Skip to contentSkip to conversation

minor netfilter fixes for 2.6.30-rc

PATCH Applied Pablo Neira Ayuso 2 participants 6 messages Apr 17, 2009

Subsystems (2)
lore
Share & tools
lore

Pablo Neira Ayuso

author maintainer
#0 lore raw

cover minor netfilter fixes for 2.6.30-rc

Hi Patrick,

The following patches are a couple of minor fixes for 2.6.30-rc:

* fix missing error report to user-space in ctnetlink if we fail to allocate the message.
* fix misleading error in nfnetlink (-EPERM) if we fail to load nfnetlink. This patch replaces it with ENOMEM as it does rtnetlink.

---

Pablo Neira Ayuso (2):
netfilter: nfnetlink: return ENOMEM if we fail to create netlink socket
netfilter: ctnetlink: report error if event message allocation fails

 net/netfilter/nf_conntrack_netlink.c |    5 +++--
 net/netfilter/nfnetlink.c            |    2 +-
 2 files changed, 4 insertions(+), 3 deletions(-)
author maintainer
#1 lore raw

PATCH 1/2 netfilter: ctnetlink: report error if event message allocation fails

This patch fixes an inconsistency that results in no error reports to user-space listeners if we fail to allocate the event message.

---

 net/netfilter/nf_conntrack_netlink.c |    5 +++--
 1 files changed, 3 insertions(+), 2 deletions(-)
net/netfilter/nf_conntrack_netlink.c +3 -2
index c6439c7..5e8503c 100644
@@ -512,7 +512,7 @@ static int ctnetlink_conntrack_event(struct notifier_block *this,
 
 	skb = ctnetlink_alloc_skb(tuple(ct, IP_CT_DIR_ORIGINAL), GFP_ATOMIC);
 	if (!skb)
-		return NOTIFY_DONE;
+		goto errout;
 
 	b = skb->tail;
 
@@ -591,8 +591,9 @@ static int ctnetlink_conntrack_event(struct notifier_block *this,
 nla_put_failure:
 	rcu_read_unlock();
 nlmsg_failure:
-	nfnetlink_set_err(0, group, -ENOBUFS);
 	kfree_skb(skb);
+errout:
+	nfnetlink_set_err(0, group, -ENOBUFS);
 	return NOTIFY_DONE;
 }
 #endif /* CONFIG_NF_CONNTRACK_EVENTS */
author maintainer
#4 lore raw

PATCH 2/2 netfilter: nfnetlink: return ENOMEM if we fail to create netlink socket

In reply to Pablo Neira Ayuso

With this patch, nfnetlink returns -ENOMEM instead of -EPERM if we fail to create the nfnetlink netlink socket during the module loading. This is exactly what rtnetlink does in this case.

Ideally, it would be better if we propagate the error that has happened in netlink_kernel_create(), however, this function still does not implement this yet.

---

 net/netfilter/nfnetlink.c |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)
net/netfilter/nfnetlink.c +1 -1
index 2785d66..b8ab37a 100644
@@ -203,7 +203,7 @@ static int __init nfnetlink_init(void)
 				     nfnetlink_rcv, NULL, THIS_MODULE);
 	if (!nfnl) {
 		printk(KERN_ERR "cannot initialize nfnetlink!\n");
-		return -1;
+		return -ENOMEM;
 	}
 
 	return 0;
author maintainer
#2 lore raw

Re: [PATCH 1/2] netfilter: ctnetlink: report error if event message allocation fails

In reply to Pablo Neira Ayuso
Quoted 2 lines · Pablo Neira Ayuso

This patch fixes an inconsistency that results in no error reports to user-space listeners if we fail to allocate the event message.

I was missing the expectation part. New patch attached.

Signature · 1 line
"Los honestos son inadaptados sociales" -- Les Luthiers

netfilter: ctnetlink: report error if event message allocation fails

This patch fixes an inconsistency that results in no error reports to user-space listeners if we fail to allocate the event message.

---

 net/netfilter/nf_conntrack_netlink.c |   10 ++++++----
 1 files changed, 6 insertions(+), 4 deletions(-)
net/netfilter/nf_conntrack_netlink.c +6 -4
index c6439c7..0ea36e0 100644
@@ -512,7 +512,7 @@ static int ctnetlink_conntrack_event(struct notifier_block *this,
 
 	skb = ctnetlink_alloc_skb(tuple(ct, IP_CT_DIR_ORIGINAL), GFP_ATOMIC);
 	if (!skb)
-		return NOTIFY_DONE;
+		goto errout;
 
 	b = skb->tail;
 
@@ -591,8 +591,9 @@ static int ctnetlink_conntrack_event(struct notifier_block *this,
 nla_put_failure:
 	rcu_read_unlock();
 nlmsg_failure:
-	nfnetlink_set_err(0, group, -ENOBUFS);
 	kfree_skb(skb);
+errout:
+	nfnetlink_set_err(0, group, -ENOBUFS);
 	return NOTIFY_DONE;
 }
 #endif /* CONFIG_NF_CONNTRACK_EVENTS */
@@ -1564,7 +1565,7 @@ static int ctnetlink_expect_event(struct notifier_block *this,
 
 	skb = alloc_skb(NLMSG_GOODSIZE, GFP_ATOMIC);
 	if (!skb)
-		return NOTIFY_DONE;
+		goto errout;
 
 	b = skb->tail;
 
@@ -1589,8 +1590,9 @@ static int ctnetlink_expect_event(struct notifier_block *this,
 nla_put_failure:
 	rcu_read_unlock();
 nlmsg_failure:
-	nfnetlink_set_err(0, 0, -ENOBUFS);
 	kfree_skb(skb);
+errout:
+	nfnetlink_set_err(0, 0, -ENOBUFS);
 	return NOTIFY_DONE;
 }
 #endif
#3 lore raw

Re: [PATCH 1/2] netfilter: ctnetlink: report error if event message allocation fails

Quoted 5 lines · Pablo Neira Ayuso
Quoted 2 lines · Pablo Neira Ayuso

This patch fixes an inconsistency that results in no error reports to user-space listeners if we fail to allocate the event message.

I was missing the expectation part. New patch attached.

Applied, thanks Pablo.

#5 lore raw

Re: [PATCH 2/2] netfilter: nfnetlink: return ENOMEM if we fail to create netlink socket

In reply to Pablo Neira Ayuso
Quoted 3 lines · Pablo Neira Ayuso

With this patch, nfnetlink returns -ENOMEM instead of -EPERM if we fail to create the nfnetlink netlink socket during the module loading. This is exactly what rtnetlink does in this case.

Also applied, thanks.

Quoted 3 lines

Ideally, it would be better if we propagate the error that has happened in netlink_kernel_create(), however, this function still does not implement this yet.

Agreed, but AFAICS netlink_kernel_create() only returns an error on memory shortage, so this is fine currently.