On Tue, Jan 06, 2026 at 08:48:31AM +0100, Laurent Vivier wrote:
> On 1/6/26 00:28, David Gibson wrote:
> > On Mon, Jan 05, 2026 at 03:50:16PM +0100, Laurent Vivier wrote:
> > > During vhost-user device initialization, UDP datagrams may arrive on
> > > listening sockets before the guest has enabled the RX virtqueue.
> > > 
> > > When this happens, udp_vu_sock_recv() returns 0 without consuming
> > > the datagram from the socket. The caller, udp_sock_fwd(), uses a
> > > while loop with udp_peek_addr() to process pending datagrams. Since
> > > the datagram remains in the socket buffer, udp_peek_addr() keeps
> > > returning data available, causing a busy loop with 100% CPU usage.
> > > 
> > > Add an early check for virtqueue readiness in udp_vu_sock_to_tap(),
> > > mirroring tcp_vu_data_from_sock(). When the queue is not enabled or
> > > not started, explicitly discard the datagram with recvmsg() and
> > > return. The caller will drain remaining datagrams through repeated
> > > calls.
> > > 
> > > Fixes: 28997fcb29b5 ("vhost-user: add vhost-user")
> > > Link: https://bugs.passt.top/show_bug.cgi?id=185
> > > Signed-off-by: Laurent Vivier <lvivier@redhat.com>
> > 
> > My only concern here is that this will only consume a single datagram,
> > whereas it looks like the loop could consume multiple datagrams.  I'm
> > guessing something higher up will keep calling this until the queue is
> > empty, but I'm not certain.  Otherwise LGTM.
> 
> I can move the change into udp_vu_sock_recv() (inside the datagram loop), it
> will be closer to what udp_buf_sock_to_tap() does: reads data with
> udp_sock_recv() and if fd_tap is not initialized tap_send_frames() drops
> them.

That seems a little safer (or at least, more obviously safe) to me.

-- 
David Gibson (he or they)	| I'll have my music baroque, and my code
david AT gibson.dropbear.id.au	| minimalist, thank you, not the other way
				| around.
http://www.ozlabs.org/~dgibson