On Mon, Oct 21, 2024 at 05:53:17PM +0100, Peter Maydell wrote:
> On Mon, 21 Oct 2024 at 15:54, Daniel P. Berrangé <berra...@redhat.com> wrote:
> >
> > Setting the SO_REUSEADDR property on a socket allows binding to a port
> > number that is in the TIMED_WAIT state. This is usually done on listener
> > sockets, to enable a server to restart itself without having to wait for
> > the completion of TIMED_WAIT on the port.
>
> [...]
>
> > diff --git a/util/qemu-sockets.c b/util/qemu-sockets.c
> > index 60c44b2b56..80594ecad5 100644
> > --- a/util/qemu-sockets.c
> > +++ b/util/qemu-sockets.c
> > @@ -367,7 +367,6 @@ static int inet_connect_addr(const InetSocketAddress
> > *saddr,
> > addr->ai_family);
> > return -1;
> > }
> > - socket_set_fast_reuse(sock);
> >
> > /* connect to peer */
> > do {
>
> We definitely want to keep the socket_set_fast_reuse()
> call in create_fast_reuse_socket() as that function is
> used (only) in the "create socket, listen, bind" server
> socket code path. (Arguably create_fast_reuse_socket()
> is a bit unnecessary as it has only one callsite.)
>
> The one in inet_connect_addr() is clearly wrong as that's
> the client end (fixed in this patch).
>
> How about the call in inet_dgram_saddr() ? I'm not sure how
> SO_REUSEADDR interacts with UDP sockets... (I'm assuming
> the answer is "we need it there" so I'm kind of asking for
> the code-review record really.)
We need the one in inet_dgram_saddr, because there is an
explicit bind() call there, for the situation where the
local UDP address is set.
> In net/socket.c we already set SO_REUSEADDR for dgram
> and for listening sockets but not for client ones, so
> we're now consistent there.
>
> thanks
> -- PMM
>
With regards,
Daniel
--
|: https://berrange.com -o- https://www.flickr.com/photos/dberrange :|
|: https://libvirt.org -o- https://fstop138.berrange.com :|
|: https://entangle-photo.org -o- https://www.instagram.com/dberrange :|
