> -----Original Message----- > From: Linhaifeng [mailto:haifeng.lin at huawei.com] > Sent: Friday, April 24, 2015 9:01 AM > To: Xie, Huawei; dev at dpdk.org > Cc: luke at snabb.co; mst at redhat.com > Subject: Re: [PATCH] vhost: flush used->idx update before reading avail- > >flags > > > > On 2015/4/23 0:33, Huawei Xie wrote: > > update of used->idx and read of avail->flags could be reordered. > > memory fence should be used to ensure the order, otherwise guest could > see a stale used->idx value after it toggles the interrupt suppression flag. > > > > Signed-off-by: Huawei Xie <huawei.xie at intel.com> > > --- > > lib/librte_vhost/vhost_rxtx.c | 3 +++ > > 1 file changed, 3 insertions(+) > > > > diff --git a/lib/librte_vhost/vhost_rxtx.c b/lib/librte_vhost/vhost_rxtx.c > > index 510ffe8..6afba35 100644 > > --- a/lib/librte_vhost/vhost_rxtx.c > > +++ b/lib/librte_vhost/vhost_rxtx.c > > @@ -178,6 +178,9 @@ virtio_dev_rx(struct virtio_net *dev, uint16_t > queue_id, > > *(volatile uint16_t *)&vq->used->idx += count; > > vq->last_used_idx = res_end_idx; > > > > + /* flush used->idx update before we read avail->flags. */ > > + rte_mb(); > > + > > /* Kick the guest if necessary. */ > > if (!(vq->avail->flags & VRING_AVAIL_F_NO_INTERRUPT)) > > eventfd_write((int)vq->callfd, 1); > > > > If not add memory fence what would happen? Packets loss or interrupt > loss?How to test it ?
If those two store and load are reordered, guest toggles the interrupt suppression flag, then it checks that there is no more work to do. Actually there is.
