Re: [NSE] Raw sockets support

[majek04 <nmap () forest one pl>]

majek04 wrote:
> Diman Todorov wrote:
> > I personally think that modifying nsock is preferable to introducing  
> > a thread library dependence into nmap.
>
> In summary we can add pcap support to nsock or
> use some ugly and not portable hack.

Hi again.

I have next question.
I'm not pcap expert yet.

As far as I understand Fyodor's patch http://seclists.org/nmap-dev/2006/q4/att-0128/nmap-osx-fix_diff
means that on some architectures we must use blocking pcap_next.

Integrating some pcap/dnet features into nsock makes sense only if pcap descriptors
could be used in the same 'select()' as nsock descriptors.

If I'm not wrong we're left with only two options
 -  running pcap_next in other thread than nsock_loop or
 -  running pcap_next after nsock_loop with very short timeouts.
    The problem here is that pcap timeouts are broken. The alternative
    would be to set pcap descriptor in nonblocking mode. But I don't
    know if it's portable.


So what do you guys think about it?

Cheers,
Marek Majkowski

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