Re: [NSE] Catching socket errors, documentation discrepancy

[David Fifield <david () bamsoftware com>]

On Mon, May 28, 2012 at 04:18:59PM -0700, David Fifield wrote:
> On Wed, Sep 21, 2011 at 10:33:37PM -0500, Daniel Miller wrote:
> > On Wed, Sep 21, 2011 at 5:26 PM, David Fifield <david () bamsoftware com> wrote:
> > > On Wed, Sep 21, 2011 at 05:23:25PM -0500, Daniel Miller wrote:
> > > > While working on a script, I found that the socket I/O described on
> > > > the NSE API page (http://nmap.org/book/nse-api.html) and in the nmap
> > > > library documentation (http://nmap.org/nsedoc/lib/nmap.html#send) is
> > > > not how it is currently implemented. Specifically, socket operations
> > > > are now throwing Lua errors instead of returning a false value and
> > > > an error string. Is there a current reference for how errors should
> > > > be treated? Is there a plan to change the existing documentation to
> > > > match the current implementation? I could source-dive and find how
> > > > to do it myself, but I think we should be a little more
> > > > user-friendly for script authors who take the documentation at its
> > > > word.
> > >
> > > Please be more specific about the socket errors that are causing Lua
> > > errors for you. Can you show us example code?
> >
> > Specifically, trying to send on an unconnected socket. There's
> > probably a better way to do things in general, but I'm working with
> > someone else's code and trying not to make sweeping changes. A minimal
> > test script is attached that exhibits the behavior.
>
> You are right, it shouldn't work like that. Here is what I see:
>
> Initiating NSE at 10:03
> NSE: test threw an error!
> /home/david/nmap-git/scripts/test.nse:15: socket must be connected
>
> stack traceback:
>         [C]: in function 'send'
>         /home/david/nmap-git/scripts/test.nse:15: in function </home/david/nmap-git/scripts/test.nse:13>
>         (...tail calls...)
>
> I'm attaching a patch that turns these into normal error returns. With
> it, I get this.
>
> Initiating NSE at 16:16
> NSE: Finished 'test' (thread: 0x1bf3b60).
> Completed NSE at 16:16, 0.00s elapsed
> Pre-scan script results:
> |_test: socket must be connected
>
> The patch is mostly a mechanical replacement of
>       check_nsock_udata(L, 1, 1);
> width
>       nse_nsock_udata *nu = check_nsock_udata(L, 1);
>       NSOCK_UDATA_ENSURE_OPEN(L, nu);
> where the third argument to check_nsock_udata is a boolean controlling
> whether the socket must be open, and NSOCK_UDATA_ENSURE_OPEN is a macro
> that returns the enclosing function with an error if the socket is not
> open.

I committed this patch as r29016.

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