Re: [BUG] VPN broken in net-next

[David Miller <davem@davemloft.net>]

From: Julian Anastasov <ja@ssi.bg>
Date: Thu, 3 Mar 2011 15:09:22 +0200 (EET)

> On Thu, 3 Mar 2011, Julian Anastasov wrote:
> 
>> 	May be the problem is in inet_hash_insert(), it should
>> hash ifa_local, not ifa_address. May be they are equal for
> 
> 	... and of course the new __ip_dev_find should use
> ifa_local too.

Thanks for looking into this Julian.  I will look at the other
cases you found later.

Stephen, is this sufficient to fix your problem?  I suspect it is
not because fib_add_addr() adds prefixes with RTN_LOCAL to the
local routing table too :-/

I suspect that even if we need to handle prefixes, we can still use
the hash for optimistic lookup, and fallback to a local table FIB
inspection if that fails.

--------------------
ipv4: Fix __ip_dev_find() to use ifa_local instead of ifa_address.

Reported-by: Stephen Hemminger <shemminger@vyatta.com>
Reported-by: Julian Anastasov <ja@ssi.bg>
Signed-off-by: David S. Miller <davem@davemloft.net>

diff --git a/net/ipv4/devinet.c b/net/ipv4/devinet.c
index 9038928..ff53860 100644
--- a/net/ipv4/devinet.c
+++ b/net/ipv4/devinet.c
@@ -111,7 +111,7 @@ static inline unsigned int inet_addr_hash(struct net *net, __be32 addr)
 
 static void inet_hash_insert(struct net *net, struct in_ifaddr *ifa)
 {
-	unsigned int hash = inet_addr_hash(net, ifa->ifa_address);
+	unsigned int hash = inet_addr_hash(net, ifa->ifa_local);
 
 	spin_lock(&inet_addr_hash_lock);
 	hlist_add_head_rcu(&ifa->hash, &inet_addr_lst[hash]);
@@ -146,7 +146,7 @@ struct net_device *__ip_dev_find(struct net *net, __be32 addr, bool devref)
 
 		if (!net_eq(dev_net(dev), net))
 			continue;
-		if (ifa->ifa_address == addr) {
+		if (ifa->ifa_local == addr) {
 			result = dev;
 			break;
 		}