From mboxrd@z Thu Jan 1 00:00:00 1970 From: Tiwei Bie Subject: Re: [PATCH] vhost: fix notification for packed ring Date: Thu, 11 Oct 2018 22:10:28 +0800 Message-ID: <20181011141028.GA11493@debian> References: <20181011130655.5485-1-tiwei.bie@intel.com> <20181011093628-mutt-send-email-mst@kernel.org> Mime-Version: 1.0 Content-Type: text/plain; charset=utf-8 Content-Transfer-Encoding: 8bit Cc: Jason Wang , maxime.coquelin@redhat.com, zhihong.wang@intel.com, dev@dpdk.org, stable@dpdk.org To: "Michael S. Tsirkin" Return-path: Content-Disposition: inline In-Reply-To: <20181011093628-mutt-send-email-mst@kernel.org> List-Id: DPDK patches and discussions List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: dev-bounces@dpdk.org Sender: "dev" On Thu, Oct 11, 2018 at 09:36:48AM -0400, Michael S. Tsirkin wrote: > On Thu, Oct 11, 2018 at 09:34:06PM +0800, Jason Wang wrote: > > > > > > On 2018年10月11日 21:06, Tiwei Bie wrote: > > > The notification can't be disabled in packed ring when > > > application tries to disable notification, because the > > > device event flags field is overwritten by an unexpected > > > value. This patch fixes this issue. > > > > > > Fixes: b1cce26af1dc ("vhost: add notification for packed ring") > > > Cc: stable@dpdk.org > > > > > > Signed-off-by: Tiwei Bie > > > --- > > > lib/librte_vhost/vhost.c | 7 +++++-- > > > 1 file changed, 5 insertions(+), 2 deletions(-) > > > > > > diff --git a/lib/librte_vhost/vhost.c b/lib/librte_vhost/vhost.c > > > index e62f4c594..c9270bdec 100644 > > > --- a/lib/librte_vhost/vhost.c > > > +++ b/lib/librte_vhost/vhost.c > > > @@ -667,8 +667,10 @@ vhost_enable_notify_packed(struct virtio_net *dev, > > > { > > > uint16_t flags; > > > - if (!enable) > > > - vq->device_event->flags = VRING_EVENT_F_DISABLE; > > > + if (!enable) { > > > + flags = VRING_EVENT_F_DISABLE; > > > + goto out; > > > + } > > > flags = VRING_EVENT_F_ENABLE; > > > if (dev->features & (1ULL << VIRTIO_RING_F_EVENT_IDX)) { > > > @@ -677,6 +679,7 @@ vhost_enable_notify_packed(struct virtio_net *dev, > > > vq->avail_wrap_counter << 15; > > > } > > > +out: > > > rte_smp_wmb(); > > > > It looks to me the barrier is even not needed for !enable. > > > > Thanks > > Good point, I agree. > Besides that > > Reviewed-by: Michael S. Tsirkin > Yeah, it's better to return directly after disabling the interrupt. I will send a new version to do this change. Thanks! > > > > > vq->device_event->flags = flags;