;
+
+	ctx = dev->ctx;
+	s = bcg_get_udata(bc);
+	idx = dev - ctx->dev;
+
+	/* mark cage as closed to the stream. */
+	rte_spinlock_lock(&s->tx.lock);
+	if (bc == s->tx.cg[idx])
+		s->tx.cg[idx] = NULL;
+	rte_spinlock_unlock(&s->tx.lock);
+}
+
+uint16_t
+tle_udp_tx_bulk(struct tle_udp_dev *dev, struct rte_mbuf *pkt[], uint16_t num)
+{
+	struct buf_cage *bc;
+	uint32_t i, n;
+
+	for (i = 0; i != num; i += n) {
+
+		bc = dev->tx.bc;
+		if (bc == NULL) {
+			if (dev->tx.beq.num == 0)
+				bcg_queue_append(&dev->tx.beq, &dev->tx.feq);
+			bc = __bcg_dequeue_head(&dev->tx.beq);
+			if (bc == NULL)
+				break;
+			tx_cage_update(dev, bc);
+			dev->tx.bc = bc;
+		}
+
+		n = bcg_get(bc, (const void **)(uintptr_t)&pkt[i], num - i);
+
+		/* cage is empty, need to free it and notify related stream. */
+		if (bcg_fill_count(bc) == 0) {
+			tx_cage_release(bc);
+			dev->tx.bc = NULL;
+		}
+	}
+
+	return i;
+}
+
+static int
+check_pkt_csum(const struct rte_mbuf *m, uint32_t type)
+{
+	const struct ipv4_hdr *l3h4;
+	const struct ipv6_hdr *l3h6;
+	const struct udp_hdr *l4h;
+	int32_t ret;
+	uint16_t csum;
+
+	ret = 0;
+	l3h4 = rte_pktmbuf_mtod_offset(m, const struct ipv4_hdr *, m->l2_len);
+	l3h6 = rte_pktmbuf_mtod_offset(m, const struct ipv6_hdr *, m->l2_len);
+
+	if ((m->ol_flags & PKT_RX_IP_CKSUM_BAD) != 0) {
+		csum = _ipv4x_cksum(l3h4, m->l3_len);
+		ret = (csum != UINT16_MAX);
+	}
+
+	if (ret == 0 && (m->ol_flags & PKT_RX_L4_CKSUM_BAD) != 0) {
+
+		/*
+		 * for IPv4 it is allowed to have zero UDP cksum,
+		 * for IPv6 valid UDP cksum is mandatory.
+		 */
+		if (type == TLE_UDP_V4) {
+			l4h = (const struct udp_hdr *)((uintptr_t)l3h4 +
+				m->l3_len);
+			csum = (l4h->dgram_cksum == 0) ? UINT16_MAX :
+				_ipv4_udptcp_mbuf_cksum(m,
+				m->l2_len + m->l3_len, l3h4);
+		} else
+			csum = _ipv6_udptcp_mbuf_cksum(m,
+				m->l2_len + m->l3_len, l3h6);
+
+		ret = (csum != UINT16_MAX);
+	}
+
+	return ret;
+}
+
+/* exclude NULLs from the final list of packets. */
+static inline uint32_t
+compress_pkt_list(struct rte_mbuf *pkt[], uint32_t nb_pkt, uint32_t nb_zero)
+{
+	uint32_t i, j, k, l;
+
+	for (j = nb_pkt; nb_zero != 0 && j-- != 0; ) {
+
+		/* found a hole. */
+		if (pkt[j] == NULL) {
+
+			/* find how big is it. */
+			for (i = j; i-- != 0 && pkt[i] == NULL; )
+				;
+			/* fill the hole. */
+			for (k = j + 1, l = i + 1; k != nb_pkt; k++, l++)
+				pkt[l] = pkt[k];
+
+			nb_pkt -= j - i;
+			nb_zero -= j - i;
+		}
+	}
+
+	return nb_pkt;
+}
+
+/*
+ * helper function, do the necessary pre-processing for the received packets
+ * before handiing them to the strem_recv caller.
+ */
+static inline struct rte_mbuf *
+recv_pkt_process(struct rte_mbuf *m, uint32_t type)
+{
+	uint64_t f;
+
+	f = m->ol_flags & (PKT_RX_IP_CKSUM_BAD | PKT_RX_L4_CKSUM_BAD);
+	if (f != 0) {
+		if (check_pkt_csum(m, type) == 0)
+			m->ol_flags ^= f;
+		else {
+			rte_pktmbuf_free(m);
+			return NULL;
+		}
+	}
+
+	rte_pktmbuf_adj(m, m->l2_len + m->l3_len + m->l4_len);
+	return m;
+}
+
+uint16_t
+tle_udp_stream_recv(struct tle_udp_stream *s, struct rte_mbuf *pkt[],
+	uint16_t num)
+{
+	uint32_t i, k, n;
+
+	n = rte_ring_mc_dequeue_burst(s->rx.q, (void **)pkt, num);
+	if (n == 0)
+		return 0;
+
+	/*
+	 * if we still have packets to read,
+	 * then rearm stream RX event.
+	 */
+	if (n == num && rte_ring_count(s->rx.q) != 0) {
+		if (rwl_try_acquire(&s->rx.use) > 0 && s->rx.ev != NULL)
+			tle_event_raise(s->rx.ev);
+		rwl_release(&s->rx.use);
+	}
+
+	k = 0;
+	for (i = 0; i != RTE_ALIGN_FLOOR(n, 4); i += 4) {
+		pkt[i] = recv_pkt_process(pkt[i], s->type);
+		pkt[i + 1] = recv_pkt_process(pkt[i + 1], s->type);
+		pkt[i + 2] = recv_pkt_process(pkt[i + 2], s->type);
+		pkt[i + 3] = recv_pkt_process(pkt[i + 3], s->type);
+		k += (pkt[i] == NULL) + (pkt[i + 1] == NULL) +
+			(pkt[i + 2] == NULL) + (pkt[i + 3] == NULL);
+	}
+
+	switch (n % 4) {
+	case 3:
+		pkt[i + 2] = recv_pkt_process(pkt[i + 2], s->type);
+		k += (pkt[i + 2] == NULL);
+	case 2:
+		pkt[i + 1] = recv_pkt_process(pkt[i + 1], s->type);
+		k += (pkt[i + 1] == NULL);
+	case 1:
+		pkt[i] = recv_pkt_process(pkt[i], s->type);
+		k += (pkt[i] == NULL);
+	}
+
+	return compress_pkt_list(pkt, n, k);
+}
+
+static int32_t
+udp_get_dest(struct tle_udp_stream *s, const void *dst_addr,
+	struct tle_udp_dest *dst)
+{
+	int32_t rc;
+	const struct in_addr *d4;
+	const struct in6_addr *d6;
+	struct tle_udp_ctx *ctx;
+	struct tle_udp_dev *dev;
+
+	ctx = s->ctx;
+
+	/* it is here just to keep gcc happy. */
+	d4 = NULL;
+
+	if (s->type == TLE_UDP_V4) {
+		d4 = dst_addr;
+		rc = ctx->prm.lookup4(ctx->prm.lookup4_data, d4, dst);
+	} else if (s->type == TLE_UDP_V6) {
+		d6 = dst_addr;
+		rc = ctx->prm.lookup6(ctx->prm.lookup6_data, d6, dst);
+	} else
+		rc = -ENOENT;
+
+	if (rc < 0 || dst->dev == NULL || dst->dev->ctx != ctx)
+		return -ENOENT;
+
+	dev = dst->dev;
+	if (s->type == TLE_UDP_V4) {
+		struct ipv4_hdr *l3h;
+		l3h = (struct ipv4_hdr *)(dst->hdr + dst->l2_len);
+		l3h->src_addr = dev->prm.local_addr4.s_addr;
+		l3h->dst_addr = d4->s_addr;
+	} else {
+		struct ipv6_hdr *l3h;
+		l3h = (struct ipv6_hdr *)(dst->hdr + dst->l2_len);
+		rte_memcpy(l3h->src_addr, &dev->prm.local_addr6,
+			sizeof(l3h->src_addr));
+		rte_memcpy(l3h->dst_addr, d6, sizeof(l3h->dst_addr));
+	}
+
+	return dev - ctx->dev;
+}
+
+static inline int
+udp_fill_mbuf(struct rte_mbuf *m,
+	uint32_t type, uint64_t ol_flags, uint32_t pid,
+	union udph udph, const struct tle_udp_dest *dst)
+{
+	uint32_t len, plen;
+	char *l2h;
+	union udph *l4h;
+
+	len = dst->l2_len + dst->l3_len;
+	plen = m->pkt_len;
+
+	/* copy to mbuf L2/L3 header template. */
+
+	l2h = rte_pktmbuf_prepend(m, len + sizeof(*l4h));
+	if (l2h == NULL)
+		return -ENOBUFS;
+
+	/* copy L2/L3 header */
+	rte_memcpy(l2h, dst->hdr, len);
+
+	/* copy UDP header */
+	l4h = (union udph *)(l2h + len);
+	l4h->raw = udph.raw;
+
+	/* setup mbuf TX offload related fields. */
+	m->tx_offload = _mbuf_tx_offload(dst->l2_len, dst->l3_len,
+		sizeof(*l4h), 0, 0, 0);
+	m->ol_flags |= ol_flags;
+
+	l4h->len = rte_cpu_to_be_16(plen + sizeof(*l4h));
+
+	/* update proto specific fields. */
+
+	if (type == TLE_UDP_V4) {
+		struct ipv4_hdr *l3h;
+		l3h = (struct ipv4_hdr *)(l2h + dst->l2_len);
+		l3h->packet_id = rte_cpu_to_be_16(pid);
+		l3h->total_length = rte_cpu_to_be_16(plen + dst->l3_len +
+			sizeof(*l4h));
+
+		if ((ol_flags & PKT_TX_UDP_CKSUM) != 0)
+			l4h->cksum = _ipv4x_phdr_cksum(l3h, m->l3_len,
+				ol_flags);
+		else
+			l4h->cksum = _ipv4_udptcp_mbuf_cksum(m, len, l3h);
+
+		if ((ol_flags & PKT_TX_IP_CKSUM) == 0)
+			l3h->hdr_checksum = _ipv4x_cksum(l3h, m->l3_len);
+	} else {
+		struct ipv6_hdr *l3h;
+		l3h = (struct ipv6_hdr *)(l2h + dst->l2_len);
+		l3h->payload_len = rte_cpu_to_be_16(plen + sizeof(*l4h));
+		if ((ol_flags & PKT_TX_UDP_CKSUM) != 0)
+			l4h->cksum = rte_ipv6_phdr_cksum(l3h, ol_flags);
+		else
+			l4h->cksum = _ipv6_udptcp_mbuf_cksum(m, len, l3h);
+	}
+
+	return 0;
+}
+
+/* ???
+ * probably this function should be there -
+ * rte_ipv[4,6]_fragment_packet should do that.
+ */
+static inline void
+frag_fixup(const struct rte_mbuf *ms, struct rte_mbuf *mf, uint32_t type)
+{
+	struct ipv4_hdr *l3h;
+
+	mf->ol_flags = ms->ol_flags;
+	mf->tx_offload = ms->tx_offload;
+
+	if (type == TLE_UDP_V4 && (ms->ol_flags & PKT_TX_IP_CKSUM) == 0) {
+		l3h = rte_pktmbuf_mtod(mf, struct ipv4_hdr *);
+		l3h->hdr_checksum = _ipv4x_cksum(l3h, mf->l3_len);
+	}
+}
+
+/*
+ * Returns negative for failure to fragment or actual number of fragments.
+ */
+static inline int
+fragment(struct rte_mbuf *pkt, struct rte_mbuf *frag[], uint32_t num,
+	uint32_t type, const struct tle_udp_dest *dst)
+{
+	int32_t frag_num, i;
+	uint16_t mtu;
+	void *eth_hdr;
+
+	/* Remove the Ethernet header from the input packet */
+	rte_pktmbuf_adj(pkt, dst->l2_len);
+	mtu = dst->mtu - dst->l2_len;
+
+	/* fragment packet */
+	if (type == TLE_UDP_V4)
+		frag_num = rte_ipv4_fragment_packet(pkt, frag, num, mtu,
+			dst->head_mp, dst->head_mp);
+	else
+		frag_num = rte_ipv6_fragment_packet(pkt, frag, num, mtu,
+			dst->head_mp, dst->head_mp);
+
+	if (frag_num > 0) {
+		for (i = 0; i != frag_num; i++) {
+
+			frag_fixup(pkt, frag[i], type);
+
+			/* Move data_off to include l2 header first */
+			eth_hdr = rte_pktmbuf_prepend(frag[i], dst->l2_len);
+
+			/* copy l2 header into fragment */
+			rte_memcpy(eth_hdr, dst->hdr, dst->l2_len);
+		}
+	}
+
+	return frag_num;
+}
+
+/* enqueue up to num packets to the destination device queue. */
+static inline uint16_t
+queue_pkt_out(struct tle_udp_stream *s, struct bcg_queue *bq, uint32_t di,
+	const void *pkt[], uint16_t num)
+{
+	struct buf_cage *bc;
+	uint32_t i, n;
+
+	rte_spinlock_lock(&s->tx.lock);
+	bc = s->tx.cg[di];
+
+	for (i = 0; i != num; i += n) {
+		if (bc == NULL) {
+			bc = bcg_alloc(s->tx.st);
+			if (bc == NULL)
+				break;
+			n = bcg_put(bc, pkt + i, num - i);
+			bcg_enqueue_tail(bq, bc);
+		} else
+			n = bcg_put(bc, pkt + i, num - i);
+
+		if (n != num - i)
+			bc = NULL;
+	}
+
+	s->tx.cg[di] = bc;
+	rte_spinlock_unlock(&s->tx.lock);
+	return i;
+}
+
+/*
+ * etiher enqueue all num packets or none.
+ * assumes that all number of input packets not exceed size of buf_cage.
+ */
+static inline uint16_t
+queue_frg_out(struct tle_udp_stream *s, struct bcg_queue *bq, uint32_t di,
+	const void *pkt[], uint16_t num)
+{
+	struct buf_cage *bc, *bcp;
+	uint32_t n;
+
+	rte_spinlock_lock(&s->tx.lock);
+	bc = s->tx.cg[di];
+
+	n = 0;
+	if (bc == NULL || bcg_free_count(bc) < num) {
+		bcp = bc;
+		bc = bcg_alloc(s->tx.st);
+		if (bc != NULL) {
+			if (bcp != NULL)
+				n = bcg_put(bcp, pkt, num);
+			n += bcg_put(bc, pkt, num - n);
+			bcg_enqueue_tail(bq, bc);
+		}
+	} else
+		n = bcg_put(bc, pkt, num);
+
+	s->tx.cg[di] = bc;
+	rte_spinlock_unlock(&s->tx.lock);
+	return n;
+}
+
+uint16_t
+tle_udp_stream_send(struct tle_udp_stream *s, struct rte_mbuf *pkt[],
+	uint16_t num, const struct sockaddr *dst_addr)
+{
+	int32_t di, frg, rc;
+	uint64_t ol_flags;
+	uint32_t i, k, n;
+	uint32_t mtu, pid, type;
+	const struct sockaddr_in *d4;
+	const struct sockaddr_in6 *d6;
+	const void *da;
+	union udph udph;
+	struct tle_udp_dest dst;
+
+	type = s->type;
+
+	/* start filling UDP header. */
+	udph.raw = 0;
+	udph.ports.src = s->port.dst;
+
+	/* figure out what destination addr/port to use. */
+	if (dst_addr != NULL) {
+		if (dst_addr->sa_family != s->prm.remote_addr.ss_family) {
+			rte_errno = EINVAL;
+			return 0;
+		}
+		if (type == TLE_UDP_V4) {
+			d4 = (const struct sockaddr_in *)dst_addr;
+			da = &d4->sin_addr;
+			udph.ports.dst = d4->sin_port;
+		} else {
+			d6 = (const struct sockaddr_in6 *)dst_addr;
+			da = &d6->sin6_addr;
+			udph.ports.dst = d6->sin6_port;
+		}
+	} else {
+		udph.ports.dst = s->port.src;
+		if (type == TLE_UDP_V4)
+			da = &s->ipv4.addr.src;
+		else
+			da = &s->ipv6.addr.src;
+	}
+
+	di = udp_get_dest(s, da, &dst);
+	if (di < 0) {
+		rte_errno = -di;
+		return 0;
+	}
+
+	pid = rte_atomic32_add_return(&dst.dev->tx.packet_id[type], num) - num;
+	mtu = dst.mtu - dst.l2_len - dst.l3_len;
+
+	/* mark stream as not closable. */
+	if (rwl_acquire(&s->tx.use) < 0)
+		return 0;
+
+	for (i = 0, k = 0; k != num; k = i) {
+
+		/* copy L2/L3/L4 headers into mbufs, setup mbufs metadata. */
+
+		frg = 0;
+		ol_flags = dst.dev->tx.ol_flags[type];
+
+		while (i != num && frg == 0) {
+			frg = pkt[i]->pkt_len > mtu;
+			if (frg != 0)
+				ol_flags &= ~PKT_TX_UDP_CKSUM;
+			rc = udp_fill_mbuf(pkt[i], type, ol_flags, pid + i,
+				udph, &dst);
+			if (rc != 0) {
+				rte_errno = -rc;
+				goto out;
+			}
+			i += (frg == 0);
+		}
+
+		/* enqueue non-fragment packets to the destination device. */
+		if (k != i) {
+			k += queue_pkt_out(s, &dst.dev->tx.feq, di,
+				(const void **)(uintptr_t)&pkt[k], i - k);
+
+			/* stream TX queue is full. */
+			if (k != i)
+				break;
+		}
+
+		/* enqueue packet that need to be fragmented */
+		if (i != num) {
+
+			struct rte_mbuf *frag[RTE_LIBRTE_IP_FRAG_MAX_FRAG];
+
+			/* fragment the packet. */
+			rc = fragment(pkt[i], frag, RTE_DIM(frag), type, &dst);
+			if (rc < 0) {
+				rte_errno = -rc;
+				break;
+			}
+
+			n = queue_frg_out(s, &dst.dev->tx.feq, di,
+				(const void **)(uintptr_t)frag, rc);
+			if (n == 0) {
+				while (rc-- != 0)
+					rte_pktmbuf_free(frag[rc]);
+				break;
+			}
+
+			/* all fragments enqueued, free the original packet. */
+			rte_pktmbuf_free(pkt[i]);
+			i++;
+		}
+	}
+
+	/* if possible, rearm socket write event. */
+	if (k == num && s->tx.ev != NULL)
+		tle_event_raise(s->tx.ev);
+
+out:
+	rwl_release(&s->tx.use);
+
+	/*
+	 * remove pkt l2/l3 headers, restore ol_flags for unsent, but
+	 * already modified packets.
+	 */
+	ol_flags = ~dst.dev->tx.ol_flags[type];
+	for (n = k; n != i; n++) {
+		rte_pktmbuf_adj(pkt[n], dst.l2_len + dst.l3_len + sizeof(udph));
+		pkt[n]->ol_flags &= ol_flags;
+	}
+
+	return k;
+}