Re: [PATCH] Dramatically reduce # of fopen()s and fclose()s in nbase_rnd.c

[Kris Katterjohn <katterjohn () gmail com>]

Fyodor wrote:
> On Fri, Dec 15, 2006 at 11:43:38AM -0600, Kris Katterjohn wrote:
> > This simple patch dramatically reduces the number of fopen()s and
> > fclose()s used when getting random numbers from nbase_rnd.c by keeping
> > the rng device open.
>
> It seems like a tough call, since keeping file descriptors open has
> its own problems.  Particularly since Nmap often uses so many
> concurrently anywa.  Have you measured how much time the 512 opens
> take up?  Do you see any performance difference from your patch?
>
> Cheers,
> -F

Based on the tests I've done (not very extensive because I haven't had
to do anything very.. extensive), the open()s take from fractions of a
second (1024 ports on localhost) to 12 seconds (all ports on google).
The patch usually cuts off about 4-8 seconds.

Can you think of good options etc. to use when testing? I did bare SYN
scans.

Thanks,
Kris Katterjohn

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