From: Jens Freimann Date: Mon, 17 Dec 2018 21:31:35 +0000 (+0100) Subject: net/virtio: implement Rx path for packed queues X-Git-Url: http://git.droids-corp.org/?a=commitdiff_plain;h=a76290c8f1cf;p=dpdk.git net/virtio: implement Rx path for packed queues Implement the receive part. Signed-off-by: Jens Freimann Signed-off-by: Tiwei Bie Reviewed-by: Maxime Coquelin --- diff --git a/drivers/net/virtio/virtio_ethdev.c b/drivers/net/virtio/virtio_ethdev.c index 53dcbb75f0..7b32b29809 100644 --- a/drivers/net/virtio/virtio_ethdev.c +++ b/drivers/net/virtio/virtio_ethdev.c @@ -1365,24 +1365,40 @@ set_rxtx_funcs(struct rte_eth_dev *eth_dev) } } - if (hw->use_simple_rx) { - PMD_INIT_LOG(INFO, "virtio: using simple Rx path on port %u", - eth_dev->data->port_id); - eth_dev->rx_pkt_burst = virtio_recv_pkts_vec; - } else if (hw->use_inorder_rx) { - PMD_INIT_LOG(INFO, - "virtio: using inorder mergeable buffer Rx path on port %u", - eth_dev->data->port_id); - eth_dev->rx_pkt_burst = &virtio_recv_mergeable_pkts_inorder; - } else if (vtpci_with_feature(hw, VIRTIO_NET_F_MRG_RXBUF)) { - PMD_INIT_LOG(INFO, - "virtio: using mergeable buffer Rx path on port %u", - eth_dev->data->port_id); - eth_dev->rx_pkt_burst = &virtio_recv_mergeable_pkts; + if (vtpci_packed_queue(hw)) { + if (vtpci_with_feature(hw, VIRTIO_NET_F_MRG_RXBUF)) { + PMD_INIT_LOG(INFO, + "virtio: using packed ring mergeable buffer Rx path on port %u", + eth_dev->data->port_id); + eth_dev->rx_pkt_burst = + &virtio_recv_mergeable_pkts_packed; + } else { + PMD_INIT_LOG(INFO, + "virtio: using packed ring standard Rx path on port %u", + eth_dev->data->port_id); + eth_dev->rx_pkt_burst = &virtio_recv_pkts_packed; + } } else { - PMD_INIT_LOG(INFO, "virtio: using standard Rx path on port %u", - eth_dev->data->port_id); - eth_dev->rx_pkt_burst = &virtio_recv_pkts; + if (hw->use_simple_rx) { + PMD_INIT_LOG(INFO, "virtio: using simple Rx path on port %u", + eth_dev->data->port_id); + eth_dev->rx_pkt_burst = virtio_recv_pkts_vec; + } else if (hw->use_inorder_rx) { + PMD_INIT_LOG(INFO, + "virtio: using inorder mergeable buffer Rx path on port %u", + eth_dev->data->port_id); + eth_dev->rx_pkt_burst = + &virtio_recv_mergeable_pkts_inorder; + } else if (vtpci_with_feature(hw, VIRTIO_NET_F_MRG_RXBUF)) { + PMD_INIT_LOG(INFO, + "virtio: using mergeable buffer Rx path on port %u", + eth_dev->data->port_id); + eth_dev->rx_pkt_burst = &virtio_recv_mergeable_pkts; + } else { + PMD_INIT_LOG(INFO, "virtio: using standard Rx path on port %u", + eth_dev->data->port_id); + eth_dev->rx_pkt_burst = &virtio_recv_pkts; + } } } @@ -1946,6 +1962,12 @@ virtio_dev_configure(struct rte_eth_dev *dev) } } + if (vtpci_packed_queue(hw)) { + hw->use_simple_rx = 0; + hw->use_inorder_rx = 0; + hw->use_inorder_tx = 0; + } + #if defined RTE_ARCH_ARM64 || defined RTE_ARCH_ARM if (!rte_cpu_get_flag_enabled(RTE_CPUFLAG_NEON)) { hw->use_simple_rx = 0; diff --git a/drivers/net/virtio/virtio_ethdev.h b/drivers/net/virtio/virtio_ethdev.h index 05d3551804..88b8c42a36 100644 --- a/drivers/net/virtio/virtio_ethdev.h +++ b/drivers/net/virtio/virtio_ethdev.h @@ -73,10 +73,15 @@ int virtio_dev_tx_queue_setup_finish(struct rte_eth_dev *dev, uint16_t virtio_recv_pkts(void *rx_queue, struct rte_mbuf **rx_pkts, uint16_t nb_pkts); +uint16_t virtio_recv_pkts_packed(void *rx_queue, struct rte_mbuf **rx_pkts, + uint16_t nb_pkts); uint16_t virtio_recv_mergeable_pkts(void *rx_queue, struct rte_mbuf **rx_pkts, uint16_t nb_pkts); +uint16_t virtio_recv_mergeable_pkts_packed(void *rx_queue, + struct rte_mbuf **rx_pkts, uint16_t nb_pkts); + uint16_t virtio_recv_mergeable_pkts_inorder(void *rx_queue, struct rte_mbuf **rx_pkts, uint16_t nb_pkts); diff --git a/drivers/net/virtio/virtio_rxtx.c b/drivers/net/virtio/virtio_rxtx.c index 9cc7ad2d0c..8564f18a7c 100644 --- a/drivers/net/virtio/virtio_rxtx.c +++ b/drivers/net/virtio/virtio_rxtx.c @@ -31,6 +31,7 @@ #include "virtqueue.h" #include "virtio_rxtx.h" #include "virtio_rxtx_simple.h" +#include "virtio_ring.h" #ifdef RTE_LIBRTE_VIRTIO_DEBUG_DUMP #define VIRTIO_DUMP_PACKET(m, len) rte_pktmbuf_dump(stdout, m, len) @@ -105,6 +106,47 @@ vq_ring_free_id_packed(struct virtqueue *vq, uint16_t id) dxp->next = VQ_RING_DESC_CHAIN_END; } +static uint16_t +virtqueue_dequeue_burst_rx_packed(struct virtqueue *vq, + struct rte_mbuf **rx_pkts, + uint32_t *len, + uint16_t num) +{ + struct rte_mbuf *cookie; + uint16_t used_idx; + uint16_t id; + struct vring_packed_desc *desc; + uint16_t i; + + desc = vq->ring_packed.desc_packed; + + for (i = 0; i < num; i++) { + used_idx = vq->vq_used_cons_idx; + if (!desc_is_used(&desc[used_idx], vq)) + return i; + len[i] = desc[used_idx].len; + id = desc[used_idx].id; + cookie = (struct rte_mbuf *)vq->vq_descx[id].cookie; + if (unlikely(cookie == NULL)) { + PMD_DRV_LOG(ERR, "vring descriptor with no mbuf cookie at %u", + vq->vq_used_cons_idx); + break; + } + rte_prefetch0(cookie); + rte_packet_prefetch(rte_pktmbuf_mtod(cookie, void *)); + rx_pkts[i] = cookie; + + vq->vq_free_cnt++; + vq->vq_used_cons_idx++; + if (vq->vq_used_cons_idx >= vq->vq_nentries) { + vq->vq_used_cons_idx -= vq->vq_nentries; + vq->used_wrap_counter ^= 1; + } + } + + return i; +} + static uint16_t virtqueue_dequeue_burst_rx(struct virtqueue *vq, struct rte_mbuf **rx_pkts, uint32_t *len, uint16_t num) @@ -350,6 +392,51 @@ virtqueue_enqueue_recv_refill(struct virtqueue *vq, struct rte_mbuf *cookie) return 0; } +static inline int +virtqueue_enqueue_recv_refill_packed(struct virtqueue *vq, + struct rte_mbuf **cookie, uint16_t num) +{ + struct vring_packed_desc *start_dp = vq->ring_packed.desc_packed; + uint16_t flags = VRING_DESC_F_WRITE | vq->avail_used_flags; + struct virtio_hw *hw = vq->hw; + struct vq_desc_extra *dxp; + uint16_t idx; + int i; + + if (unlikely(vq->vq_free_cnt == 0)) + return -ENOSPC; + if (unlikely(vq->vq_free_cnt < num)) + return -EMSGSIZE; + + for (i = 0; i < num; i++) { + idx = vq->vq_avail_idx; + dxp = &vq->vq_descx[idx]; + dxp->cookie = (void *)cookie[i]; + dxp->ndescs = 1; + + start_dp[idx].addr = VIRTIO_MBUF_ADDR(cookie[i], vq) + + RTE_PKTMBUF_HEADROOM - hw->vtnet_hdr_size; + start_dp[idx].len = cookie[i]->buf_len - RTE_PKTMBUF_HEADROOM + + hw->vtnet_hdr_size; + + vq->vq_desc_head_idx = dxp->next; + if (vq->vq_desc_head_idx == VQ_RING_DESC_CHAIN_END) + vq->vq_desc_tail_idx = vq->vq_desc_head_idx; + rte_smp_wmb(); + start_dp[idx].flags = flags; + if (++vq->vq_avail_idx >= vq->vq_nentries) { + vq->vq_avail_idx -= vq->vq_nentries; + vq->avail_wrap_counter ^= 1; + vq->avail_used_flags = + VRING_DESC_F_AVAIL(vq->avail_wrap_counter) | + VRING_DESC_F_USED(!vq->avail_wrap_counter); + flags = VRING_DESC_F_WRITE | vq->avail_used_flags; + } + } + vq->vq_free_cnt = (uint16_t)(vq->vq_free_cnt - num); + return 0; +} + /* When doing TSO, the IP length is not included in the pseudo header * checksum of the packet given to the PMD, but for virtio it is * expected. @@ -801,7 +888,11 @@ virtio_dev_rx_queue_setup_finish(struct rte_eth_dev *dev, uint16_t queue_idx) break; /* Enqueue allocated buffers */ - error = virtqueue_enqueue_recv_refill(vq, m); + if (vtpci_packed_queue(vq->hw)) + error = virtqueue_enqueue_recv_refill_packed(vq, + &m, 1); + else + error = virtqueue_enqueue_recv_refill(vq, m); if (error) { rte_pktmbuf_free(m); break; @@ -809,7 +900,8 @@ virtio_dev_rx_queue_setup_finish(struct rte_eth_dev *dev, uint16_t queue_idx) nbufs++; } - vq_update_avail_idx(vq); + if (!vtpci_packed_queue(vq->hw)) + vq_update_avail_idx(vq); } PMD_INIT_LOG(DEBUG, "Allocated %d bufs", nbufs); @@ -896,7 +988,10 @@ virtio_discard_rxbuf(struct virtqueue *vq, struct rte_mbuf *m) * Requeue the discarded mbuf. This should always be * successful since it was just dequeued. */ - error = virtqueue_enqueue_recv_refill(vq, m); + if (vtpci_packed_queue(vq->hw)) + error = virtqueue_enqueue_recv_refill_packed(vq, &m, 1); + else + error = virtqueue_enqueue_recv_refill(vq, m); if (unlikely(error)) { RTE_LOG(ERR, PMD, "cannot requeue discarded mbuf"); @@ -1136,6 +1231,104 @@ virtio_recv_pkts(void *rx_queue, struct rte_mbuf **rx_pkts, uint16_t nb_pkts) return nb_rx; } +uint16_t +virtio_recv_pkts_packed(void *rx_queue, struct rte_mbuf **rx_pkts, + uint16_t nb_pkts) +{ + struct virtnet_rx *rxvq = rx_queue; + struct virtqueue *vq = rxvq->vq; + struct virtio_hw *hw = vq->hw; + struct rte_mbuf *rxm, *new_mbuf; + uint16_t num, nb_rx; + uint32_t len[VIRTIO_MBUF_BURST_SZ]; + struct rte_mbuf *rcv_pkts[VIRTIO_MBUF_BURST_SZ]; + int error; + uint32_t i, nb_enqueued; + uint32_t hdr_size; + struct virtio_net_hdr *hdr; + + nb_rx = 0; + if (unlikely(hw->started == 0)) + return nb_rx; + + num = RTE_MIN(VIRTIO_MBUF_BURST_SZ, nb_pkts); + if (likely(num > DESC_PER_CACHELINE)) + num = num - ((vq->vq_used_cons_idx + num) % DESC_PER_CACHELINE); + + num = virtqueue_dequeue_burst_rx_packed(vq, rcv_pkts, len, num); + PMD_RX_LOG(DEBUG, "dequeue:%d", num); + + nb_enqueued = 0; + hdr_size = hw->vtnet_hdr_size; + + for (i = 0; i < num; i++) { + rxm = rcv_pkts[i]; + + PMD_RX_LOG(DEBUG, "packet len:%d", len[i]); + + if (unlikely(len[i] < hdr_size + ETHER_HDR_LEN)) { + PMD_RX_LOG(ERR, "Packet drop"); + nb_enqueued++; + virtio_discard_rxbuf(vq, rxm); + rxvq->stats.errors++; + continue; + } + + rxm->port = rxvq->port_id; + rxm->data_off = RTE_PKTMBUF_HEADROOM; + rxm->ol_flags = 0; + rxm->vlan_tci = 0; + + rxm->pkt_len = (uint32_t)(len[i] - hdr_size); + rxm->data_len = (uint16_t)(len[i] - hdr_size); + + hdr = (struct virtio_net_hdr *)((char *)rxm->buf_addr + + RTE_PKTMBUF_HEADROOM - hdr_size); + + if (hw->vlan_strip) + rte_vlan_strip(rxm); + + if (hw->has_rx_offload && virtio_rx_offload(rxm, hdr) < 0) { + virtio_discard_rxbuf(vq, rxm); + rxvq->stats.errors++; + continue; + } + + virtio_rx_stats_updated(rxvq, rxm); + + rx_pkts[nb_rx++] = rxm; + } + + rxvq->stats.packets += nb_rx; + + /* Allocate new mbuf for the used descriptor */ + while (likely(!virtqueue_full(vq))) { + new_mbuf = rte_mbuf_raw_alloc(rxvq->mpool); + if (unlikely(new_mbuf == NULL)) { + struct rte_eth_dev *dev = + &rte_eth_devices[rxvq->port_id]; + dev->data->rx_mbuf_alloc_failed++; + break; + } + error = virtqueue_enqueue_recv_refill_packed(vq, &new_mbuf, 1); + if (unlikely(error)) { + rte_pktmbuf_free(new_mbuf); + break; + } + nb_enqueued++; + } + + if (likely(nb_enqueued)) { + if (unlikely(virtqueue_kick_prepare_packed(vq))) { + virtqueue_notify(vq); + PMD_RX_LOG(DEBUG, "Notified"); + } + } + + return nb_rx; +} + + uint16_t virtio_recv_mergeable_pkts_inorder(void *rx_queue, struct rte_mbuf **rx_pkts, @@ -1493,6 +1686,182 @@ virtio_recv_mergeable_pkts(void *rx_queue, return nb_rx; } +uint16_t +virtio_recv_mergeable_pkts_packed(void *rx_queue, + struct rte_mbuf **rx_pkts, + uint16_t nb_pkts) +{ + struct virtnet_rx *rxvq = rx_queue; + struct virtqueue *vq = rxvq->vq; + struct virtio_hw *hw = vq->hw; + struct rte_mbuf *rxm; + struct rte_mbuf *prev = NULL; + uint16_t num, nb_rx = 0; + uint32_t len[VIRTIO_MBUF_BURST_SZ]; + struct rte_mbuf *rcv_pkts[VIRTIO_MBUF_BURST_SZ]; + uint32_t nb_enqueued = 0; + uint32_t seg_num = 0; + uint32_t seg_res = 0; + uint32_t hdr_size = hw->vtnet_hdr_size; + int32_t i; + int error; + + if (unlikely(hw->started == 0)) + return nb_rx; + + + num = nb_pkts; + if (unlikely(num > VIRTIO_MBUF_BURST_SZ)) + num = VIRTIO_MBUF_BURST_SZ; + if (likely(num > DESC_PER_CACHELINE)) + num = num - ((vq->vq_used_cons_idx + num) % DESC_PER_CACHELINE); + + num = virtqueue_dequeue_burst_rx_packed(vq, rcv_pkts, len, num); + + for (i = 0; i < num; i++) { + struct virtio_net_hdr_mrg_rxbuf *header; + + PMD_RX_LOG(DEBUG, "dequeue:%d", num); + PMD_RX_LOG(DEBUG, "packet len:%d", len[i]); + + rxm = rcv_pkts[i]; + + if (unlikely(len[i] < hdr_size + ETHER_HDR_LEN)) { + PMD_RX_LOG(ERR, "Packet drop"); + nb_enqueued++; + virtio_discard_rxbuf(vq, rxm); + rxvq->stats.errors++; + continue; + } + + header = (struct virtio_net_hdr_mrg_rxbuf *)((char *) + rxm->buf_addr + RTE_PKTMBUF_HEADROOM - hdr_size); + seg_num = header->num_buffers; + + if (seg_num == 0) + seg_num = 1; + + rxm->data_off = RTE_PKTMBUF_HEADROOM; + rxm->nb_segs = seg_num; + rxm->ol_flags = 0; + rxm->vlan_tci = 0; + rxm->pkt_len = (uint32_t)(len[i] - hdr_size); + rxm->data_len = (uint16_t)(len[i] - hdr_size); + + rxm->port = rxvq->port_id; + rx_pkts[nb_rx] = rxm; + prev = rxm; + + if (hw->has_rx_offload && + virtio_rx_offload(rxm, &header->hdr) < 0) { + virtio_discard_rxbuf(vq, rxm); + rxvq->stats.errors++; + continue; + } + + if (hw->vlan_strip) + rte_vlan_strip(rx_pkts[nb_rx]); + + seg_res = seg_num - 1; + + /* Merge remaining segments */ + while (seg_res != 0 && i < (num - 1)) { + i++; + + rxm = rcv_pkts[i]; + rxm->data_off = RTE_PKTMBUF_HEADROOM - hdr_size; + rxm->pkt_len = (uint32_t)(len[i]); + rxm->data_len = (uint16_t)(len[i]); + + rx_pkts[nb_rx]->pkt_len += (uint32_t)(len[i]); + rx_pkts[nb_rx]->data_len += (uint16_t)(len[i]); + + if (prev) + prev->next = rxm; + + prev = rxm; + seg_res -= 1; + } + + if (!seg_res) { + virtio_rx_stats_updated(rxvq, rx_pkts[nb_rx]); + nb_rx++; + } + } + + /* Last packet still need merge segments */ + while (seg_res != 0) { + uint16_t rcv_cnt = RTE_MIN((uint16_t)seg_res, + VIRTIO_MBUF_BURST_SZ); + if (likely(vq->vq_free_cnt >= rcv_cnt)) { + num = virtqueue_dequeue_burst_rx_packed(vq, rcv_pkts, + len, rcv_cnt); + uint16_t extra_idx = 0; + + rcv_cnt = num; + + while (extra_idx < rcv_cnt) { + rxm = rcv_pkts[extra_idx]; + + rxm->data_off = + RTE_PKTMBUF_HEADROOM - hdr_size; + rxm->pkt_len = (uint32_t)(len[extra_idx]); + rxm->data_len = (uint16_t)(len[extra_idx]); + + prev->next = rxm; + prev = rxm; + rx_pkts[nb_rx]->pkt_len += len[extra_idx]; + rx_pkts[nb_rx]->data_len += len[extra_idx]; + extra_idx += 1; + } + seg_res -= rcv_cnt; + if (!seg_res) { + virtio_rx_stats_updated(rxvq, rx_pkts[nb_rx]); + nb_rx++; + } + } else { + PMD_RX_LOG(ERR, + "No enough segments for packet."); + if (prev) + virtio_discard_rxbuf(vq, prev); + rxvq->stats.errors++; + break; + } + } + + rxvq->stats.packets += nb_rx; + + /* Allocate new mbuf for the used descriptor */ + if (likely(!virtqueue_full(vq))) { + /* free_cnt may include mrg descs */ + uint16_t free_cnt = vq->vq_free_cnt; + struct rte_mbuf *new_pkts[free_cnt]; + + if (!rte_pktmbuf_alloc_bulk(rxvq->mpool, new_pkts, free_cnt)) { + error = virtqueue_enqueue_recv_refill_packed(vq, + new_pkts, free_cnt); + if (unlikely(error)) { + for (i = 0; i < free_cnt; i++) + rte_pktmbuf_free(new_pkts[i]); + } + nb_enqueued += free_cnt; + } else { + struct rte_eth_dev *dev = + &rte_eth_devices[rxvq->port_id]; + dev->data->rx_mbuf_alloc_failed += free_cnt; + } + } + + if (likely(nb_enqueued)) { + if (unlikely(virtqueue_kick_prepare_packed(vq))) { + virtqueue_notify(vq); + PMD_RX_LOG(DEBUG, "Notified"); + } + } + + return nb_rx; +} + uint16_t virtio_xmit_pkts_packed(void *tx_queue, struct rte_mbuf **tx_pkts, uint16_t nb_pkts) diff --git a/drivers/net/virtio/virtqueue.c b/drivers/net/virtio/virtqueue.c index 56a77cc71d..5b03f7a27b 100644 --- a/drivers/net/virtio/virtqueue.c +++ b/drivers/net/virtio/virtqueue.c @@ -54,9 +54,36 @@ virtqueue_detach_unused(struct virtqueue *vq) return NULL; } +/* Flush used descs */ +static void +virtqueue_rxvq_flush_packed(struct virtqueue *vq) +{ + struct vq_desc_extra *dxp; + uint16_t i; + + struct vring_packed_desc *descs = vq->ring_packed.desc_packed; + int cnt = 0; + + i = vq->vq_used_cons_idx; + while (desc_is_used(&descs[i], vq) && cnt++ < vq->vq_nentries) { + dxp = &vq->vq_descx[descs[i].id]; + if (dxp->cookie != NULL) { + rte_pktmbuf_free(dxp->cookie); + dxp->cookie = NULL; + } + vq->vq_free_cnt++; + vq->vq_used_cons_idx++; + if (vq->vq_used_cons_idx >= vq->vq_nentries) { + vq->vq_used_cons_idx -= vq->vq_nentries; + vq->used_wrap_counter ^= 1; + } + i = vq->vq_used_cons_idx; + } +} + /* Flush the elements in the used ring. */ -void -virtqueue_rxvq_flush(struct virtqueue *vq) +static void +virtqueue_rxvq_flush_split(struct virtqueue *vq) { struct virtnet_rx *rxq = &vq->rxq; struct virtio_hw *hw = vq->hw; @@ -102,3 +129,15 @@ virtqueue_rxvq_flush(struct virtqueue *vq) } } } + +/* Flush the elements in the used ring. */ +void +virtqueue_rxvq_flush(struct virtqueue *vq) +{ + struct virtio_hw *hw = vq->hw; + + if (vtpci_packed_queue(hw)) + virtqueue_rxvq_flush_packed(vq); + else + virtqueue_rxvq_flush_split(vq); +}