Re: [PATCH 0/6] MR1111: winsock: Always use Unix SO_REUSEADDR on TCP sockets.

[Paul Gofman]

On 10/25/22 19:04, Zebediah Figura (@zfigura) wrote:
> Zebediah Figura (@zfigura) commented about server/sock.c:
> > -
> >
> > +#ifdef IPV6_V6ONLY
> > - if (sock->family == WS_AF_INET6)
> > - {
> > -
> >    socklen_t len = sizeof(*v6only);
> > -
> > -
> >    getsockopt( get_unix_fd(sock->fd), IPPROTO_IPV6, IPV6_V6ONLY, v6only, &len );
> > - }
> >
> > +#endif
> > -
> > - if (!is_blocking_addr( sock, addr )) return 0;
> > -
> > - if ((entry = rb_get( &bound_addresses_tree, addr )))
> > - {
> > -
> >    bound_addr = WINE_RB_ENTRY_VALUE(entry, struct bound_addr, entry);
> > -
> >    if (bound_addr->reuse_count == -1 || !sock->reuseaddr) return 1;
>
> This implies that if we bind a socket without REUSEADDR, and then one with REUSEADDR, then we get STATUS_ACCESS_DENIED (i.e. WSAEACCES). That doesn't match what [1] says, and I don't think we have tests for it either. (Nor tests for binding the first socket with REUSEADDR and the second without, for that matter.)

Eh, I guess that was exactly the case at least in the 
test_so_reuseaddr() before my changes. Should I probably bring this case 
back additionally?