Re: [NSE] Dual-stack (IPv4/6) bug/feature?

[David Fifield <david () bamsoftware com>]

On Wed, Jun 06, 2012 at 02:49:49PM -0500, Daniel Miller wrote:
> I tested this using this tiny test script:
>
> > portrule = shortport.http
> >
> > action = function(host, port)
> >  local uri = "/"
> >  local ret = http.get(host.targetname, port.number, uri)
> >  return ret['status-line']
> > end
>
> I can confirm that on Linux 3.2.0, Nmap tries to connect via IPv6
> for this script when a AAAA record is available, even when a IPv4
> scan is requested. However, changing the http.get line to this:
>
> local ret = http.get(host, port, uri)
>
> ...fixes the issue. Remains to be seen whether a check could be put
> in place (guessing in the Nsock bindings) to prevent this issue for
> other cases.

http.get(host, port, uri) is the better way to make the request.

Maybe we should add a getaddrinfo hint for ai_family=o.af() in
l_connect?

David Fifield
_______________________________________________
Sent through the nmap-dev mailing list
http://cgi.insecure.org/mailman/listinfo/nmap-dev
Archived at http://seclists.org/nmap-dev/