netdev - Re: [PATCH v2 net 1/5] icmp: introduce helper for NAT'd source address in network device context

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:   Mon, 10 Feb 2020 23:26:14 +0100
From:   Florian Westphal <fw@...len.de>
To:     "Jason A. Donenfeld" <Jason@...c4.com>
Cc:     Florian Westphal <fw@...len.de>, Netdev <netdev@...r.kernel.org>,
        David Miller <davem@...emloft.net>
Subject: Re: [PATCH v2 net 1/5] icmp: introduce helper for NAT'd source
 address in network device context

Jason A. Donenfeld <Jason@...c4.com> wrote:
> Hi Florian,
> 
> On Mon, Feb 10, 2020 at 10:33 PM Florian Westphal <fw@...len.de> wrote:
> >
> > Jason A. Donenfeld <Jason@...c4.com> wrote:
> > > On Mon, Feb 10, 2020 at 3:15 PM Jason A. Donenfeld <Jason@...c4.com> wrote:
> > > > +               ip_hdr(skb_in)->saddr = ct->tuplehash[0].tuple.src.u3.ip;
> > > > +       }
> > > > +       icmp_send(skb_in, type, code, info);
> > >
> > > According to the comments in icmp_send, access to
> > > ip_hdr(skb_in)->saddr requires first checking for `if
> > > (skb_network_header(skb_in) < skb_in->head ||
> > > (skb_network_header(skb_in) + sizeof(struct iphdr)) >
> > > skb_tail_pointer(skb_in))` first to be safe.
> >
> > You will probably also need skb_ensure_writable() to handle cloned skbs.
> >
> > I also suggest to check "ct->status & IPS_NAT_MASK", nat is only done if
> > those bits are set.
> 
> Thanks for the suggestions. I've made these changes and they're queued
> up for a v3, currently staged in wireguard-linux.git's stable branch:
> https://git.zx2c4.com/wireguard-linux/log/?h=stable

I think this is a bit too conservative, f.e. i don't see how
ndo-called skbs could be shared (tx path needs to be able to change skb
list pointers)?

If its needed it looks ok.

Otherwise, I would suggest something like this:

void icmp_ndo_send(struct sk_buff *skb_in, int type, int code, __be32 info)
{
   enum ip_conntrack_info ctinfo;
   struct nf_conn *ct;
   __be32 orig_ip;

   ct = nf_ct_get(skb_in, &ctinfo);
   if (!ct || ((ct->status & IPS_NAT_MASK) == 0) {
      icmp_send(skb_in, type, code, info);
      return;
    }

   /* avoid reallocations */
   if (skb_network_header(skb_in) < skb_in->head ||
       (skb_network_header(skb_in) + sizeof(struct iphdr)) >
        skb_tail_pointer(skb_in))
       return;

    /* handle clones. NB: if reallocations are to be avoided, then
     * if (skb_cloned(skb_in) &&
     * !skb_clone_writable(skb_in, skb_network_offset(skb_in) + iphlen))
     *
     * ... should be placed here instead:
     */
    if (unlikely(skb_ensure_writable(skb_in,
                 skb_network_offset(skb_in) + sizeof(struct iphdr))))
        return;

    orig_ip = ip_hdr(skb_in)->saddr;
    ip_hdr(skb_in)->saddr = ct->tuplehash[0].tuple.src.u3.ip;
    icmp_send(skb_in, type, code, info);
    ip_hdr(skb_in)->saddr = orig_ip;
}