[PATCH nf] netfilter: add back stackpointer size checks

[PATCH nf] netfilter: add back stackpointer size checks

[Florian Westphal]

The rationale for removing the check is only correct for rulesets
generated by ip(6)tables.

In iptables, a jump can only occur to a user-defined chain, i.e.
because we size the stack based on number of user-defined chains we
cannot exceed stack size.

However, the underlying binary format has no such restriction,
and the validation step only ensures that the jump target is a
valid rule start point.

IOW, its possible to build a rule blob that has no user-defined
chains but does contain a jump.

If this happens, no jump stack gets allocated and crash occurs
because no jumpstack was allocated.

Fixes: 7814b6ec6d0d6 ("netfilter: xtables: don't save/restore jumpstack offset")
Reported-by: syzbot+e783f671527912cd9403@syzkaller.appspotmail.com
Signed-off-by: Florian Westphal <fw@strlen.de>
---
 net/ipv4/netfilter/arp_tables.c | 4 ++++
 net/ipv4/netfilter/ip_tables.c  | 7 ++++++-
 net/ipv6/netfilter/ip6_tables.c | 4 ++++
 3 files changed, 14 insertions(+), 1 deletion(-)

diff --git a/net/ipv4/netfilter/arp_tables.c b/net/ipv4/netfilter/arp_tables.c
index 4ffe302f9b82..e3e420f3ba7b 100644
--- a/net/ipv4/netfilter/arp_tables.c
+++ b/net/ipv4/netfilter/arp_tables.c
@@ -252,6 +252,10 @@ unsigned int arpt_do_table(struct sk_buff *skb,
 			}
 			if (table_base + v
 			    != arpt_next_entry(e)) {
+				if (unlikely(stackidx >= private->stacksize)) {
+					verdict = NF_DROP;
+					break;
+				}
 				jumpstack[stackidx++] = e;
 			}
 
diff --git a/net/ipv4/netfilter/ip_tables.c b/net/ipv4/netfilter/ip_tables.c
index 9a71f3149507..e38395a8dcf2 100644
--- a/net/ipv4/netfilter/ip_tables.c
+++ b/net/ipv4/netfilter/ip_tables.c
@@ -330,8 +330,13 @@ ipt_do_table(struct sk_buff *skb,
 				continue;
 			}
 			if (table_base + v != ipt_next_entry(e) &&
-			    !(e->ip.flags & IPT_F_GOTO))
+			    !(e->ip.flags & IPT_F_GOTO)) {
+				if (unlikely(stackidx >= private->stacksize)) {
+					verdict = NF_DROP;
+					break;
+				}
 				jumpstack[stackidx++] = e;
+			}
 
 			e = get_entry(table_base, v);
 			continue;
diff --git a/net/ipv6/netfilter/ip6_tables.c b/net/ipv6/netfilter/ip6_tables.c
index af4c917e0836..62358b93bbac 100644
--- a/net/ipv6/netfilter/ip6_tables.c
+++ b/net/ipv6/netfilter/ip6_tables.c
@@ -352,6 +352,10 @@ ip6t_do_table(struct sk_buff *skb,
 			}
 			if (table_base + v != ip6t_next_entry(e) &&
 			    !(e->ipv6.flags & IP6T_F_GOTO)) {
+				if (unlikely(stackidx >= private->stacksize)) {
+					verdict = NF_DROP;
+					break;
+				}
 				jumpstack[stackidx++] = e;
 			}
 
-- 
2.13.6

Re: [PATCH nf] netfilter: add back stackpointer size checks

[Pablo Neira Ayuso]

On Wed, Feb 07, 2018 at 01:46:25PM +0100, Florian Westphal wrote:
> The rationale for removing the check is only correct for rulesets
> generated by ip(6)tables.
> 
> In iptables, a jump can only occur to a user-defined chain, i.e.
> because we size the stack based on number of user-defined chains we
> cannot exceed stack size.
> 
> However, the underlying binary format has no such restriction,
> and the validation step only ensures that the jump target is a
> valid rule start point.
> 
> IOW, its possible to build a rule blob that has no user-defined
> chains but does contain a jump.
> 
> If this happens, no jump stack gets allocated and crash occurs
> because no jumpstack was allocated.

Applied, thanks.