Re: [NSE] Strange bug in broadcast-ms-sql-discover

[Patrik Karlsson <patrik () cqure net>]

On Tue, Oct 25, 2011 at 12:09 PM, Duarte Silva
<duarte.silva () serializing me>wrote:

> Hi,
>
> I agree, with that, but I think a generic error message (without stack
> traces)
> would be better in order not to pollute the nmap output. Something like:
>
> |_broadcast-ms-sql-discover: FAILURE: An unexpected failure occurred,
> please
> run with -d2 and --script-trace and report to ...
>
> We could also think about generating a dump file (using a specific option
> or in
> a automatic fashion when a script failed with an exception) with the stack
> trace and any useful information for tracking down the problem cause. Then
> the
> user would only have to submit that file.
>
> Best regards,
> Duarte Silva
>
> On Monday 24 October 2011 22:32:08 David Fifield wrote:
> > On Mon, Oct 24, 2011 at 06:44:43PM +0200, Patrik Karlsson wrote:
> > > Hi all,
> > >
> > > I'm seeing the following error when running the
> broadcast-ms-sql-discover
> > > script:
> > >
> > > NSE: 'broadcast-ms-sql-discover' (thread: 0x7feb3b19df40) threw an
> error!
> > > ...sers/patrik/hacktools/rd/nmap-dev/./nselib/mssql.lua:2012: bad
> > > argument #1 to 'set_port_version' (host is not being processed right
> > > now)
> > >
> > > I'm quite sure that this error did not occur in the past. Does anyone
> > > have an idea why it's behaving this way, except for the obvious error
> > > message? As a temporary fix I've commented out the line causing the
> > > problem an committed it as r26928.
> >
> > broadcast-ms-sql-discover doesn't have any real host tables to work
> > with, so it creates it own dummy tables.
> >         local host = { ip = "255.255.255.255" }
> >         local port = { number = 1434, protocol = "udp" }
> > But as it's a prerule script, Nmap hasn't actually created any Target
> > objects yet. I don't see how it could have worked before. Commenting it
> > out seems fine.
> >
> > > I noticed this error only after I enabled debugging as I was quite
> > > certain there were SQL servers on the network.
> > > I'm not all that happy with scripts silently failing unless running in
> > > debug mode.
> > > Does anyone else feel that this isn't all that great either?
> >
> > I tend to agree that errors and stack traces should be visible by
> > default.
> >
> > David Fifield
> > _______________________________________________
> > Sent through the nmap-dev mailing list
> > http://cgi.insecure.org/mailman/listinfo/nmap-dev
> > Archived at http://seclists.org/nmap-dev/

The generic error message suggesting to do -d2 and --script-trace works for
me.
I think the dump file may be overkill for now and that copying the output to
a mail would be sufficient.
David, any thoughts?

-- 
Patrik Karlsson
http://www.cqure.net
http://twitter.com/nevdull77
_______________________________________________
Sent through the nmap-dev mailing list
http://cgi.insecure.org/mailman/listinfo/nmap-dev
Archived at http://seclists.org/nmap-dev/