Pavel Emelyanov
2007-Nov-27 08:21 UTC
[PATCH (resubmit)][BRIDGE] Properly dereference the br_should_route_hook
This hook is protected with the RCU, so simple
if (br_should_route_hook)
br_should_route_hook(...)
is not enough on some architectures.
Use the rcu_dereference/rcu_assign_pointer in this case.
Fixed Stephen's comment concerning using the typeof().
Signed-off-by: Pavel Emelyanov <xemul@openvz.org>
---
diff --git a/net/bridge/br_input.c b/net/bridge/br_input.c
index 3cedd4e..0ee79a7 100644
--- a/net/bridge/br_input.c
+++ b/net/bridge/br_input.c
@@ -122,6 +122,7 @@ static inline int is_link_local(const unsigned char *dest)
struct sk_buff *br_handle_frame(struct net_bridge_port *p, struct sk_buff *skb)
{
const unsigned char *dest = eth_hdr(skb)->h_dest;
+ int (*rhook)(struct sk_buff *skb);
if (!is_valid_ether_addr(eth_hdr(skb)->h_source))
goto drop;
@@ -147,9 +148,9 @@ struct sk_buff *br_handle_frame(struct net_bridge_port *p,
struct sk_buff *skb)
switch (p->state) {
case BR_STATE_FORWARDING:
-
- if (br_should_route_hook) {
- if (br_should_route_hook(skb))
+ rhook = rcu_dereference(br_should_route_hook);
+ if (rhook != NULL) {
+ if (rhook(skb))
return skb;
dest = eth_hdr(skb)->h_dest;
}
diff --git a/net/bridge/netfilter/ebtable_broute.c
b/net/bridge/netfilter/ebtable_broute.c
index e44519e..be6f186 100644
--- a/net/bridge/netfilter/ebtable_broute.c
+++ b/net/bridge/netfilter/ebtable_broute.c
@@ -70,13 +70,13 @@ static int __init ebtable_broute_init(void)
if (ret < 0)
return ret;
/* see br_input.c */
- br_should_route_hook = ebt_broute;
+ rcu_assign_pointer(br_should_route_hook, ebt_broute);
return ret;
}
static void __exit ebtable_broute_fini(void)
{
- br_should_route_hook = NULL;
+ rcu_assign_pointer(br_should_route_hook, NULL);
synchronize_net();
ebt_unregister_table(&broute_table);
}
Herbert Xu
2007-Nov-29 05:04 UTC
[PATCH (resubmit)][BRIDGE] Properly dereference the br_should_route_hook
On Tue, Nov 27, 2007 at 07:21:08PM +0300, Pavel Emelyanov wrote:> This hook is protected with the RCU, so simple > > if (br_should_route_hook) > br_should_route_hook(...) > > is not enough on some architectures. > > Use the rcu_dereference/rcu_assign_pointer in this case. > > Fixed Stephen's comment concerning using the typeof(). > > Signed-off-by: Pavel Emelyanov <xemul@openvz.org>Applied to net-2.6. Thanks Pavel!> static void __exit ebtable_broute_fini(void) > { > - br_should_route_hook = NULL; > + rcu_assign_pointer(br_should_route_hook, NULL);Just for the record, rcu_assign_pointer is never necessary when you're assigning NULL. The reason is that rcu_assign_pointer serves as a barrier between the initialisation of the content of what you're assigning and the actual assignment. Since NULL does not need to be initialised you don't need the barrier :) Hmm, perhaps we could even build this logic into rcu_assign_pointer. Then again, who still uses an Alpha? Mine died years ago :) Cheers, -- Visit Openswan at http://www.openswan.org/ Email: Herbert Xu ~{PmV>HI~} <herbert@gondor.apana.org.au> Home Page: http://gondor.apana.org.au/~herbert/ PGP Key: http://gondor.apana.org.au/~herbert/pubkey.txt