From patchwork Mon Dec 11 05:13:29 2017 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Tiwei Bie X-Patchwork-Id: 32061 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 [127.0.0.1]) by dpdk.org (Postfix) with ESMTP id E9C551AEF0; Mon, 11 Dec 2017 06:14:59 +0100 (CET) Received: from mga18.intel.com (mga18.intel.com [134.134.136.126]) by dpdk.org (Postfix) with ESMTP id 5DDB37D52; Mon, 11 Dec 2017 06:14:57 +0100 (CET) X-Amp-Result: SKIPPED(no attachment in message) X-Amp-File-Uploaded: False Received: from orsmga005.jf.intel.com ([10.7.209.41]) by orsmga106.jf.intel.com with ESMTP/TLS/DHE-RSA-AES256-GCM-SHA384; 10 Dec 2017 21:14:56 -0800 X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="5.45,391,1508828400"; d="scan'208";a="183131801" Received: from deepin-15.sh.intel.com (HELO debian-xvivbkq.sh.intel.com) ([10.67.104.165]) by orsmga005.jf.intel.com with ESMTP; 10 Dec 2017 21:14:55 -0800 From: Tiwei Bie To: yliu@fridaylinux.org, maxime.coquelin@redhat.com, dev@dpdk.org Cc: stable@dpdk.org Date: Mon, 11 Dec 2017 13:13:29 +0800 Message-Id: <20171211051332.31888-2-tiwei.bie@intel.com> X-Mailer: git-send-email 2.13.3 In-Reply-To: <20171211051332.31888-1-tiwei.bie@intel.com> References: <20171207053059.19487-1-tiwei.bie@intel.com> <20171211051332.31888-1-tiwei.bie@intel.com> Subject: [dpdk-dev] [PATCH v2 1/4] net/virtio: fix vector Rx break caused by rxq flushing X-BeenThere: dev@dpdk.org X-Mailman-Version: 2.1.15 Precedence: list List-Id: DPDK patches and discussions List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: dev-bounces@dpdk.org Sender: "dev" The vector Rx will be broken if backend has consumed all the descs in the avail ring before the device is started. Because in current implementation, vector Rx will return immediately without refilling the avail ring if the used ring is empty. So we have to refill the avail ring after flushing the elements in the used ring for vector Rx. Besides, vector Rx has a different ring layout assumption and mbuf management. So we need to handle it differently. Fixes: d8227497ec5c ("net/virtio: flush Rx queues on start") Cc: stable@dpdk.org Reported-by: Antonio Fischetti Signed-off-by: Tiwei Bie Reviewed-by: Maxime Coquelin Tested-by: Antonio Fischetti --- drivers/net/virtio/virtio_ethdev.c | 2 +- drivers/net/virtio/virtqueue.c | 31 ++++++++++++++++++++++++------- drivers/net/virtio/virtqueue.h | 2 +- 3 files changed, 26 insertions(+), 9 deletions(-) diff --git a/drivers/net/virtio/virtio_ethdev.c b/drivers/net/virtio/virtio_ethdev.c index e0328f61d..64a0cc608 100644 --- a/drivers/net/virtio/virtio_ethdev.c +++ b/drivers/net/virtio/virtio_ethdev.c @@ -1860,7 +1860,7 @@ virtio_dev_start(struct rte_eth_dev *dev) for (i = 0; i < dev->data->nb_rx_queues; i++) { rxvq = dev->data->rx_queues[i]; /* Flush the old packets */ - virtqueue_flush(rxvq->vq); + virtqueue_rxvq_flush(rxvq->vq); virtqueue_notify(rxvq->vq); } diff --git a/drivers/net/virtio/virtqueue.c b/drivers/net/virtio/virtqueue.c index c3a536f8a..696d0e4a4 100644 --- a/drivers/net/virtio/virtqueue.c +++ b/drivers/net/virtio/virtqueue.c @@ -37,6 +37,7 @@ #include "virtqueue.h" #include "virtio_logs.h" #include "virtio_pci.h" +#include "virtio_rxtx_simple.h" /* * Two types of mbuf to be cleaned: @@ -62,8 +63,10 @@ virtqueue_detatch_unused(struct virtqueue *vq) /* Flush the elements in the used ring. */ void -virtqueue_flush(struct virtqueue *vq) +virtqueue_rxvq_flush(struct virtqueue *vq) { + struct virtnet_rx *rxq = &vq->rxq; + struct virtio_hw *hw = vq->hw; struct vring_used_elem *uep; struct vq_desc_extra *dxp; uint16_t used_idx, desc_idx; @@ -74,13 +77,27 @@ virtqueue_flush(struct virtqueue *vq) for (i = 0; i < nb_used; i++) { used_idx = vq->vq_used_cons_idx & (vq->vq_nentries - 1); uep = &vq->vq_ring.used->ring[used_idx]; - desc_idx = (uint16_t)uep->id; - dxp = &vq->vq_descx[desc_idx]; - if (dxp->cookie != NULL) { - rte_pktmbuf_free(dxp->cookie); - dxp->cookie = NULL; + if (hw->use_simple_rx) { + desc_idx = used_idx; + rte_pktmbuf_free(vq->sw_ring[desc_idx]); + vq->vq_free_cnt++; + } else { + desc_idx = (uint16_t)uep->id; + dxp = &vq->vq_descx[desc_idx]; + if (dxp->cookie != NULL) { + rte_pktmbuf_free(dxp->cookie); + dxp->cookie = NULL; + } + vq_ring_free_chain(vq, desc_idx); } vq->vq_used_cons_idx++; - vq_ring_free_chain(vq, desc_idx); + } + + if (hw->use_simple_rx) { + while (vq->vq_free_cnt >= RTE_VIRTIO_VPMD_RX_REARM_THRESH) { + virtio_rxq_rearm_vec(rxq); + if (virtqueue_kick_prepare(vq)) + virtqueue_notify(vq); + } } } diff --git a/drivers/net/virtio/virtqueue.h b/drivers/net/virtio/virtqueue.h index 2305d91a4..ab466c2db 100644 --- a/drivers/net/virtio/virtqueue.h +++ b/drivers/net/virtio/virtqueue.h @@ -304,7 +304,7 @@ void virtqueue_dump(struct virtqueue *vq); struct rte_mbuf *virtqueue_detatch_unused(struct virtqueue *vq); /* Flush the elements in the used ring. */ -void virtqueue_flush(struct virtqueue *vq); +void virtqueue_rxvq_flush(struct virtqueue *vq); static inline int virtqueue_full(const struct virtqueue *vq)