[PATCH nf-next 0/2] fixes for recent nf_compact hooks

Two possible error conditions were caught during an extended testing
session, and by a build robot.  These patches fix the two issues (a
missing handler when config is changed, and a potential NULL
dereference).

Aaron Conole (2):
  netfilter: Fix potential null pointer dereference
  nf_set_hooks_head: acommodate different kconfig

 net/netfilter/core.c | 6 +++++-
 1 file changed, 5 insertions(+), 1 deletion(-)

-- 
2.5.5

It's possible for nf_hook_entry_head to return NULL if two
nf_unregister_net_hook calls happen simultaneously with a single hook
entry in the list.  This fix ensures that no null pointer dereference
could occur when such a race happens.

Signed-off-by: Aaron Conole <redacted>
---
 net/netfilter/core.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/net/netfilter/core.c b/net/netfilter/core.c
index 360c63d..e58e420 100644
--- a/net/netfilter/core.c
+++ b/net/netfilter/core.c

@@ -160,7 +160,7 @@ void nf_unregister_net_hook(struct net *net, const struct nf_hook_ops *reg)
 
 	mutex_lock(&nf_hook_mutex);
 	hooks_entry = nf_hook_entry_head(net, reg);
-	if (hooks_entry->orig_ops == reg) {
+	if (hooks_entry && hooks_entry->orig_ops == reg) {
 		nf_set_hooks_head(net, reg,
 				  nf_entry_dereference(hooks_entry->next));
 		goto unlock;

-- 
2.5.5

When CONFIG_NETFILTER_INGRESS is unset (or no), we need to handle
the request for registration properly by dropping the hook.  This
releases the entry during the set.

Signed-off-by: Aaron Conole <redacted>
---
 net/netfilter/core.c | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/net/netfilter/core.c b/net/netfilter/core.c
index e58e420..1d0a4c9 100644
--- a/net/netfilter/core.c
+++ b/net/netfilter/core.c

@@ -90,10 +90,14 @@ static void nf_set_hooks_head(struct net *net, const struct nf_hook_ops *reg,
 {
 	switch (reg->pf) {
 	case NFPROTO_NETDEV:
+#ifdef CONFIG_NETFILTER_INGRESS
 		/* We already checked in nf_register_net_hook() that this is
 		 * used from ingress.
 		 */
 		rcu_assign_pointer(reg->dev->nf_hooks_ingress, entry);
+#else
+		kfree(entry);
+#endif
 		break;
 	default:
 		rcu_assign_pointer(net->nf.hooks[reg->pf][reg->hooknum],

-- 
2.5.5

Aaron Conole [off-list ref] wrote:

This looks dodgy (its correct though).

I'd propose to add a test to nf_register_net_hook()
to bail with -EOPNOSTUPP instead of this "#else kfree()" if we get
NFPROTO_NETDEV pf with CONFIG_NETFILTER_INGRESS=n build instead.

Florian Westphal [off-list ref] writes:

Aaron Conole [off-list ref] wrote:

quoted

When CONFIG_NETFILTER_INGRESS is unset (or no), we need to handle
the request for registration properly by dropping the hook.  This
releases the entry during the set.

Signed-off-by: Aaron Conole <redacted>
---
 net/netfilter/core.c | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/net/netfilter/core.c b/net/netfilter/core.c
index e58e420..1d0a4c9 100644
--- a/net/netfilter/core.c
+++ b/net/netfilter/core.c

@@ -90,10 +90,14 @@ static void nf_set_hooks_head(struct net *net, const struct nf_hook_ops *reg,
 {
 	switch (reg->pf) {
 	case NFPROTO_NETDEV:
+#ifdef CONFIG_NETFILTER_INGRESS
 		/* We already checked in nf_register_net_hook() that this is
 		 * used from ingress.
 		 */
 		rcu_assign_pointer(reg->dev->nf_hooks_ingress, entry);
+#else
+		kfree(entry);
+#endif
 		break;

This looks dodgy (its correct though).

I'd propose to add a test to nf_register_net_hook()
to bail with -EOPNOSTUPP instead of this "#else kfree()" if we get
NFPROTO_NETDEV pf with CONFIG_NETFILTER_INGRESS=n build instead.

Okay, I'll spin a new version.

Thanks for the review, Florian!

-Aaron