mbuf: add rte prefix to offload flags
[dpdk.git] / lib / vhost / virtio_net.c
index abfd67d..e16a8be 100644 (file)
@@ -218,13 +218,11 @@ vhost_flush_enqueue_batch_packed(struct virtio_net *dev,
 {
        uint16_t i;
        uint16_t flags;
-       uint16_t last_used_idx = vq->last_used_idx;
-       struct vring_packed_desc *desc_base = &vq->desc_packed[last_used_idx];
+       uint16_t last_used_idx;
+       struct vring_packed_desc *desc_base;
 
-       if (vq->shadow_used_idx) {
-               do_data_copy_enqueue(dev, vq);
-               vhost_flush_enqueue_shadow_packed(dev, vq);
-       }
+       last_used_idx = vq->last_used_idx;
+       desc_base = &vq->desc_packed[last_used_idx];
 
        flags = PACKED_DESC_ENQUEUE_USED_FLAG(vq->used_wrap_counter);
 
@@ -413,25 +411,25 @@ vhost_shadow_enqueue_single_packed(struct virtio_net *dev,
 static __rte_always_inline void
 virtio_enqueue_offload(struct rte_mbuf *m_buf, struct virtio_net_hdr *net_hdr)
 {
-       uint64_t csum_l4 = m_buf->ol_flags & PKT_TX_L4_MASK;
+       uint64_t csum_l4 = m_buf->ol_flags & RTE_MBUF_F_TX_L4_MASK;
 
-       if (m_buf->ol_flags & PKT_TX_TCP_SEG)
-               csum_l4 |= PKT_TX_TCP_CKSUM;
+       if (m_buf->ol_flags & RTE_MBUF_F_TX_TCP_SEG)
+               csum_l4 |= RTE_MBUF_F_TX_TCP_CKSUM;
 
        if (csum_l4) {
                net_hdr->flags = VIRTIO_NET_HDR_F_NEEDS_CSUM;
                net_hdr->csum_start = m_buf->l2_len + m_buf->l3_len;
 
                switch (csum_l4) {
-               case PKT_TX_TCP_CKSUM:
+               case RTE_MBUF_F_TX_TCP_CKSUM:
                        net_hdr->csum_offset = (offsetof(struct rte_tcp_hdr,
                                                cksum));
                        break;
-               case PKT_TX_UDP_CKSUM:
+               case RTE_MBUF_F_TX_UDP_CKSUM:
                        net_hdr->csum_offset = (offsetof(struct rte_udp_hdr,
                                                dgram_cksum));
                        break;
-               case PKT_TX_SCTP_CKSUM:
+               case RTE_MBUF_F_TX_SCTP_CKSUM:
                        net_hdr->csum_offset = (offsetof(struct rte_sctp_hdr,
                                                cksum));
                        break;
@@ -443,7 +441,7 @@ virtio_enqueue_offload(struct rte_mbuf *m_buf, struct virtio_net_hdr *net_hdr)
        }
 
        /* IP cksum verification cannot be bypassed, then calculate here */
-       if (m_buf->ol_flags & PKT_TX_IP_CKSUM) {
+       if (m_buf->ol_flags & RTE_MBUF_F_TX_IP_CKSUM) {
                struct rte_ipv4_hdr *ipv4_hdr;
 
                ipv4_hdr = rte_pktmbuf_mtod_offset(m_buf, struct rte_ipv4_hdr *,
@@ -452,15 +450,15 @@ virtio_enqueue_offload(struct rte_mbuf *m_buf, struct virtio_net_hdr *net_hdr)
                ipv4_hdr->hdr_checksum = rte_ipv4_cksum(ipv4_hdr);
        }
 
-       if (m_buf->ol_flags & PKT_TX_TCP_SEG) {
-               if (m_buf->ol_flags & PKT_TX_IPV4)
+       if (m_buf->ol_flags & RTE_MBUF_F_TX_TCP_SEG) {
+               if (m_buf->ol_flags & RTE_MBUF_F_TX_IPV4)
                        net_hdr->gso_type = VIRTIO_NET_HDR_GSO_TCPV4;
                else
                        net_hdr->gso_type = VIRTIO_NET_HDR_GSO_TCPV6;
                net_hdr->gso_size = m_buf->tso_segsz;
                net_hdr->hdr_len = m_buf->l2_len + m_buf->l3_len
                                        + m_buf->l4_len;
-       } else if (m_buf->ol_flags & PKT_TX_UDP_SEG) {
+       } else if (m_buf->ol_flags & RTE_MBUF_F_TX_UDP_SEG) {
                net_hdr->gso_type = VIRTIO_NET_HDR_GSO_UDP;
                net_hdr->gso_size = m_buf->tso_segsz;
                net_hdr->hdr_len = m_buf->l2_len + m_buf->l3_len +
@@ -1255,18 +1253,16 @@ virtio_dev_rx_split(struct virtio_net *dev, struct vhost_virtqueue *vq,
 }
 
 static __rte_always_inline int
-virtio_dev_rx_batch_packed(struct virtio_net *dev,
+virtio_dev_rx_sync_batch_check(struct virtio_net *dev,
                           struct vhost_virtqueue *vq,
-                          struct rte_mbuf **pkts)
+                          struct rte_mbuf **pkts,
+                          uint64_t *desc_addrs,
+                          uint64_t *lens)
 {
        bool wrap_counter = vq->avail_wrap_counter;
        struct vring_packed_desc *descs = vq->desc_packed;
        uint16_t avail_idx = vq->last_avail_idx;
-       uint64_t desc_addrs[PACKED_BATCH_SIZE];
-       struct virtio_net_hdr_mrg_rxbuf *hdrs[PACKED_BATCH_SIZE];
        uint32_t buf_offset = sizeof(struct virtio_net_hdr_mrg_rxbuf);
-       uint64_t lens[PACKED_BATCH_SIZE];
-       uint16_t ids[PACKED_BATCH_SIZE];
        uint16_t i;
 
        if (unlikely(avail_idx & PACKED_BATCH_MASK))
@@ -1304,6 +1300,84 @@ virtio_dev_rx_batch_packed(struct virtio_net *dev,
                        return -1;
        }
 
+       return 0;
+}
+
+static __rte_always_inline int
+virtio_dev_rx_async_batch_check(struct virtio_net *dev,
+                          struct vhost_virtqueue *vq,
+                          struct rte_mbuf **pkts,
+                          uint64_t *desc_addrs,
+                          uint64_t *lens)
+{
+       bool wrap_counter = vq->avail_wrap_counter;
+       struct vring_packed_desc *descs = vq->desc_packed;
+       uint16_t avail_idx = vq->last_avail_idx;
+       uint16_t used_idx = vq->last_used_idx;
+       uint32_t buf_offset = sizeof(struct virtio_net_hdr_mrg_rxbuf);
+       uint32_t cpy_threshold = vq->async_threshold;
+       uint16_t i;
+
+       vhost_for_each_try_unroll(i, 0, PACKED_BATCH_SIZE) {
+               if (unlikely(pkts[i]->data_len >= cpy_threshold))
+                       return -1;
+       }
+
+       if (unlikely(avail_idx & PACKED_BATCH_MASK))
+               return -1;
+
+       if (unlikely((avail_idx + PACKED_BATCH_SIZE) > vq->size))
+               return -1;
+
+       if (unlikely((used_idx + PACKED_BATCH_SIZE) > vq->size))
+               return -1;
+
+       vhost_for_each_try_unroll(i, 0, PACKED_BATCH_SIZE) {
+               if (unlikely(pkts[i]->next != NULL))
+                       return -1;
+               if (unlikely(!desc_is_avail(&descs[avail_idx + i],
+                                           wrap_counter)))
+                       return -1;
+       }
+
+       vhost_for_each_try_unroll(i, 0, PACKED_BATCH_SIZE)
+               lens[i] = descs[avail_idx + i].len;
+
+       vhost_for_each_try_unroll(i, 0, PACKED_BATCH_SIZE) {
+               if (unlikely(pkts[i]->pkt_len > (lens[i] - buf_offset)))
+                       return -1;
+       }
+
+       vhost_for_each_try_unroll(i, 0, PACKED_BATCH_SIZE)
+               desc_addrs[i] = vhost_iova_to_vva(dev, vq,
+                                                 descs[avail_idx + i].addr,
+                                                 &lens[i],
+                                                 VHOST_ACCESS_RW);
+
+       vhost_for_each_try_unroll(i, 0, PACKED_BATCH_SIZE) {
+               if (unlikely(!desc_addrs[i]))
+                       return -1;
+               if (unlikely(lens[i] != descs[avail_idx + i].len))
+                       return -1;
+       }
+
+       return 0;
+}
+
+static __rte_always_inline void
+virtio_dev_rx_batch_packed_copy(struct virtio_net *dev,
+                          struct vhost_virtqueue *vq,
+                          struct rte_mbuf **pkts,
+                          uint64_t *desc_addrs,
+                          uint64_t *lens)
+{
+       uint32_t buf_offset = sizeof(struct virtio_net_hdr_mrg_rxbuf);
+       struct virtio_net_hdr_mrg_rxbuf *hdrs[PACKED_BATCH_SIZE];
+       struct vring_packed_desc *descs = vq->desc_packed;
+       uint16_t avail_idx = vq->last_avail_idx;
+       uint16_t ids[PACKED_BATCH_SIZE];
+       uint16_t i;
+
        vhost_for_each_try_unroll(i, 0, PACKED_BATCH_SIZE) {
                rte_prefetch0((void *)(uintptr_t)desc_addrs[i]);
                hdrs[i] = (struct virtio_net_hdr_mrg_rxbuf *)
@@ -1331,6 +1405,51 @@ virtio_dev_rx_batch_packed(struct virtio_net *dev,
                ids[i] = descs[avail_idx + i].id;
 
        vhost_flush_enqueue_batch_packed(dev, vq, lens, ids);
+}
+
+static __rte_always_inline int
+virtio_dev_rx_sync_batch_packed(struct virtio_net *dev,
+                          struct vhost_virtqueue *vq,
+                          struct rte_mbuf **pkts)
+{
+       uint64_t desc_addrs[PACKED_BATCH_SIZE];
+       uint64_t lens[PACKED_BATCH_SIZE];
+
+       if (virtio_dev_rx_sync_batch_check(dev, vq, pkts, desc_addrs, lens) == -1)
+               return -1;
+
+       if (vq->shadow_used_idx) {
+               do_data_copy_enqueue(dev, vq);
+               vhost_flush_enqueue_shadow_packed(dev, vq);
+       }
+
+       virtio_dev_rx_batch_packed_copy(dev, vq, pkts, desc_addrs, lens);
+
+       return 0;
+}
+
+static __rte_always_inline int
+virtio_dev_rx_async_batch_packed(struct virtio_net *dev,
+                          struct vhost_virtqueue *vq,
+                          struct rte_mbuf **pkts,
+                          struct rte_mbuf **comp_pkts, uint32_t *pkt_done)
+{
+       uint16_t i;
+       uint64_t desc_addrs[PACKED_BATCH_SIZE];
+       uint64_t lens[PACKED_BATCH_SIZE];
+
+       if (virtio_dev_rx_async_batch_check(dev, vq, pkts, desc_addrs, lens) == -1)
+               return -1;
+
+       virtio_dev_rx_batch_packed_copy(dev, vq, pkts, desc_addrs, lens);
+
+       if (vq->shadow_used_idx) {
+               do_data_copy_enqueue(dev, vq);
+               vhost_flush_enqueue_shadow_packed(dev, vq);
+       }
+
+       vhost_for_each_try_unroll(i, 0, PACKED_BATCH_SIZE)
+               comp_pkts[(*pkt_done)++] = pkts[i];
 
        return 0;
 }
@@ -1372,7 +1491,7 @@ virtio_dev_rx_packed(struct virtio_net *dev,
                rte_prefetch0(&vq->desc_packed[vq->last_avail_idx]);
 
                if (count - pkt_idx >= PACKED_BATCH_SIZE) {
-                       if (!virtio_dev_rx_batch_packed(dev, vq,
+                       if (!virtio_dev_rx_sync_batch_packed(dev, vq,
                                                        &pkts[pkt_idx])) {
                                pkt_idx += PACKED_BATCH_SIZE;
                                continue;
@@ -1467,7 +1586,7 @@ virtio_dev_rx_async_get_info_idx(uint16_t pkts_idx,
        uint16_t vq_size, uint16_t n_inflight)
 {
        return pkts_idx > n_inflight ? (pkts_idx - n_inflight) :
-               (vq_size - n_inflight + pkts_idx) & (vq_size - 1);
+               (vq_size - n_inflight + pkts_idx) % vq_size;
 }
 
 static __rte_always_inline void
@@ -1525,6 +1644,7 @@ virtio_dev_rx_async_submit_split(struct virtio_net *dev,
        struct async_inflight_info *pkts_info = vq->async_pkts_info;
        uint32_t n_pkts = 0, pkt_err = 0;
        uint32_t num_async_pkts = 0, num_done_pkts = 0;
+       int32_t n_xfer;
        struct {
                uint16_t pkt_idx;
                uint16_t last_avail_idx;
@@ -1605,8 +1725,17 @@ virtio_dev_rx_async_submit_split(struct virtio_net *dev,
                if (unlikely(pkt_burst_idx >= VHOST_ASYNC_BATCH_THRESHOLD ||
                        ((VHOST_MAX_ASYNC_VEC >> 1) - segs_await <
                        BUF_VECTOR_MAX))) {
-                       n_pkts = vq->async_ops.transfer_data(dev->vid,
+                       n_xfer = vq->async_ops.transfer_data(dev->vid,
                                        queue_id, tdes, 0, pkt_burst_idx);
+                       if (n_xfer >= 0) {
+                               n_pkts = n_xfer;
+                       } else {
+                               VHOST_LOG_DATA(ERR,
+                                       "(%d) %s: failed to transfer data for queue id %d.\n",
+                                       dev->vid, __func__, queue_id);
+                               n_pkts = 0;
+                       }
+
                        iovec_idx = 0;
                        it_idx = 0;
 
@@ -1629,8 +1758,15 @@ virtio_dev_rx_async_submit_split(struct virtio_net *dev,
        }
 
        if (pkt_burst_idx) {
-               n_pkts = vq->async_ops.transfer_data(dev->vid,
-                               queue_id, tdes, 0, pkt_burst_idx);
+               n_xfer = vq->async_ops.transfer_data(dev->vid, queue_id, tdes, 0, pkt_burst_idx);
+               if (n_xfer >= 0) {
+                       n_pkts = n_xfer;
+               } else {
+                       VHOST_LOG_DATA(ERR, "(%d) %s: failed to transfer data for queue id %d.\n",
+                               dev->vid, __func__, queue_id);
+                       n_pkts = 0;
+               }
+
                vq->async_pkts_inflight_n += n_pkts;
 
                if (unlikely(n_pkts < pkt_burst_idx))
@@ -1725,29 +1861,6 @@ vhost_update_used_packed(struct vhost_virtqueue *vq,
        vq->desc_packed[head_idx].flags = head_flags;
 }
 
-static __rte_always_inline int
-virtio_dev_rx_async_batch_packed(struct virtio_net *dev,
-                          struct vhost_virtqueue *vq,
-                          struct rte_mbuf **pkts,
-                          struct rte_mbuf **comp_pkts, uint32_t *pkt_done)
-{
-       uint16_t i;
-       uint32_t cpy_threshold = vq->async_threshold;
-
-       vhost_for_each_try_unroll(i, 0, PACKED_BATCH_SIZE) {
-               if (unlikely(pkts[i]->pkt_len >= cpy_threshold))
-                       return -1;
-       }
-       if (!virtio_dev_rx_batch_packed(dev, vq, pkts)) {
-               vhost_for_each_try_unroll(i, 0, PACKED_BATCH_SIZE)
-                       comp_pkts[(*pkt_done)++] = pkts[i];
-
-               return 0;
-       }
-
-       return -1;
-}
-
 static __rte_always_inline int
 vhost_enqueue_async_single_packed(struct virtio_net *dev,
                            struct vhost_virtqueue *vq,
@@ -1900,6 +2013,7 @@ virtio_dev_rx_async_submit_packed(struct virtio_net *dev,
        uint16_t async_descs_idx = 0;
        uint16_t num_buffers;
        uint16_t num_descs;
+       int32_t n_xfer;
 
        struct rte_vhost_iov_iter *it_pool = vq->it_pool;
        struct iovec *vec_pool = vq->vec_pool;
@@ -1940,7 +2054,7 @@ virtio_dev_rx_async_submit_packed(struct virtio_net *dev,
 
                slot_idx = (vq->async_pkts_idx + num_async_pkts) % vq->size;
                if (it_pool[it_idx].count) {
-                       uint16_t from, to;
+                       uint16_t from;
 
                        async_descs_idx += num_descs;
                        async_fill_desc(&tdes[pkt_burst_idx++],
@@ -1959,11 +2073,13 @@ virtio_dev_rx_async_submit_packed(struct virtio_net *dev,
                         * descriptors.
                         */
                        from = vq->shadow_used_idx - num_buffers;
-                       to = vq->async_buffer_idx_packed % vq->size;
                        store_dma_desc_info_packed(vq->shadow_used_packed,
-                                       vq->async_buffers_packed, vq->size, from, to, num_buffers);
+                                       vq->async_buffers_packed, vq->size, from,
+                                       vq->async_buffer_idx_packed, num_buffers);
 
                        vq->async_buffer_idx_packed += num_buffers;
+                       if (vq->async_buffer_idx_packed >= vq->size)
+                               vq->async_buffer_idx_packed -= vq->size;
                        vq->shadow_used_idx -= num_buffers;
                } else {
                        comp_pkts[num_done_pkts++] = pkts[pkt_idx];
@@ -1980,8 +2096,17 @@ virtio_dev_rx_async_submit_packed(struct virtio_net *dev,
                 */
                if (unlikely(pkt_burst_idx >= VHOST_ASYNC_BATCH_THRESHOLD ||
                        ((VHOST_MAX_ASYNC_VEC >> 1) - segs_await < BUF_VECTOR_MAX))) {
-                       n_pkts = vq->async_ops.transfer_data(dev->vid, queue_id,
-                               tdes, 0, pkt_burst_idx);
+                       n_xfer = vq->async_ops.transfer_data(dev->vid,
+                                       queue_id, tdes, 0, pkt_burst_idx);
+                       if (n_xfer >= 0) {
+                               n_pkts = n_xfer;
+                       } else {
+                               VHOST_LOG_DATA(ERR,
+                                       "(%d) %s: failed to transfer data for queue id %d.\n",
+                                       dev->vid, __func__, queue_id);
+                               n_pkts = 0;
+                       }
+
                        iovec_idx = 0;
                        it_idx = 0;
                        segs_await = 0;
@@ -2003,7 +2128,15 @@ virtio_dev_rx_async_submit_packed(struct virtio_net *dev,
        } while (pkt_idx < count);
 
        if (pkt_burst_idx) {
-               n_pkts = vq->async_ops.transfer_data(dev->vid, queue_id, tdes, 0, pkt_burst_idx);
+               n_xfer = vq->async_ops.transfer_data(dev->vid, queue_id, tdes, 0, pkt_burst_idx);
+               if (n_xfer >= 0) {
+                       n_pkts = n_xfer;
+               } else {
+                       VHOST_LOG_DATA(ERR, "(%d) %s: failed to transfer data for queue id %d.\n",
+                               dev->vid, __func__, queue_id);
+                       n_pkts = 0;
+               }
+
                vq->async_pkts_inflight_n += n_pkts;
 
                if (unlikely(n_pkts < pkt_burst_idx))
@@ -2016,6 +2149,8 @@ virtio_dev_rx_async_submit_packed(struct virtio_net *dev,
                dma_error_handler_packed(vq, async_descs, async_descs_idx, slot_idx, pkt_err,
                                        &pkt_idx, &num_async_pkts, &num_done_pkts);
        vq->async_pkts_idx += num_async_pkts;
+       if (vq->async_pkts_idx >= vq->size)
+               vq->async_pkts_idx -= vq->size;
        *comp_count = num_done_pkts;
 
        if (likely(vq->shadow_used_idx)) {
@@ -2064,7 +2199,7 @@ write_back_completed_descs_packed(struct vhost_virtqueue *vq,
        uint16_t from, to;
 
        do {
-               from = vq->last_async_buffer_idx_packed % vq->size;
+               from = vq->last_async_buffer_idx_packed;
                to = (from + nr_left) % vq->size;
                if (to > from) {
                        vhost_update_used_packed(vq, vq->async_buffers_packed + from, to - from);
@@ -2073,62 +2208,54 @@ write_back_completed_descs_packed(struct vhost_virtqueue *vq,
                } else {
                        vhost_update_used_packed(vq, vq->async_buffers_packed + from,
                                vq->size - from);
-                       vq->last_async_buffer_idx_packed += vq->size - from;
+                       vq->last_async_buffer_idx_packed = 0;
                        nr_left -= vq->size - from;
                }
        } while (nr_left > 0);
 }
 
-uint16_t rte_vhost_poll_enqueue_completed(int vid, uint16_t queue_id,
+static __rte_always_inline uint16_t
+vhost_poll_enqueue_completed(struct virtio_net *dev, uint16_t queue_id,
                struct rte_mbuf **pkts, uint16_t count)
 {
-       struct virtio_net *dev = get_device(vid);
        struct vhost_virtqueue *vq;
        uint16_t n_pkts_cpl = 0, n_pkts_put = 0, n_descs = 0, n_buffers = 0;
        uint16_t start_idx, pkts_idx, vq_size;
        struct async_inflight_info *pkts_info;
        uint16_t from, i;
-
-       if (!dev)
-               return 0;
-
-       VHOST_LOG_DATA(DEBUG, "(%d) %s\n", dev->vid, __func__);
-       if (unlikely(!is_valid_virt_queue_idx(queue_id, 0, dev->nr_vring))) {
-               VHOST_LOG_DATA(ERR, "(%d) %s: invalid virtqueue idx %d.\n",
-                       dev->vid, __func__, queue_id);
-               return 0;
-       }
+       int32_t n_cpl;
 
        vq = dev->virtqueue[queue_id];
 
-       if (unlikely(!vq->async_registered)) {
-               VHOST_LOG_DATA(ERR, "(%d) %s: async not registered for queue id %d.\n",
-                       dev->vid, __func__, queue_id);
-               return 0;
-       }
-
-       rte_spinlock_lock(&vq->access_lock);
-
        pkts_idx = vq->async_pkts_idx % vq->size;
        pkts_info = vq->async_pkts_info;
        vq_size = vq->size;
        start_idx = virtio_dev_rx_async_get_info_idx(pkts_idx,
                vq_size, vq->async_pkts_inflight_n);
 
-       if (count > vq->async_last_pkts_n)
-               n_pkts_cpl = vq->async_ops.check_completed_copies(vid,
+       if (count > vq->async_last_pkts_n) {
+               n_cpl = vq->async_ops.check_completed_copies(dev->vid,
                        queue_id, 0, count - vq->async_last_pkts_n);
+               if (n_cpl >= 0) {
+                       n_pkts_cpl = n_cpl;
+               } else {
+                       VHOST_LOG_DATA(ERR,
+                               "(%d) %s: failed to check completed copies for queue id %d.\n",
+                               dev->vid, __func__, queue_id);
+                       n_pkts_cpl = 0;
+               }
+       }
        n_pkts_cpl += vq->async_last_pkts_n;
 
        n_pkts_put = RTE_MIN(count, n_pkts_cpl);
        if (unlikely(n_pkts_put == 0)) {
                vq->async_last_pkts_n = n_pkts_cpl;
-               goto done;
+               return 0;
        }
 
        if (vq_is_packed(dev)) {
                for (i = 0; i < n_pkts_put; i++) {
-                       from = (start_idx + i) & (vq_size - 1);
+                       from = (start_idx + i) % vq_size;
                        n_buffers += pkts_info[from].nr_buffers;
                        pkts[i] = pkts_info[from].mbuf;
                }
@@ -2156,16 +2283,82 @@ uint16_t rte_vhost_poll_enqueue_completed(int vid, uint16_t queue_id,
                        vhost_vring_call_split(dev, vq);
                }
        } else {
-               if (vq_is_packed(dev))
+               if (vq_is_packed(dev)) {
                        vq->last_async_buffer_idx_packed += n_buffers;
-               else
+                       if (vq->last_async_buffer_idx_packed >= vq->size)
+                               vq->last_async_buffer_idx_packed -= vq->size;
+               } else {
                        vq->last_async_desc_idx_split += n_descs;
+               }
+       }
+
+       return n_pkts_put;
+}
+
+uint16_t
+rte_vhost_poll_enqueue_completed(int vid, uint16_t queue_id,
+               struct rte_mbuf **pkts, uint16_t count)
+{
+       struct virtio_net *dev = get_device(vid);
+       struct vhost_virtqueue *vq;
+       uint16_t n_pkts_cpl = 0;
+
+       if (!dev)
+               return 0;
+
+       VHOST_LOG_DATA(DEBUG, "(%d) %s\n", dev->vid, __func__);
+       if (unlikely(!is_valid_virt_queue_idx(queue_id, 0, dev->nr_vring))) {
+               VHOST_LOG_DATA(ERR, "(%d) %s: invalid virtqueue idx %d.\n",
+                       dev->vid, __func__, queue_id);
+               return 0;
+       }
+
+       vq = dev->virtqueue[queue_id];
+
+       if (unlikely(!vq->async_registered)) {
+               VHOST_LOG_DATA(ERR, "(%d) %s: async not registered for queue id %d.\n",
+                       dev->vid, __func__, queue_id);
+               return 0;
        }
 
-done:
+       rte_spinlock_lock(&vq->access_lock);
+
+       n_pkts_cpl = vhost_poll_enqueue_completed(dev, queue_id, pkts, count);
+
        rte_spinlock_unlock(&vq->access_lock);
 
-       return n_pkts_put;
+       return n_pkts_cpl;
+}
+
+uint16_t
+rte_vhost_clear_queue_thread_unsafe(int vid, uint16_t queue_id,
+               struct rte_mbuf **pkts, uint16_t count)
+{
+       struct virtio_net *dev = get_device(vid);
+       struct vhost_virtqueue *vq;
+       uint16_t n_pkts_cpl = 0;
+
+       if (!dev)
+               return 0;
+
+       VHOST_LOG_DATA(DEBUG, "(%d) %s\n", dev->vid, __func__);
+       if (unlikely(!is_valid_virt_queue_idx(queue_id, 0, dev->nr_vring))) {
+               VHOST_LOG_DATA(ERR, "(%d) %s: invalid virtqueue idx %d.\n",
+                       dev->vid, __func__, queue_id);
+               return 0;
+       }
+
+       vq = dev->virtqueue[queue_id];
+
+       if (unlikely(!vq->async_registered)) {
+               VHOST_LOG_DATA(ERR, "(%d) %s: async not registered for queue id %d.\n",
+                       dev->vid, __func__, queue_id);
+               return 0;
+       }
+
+       n_pkts_cpl = vhost_poll_enqueue_completed(dev, queue_id, pkts, count);
+
+       return n_pkts_cpl;
 }
 
 static __rte_always_inline uint32_t
@@ -2256,14 +2449,17 @@ virtio_net_with_host_offload(struct virtio_net *dev)
        return false;
 }
 
-static void
-parse_ethernet(struct rte_mbuf *m, uint16_t *l4_proto, void **l4_hdr)
+static int
+parse_headers(struct rte_mbuf *m, uint8_t *l4_proto)
 {
        struct rte_ipv4_hdr *ipv4_hdr;
        struct rte_ipv6_hdr *ipv6_hdr;
-       void *l3_hdr = NULL;
        struct rte_ether_hdr *eth_hdr;
        uint16_t ethertype;
+       uint16_t data_len = rte_pktmbuf_data_len(m);
+
+       if (data_len < sizeof(struct rte_ether_hdr))
+               return -EINVAL;
 
        eth_hdr = rte_pktmbuf_mtod(m, struct rte_ether_hdr *);
 
@@ -2271,6 +2467,10 @@ parse_ethernet(struct rte_mbuf *m, uint16_t *l4_proto, void **l4_hdr)
        ethertype = rte_be_to_cpu_16(eth_hdr->ether_type);
 
        if (ethertype == RTE_ETHER_TYPE_VLAN) {
+               if (data_len < sizeof(struct rte_ether_hdr) +
+                               sizeof(struct rte_vlan_hdr))
+                       goto error;
+
                struct rte_vlan_hdr *vlan_hdr =
                        (struct rte_vlan_hdr *)(eth_hdr + 1);
 
@@ -2278,80 +2478,134 @@ parse_ethernet(struct rte_mbuf *m, uint16_t *l4_proto, void **l4_hdr)
                ethertype = rte_be_to_cpu_16(vlan_hdr->eth_proto);
        }
 
-       l3_hdr = (char *)eth_hdr + m->l2_len;
-
        switch (ethertype) {
        case RTE_ETHER_TYPE_IPV4:
-               ipv4_hdr = l3_hdr;
-               *l4_proto = ipv4_hdr->next_proto_id;
+               if (data_len < m->l2_len + sizeof(struct rte_ipv4_hdr))
+                       goto error;
+               ipv4_hdr = rte_pktmbuf_mtod_offset(m, struct rte_ipv4_hdr *,
+                               m->l2_len);
                m->l3_len = rte_ipv4_hdr_len(ipv4_hdr);
-               *l4_hdr = (char *)l3_hdr + m->l3_len;
-               m->ol_flags |= PKT_TX_IPV4;
+               if (data_len < m->l2_len + m->l3_len)
+                       goto error;
+               m->ol_flags |= RTE_MBUF_F_TX_IPV4;
+               *l4_proto = ipv4_hdr->next_proto_id;
                break;
        case RTE_ETHER_TYPE_IPV6:
-               ipv6_hdr = l3_hdr;
-               *l4_proto = ipv6_hdr->proto;
+               if (data_len < m->l2_len + sizeof(struct rte_ipv6_hdr))
+                       goto error;
+               ipv6_hdr = rte_pktmbuf_mtod_offset(m, struct rte_ipv6_hdr *,
+                               m->l2_len);
                m->l3_len = sizeof(struct rte_ipv6_hdr);
-               *l4_hdr = (char *)l3_hdr + m->l3_len;
-               m->ol_flags |= PKT_TX_IPV6;
+               m->ol_flags |= RTE_MBUF_F_TX_IPV6;
+               *l4_proto = ipv6_hdr->proto;
                break;
        default:
-               m->l3_len = 0;
-               *l4_proto = 0;
-               *l4_hdr = NULL;
+               /* a valid L3 header is needed for further L4 parsing */
+               goto error;
+       }
+
+       /* both CSUM and GSO need a valid L4 header */
+       switch (*l4_proto) {
+       case IPPROTO_TCP:
+               if (data_len < m->l2_len + m->l3_len +
+                               sizeof(struct rte_tcp_hdr))
+                       goto error;
+               break;
+       case IPPROTO_UDP:
+               if (data_len < m->l2_len + m->l3_len +
+                               sizeof(struct rte_udp_hdr))
+                       goto error;
                break;
+       case IPPROTO_SCTP:
+               if (data_len < m->l2_len + m->l3_len +
+                               sizeof(struct rte_sctp_hdr))
+                       goto error;
+               break;
+       default:
+               goto error;
        }
+
+       return 0;
+
+error:
+       m->l2_len = 0;
+       m->l3_len = 0;
+       m->ol_flags = 0;
+       return -EINVAL;
 }
 
 static __rte_always_inline void
 vhost_dequeue_offload_legacy(struct virtio_net_hdr *hdr, struct rte_mbuf *m)
 {
-       uint16_t l4_proto = 0;
-       void *l4_hdr = NULL;
+       uint8_t l4_proto = 0;
        struct rte_tcp_hdr *tcp_hdr = NULL;
+       uint16_t tcp_len;
+       uint16_t data_len = rte_pktmbuf_data_len(m);
+
+       if (parse_headers(m, &l4_proto) < 0)
+               return;
 
-       parse_ethernet(m, &l4_proto, &l4_hdr);
        if (hdr->flags == VIRTIO_NET_HDR_F_NEEDS_CSUM) {
                if (hdr->csum_start == (m->l2_len + m->l3_len)) {
                        switch (hdr->csum_offset) {
                        case (offsetof(struct rte_tcp_hdr, cksum)):
-                               if (l4_proto == IPPROTO_TCP)
-                                       m->ol_flags |= PKT_TX_TCP_CKSUM;
+                               if (l4_proto != IPPROTO_TCP)
+                                       goto error;
+                               m->ol_flags |= RTE_MBUF_F_TX_TCP_CKSUM;
                                break;
                        case (offsetof(struct rte_udp_hdr, dgram_cksum)):
-                               if (l4_proto == IPPROTO_UDP)
-                                       m->ol_flags |= PKT_TX_UDP_CKSUM;
+                               if (l4_proto != IPPROTO_UDP)
+                                       goto error;
+                               m->ol_flags |= RTE_MBUF_F_TX_UDP_CKSUM;
                                break;
                        case (offsetof(struct rte_sctp_hdr, cksum)):
-                               if (l4_proto == IPPROTO_SCTP)
-                                       m->ol_flags |= PKT_TX_SCTP_CKSUM;
+                               if (l4_proto != IPPROTO_SCTP)
+                                       goto error;
+                               m->ol_flags |= RTE_MBUF_F_TX_SCTP_CKSUM;
                                break;
                        default:
-                               break;
+                               goto error;
                        }
+               } else {
+                       goto error;
                }
        }
 
-       if (l4_hdr && hdr->gso_type != VIRTIO_NET_HDR_GSO_NONE) {
+       if (hdr->gso_type != VIRTIO_NET_HDR_GSO_NONE) {
                switch (hdr->gso_type & ~VIRTIO_NET_HDR_GSO_ECN) {
                case VIRTIO_NET_HDR_GSO_TCPV4:
                case VIRTIO_NET_HDR_GSO_TCPV6:
-                       tcp_hdr = l4_hdr;
-                       m->ol_flags |= PKT_TX_TCP_SEG;
+                       if (l4_proto != IPPROTO_TCP)
+                               goto error;
+                       tcp_hdr = rte_pktmbuf_mtod_offset(m,
+                                       struct rte_tcp_hdr *,
+                                       m->l2_len + m->l3_len);
+                       tcp_len = (tcp_hdr->data_off & 0xf0) >> 2;
+                       if (data_len < m->l2_len + m->l3_len + tcp_len)
+                               goto error;
+                       m->ol_flags |= RTE_MBUF_F_TX_TCP_SEG;
                        m->tso_segsz = hdr->gso_size;
-                       m->l4_len = (tcp_hdr->data_off & 0xf0) >> 2;
+                       m->l4_len = tcp_len;
                        break;
                case VIRTIO_NET_HDR_GSO_UDP:
-                       m->ol_flags |= PKT_TX_UDP_SEG;
+                       if (l4_proto != IPPROTO_UDP)
+                               goto error;
+                       m->ol_flags |= RTE_MBUF_F_TX_UDP_SEG;
                        m->tso_segsz = hdr->gso_size;
                        m->l4_len = sizeof(struct rte_udp_hdr);
                        break;
                default:
                        VHOST_LOG_DATA(WARNING,
                                "unsupported gso type %u.\n", hdr->gso_type);
-                       break;
+                       goto error;
                }
        }
+       return;
+
+error:
+       m->l2_len = 0;
+       m->l3_len = 0;
+       m->ol_flags = 0;
 }
 
 static __rte_always_inline void
@@ -2370,7 +2624,7 @@ vhost_dequeue_offload(struct virtio_net_hdr *hdr, struct rte_mbuf *m,
                return;
        }
 
-       m->ol_flags |= PKT_RX_IP_CKSUM_UNKNOWN;
+       m->ol_flags |= RTE_MBUF_F_RX_IP_CKSUM_UNKNOWN;
 
        ptype = rte_net_get_ptype(m, &hdr_lens, RTE_PTYPE_ALL_MASK);
        m->packet_type = ptype;
@@ -2397,7 +2651,7 @@ vhost_dequeue_offload(struct virtio_net_hdr *hdr, struct rte_mbuf *m,
 
                hdrlen = hdr_lens.l2_len + hdr_lens.l3_len + hdr_lens.l4_len;
                if (hdr->csum_start <= hdrlen && l4_supported != 0) {
-                       m->ol_flags |= PKT_RX_L4_CKSUM_NONE;
+                       m->ol_flags |= RTE_MBUF_F_RX_L4_CKSUM_NONE;
                } else {
                        /* Unknown proto or tunnel, do sw cksum. We can assume
                         * the cksum field is in the first segment since the
@@ -2427,13 +2681,13 @@ vhost_dequeue_offload(struct virtio_net_hdr *hdr, struct rte_mbuf *m,
                case VIRTIO_NET_HDR_GSO_TCPV6:
                        if ((ptype & RTE_PTYPE_L4_MASK) != RTE_PTYPE_L4_TCP)
                                break;
-                       m->ol_flags |= PKT_RX_LRO | PKT_RX_L4_CKSUM_NONE;
+                       m->ol_flags |= RTE_MBUF_F_RX_LRO | RTE_MBUF_F_RX_L4_CKSUM_NONE;
                        m->tso_segsz = hdr->gso_size;
                        break;
                case VIRTIO_NET_HDR_GSO_UDP:
                        if ((ptype & RTE_PTYPE_L4_MASK) != RTE_PTYPE_L4_UDP)
                                break;
-                       m->ol_flags |= PKT_RX_LRO | PKT_RX_L4_CKSUM_NONE;
+                       m->ol_flags |= RTE_MBUF_F_RX_LRO | RTE_MBUF_F_RX_L4_CKSUM_NONE;
                        m->tso_segsz = hdr->gso_size;
                        break;
                default:
@@ -2649,6 +2903,9 @@ virtio_dev_extbuf_alloc(struct rte_mbuf *pkt, uint32_t size)
        return 0;
 }
 
+/*
+ * Prepare a host supported pktmbuf.
+ */
 static __rte_always_inline int
 virtio_dev_pktmbuf_prep(struct virtio_net *dev, struct rte_mbuf *pkt,
                         uint32_t data_len)
@@ -2667,32 +2924,6 @@ virtio_dev_pktmbuf_prep(struct virtio_net *dev, struct rte_mbuf *pkt,
        return -1;
 }
 
-/*
- * Allocate a host supported pktmbuf.
- */
-static __rte_always_inline struct rte_mbuf *
-virtio_dev_pktmbuf_alloc(struct virtio_net *dev, struct rte_mempool *mp,
-                        uint32_t data_len)
-{
-       struct rte_mbuf *pkt = rte_pktmbuf_alloc(mp);
-
-       if (unlikely(pkt == NULL)) {
-               VHOST_LOG_DATA(ERR,
-                       "Failed to allocate memory for mbuf.\n");
-               return NULL;
-       }
-
-       if (virtio_dev_pktmbuf_prep(dev, pkt, data_len)) {
-               /* Data doesn't fit into the buffer and the host supports
-                * only linear buffers
-                */
-               rte_pktmbuf_free(pkt);
-               return NULL;
-       }
-
-       return pkt;
-}
-
 __rte_always_inline
 static uint16_t
 virtio_dev_tx_split(struct virtio_net *dev, struct vhost_virtqueue *vq,
@@ -2722,6 +2953,9 @@ virtio_dev_tx_split(struct virtio_net *dev, struct vhost_virtqueue *vq,
        VHOST_LOG_DATA(DEBUG, "(%d) about to dequeue %u buffers\n",
                        dev->vid, count);
 
+       if (rte_pktmbuf_alloc_bulk(mbuf_pool, pkts, count))
+               return 0;
+
        for (i = 0; i < count; i++) {
                struct buf_vector buf_vec[BUF_VECTOR_MAX];
                uint16_t head_idx;
@@ -2738,8 +2972,8 @@ virtio_dev_tx_split(struct virtio_net *dev, struct vhost_virtqueue *vq,
 
                update_shadow_used_ring_split(vq, head_idx, 0);
 
-               pkts[i] = virtio_dev_pktmbuf_alloc(dev, mbuf_pool, buf_len);
-               if (unlikely(pkts[i] == NULL)) {
+               err = virtio_dev_pktmbuf_prep(dev, pkts[i], buf_len);
+               if (unlikely(err)) {
                        /*
                         * mbuf allocation fails for jumbo packets when external
                         * buffer allocation is not allowed and linear buffer
@@ -2759,7 +2993,6 @@ virtio_dev_tx_split(struct virtio_net *dev, struct vhost_virtqueue *vq,
                err = copy_desc_to_mbuf(dev, vq, buf_vec, nr_vec, pkts[i],
                                mbuf_pool, legacy_ol_flags);
                if (unlikely(err)) {
-                       rte_pktmbuf_free(pkts[i]);
                        if (!allocerr_warned) {
                                VHOST_LOG_DATA(ERR,
                                        "Failed to copy desc to mbuf on %s.\n",
@@ -2772,6 +3005,9 @@ virtio_dev_tx_split(struct virtio_net *dev, struct vhost_virtqueue *vq,
                }
        }
 
+       if (dropped)
+               rte_pktmbuf_free_bulk(&pkts[i - 1], count - i + 1);
+
        vq->last_avail_idx += i;
 
        do_data_copy_dequeue(vq);