Re: [netfilter-core] Re: [lockup] 2.6.17-rc3: netfilter/sctp: lockup in sctp_new(), do_basic_checks()

[Ingo Molnar <mingo@elte.hu>]

* Patrick McHardy <kaber@trash.net> wrote:

> Ingo Molnar wrote:
> > thinking about it, what prevents the SCTP chunk's len field from being 
> > zero, and thus causing an infinite loop in for_each_sctp_chunk()? The 
> > patch below should fix that.
> > 
> > 	Ingo
> > 
> > ----
> > From: Ingo Molnar <mingo@elte.hu>
> > 
> > fix infinite loop in the SCTP-netfilter code: check SCTP chunk size to 
> > guarantee progress of for_each_sctp_chunk(). (all other uses of 
> > for_each_sctp_chunk() are preceded by do_basic_checks(), so this fix 
> > should be complete.)
> > 
> > Signed-off-by: Ingo Molnar <mingo@elte.hu>
> > 
> > Index: linux/net/ipv4/netfilter/ip_conntrack_proto_sctp.c
> > ===================================================================
> > --- linux.orig/net/ipv4/netfilter/ip_conntrack_proto_sctp.c
> > +++ linux/net/ipv4/netfilter/ip_conntrack_proto_sctp.c
> > @@ -227,6 +227,15 @@ static int do_basic_checks(struct ip_con
> >  	flag = 0;
> >  
> >  	for_each_sctp_chunk (skb, sch, _sch, offset, count) {
> > +		unsigned int len = (htons(sch->length) + 3) & ~3;
> > +
> > +		/*
> > +		 * Dont get into a loop with zero-sized or negative
> > +		 * length values:
> > +		 */
> > +		if (!len || len >= skb->len)
> > +			goto fail;
> > +
> 
> I just came up with a similar fix :) I think I'm going to take my own 
> patch though because its IMO slightly nicer. Thanks anyway.

could you send your patch so that i can start using it instead of mine?

	Ingo