[PATCH 0/2] Netfilter updates for 3.5-rc5

From: Pablo Neira Ayuso <pablo@netfilter.org>

Hi David,

The following patches provide two fixes:

* One to get the timeout special parameter for the SET target back working
  (this was introduced while trying to fix another bug in 3.4) from
  Jozsef Kadlecsik.

* One crash fix if containers and nf_conntrack are used reported by Hans
  Schillstrom by myself.

You can pull these fixes from:

git://1984.lsi.us.es/nf master

Thanks.

little notice: I forgot to add my Signed-off-by while manually applying
Jozsef's patch, sorry. It was a bit too late to fix, I already pushed out
to my master branch.

Jozsef Kadlecsik (1):
  netfilter: ipset: timeout fixing bug broke SET target special timeout value

Pablo Neira Ayuso (1):
  netfilter: nf_ct_ecache: fix crash with multiple containers, one shutting down

 include/net/netfilter/nf_conntrack_ecache.h |    2 +-
 net/netfilter/xt_set.c                      |    4 +++-
 2 files changed, 4 insertions(+), 2 deletions(-)

-- 
1.7.10

From: Jozsef Kadlecsik <redacted>

The patch "127f559 netfilter: ipset: fix timeout value overflow bug"
broke the SET target when no timeout was specified.

Reported-by: Jean-Philippe Menil <redacted>
Signed-off-by: Jozsef Kadlecsik <redacted>
---
 net/netfilter/xt_set.c |    4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/net/netfilter/xt_set.c b/net/netfilter/xt_set.c
index 035960e..b172cbc 100644
--- a/net/netfilter/xt_set.c
+++ b/net/netfilter/xt_set.c

@@ -16,6 +16,7 @@
 
 #include <linux/netfilter/x_tables.h>
 #include <linux/netfilter/xt_set.h>
+#include <linux/netfilter/ipset/ip_set_timeout.h>
 
 MODULE_LICENSE("GPL");
 MODULE_AUTHOR("Jozsef Kadlecsik <kadlec@blackhole.kfki.hu>");

@@ -310,7 +311,8 @@ set_target_v2(struct sk_buff *skb, const struct xt_action_param *par)
 		info->del_set.flags, 0, UINT_MAX);
 
 	/* Normalize to fit into jiffies */
-	if (add_opt.timeout > UINT_MAX/MSEC_PER_SEC)
+	if (add_opt.timeout != IPSET_NO_TIMEOUT
+	    && add_opt.timeout > UINT_MAX/MSEC_PER_SEC)
 		add_opt.timeout = UINT_MAX/MSEC_PER_SEC;
 	if (info->add_set.index != IPSET_INVALID_ID)
 		ip_set_add(info->add_set.index, skb, par, &add_opt);

-- 
1.7.10

From: Pablo Neira Ayuso <pablo@netfilter.org>

Hans reports that he's still hitting:

BUG: unable to handle kernel NULL pointer dereference at 000000000000027c
IP: [<ffffffff813615db>] netlink_has_listeners+0xb/0x60
PGD 0
Oops: 0000 [#3] PREEMPT SMP
CPU 0

It happens when adding a number of containers with do:

nfct_query(h, NFCT_Q_CREATE, ct);

and most likely one namespace shuts down.

this problem was supposed to be fixed by:
70e9942 netfilter: nf_conntrack: make event callback registration per-netns

Still, it was missing one rcu_access_pointer to check if the callback
is set or not.

Reported-by: Hans Schillstrom <redacted>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
---
 include/net/netfilter/nf_conntrack_ecache.h |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/include/net/netfilter/nf_conntrack_ecache.h b/include/net/netfilter/nf_conntrack_ecache.h
index a88fb69..e1ce104 100644
--- a/include/net/netfilter/nf_conntrack_ecache.h
+++ b/include/net/netfilter/nf_conntrack_ecache.h

@@ -78,7 +78,7 @@ nf_conntrack_event_cache(enum ip_conntrack_events event, struct nf_conn *ct)
 	struct net *net = nf_ct_net(ct);
 	struct nf_conntrack_ecache *e;
 
-	if (net->ct.nf_conntrack_event_cb == NULL)
+	if (!rcu_access_pointer(net->ct.nf_conntrack_event_cb))
 		return;
 
 	e = nf_ct_ecache_find(ct);

-- 
1.7.10

From: pablo@netfilter.org
Date: Fri,  6 Jul 2012 13:39:38 +0200

+	if (add_opt.timeout != IPSET_NO_TIMEOUT
+	    && add_opt.timeout > UINT_MAX/MSEC_PER_SEC)

We do not write conditionals like this, with operators beginning
a continued line.  Instead, write this as:

	if (a &&
	    b)

Thanks.

On Mon, Jul 09, 2012 at 12:29:03AM -0700, David Miller wrote:

From: pablo@netfilter.org
Date: Fri,  6 Jul 2012 13:39:38 +0200

quoted

+	if (add_opt.timeout != IPSET_NO_TIMEOUT
+	    && add_opt.timeout > UINT_MAX/MSEC_PER_SEC)

We do not write conditionals like this, with operators beginning
a continued line.  Instead, write this as:

	if (a &&
	    b)

Oops, indeed, sorry. New patch attached.

I've also rebased my tree to include this change. Should I send a new
pull request?

Let me know what you prefer.

From: Pablo Neira Ayuso <pablo@netfilter.org>
Date: Mon, 9 Jul 2012 10:58:27 +0200

I've also rebased my tree to include this change. Should I send a new
pull request?

Next time send a new pull request.

This time, I re-pulled, thanks.