From f412f14facc8c30b0b89e2cf4a73faca330001e6 Mon Sep 17 00:00:00 2001 From: Jerin Jacob Date: Tue, 5 Jul 2016 18:19:24 +0530 Subject: [PATCH] net/virtio: move SSE based Rx code to separate file Split out SSE instruction based virtio simple Rx implementation to a separate file Signed-off-by: Jerin Jacob Acked-by: Yuanhan Liu --- drivers/net/virtio/Makefile | 4 + drivers/net/virtio/virtio_rxtx_simple.c | 276 +------------------- drivers/net/virtio/virtio_rxtx_simple.h | 136 ++++++++++ drivers/net/virtio/virtio_rxtx_simple_sse.c | 222 ++++++++++++++++ 4 files changed, 373 insertions(+), 265 deletions(-) create mode 100644 drivers/net/virtio/virtio_rxtx_simple.h create mode 100644 drivers/net/virtio/virtio_rxtx_simple_sse.c diff --git a/drivers/net/virtio/Makefile b/drivers/net/virtio/Makefile index b9b0d8dd45..c4103b748f 100644 --- a/drivers/net/virtio/Makefile +++ b/drivers/net/virtio/Makefile @@ -52,6 +52,10 @@ SRCS-$(CONFIG_RTE_LIBRTE_VIRTIO_PMD) += virtio_rxtx.c SRCS-$(CONFIG_RTE_LIBRTE_VIRTIO_PMD) += virtio_ethdev.c SRCS-$(CONFIG_RTE_LIBRTE_VIRTIO_PMD) += virtio_rxtx_simple.c +ifeq ($(CONFIG_RTE_ARCH_X86),y) +SRCS-$(CONFIG_RTE_LIBRTE_VIRTIO_PMD) += virtio_rxtx_simple_sse.c +endif + ifeq ($(CONFIG_RTE_VIRTIO_USER),y) SRCS-$(CONFIG_RTE_LIBRTE_VIRTIO_PMD) += virtio_user/vhost_user.c SRCS-$(CONFIG_RTE_LIBRTE_VIRTIO_PMD) += virtio_user/virtio_user_dev.c diff --git a/drivers/net/virtio/virtio_rxtx_simple.c b/drivers/net/virtio/virtio_rxtx_simple.c index b7f460d572..b651e53b24 100644 --- a/drivers/net/virtio/virtio_rxtx_simple.c +++ b/drivers/net/virtio/virtio_rxtx_simple.c @@ -51,14 +51,7 @@ #include #include -#include "virtio_logs.h" -#include "virtio_ethdev.h" -#include "virtqueue.h" -#include "virtio_rxtx.h" - -#define RTE_VIRTIO_VPMD_RX_BURST 32 -#define RTE_VIRTIO_DESC_PER_LOOP 8 -#define RTE_VIRTIO_VPMD_RX_REARM_THRESH RTE_VIRTIO_VPMD_RX_BURST +#include "virtio_rxtx_simple.h" #ifndef __INTEL_COMPILER #pragma GCC diagnostic ignored "-Wcast-qual" @@ -90,263 +83,6 @@ virtqueue_enqueue_recv_refill_simple(struct virtqueue *vq, return 0; } -static inline void -virtio_rxq_rearm_vec(struct virtnet_rx *rxvq) -{ - int i; - uint16_t desc_idx; - struct rte_mbuf **sw_ring; - struct vring_desc *start_dp; - int ret; - struct virtqueue *vq = rxvq->vq; - - desc_idx = vq->vq_avail_idx & (vq->vq_nentries - 1); - sw_ring = &vq->sw_ring[desc_idx]; - start_dp = &vq->vq_ring.desc[desc_idx]; - - ret = rte_mempool_get_bulk(rxvq->mpool, (void **)sw_ring, - RTE_VIRTIO_VPMD_RX_REARM_THRESH); - if (unlikely(ret)) { - rte_eth_devices[rxvq->port_id].data->rx_mbuf_alloc_failed += - RTE_VIRTIO_VPMD_RX_REARM_THRESH; - return; - } - - for (i = 0; i < RTE_VIRTIO_VPMD_RX_REARM_THRESH; i++) { - uintptr_t p; - - p = (uintptr_t)&sw_ring[i]->rearm_data; - *(uint64_t *)p = rxvq->mbuf_initializer; - - start_dp[i].addr = - VIRTIO_MBUF_ADDR(sw_ring[i], vq) + - RTE_PKTMBUF_HEADROOM - vq->hw->vtnet_hdr_size; - start_dp[i].len = sw_ring[i]->buf_len - - RTE_PKTMBUF_HEADROOM + vq->hw->vtnet_hdr_size; - } - - vq->vq_avail_idx += RTE_VIRTIO_VPMD_RX_REARM_THRESH; - vq->vq_free_cnt -= RTE_VIRTIO_VPMD_RX_REARM_THRESH; - vq_update_avail_idx(vq); -} - -#ifdef RTE_MACHINE_CPUFLAG_SSSE3 - -#include - -/* virtio vPMD receive routine, only accept(nb_pkts >= RTE_VIRTIO_DESC_PER_LOOP) - * - * This routine is for non-mergeable RX, one desc for each guest buffer. - * This routine is based on the RX ring layout optimization. Each entry in the - * avail ring points to the desc with the same index in the desc ring and this - * will never be changed in the driver. - * - * - nb_pkts < RTE_VIRTIO_DESC_PER_LOOP, just return no packet - */ -uint16_t -virtio_recv_pkts_vec(void *rx_queue, struct rte_mbuf **rx_pkts, - uint16_t nb_pkts) -{ - struct virtnet_rx *rxvq = rx_queue; - struct virtqueue *vq = rxvq->vq; - uint16_t nb_used; - uint16_t desc_idx; - struct vring_used_elem *rused; - struct rte_mbuf **sw_ring; - struct rte_mbuf **sw_ring_end; - uint16_t nb_pkts_received; - __m128i shuf_msk1, shuf_msk2, len_adjust; - - shuf_msk1 = _mm_set_epi8( - 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, /* vlan tci */ - 5, 4, /* dat len */ - 0xFF, 0xFF, 5, 4, /* pkt len */ - 0xFF, 0xFF, 0xFF, 0xFF /* packet type */ - - ); - - shuf_msk2 = _mm_set_epi8( - 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, /* vlan tci */ - 13, 12, /* dat len */ - 0xFF, 0xFF, 13, 12, /* pkt len */ - 0xFF, 0xFF, 0xFF, 0xFF /* packet type */ - ); - - /* Subtract the header length. - * In which case do we need the header length in used->len ? - */ - len_adjust = _mm_set_epi16( - 0, 0, - 0, - (uint16_t)-vq->hw->vtnet_hdr_size, - 0, (uint16_t)-vq->hw->vtnet_hdr_size, - 0, 0); - - if (unlikely(nb_pkts < RTE_VIRTIO_DESC_PER_LOOP)) - return 0; - - nb_used = VIRTQUEUE_NUSED(vq); - - rte_compiler_barrier(); - - if (unlikely(nb_used == 0)) - return 0; - - nb_pkts = RTE_ALIGN_FLOOR(nb_pkts, RTE_VIRTIO_DESC_PER_LOOP); - nb_used = RTE_MIN(nb_used, nb_pkts); - - desc_idx = (uint16_t)(vq->vq_used_cons_idx & (vq->vq_nentries - 1)); - rused = &vq->vq_ring.used->ring[desc_idx]; - sw_ring = &vq->sw_ring[desc_idx]; - sw_ring_end = &vq->sw_ring[vq->vq_nentries]; - - rte_prefetch0(rused); - - if (vq->vq_free_cnt >= RTE_VIRTIO_VPMD_RX_REARM_THRESH) { - virtio_rxq_rearm_vec(rxvq); - if (unlikely(virtqueue_kick_prepare(vq))) - virtqueue_notify(vq); - } - - for (nb_pkts_received = 0; - nb_pkts_received < nb_used;) { - __m128i desc[RTE_VIRTIO_DESC_PER_LOOP / 2]; - __m128i mbp[RTE_VIRTIO_DESC_PER_LOOP / 2]; - __m128i pkt_mb[RTE_VIRTIO_DESC_PER_LOOP]; - - mbp[0] = _mm_loadu_si128((__m128i *)(sw_ring + 0)); - desc[0] = _mm_loadu_si128((__m128i *)(rused + 0)); - _mm_storeu_si128((__m128i *)&rx_pkts[0], mbp[0]); - - mbp[1] = _mm_loadu_si128((__m128i *)(sw_ring + 2)); - desc[1] = _mm_loadu_si128((__m128i *)(rused + 2)); - _mm_storeu_si128((__m128i *)&rx_pkts[2], mbp[1]); - - mbp[2] = _mm_loadu_si128((__m128i *)(sw_ring + 4)); - desc[2] = _mm_loadu_si128((__m128i *)(rused + 4)); - _mm_storeu_si128((__m128i *)&rx_pkts[4], mbp[2]); - - mbp[3] = _mm_loadu_si128((__m128i *)(sw_ring + 6)); - desc[3] = _mm_loadu_si128((__m128i *)(rused + 6)); - _mm_storeu_si128((__m128i *)&rx_pkts[6], mbp[3]); - - pkt_mb[1] = _mm_shuffle_epi8(desc[0], shuf_msk2); - pkt_mb[0] = _mm_shuffle_epi8(desc[0], shuf_msk1); - pkt_mb[1] = _mm_add_epi16(pkt_mb[1], len_adjust); - pkt_mb[0] = _mm_add_epi16(pkt_mb[0], len_adjust); - _mm_storeu_si128((void *)&rx_pkts[1]->rx_descriptor_fields1, - pkt_mb[1]); - _mm_storeu_si128((void *)&rx_pkts[0]->rx_descriptor_fields1, - pkt_mb[0]); - - pkt_mb[3] = _mm_shuffle_epi8(desc[1], shuf_msk2); - pkt_mb[2] = _mm_shuffle_epi8(desc[1], shuf_msk1); - pkt_mb[3] = _mm_add_epi16(pkt_mb[3], len_adjust); - pkt_mb[2] = _mm_add_epi16(pkt_mb[2], len_adjust); - _mm_storeu_si128((void *)&rx_pkts[3]->rx_descriptor_fields1, - pkt_mb[3]); - _mm_storeu_si128((void *)&rx_pkts[2]->rx_descriptor_fields1, - pkt_mb[2]); - - pkt_mb[5] = _mm_shuffle_epi8(desc[2], shuf_msk2); - pkt_mb[4] = _mm_shuffle_epi8(desc[2], shuf_msk1); - pkt_mb[5] = _mm_add_epi16(pkt_mb[5], len_adjust); - pkt_mb[4] = _mm_add_epi16(pkt_mb[4], len_adjust); - _mm_storeu_si128((void *)&rx_pkts[5]->rx_descriptor_fields1, - pkt_mb[5]); - _mm_storeu_si128((void *)&rx_pkts[4]->rx_descriptor_fields1, - pkt_mb[4]); - - pkt_mb[7] = _mm_shuffle_epi8(desc[3], shuf_msk2); - pkt_mb[6] = _mm_shuffle_epi8(desc[3], shuf_msk1); - pkt_mb[7] = _mm_add_epi16(pkt_mb[7], len_adjust); - pkt_mb[6] = _mm_add_epi16(pkt_mb[6], len_adjust); - _mm_storeu_si128((void *)&rx_pkts[7]->rx_descriptor_fields1, - pkt_mb[7]); - _mm_storeu_si128((void *)&rx_pkts[6]->rx_descriptor_fields1, - pkt_mb[6]); - - if (unlikely(nb_used <= RTE_VIRTIO_DESC_PER_LOOP)) { - if (sw_ring + nb_used <= sw_ring_end) - nb_pkts_received += nb_used; - else - nb_pkts_received += sw_ring_end - sw_ring; - break; - } else { - if (unlikely(sw_ring + RTE_VIRTIO_DESC_PER_LOOP >= - sw_ring_end)) { - nb_pkts_received += sw_ring_end - sw_ring; - break; - } else { - nb_pkts_received += RTE_VIRTIO_DESC_PER_LOOP; - - rx_pkts += RTE_VIRTIO_DESC_PER_LOOP; - sw_ring += RTE_VIRTIO_DESC_PER_LOOP; - rused += RTE_VIRTIO_DESC_PER_LOOP; - nb_used -= RTE_VIRTIO_DESC_PER_LOOP; - } - } - } - - vq->vq_used_cons_idx += nb_pkts_received; - vq->vq_free_cnt += nb_pkts_received; - rxvq->stats.packets += nb_pkts_received; - return nb_pkts_received; -} - -#endif - -#define VIRTIO_TX_FREE_THRESH 32 -#define VIRTIO_TX_MAX_FREE_BUF_SZ 32 -#define VIRTIO_TX_FREE_NR 32 -/* TODO: vq->tx_free_cnt could mean num of free slots so we could avoid shift */ -static inline void -virtio_xmit_cleanup(struct virtqueue *vq) -{ - uint16_t i, desc_idx; - uint32_t nb_free = 0; - struct rte_mbuf *m, *free[VIRTIO_TX_MAX_FREE_BUF_SZ]; - - desc_idx = (uint16_t)(vq->vq_used_cons_idx & - ((vq->vq_nentries >> 1) - 1)); - m = (struct rte_mbuf *)vq->vq_descx[desc_idx++].cookie; - m = __rte_pktmbuf_prefree_seg(m); - if (likely(m != NULL)) { - free[0] = m; - nb_free = 1; - for (i = 1; i < VIRTIO_TX_FREE_NR; i++) { - m = (struct rte_mbuf *)vq->vq_descx[desc_idx++].cookie; - m = __rte_pktmbuf_prefree_seg(m); - if (likely(m != NULL)) { - if (likely(m->pool == free[0]->pool)) - free[nb_free++] = m; - else { - rte_mempool_put_bulk(free[0]->pool, - (void **)free, - RTE_MIN(RTE_DIM(free), - nb_free)); - free[0] = m; - nb_free = 1; - } - } - } - rte_mempool_put_bulk(free[0]->pool, (void **)free, - RTE_MIN(RTE_DIM(free), nb_free)); - } else { - for (i = 1; i < VIRTIO_TX_FREE_NR; i++) { - m = (struct rte_mbuf *)vq->vq_descx[desc_idx++].cookie; - m = __rte_pktmbuf_prefree_seg(m); - if (m != NULL) - rte_mempool_put(m->pool, m); - } - } - - vq->vq_used_cons_idx += VIRTIO_TX_FREE_NR; - vq->vq_free_cnt += (VIRTIO_TX_FREE_NR << 1); -} - uint16_t virtio_xmit_pkts_simple(void *tx_queue, struct rte_mbuf **tx_pkts, uint16_t nb_pkts) @@ -427,3 +163,13 @@ virtio_rxq_vec_setup(struct virtnet_rx *rxq) return 0; } + +/* Stub for linkage when arch specific implementation is not available */ +uint16_t __attribute__((weak)) +virtio_recv_pkts_vec(void *rx_queue __rte_unused, + struct rte_mbuf **rx_pkts __rte_unused, + uint16_t nb_pkts __rte_unused) +{ + rte_panic("Wrong weak function linked by linker\n"); + return 0; +} diff --git a/drivers/net/virtio/virtio_rxtx_simple.h b/drivers/net/virtio/virtio_rxtx_simple.h new file mode 100644 index 0000000000..b08f859488 --- /dev/null +++ b/drivers/net/virtio/virtio_rxtx_simple.h @@ -0,0 +1,136 @@ +/*- + * BSD LICENSE + * + * Copyright(c) 2010-2016 Intel Corporation. All rights reserved. + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * + * * Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * * Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in + * the documentation and/or other materials provided with the + * distribution. + * * Neither the name of Intel Corporation nor the names of its + * contributors may be used to endorse or promote products derived + * from this software without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS + * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT + * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR + * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT + * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, + * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT + * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE + * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#ifndef _VIRTIO_RXTX_SIMPLE_H_ +#define _VIRTIO_RXTX_SIMPLE_H_ + +#include + +#include "virtio_logs.h" +#include "virtio_ethdev.h" +#include "virtqueue.h" +#include "virtio_rxtx.h" + +#define RTE_VIRTIO_VPMD_RX_BURST 32 +#define RTE_VIRTIO_VPMD_RX_REARM_THRESH RTE_VIRTIO_VPMD_RX_BURST + +static inline void +virtio_rxq_rearm_vec(struct virtnet_rx *rxvq) +{ + int i; + uint16_t desc_idx; + struct rte_mbuf **sw_ring; + struct vring_desc *start_dp; + int ret; + struct virtqueue *vq = rxvq->vq; + + desc_idx = vq->vq_avail_idx & (vq->vq_nentries - 1); + sw_ring = &vq->sw_ring[desc_idx]; + start_dp = &vq->vq_ring.desc[desc_idx]; + + ret = rte_mempool_get_bulk(rxvq->mpool, (void **)sw_ring, + RTE_VIRTIO_VPMD_RX_REARM_THRESH); + if (unlikely(ret)) { + rte_eth_devices[rxvq->port_id].data->rx_mbuf_alloc_failed += + RTE_VIRTIO_VPMD_RX_REARM_THRESH; + return; + } + + for (i = 0; i < RTE_VIRTIO_VPMD_RX_REARM_THRESH; i++) { + uintptr_t p; + + p = (uintptr_t)&sw_ring[i]->rearm_data; + *(uint64_t *)p = rxvq->mbuf_initializer; + + start_dp[i].addr = + VIRTIO_MBUF_ADDR(sw_ring[i], vq) + + RTE_PKTMBUF_HEADROOM - vq->hw->vtnet_hdr_size; + start_dp[i].len = sw_ring[i]->buf_len - + RTE_PKTMBUF_HEADROOM + vq->hw->vtnet_hdr_size; + } + + vq->vq_avail_idx += RTE_VIRTIO_VPMD_RX_REARM_THRESH; + vq->vq_free_cnt -= RTE_VIRTIO_VPMD_RX_REARM_THRESH; + vq_update_avail_idx(vq); +} + +#define VIRTIO_TX_FREE_THRESH 32 +#define VIRTIO_TX_MAX_FREE_BUF_SZ 32 +#define VIRTIO_TX_FREE_NR 32 +/* TODO: vq->tx_free_cnt could mean num of free slots so we could avoid shift */ +static inline void +virtio_xmit_cleanup(struct virtqueue *vq) +{ + uint16_t i, desc_idx; + uint32_t nb_free = 0; + struct rte_mbuf *m, *free[VIRTIO_TX_MAX_FREE_BUF_SZ]; + + desc_idx = (uint16_t)(vq->vq_used_cons_idx & + ((vq->vq_nentries >> 1) - 1)); + m = (struct rte_mbuf *)vq->vq_descx[desc_idx++].cookie; + m = __rte_pktmbuf_prefree_seg(m); + if (likely(m != NULL)) { + free[0] = m; + nb_free = 1; + for (i = 1; i < VIRTIO_TX_FREE_NR; i++) { + m = (struct rte_mbuf *)vq->vq_descx[desc_idx++].cookie; + m = __rte_pktmbuf_prefree_seg(m); + if (likely(m != NULL)) { + if (likely(m->pool == free[0]->pool)) + free[nb_free++] = m; + else { + rte_mempool_put_bulk(free[0]->pool, + (void **)free, + RTE_MIN(RTE_DIM(free), + nb_free)); + free[0] = m; + nb_free = 1; + } + } + } + rte_mempool_put_bulk(free[0]->pool, (void **)free, + RTE_MIN(RTE_DIM(free), nb_free)); + } else { + for (i = 1; i < VIRTIO_TX_FREE_NR; i++) { + m = (struct rte_mbuf *)vq->vq_descx[desc_idx++].cookie; + m = __rte_pktmbuf_prefree_seg(m); + if (m != NULL) + rte_mempool_put(m->pool, m); + } + } + + vq->vq_used_cons_idx += VIRTIO_TX_FREE_NR; + vq->vq_free_cnt += (VIRTIO_TX_FREE_NR << 1); +} + +#endif /* _VIRTIO_RXTX_SIMPLE_H_ */ diff --git a/drivers/net/virtio/virtio_rxtx_simple_sse.c b/drivers/net/virtio/virtio_rxtx_simple_sse.c new file mode 100644 index 0000000000..87bb5c6336 --- /dev/null +++ b/drivers/net/virtio/virtio_rxtx_simple_sse.c @@ -0,0 +1,222 @@ +/*- + * BSD LICENSE + * + * Copyright(c) 2010-2015 Intel Corporation. All rights reserved. + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * + * * Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * * Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in + * the documentation and/or other materials provided with the + * distribution. + * * Neither the name of Intel Corporation nor the names of its + * contributors may be used to endorse or promote products derived + * from this software without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS + * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT + * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR + * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT + * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, + * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT + * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE + * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include +#include +#include +#include +#include + +#include + +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include + +#include "virtio_rxtx_simple.h" + +#define RTE_VIRTIO_VPMD_RX_BURST 32 +#define RTE_VIRTIO_DESC_PER_LOOP 8 +#define RTE_VIRTIO_VPMD_RX_REARM_THRESH RTE_VIRTIO_VPMD_RX_BURST + +/* virtio vPMD receive routine, only accept(nb_pkts >= RTE_VIRTIO_DESC_PER_LOOP) + * + * This routine is for non-mergeable RX, one desc for each guest buffer. + * This routine is based on the RX ring layout optimization. Each entry in the + * avail ring points to the desc with the same index in the desc ring and this + * will never be changed in the driver. + * + * - nb_pkts < RTE_VIRTIO_DESC_PER_LOOP, just return no packet + */ +uint16_t +virtio_recv_pkts_vec(void *rx_queue, struct rte_mbuf **rx_pkts, + uint16_t nb_pkts) +{ + struct virtnet_rx *rxvq = rx_queue; + struct virtqueue *vq = rxvq->vq; + uint16_t nb_used; + uint16_t desc_idx; + struct vring_used_elem *rused; + struct rte_mbuf **sw_ring; + struct rte_mbuf **sw_ring_end; + uint16_t nb_pkts_received; + __m128i shuf_msk1, shuf_msk2, len_adjust; + + shuf_msk1 = _mm_set_epi8( + 0xFF, 0xFF, 0xFF, 0xFF, + 0xFF, 0xFF, /* vlan tci */ + 5, 4, /* dat len */ + 0xFF, 0xFF, 5, 4, /* pkt len */ + 0xFF, 0xFF, 0xFF, 0xFF /* packet type */ + + ); + + shuf_msk2 = _mm_set_epi8( + 0xFF, 0xFF, 0xFF, 0xFF, + 0xFF, 0xFF, /* vlan tci */ + 13, 12, /* dat len */ + 0xFF, 0xFF, 13, 12, /* pkt len */ + 0xFF, 0xFF, 0xFF, 0xFF /* packet type */ + ); + + /* Subtract the header length. + * In which case do we need the header length in used->len ? + */ + len_adjust = _mm_set_epi16( + 0, 0, + 0, + (uint16_t)-vq->hw->vtnet_hdr_size, + 0, (uint16_t)-vq->hw->vtnet_hdr_size, + 0, 0); + + if (unlikely(nb_pkts < RTE_VIRTIO_DESC_PER_LOOP)) + return 0; + + nb_used = VIRTQUEUE_NUSED(vq); + + rte_compiler_barrier(); + + if (unlikely(nb_used == 0)) + return 0; + + nb_pkts = RTE_ALIGN_FLOOR(nb_pkts, RTE_VIRTIO_DESC_PER_LOOP); + nb_used = RTE_MIN(nb_used, nb_pkts); + + desc_idx = (uint16_t)(vq->vq_used_cons_idx & (vq->vq_nentries - 1)); + rused = &vq->vq_ring.used->ring[desc_idx]; + sw_ring = &vq->sw_ring[desc_idx]; + sw_ring_end = &vq->sw_ring[vq->vq_nentries]; + + rte_prefetch0(rused); + + if (vq->vq_free_cnt >= RTE_VIRTIO_VPMD_RX_REARM_THRESH) { + virtio_rxq_rearm_vec(rxvq); + if (unlikely(virtqueue_kick_prepare(vq))) + virtqueue_notify(vq); + } + + for (nb_pkts_received = 0; + nb_pkts_received < nb_used;) { + __m128i desc[RTE_VIRTIO_DESC_PER_LOOP / 2]; + __m128i mbp[RTE_VIRTIO_DESC_PER_LOOP / 2]; + __m128i pkt_mb[RTE_VIRTIO_DESC_PER_LOOP]; + + mbp[0] = _mm_loadu_si128((__m128i *)(sw_ring + 0)); + desc[0] = _mm_loadu_si128((__m128i *)(rused + 0)); + _mm_storeu_si128((__m128i *)&rx_pkts[0], mbp[0]); + + mbp[1] = _mm_loadu_si128((__m128i *)(sw_ring + 2)); + desc[1] = _mm_loadu_si128((__m128i *)(rused + 2)); + _mm_storeu_si128((__m128i *)&rx_pkts[2], mbp[1]); + + mbp[2] = _mm_loadu_si128((__m128i *)(sw_ring + 4)); + desc[2] = _mm_loadu_si128((__m128i *)(rused + 4)); + _mm_storeu_si128((__m128i *)&rx_pkts[4], mbp[2]); + + mbp[3] = _mm_loadu_si128((__m128i *)(sw_ring + 6)); + desc[3] = _mm_loadu_si128((__m128i *)(rused + 6)); + _mm_storeu_si128((__m128i *)&rx_pkts[6], mbp[3]); + + pkt_mb[1] = _mm_shuffle_epi8(desc[0], shuf_msk2); + pkt_mb[0] = _mm_shuffle_epi8(desc[0], shuf_msk1); + pkt_mb[1] = _mm_add_epi16(pkt_mb[1], len_adjust); + pkt_mb[0] = _mm_add_epi16(pkt_mb[0], len_adjust); + _mm_storeu_si128((void *)&rx_pkts[1]->rx_descriptor_fields1, + pkt_mb[1]); + _mm_storeu_si128((void *)&rx_pkts[0]->rx_descriptor_fields1, + pkt_mb[0]); + + pkt_mb[3] = _mm_shuffle_epi8(desc[1], shuf_msk2); + pkt_mb[2] = _mm_shuffle_epi8(desc[1], shuf_msk1); + pkt_mb[3] = _mm_add_epi16(pkt_mb[3], len_adjust); + pkt_mb[2] = _mm_add_epi16(pkt_mb[2], len_adjust); + _mm_storeu_si128((void *)&rx_pkts[3]->rx_descriptor_fields1, + pkt_mb[3]); + _mm_storeu_si128((void *)&rx_pkts[2]->rx_descriptor_fields1, + pkt_mb[2]); + + pkt_mb[5] = _mm_shuffle_epi8(desc[2], shuf_msk2); + pkt_mb[4] = _mm_shuffle_epi8(desc[2], shuf_msk1); + pkt_mb[5] = _mm_add_epi16(pkt_mb[5], len_adjust); + pkt_mb[4] = _mm_add_epi16(pkt_mb[4], len_adjust); + _mm_storeu_si128((void *)&rx_pkts[5]->rx_descriptor_fields1, + pkt_mb[5]); + _mm_storeu_si128((void *)&rx_pkts[4]->rx_descriptor_fields1, + pkt_mb[4]); + + pkt_mb[7] = _mm_shuffle_epi8(desc[3], shuf_msk2); + pkt_mb[6] = _mm_shuffle_epi8(desc[3], shuf_msk1); + pkt_mb[7] = _mm_add_epi16(pkt_mb[7], len_adjust); + pkt_mb[6] = _mm_add_epi16(pkt_mb[6], len_adjust); + _mm_storeu_si128((void *)&rx_pkts[7]->rx_descriptor_fields1, + pkt_mb[7]); + _mm_storeu_si128((void *)&rx_pkts[6]->rx_descriptor_fields1, + pkt_mb[6]); + + if (unlikely(nb_used <= RTE_VIRTIO_DESC_PER_LOOP)) { + if (sw_ring + nb_used <= sw_ring_end) + nb_pkts_received += nb_used; + else + nb_pkts_received += sw_ring_end - sw_ring; + break; + } else { + if (unlikely(sw_ring + RTE_VIRTIO_DESC_PER_LOOP >= + sw_ring_end)) { + nb_pkts_received += sw_ring_end - sw_ring; + break; + } else { + nb_pkts_received += RTE_VIRTIO_DESC_PER_LOOP; + + rx_pkts += RTE_VIRTIO_DESC_PER_LOOP; + sw_ring += RTE_VIRTIO_DESC_PER_LOOP; + rused += RTE_VIRTIO_DESC_PER_LOOP; + nb_used -= RTE_VIRTIO_DESC_PER_LOOP; + } + } + } + + vq->vq_used_cons_idx += nb_pkts_received; + vq->vq_free_cnt += nb_pkts_received; + rxvq->stats.packets += nb_pkts_received; + return nb_pkts_received; +} -- 2.20.1