Linux Ethernet Bridging - Sep 2019 - [Bridge] [PATCH v4 1/2] netfilter: Terminate rule eval if protocol=IPv6 and ipv6 module is disabled

Pablo Neira Ayuso <pablo at netfilter.org> wrote:
> On Tue, Sep 03, 2019 at 01:46:50PM -0300, Leonardo Bras wrote:
> > On Fri, 2019-08-30 at 22:58 +0200, Florian Westphal wrote:
> > > Leonardo Bras <leonardo at linux.ibm.com> wrote:
> > > > If IPv6 is disabled on boot (ipv6.disable=1), but
nft_fib_inet ends up
> > > > dealing with a IPv6 packet, it causes a kernel panic in
> > > > fib6_node_lookup_1(), crashing in bad_page_fault.
> > > > 
> > > > The panic is caused by trying to deference a very low
address (0x38
> > > > in ppc64le), due to ipv6.fib6_main_tbl = NULL.
> > > > BUG: Kernel NULL pointer dereference at 0x00000038
> > > > 
> > > > The kernel panic was reproduced in a host that disabled IPv6
on boot and
> > > > have to process guest packets (coming from a bridge) using
it's ip6tables.
> > > > 
> > > > Terminate rule evaluation when packet protocol is IPv6 but
the ipv6 module
> > > > is not loaded.
> > > > 
> > > > Signed-off-by: Leonardo Bras <leonardo at
linux.ibm.com>
> > > 
> > > Acked-by: Florian Westphal <fw at strlen.de>
> > > 
> > 
> > Hello Pablo,
> > 
> > Any trouble with this patch? 
> > I could see the other* one got applied, but not this one.
> > *(The other did not get acked, so i released it alone as v5)
> > 
> > Is there any fix I need to do in this one?
> 
> Hm, I see, so this one:
> 
> https://patchwork.ozlabs.org/patch/1156100/
> 
> is not enough?
No, its not.
> I was expecting we could find a way to handle this from br_netfilter
> alone itself.
We can't because we support ipv6 fib lookups from the netdev family
as well.

Alternative is to auto-accept ipv6 packets from the nf_tables eval loop,
but I think its worse.

On Tue, Sep 03, 2019 at 07:05:50PM +0200, Florian Westphal
wrote:
> Pablo Neira Ayuso <pablo at netfilter.org> wrote:
> > On Tue, Sep 03, 2019 at 01:46:50PM -0300, Leonardo Bras wrote:
> > > On Fri, 2019-08-30 at 22:58 +0200, Florian Westphal wrote:
> > > > Leonardo Bras <leonardo at linux.ibm.com> wrote:
> > > > > If IPv6 is disabled on boot (ipv6.disable=1), but
nft_fib_inet ends up
> > > > > dealing with a IPv6 packet, it causes a kernel panic in
> > > > > fib6_node_lookup_1(), crashing in bad_page_fault.
> > > > > 
> > > > > The panic is caused by trying to deference a very low
address (0x38
> > > > > in ppc64le), due to ipv6.fib6_main_tbl = NULL.
> > > > > BUG: Kernel NULL pointer dereference at 0x00000038
> > > > > 
> > > > > The kernel panic was reproduced in a host that disabled
IPv6 on boot and
> > > > > have to process guest packets (coming from a bridge)
using it's ip6tables.
> > > > > 
> > > > > Terminate rule evaluation when packet protocol is IPv6
but the ipv6 module
> > > > > is not loaded.
> > > > > 
> > > > > Signed-off-by: Leonardo Bras <leonardo at
linux.ibm.com>
> > > > 
> > > > Acked-by: Florian Westphal <fw at strlen.de>
> > > > 
> > > 
> > > Hello Pablo,
> > > 
> > > Any trouble with this patch? 
> > > I could see the other* one got applied, but not this one.
> > > *(The other did not get acked, so i released it alone as v5)
> > > 
> > > Is there any fix I need to do in this one?
> > 
> > Hm, I see, so this one:
> > 
> > https://patchwork.ozlabs.org/patch/1156100/
> > 
> > is not enough?
> 
> No, its not.
> 
> > I was expecting we could find a way to handle this from br_netfilter
> > alone itself.
> 
> We can't because we support ipv6 fib lookups from the netdev family
> as well.
> 
> Alternative is to auto-accept ipv6 packets from the nf_tables eval loop,
> but I think its worse.
Could we add a restriction for nf_tables + br_netfilter + !ipv6. I
mean, if this is an IPv6 packet, nf_tables is on and IPv6 module if
off, then drop this packet?

By dropping packet, the user could diagnose that its setup is
incomplete. I mean, if nf_tables fib ipv6 is used, then this setup is
really wrong and the user forgots to load the ipv6 module.