Re: [PATCH] Ncat: Implementation of idle timeout option for listen mode

[Tomas Hozza <thozza () redhat com>]

On 06/05/2013 04:43 PM, Tomas Hozza wrote:
> On 01/15/2013 12:25 AM, David Fifield wrote:
> > On Mon, Jan 14, 2013 at 02:53:11PM +0100, Michal Hlavinka wrote:
> > > I just found that ncat does not behave the way I'd expect.
> > >
> > > Make it listen for udp connection:
> > > ncat -l -u 1234 </dev/null
> > >
> > > and send "Hello":
> > > echo "Hello" | ncat -u localhost 1234
> > >
> > > The result is that listening ncat waits for the "Hello" message and
> > > when it comes, it terminates and prints nothing.
> > >
> > > In ncat_listen_dgram it select()s change on socket descriptor and
> > > after a few checks it goes to
> > > """/* stdin -> socket and socket -> stdout */""""
> > > while loop. Where it checks stdin first and finding it closed, it
> > > terminates. I'm not sure what is intended behaviour, but I guess it
> > > should print the message since it's waiting for it.
> >
> > I think this behavior is right. Arguably the program should quit
> > immediately upon startup since its input is closed. I would accept a
> > patch to do that.
> >
> > > And another question. Our QA engineers use netcat in their tests and
> > > one use case is to wait for connection for some time and then
> > > terminate.
> > >
> > > In old netcat there's -w option which has the same behaviour like
> > > the -i in ncat. Unfortunately, -i can't be used together with -l (it
> > > prints error message). Why is it that way? Is it just that nobody
> > > wrote the code or is there any other reason why it can't be used
> > > together?
> >
> > My guess is that it is just not implemented.
> >
> > David Fifield
>
> Hello.
>
> I implemented the idle timeout "-i" option for ncat's listen mode.
> Previously it was available only for connection mode.
>
> I also added tests for idle timeout option when used in listen mode
> into the ncat-test.pl test-suite.
>
> Patches are attached.
>
> I want to point out that when using ncat in listen mode with UDP, the
> server is unaware of the connected client until it sends some message to
> the server. This behaviour differs from connection oriented protocols
> such as TCP and SCTP.
>
> That's why I added separate tests for connection oriented protocols
> and UDP in the test-suite.
>
>
> Regards,
>
> Tomas Hozza

Hello,

I would like to check on my patches status. Did anyone look at them? If
there is any problem, please let me know so I can fix it. If not, would
it be possible to include my changes in the next nmap release?

Thank you in advance.


Regards,

Tomas Hozza
_______________________________________________
Sent through the dev mailing list
http://nmap.org/mailman/listinfo/dev
Archived at http://seclists.org/nmap-dev/