Re: [nmap-svn] r24784 - in nmap-exp/weilin/nmap-nd: . libdnet-stripped/include/dnet libdnet-stripped/src libnetutil

[Djalal Harouni <tixxdz () opendz org>]

On Sun, Jul 10, 2011 at 09:49:00AM -0700, David Fifield wrote:
> Weilin, I think this might be wrong. Unless I'm mistaken, libdnet is
> already getting IPv6 addresses for interfaces. You can see them with
> "nmap --iflist". I think they are typically stored in the
> intf_alias_addrs member.
>
> What might be the problem is that getInterfaceByName returns only the
> first matching interface, which only has a single (usually IPv4)
> address.
David, what do you think if we change the getInterfaceByName(), we can
add a second argument that will take the address family type ? in order
to return the correct interface information.

getInterfaceByName() is used in libnetutil/netutil.cc:
 route_dst_netlink() function.
 route_dst_generic() function.
 

> The problem might just be that the NSE code doesn't look far enough into
> the alias list. If you call nmap.list_interfaces, do you see the
> interfaces with IPv6 addresses?
If we do not change getInterfaceByName() behaviour we can improve the
NSE code, define nmap.list_interfaces() in nse_dnet.cc file instead of
nse_nmaplib.cc (of course register it in nse_nmaplib.cc), this way
nmap.get_interface_info() will simply call list_interfaces() which will
*construct* all the interfaces informations for us, iterate over it and
return the appropriate interface information table according to the
current address family type.


I just found that dnet:ethernet_open() function which is registred in
nse_dnet.cc file and loaded in nse_nmaplib.cc file also calls
getInterfaceByName(), this is another bug, which confirms that we must
update getInterfaceByName() function.

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