Re: Performance of nmap.

[Fyodor <fyodor () insecure org>]

On Fri, Jun 16, 2006 at 09:56:15PM +0200, majek04 wrote:
> I executed nmap with as many options as I could think of.
> And I collected the results of gcov.
>
> My tests covered most of nmaps functions, but there are
> about 50 left that I wasn't able to run by script.
> The list is there:
> http://ai.pjwstk.edu.pl/~majek/private/nmap/not-runned-functions.txt

Cool.  I went through the list and think they are all fine to stay
except for "ServiceProbe::setName, which I have removed.  I also
removed most of the TCPIP_DEBUGGING stuff from tcpip.cc (except for a
couple instances where I just want to keep them as they are in case I
need to enable it in the future).

I must admit to being a bit surprised that there wasn't more dead code
:).  We will make more when we port massping() to ultra_scan() and
maybe trim down pos_scan to just handle rpc scan.

> Maybe we can "#ifdef" these functions you suggested not to remove,
> so that they won't be compiled for ordinary user?

Well, there are only a few, and I think only one is of any significant
length.  And it would be more of a pain for me to compile them
separately when I use them with the out-of-tree tools.  So I'd
probably rather just leave them alone for now.  But if you find that
any of them actually make a meaningful difference in executable size
or runtime performance, let me know and we can find a solution.

Cheers,
-F


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