Re: Patch for random source ports with map rules - bug in random port selection?

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
 
Manuel Kasper wrote:
| Hello Darren,
|
| thanks for the quick source port randomization patch!
|
| While trying to apply it to m0n0wall today, two things struck me as 
odd with the following code:
|
|> !                 port = ipf_random() % (ntohs(np->in_pmax) -
|> !                                ntohs(np->in_pmin));
|
|
| 1. np->in_pmin isn't added to the generated random number, which means 
that the result will be
|    between 0 and (in_pmax - in_pmin - 1), rather than between in_pmin 
and in_pmax (inclusive).
|
| 2. No htons() is done on the result, and thus the resulting port 
numbers don't turn out as expected.
|
| A quick test with a map rule port range of 6000:7000 resulted in port 
numbers like the following (as seen on the network):
|     26114, 65282, 48898, 12033, 51459, 32258, ...
| all of which, when byte-swapped, are < 1000.
|
| The fix seems to be as follows:
|
| port = htons((ipf_random() %
|           (ntohs(np->in_pmax) - ntohs(np->in_pmin) + 1))
|           + ntohs(np->in_pmin));

That's what happens when you don't get enough sleep.

Darren

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (MingW32)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org
 
iEYEARECAAYFAkiLbmUACgkQP7JIXtvLbFUTkQCgpZ1aGUM+yRHoEE6V31lGjBNw
G5kAn3ReH15J7eiDK9Dkmf4yeT51CdLM
=oUlC
-----END PGP SIGNATURE-----