Nmap Development mailing list archives

Re: Patch to make --host_timeout more useful


From: Fyodor <fyodor () insecure org>
Date: Tue, 2 Jan 2001 22:18:36 -0800 (PST)

On Mon, 18 Dec 2000, Colin Phipps wrote:

For people who don't want to wait forever :-), the attached patch makes 
nmap print partial results from timed out scans. It seemed a bit silly to 
me that nmap would time out a scan then silently forget any results that 
it found.

Thanks!  That patch will certainly be useful for some users!  However, I
am not sure about whether it should be in the main distribution.   There
are a couple important reason that Nmap currently does not try to give
results for scans that are aborted due to timeout:

1) User experience -- I am worried that partial results will give people a
false sense of security.  Even though you included a little note about the
hsot timeout (a good idea!), I am worried people may not notice that.  If
the results are completely omitted, it is more obvious that something is
seriously wrong and the results cannot be trusted as complete.

2) Limited benefit -- Even if we give partial results, they generally
don't know what ports were finished being scanned and so they don't know
wich ones are left to be tested.  And even if they did, it would be
exceedingly hard to "resume" the scan with another Nmap execution since
the ports are generally randomly shuffled each time.  So will have to redo
the whole scan anyway.

3) Programmer experience -- If I print the results, I need to make sure
that all the relevant structures are consistant in every single place I
bail out of a function due to timeout.  Otherwise there could be segfaults
and other problems.  I am not sure it is entirely safe now.  I try to
avoid adding these sorts of invariants unless the advantage is very
clear.  Otherwise it gets harder and harder to add/change things in Nmap
because I have to remember all these rules like.

That being said, advanced users can certainly apply a patch like yours to
skip the timeout check and print whatever is in the result
structures.  But doing so is at their own risk :).

That being said, I could change my mind if there is overwhelming demand
for this.  So if this really helps some people/applications, speak now or
forever hold your peace!  Also note that for many scans, -v will give you
partial info as it is discovered.

Cheers,
-F


---------------------------------------------------------------------
For help using this (nmap-dev) mailing list, send a blank email to 
nmap-dev-help () insecure org . List run by ezmlm-idx (www.ezmlm.org).



Current thread: