From patchwork Fri Oct 14 08:07:07 2016 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Maxime Coquelin X-Patchwork-Id: 16574 X-Patchwork-Delegate: yuanhan.liu@linux.intel.com Return-Path: X-Original-To: patchwork@dpdk.org Delivered-To: patchwork@dpdk.org Received: from [92.243.14.124] (localhost [IPv6:::1]) by dpdk.org (Postfix) with ESMTP id 1E4B35934; Fri, 14 Oct 2016 10:07:18 +0200 (CEST) Received: from mx1.redhat.com (mx1.redhat.com [209.132.183.28]) by dpdk.org (Postfix) with ESMTP id 88F33590F for ; Fri, 14 Oct 2016 10:07:16 +0200 (CEST) Received: from int-mx13.intmail.prod.int.phx2.redhat.com (int-mx13.intmail.prod.int.phx2.redhat.com [10.5.11.26]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by mx1.redhat.com (Postfix) with ESMTPS id B4A07C05A286; Fri, 14 Oct 2016 08:07:15 +0000 (UTC) Received: from max-t460s.redhat.com (vpn1-7-225.ams2.redhat.com [10.36.7.225]) by int-mx13.intmail.prod.int.phx2.redhat.com (8.14.4/8.14.4) with ESMTP id u9E87DiT002381; Fri, 14 Oct 2016 04:07:13 -0400 From: Maxime Coquelin To: yuanhan.liu@linux.intel.com, dev@dpdk.org Cc: mst@redhat.com, jianfeng.tan@intel.com, olivier.matz@6wind.com, stephen@networkplumber.org, Maxime Coquelin Date: Fri, 14 Oct 2016 10:07:07 +0200 Message-Id: <1476432427-2724-1-git-send-email-maxime.coquelin@redhat.com> In-Reply-To: <1475773241-5714-1-git-send-email-maxime.coquelin@redhat.com> References: <1475773241-5714-1-git-send-email-maxime.coquelin@redhat.com> X-Scanned-By: MIMEDefang 2.68 on 10.5.11.26 X-Greylist: Sender IP whitelisted, not delayed by milter-greylist-4.5.16 (mx1.redhat.com [10.5.110.32]); Fri, 14 Oct 2016 08:07:16 +0000 (UTC) Subject: [dpdk-dev] [PATCH v3] vhost: Only access header if offloading is supported in dequeue path X-BeenThere: dev@dpdk.org X-Mailman-Version: 2.1.15 Precedence: list List-Id: patches and discussions about DPDK List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: dev-bounces@dpdk.org Sender: "dev" If offloading features are not negotiated, parsing the virtio header is not needed. Micro-benchmark with testpmd shows that the gain is +4% with indirect descriptors, +1% when using direct descriptors. Signed-off-by: Maxime Coquelin --- Changes since v2: ================= - Simplify code by translating first desc address unconditionnaly (Yuanhan) - Instead of checking features again, check whether hdr has been assign to call offload function. Changes since v1: ================= - Rebased - Fix early out check in vhost_dequeue_offload lib/librte_vhost/virtio_net.c | 33 +++++++++++++++++++++++++-------- 1 file changed, 25 insertions(+), 8 deletions(-) diff --git a/lib/librte_vhost/virtio_net.c b/lib/librte_vhost/virtio_net.c index 812e5d3..15ef0b0 100644 --- a/lib/librte_vhost/virtio_net.c +++ b/lib/librte_vhost/virtio_net.c @@ -555,6 +555,18 @@ rte_vhost_enqueue_burst(int vid, uint16_t queue_id, return virtio_dev_rx(dev, queue_id, pkts, count); } +static inline bool +virtio_net_with_host_offload(struct virtio_net *dev) +{ + if (dev->features & + (VIRTIO_NET_F_CSUM | VIRTIO_NET_F_HOST_ECN | + VIRTIO_NET_F_HOST_TSO4 | VIRTIO_NET_F_HOST_TSO6 | + VIRTIO_NET_F_HOST_UFO)) + return true; + + return false; +} + static void parse_ethernet(struct rte_mbuf *m, uint16_t *l4_proto, void **l4_hdr) { @@ -607,6 +619,9 @@ vhost_dequeue_offload(struct virtio_net_hdr *hdr, struct rte_mbuf *m) void *l4_hdr = NULL; struct tcp_hdr *tcp_hdr = NULL; + if (hdr->flags == 0 && hdr->gso_type == VIRTIO_NET_HDR_GSO_NONE) + 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)) { @@ -702,7 +717,7 @@ copy_desc_to_mbuf(struct virtio_net *dev, struct vring_desc *descs, uint32_t mbuf_avail, mbuf_offset; uint32_t cpy_len; struct rte_mbuf *cur = m, *prev = m; - struct virtio_net_hdr *hdr; + struct virtio_net_hdr *hdr = NULL; /* A counter to avoid desc dead loop chain */ uint32_t nr_desc = 1; @@ -715,8 +730,10 @@ copy_desc_to_mbuf(struct virtio_net *dev, struct vring_desc *descs, if (unlikely(!desc_addr)) return -1; - hdr = (struct virtio_net_hdr *)((uintptr_t)desc_addr); - rte_prefetch0(hdr); + if (virtio_net_with_host_offload(dev)) { + hdr = (struct virtio_net_hdr *)((uintptr_t)desc_addr); + rte_prefetch0(hdr); + } /* * A virtio driver normally uses at least 2 desc buffers @@ -733,18 +750,18 @@ copy_desc_to_mbuf(struct virtio_net *dev, struct vring_desc *descs, if (unlikely(!desc_addr)) return -1; - rte_prefetch0((void *)(uintptr_t)desc_addr); - desc_offset = 0; desc_avail = desc->len; nr_desc += 1; - - PRINT_PACKET(dev, (uintptr_t)desc_addr, desc->len, 0); } else { desc_avail = desc->len - dev->vhost_hlen; desc_offset = dev->vhost_hlen; } + rte_prefetch0((void *)(uintptr_t)(desc_addr + desc_offset)); + + PRINT_PACKET(dev, (uintptr_t)(desc_addr + desc_offset), desc_avail, 0); + mbuf_offset = 0; mbuf_avail = m->buf_len - RTE_PKTMBUF_HEADROOM; while (1) { @@ -831,7 +848,7 @@ copy_desc_to_mbuf(struct virtio_net *dev, struct vring_desc *descs, prev->data_len = mbuf_offset; m->pkt_len += mbuf_offset; - if (hdr->flags != 0 || hdr->gso_type != VIRTIO_NET_HDR_GSO_NONE) + if (hdr) vhost_dequeue_offload(hdr, m); return 0;