Re: [PATCH net] vhost_net: do not report a used len larger than receive buffer size
From: "Michael S. Tsirkin" <mst@redhat.com>
Date: 2014-02-12 11:47:29
Also in:
kvm, lkml, netdev
On Wed, Feb 12, 2014 at 01:57:46PM +0800, Jason Wang wrote:
Currently, even if the packet were truncated by lower socket, we still report the packet size as the used len which may confuse guest driver. Fixes this by returning the size of guest receive buffer instead. Fixes 3a4d5c94e959359ece6d6b55045c3f046677f55c (vhost_net: a kernel-level virtio server) Cc: Michael S. Tsirkin <mst@redhat.com> Signed-off-by: Jason Wang <jasowang@redhat.com>
As discussed off-line: NAK and a fixed version will be sent.
quoted hunk ↗ jump to hunk
--- drivers/vhost/net.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)diff --git a/drivers/vhost/net.c b/drivers/vhost/net.c index 9a68409..06268a0 100644 --- a/drivers/vhost/net.c +++ b/drivers/vhost/net.c@@ -525,7 +525,8 @@ static int get_rx_bufs(struct vhost_virtqueue *vq, ++headcount; seg += in; } - heads[headcount - 1].len += datalen; + if (likely(datalen < 0)) + heads[headcount - 1].len += datalen; *iovcount = seg; if (unlikely(log)) *log_num = nlogs;-- 1.8.3.2