disallow multiple matches of same type
PATCH 1/1 disallow multiple matches of same type
In latest iptables SVN, the command:
iptables -A foo -p tcp -m multiport --dport 45,47 -m multiport --sport 45:48
ends up with mangled results of:
... multiport sports multiport sports tcp spts:45:48
Since at present, iptables can only handle one match of a given type per rule. The below patch makes sure we disallow more than one.
This closes bugzilla #447
Phil
iptables.c +5 -0
@@ -2125,6 +2125,11 @@exit_error (PARAMETER_PROBLEM ,"unexpected ! flag before --match" ); +for (matchp = matches; matchp; matchp = matchp->next) { +if (strcmp (optarg, matchp->match->name) ==0 ) +exit_error (PARAMETER_PROBLEM , +"multiple matches of same type not supported" ); + } m =find_match (optarg,LOAD_MUST_SUCCEED , &matches); size =IPT_ALIGN (sizeof (struct ipt_entry_match )) + m->size;
Quoted 4 lines · Phil Oester
In latest iptables SVN, the command:
iptables -A foo -p tcp -m multiport --dport 45,47 -m multiport --sport 45:48
Quoted 8 lines
ends up with mangled results of:
... multiport sports multiport sports tcp spts:45:48
Since at present, iptables can only handle one match of a given type per rule. The below patch makes sure we disallow more than one.
This closes bugzilla #447
I think Harald already did something in this direction. Harald?
Quoted 1 line · Patrick McHardy
I think Harald already did something in this direction. Harald?
He corrected it so iptables didn't segfault, and opened a new bug for the incorrect behaviour. This fixes that new bug.
Phil
Quoted 5 lines · Phil Oester
Quoted 1 line · Patrick McHardy
I think Harald already did something in this direction. Harald?
He corrected it so iptables didn't segfault, and opened a new bug for the incorrect behaviour. This fixes that new bug.
that's correct. Haven't reviewed your bugfix yet, though.
Signature · 5 lines
- Harald Welte <[email protected]> http://netfilter.org/ ============================================================================ "Fragmentation is like classful addressing -- an interesting early architectural error that shows how much experimentation was going on while IP was being designed." -- Paul Vixie