Skip to contentSkip to conversation

nth match: using more than 1 counter

BUG Mateusz Golicz 1 participant 1 message Dec 7, 2002 netfilter

lore
Share & tools
lore
#0 lore raw

Hello,

The userspace part of the nth match (BTW, quite an useful one)
seems to be a bit broken. (at least in the CVS version from the
WWW frontend (don't know if there are any other ones ;), i'm
new to netfilter developement): http://cvs.netfilter.org/cgi-bin/cvsweb/~checkout~/netfilter/userspace/extensions/libipt_nth.c)

around line 102, in parse(), there is a fragment doing some checks concerning the --every option:

--- CUT ---

                nthinfo->packet = 0xFF;
                if(!(*flags & IPT_NTH_OPT_EVERY))
                {
                        nthinfo->counter = 0;
                }

--- CUT ---

... but actually, the IPT_NTH_OPT_EVERY flag is
set only in this function, but after this part. So, if
you specify --every after --counter, the previous nthinfo->counter will be always lost, which will make the match behave quite improperly in case you use more than a one counter... i guess this check should be something like:

--- CUT ---

libipt_nth.c Sat Dec 7 05:29:28 2002 renamed from libipt_nth.c.old Sat Dec 7 05:47:47 2002 +1 -1
@@ -100,7 +101,7 @@
                nthinfo->every = num-1;
                nthinfo->startat = 0;
                 nthinfo->packet = 0xFF;
-                if(!(*flags & IPT_NTH_OPT_EVERY))
+                if(!(*flags & IPT_NTH_OPT_COUNTER))
                 {
                         nthinfo->counter = 0;
                 }

--- CUT ---

Matthew Golicz, IP Network Dept., Cable Television "Elsat" Ltd.

Signature · 4 lines
______________________________________________________________________
mteg <[email protected]>       * MG452-RIPE *            Mateusz Golicz
* ,,I just had a thought I have seen everything in this lifetime, but*
* now I sit here and watch a cyborg diving ... in her spare time!''  *