670-ipv6-allow-rejecting-with-source-address-failed-policy.patch

From 1b5aaa4b16f6e6471ab1c07b38068197a1b4c395 Mon Sep 17 00:00:00 2001
From: Jonas Gorski <jogo@openwrt.org>
Date: Fri, 24 May 2013 14:40:54 +0200
Subject: [PATCH 1/2] ipv6: allow rejecting with "source address failed policy"

RFC6204 L-14 requires rejecting traffic from invalid addresses with
ICMPv6 Destination Unreachable, Code 5 (Source address failed ingress/
egress policy) on the LAN side, so add an appropriate rule for that.

Signed-off-by: Jonas Gorski <jogo@openwrt.org>
---
 include/net/netns/ipv6.h       |  1 +
 include/uapi/linux/fib_rules.h |  4 +++
 include/uapi/linux/rtnetlink.h |  1 +
 net/ipv4/fib_semantics.c       |  4 +++
 net/ipv4/fib_trie.c            |  1 +
 net/ipv4/ipmr.c                |  1 +
 net/ipv6/fib6_rules.c          |  4 +++
 net/ipv6/ip6mr.c               |  2 ++
 net/ipv6/route.c               | 58 +++++++++++++++++++++++++++++++++++++++++-
 9 files changed, 75 insertions(+), 1 deletion(-)

--- a/include/net/netns/ipv6.h
+++ b/include/net/netns/ipv6.h
@@ -63,6 +63,7 @@ struct netns_ipv6 {
 	unsigned long		 ip6_rt_last_gc;
 #ifdef CONFIG_IPV6_MULTIPLE_TABLES
 	struct rt6_info         *ip6_prohibit_entry;
+	struct rt6_info		*ip6_policy_failed_entry;
 	struct rt6_info         *ip6_blk_hole_entry;
 	struct fib6_table       *fib6_local_tbl;
 	struct fib_rules_ops    *fib6_rules_ops;
--- a/include/uapi/linux/fib_rules.h
+++ b/include/uapi/linux/fib_rules.h
@@ -64,6 +64,10 @@ enum {
 	FR_ACT_BLACKHOLE,	/* Drop without notification */
 	FR_ACT_UNREACHABLE,	/* Drop with ENETUNREACH */
 	FR_ACT_PROHIBIT,	/* Drop with EACCES */
+	FR_ACT_RES9,
+	FR_ACT_RES10,
+	FR_ACT_RES11,
+	FR_ACT_POLICY_FAILED,	/* Drop with EACCES */
 	__FR_ACT_MAX,
 };
 
--- a/include/uapi/linux/rtnetlink.h
+++ b/include/uapi/linux/rtnetlink.h
@@ -210,6 +210,7 @@ enum {
 	RTN_THROW,		/* Not in this table		*/
 	RTN_NAT,		/* Translate this address	*/
 	RTN_XRESOLVE,		/* Use external resolver	*/
+	RTN_POLICY_FAILED,	/* Failed ingress/egress policy */
 	__RTN_MAX
 };
 
--- a/net/ipv4/fib_semantics.c
+++ b/net/ipv4/fib_semantics.c
@@ -138,6 +138,10 @@ const struct fib_prop fib_props[RTN_MAX
 		.error	= -EINVAL,
 		.scope	= RT_SCOPE_NOWHERE,
 	},
+	[RTN_POLICY_FAILED] = {
+		.error	= -EACCES,
+		.scope	= RT_SCOPE_UNIVERSE,
+	},
 };
 
 static void rt_fibinfo_free(struct rtable __rcu **rtp)
--- a/net/ipv4/fib_trie.c
+++ b/net/ipv4/fib_trie.c
@@ -2368,6 +2368,7 @@ static const char *const rtn_type_names[
 	[RTN_THROW] = "THROW",
 	[RTN_NAT] = "NAT",
 	[RTN_XRESOLVE] = "XRESOLVE",
+	[RTN_POLICY_FAILED] = "POLICY_FAILED",
 };
 
 static inline const char *rtn_type(char *buf, size_t len, unsigned int t)
--- a/net/ipv4/ipmr.c
+++ b/net/ipv4/ipmr.c
@@ -182,6 +182,7 @@ static int ipmr_rule_action(struct fib_r
 	case FR_ACT_UNREACHABLE:
 		return -ENETUNREACH;
 	case FR_ACT_PROHIBIT:
+	case FR_ACT_POLICY_FAILED:
 		return -EACCES;
 	case FR_ACT_BLACKHOLE:
 	default:
--- a/net/ipv6/fib6_rules.c
+++ b/net/ipv6/fib6_rules.c
@@ -84,6 +84,10 @@ static int fib6_rule_action(struct fib_r
 		err = -EACCES;
 		rt = net->ipv6.ip6_prohibit_entry;
 		goto discard_pkt;
+	case FR_ACT_POLICY_FAILED:
+		err = -EACCES;
+		rt = net->ipv6.ip6_policy_failed_entry;
+		goto discard_pkt;
 	}
 
 	table = fib6_get_table(net, rule->table);
--- a/net/ipv6/ip6mr.c
+++ b/net/ipv6/ip6mr.c
@@ -167,6 +167,8 @@ static int ip6mr_rule_action(struct fib_
 		return -ENETUNREACH;
 	case FR_ACT_PROHIBIT:
 		return -EACCES;
+	case FR_ACT_POLICY_FAILED:
+		return -EACCES;
 	case FR_ACT_BLACKHOLE:
 	default:
 		return -EINVAL;
--- a/net/ipv6/route.c
+++ b/net/ipv6/route.c
@@ -90,6 +90,8 @@ static int		ip6_pkt_discard(struct sk_bu
 static int		ip6_pkt_discard_out(struct net *net, struct sock *sk, struct sk_buff *skb);
 static int		ip6_pkt_prohibit(struct sk_buff *skb);
 static int		ip6_pkt_prohibit_out(struct net *net, struct sock *sk, struct sk_buff *skb);
+static int		ip6_pkt_policy_failed(struct sk_buff *skb);
+static int		ip6_pkt_policy_failed_out(struct net *net, struct sock *sk, struct sk_buff *skb);
 static void		ip6_link_failure(struct sk_buff *skb);
 static void		ip6_rt_update_pmtu(struct dst_entry *dst, struct sock *sk,
 					   struct sk_buff *skb, u32 mtu);
@@ -297,6 +299,21 @@ static const struct rt6_info ip6_prohibi
 	.rt6i_ref	= ATOMIC_INIT(1),
 };
 
+static const struct rt6_info ip6_policy_failed_entry_template = {
+	.dst = {
+		.__refcnt	= ATOMIC_INIT(1),
+		.__use		= 1,
+		.obsolete	= DST_OBSOLETE_FORCE_CHK,
+		.error		= -EACCES,
+		.input		= ip6_pkt_policy_failed,
+		.output		= ip6_pkt_policy_failed_out,
+	},
+	.rt6i_flags	= (RTF_REJECT | RTF_NONEXTHOP),
+	.rt6i_protocol	= RTPROT_KERNEL,
+	.rt6i_metric	= ~(u32) 0,
+	.rt6i_ref	= ATOMIC_INIT(1),
+};
+
 static const struct rt6_info ip6_blk_hole_entry_template = {
 	.dst = {
 		.__refcnt	= ATOMIC_INIT(1),
@@ -1884,6 +1901,11 @@ static struct rt6_info *ip6_route_info_c
 			rt->dst.output = ip6_pkt_prohibit_out;
 			rt->dst.input = ip6_pkt_prohibit;
 			break;
+		case RTN_POLICY_FAILED:
+			rt->dst.error = -EACCES;
+			rt->dst.output = ip6_pkt_policy_failed_out;
+			rt->dst.input = ip6_pkt_policy_failed;
+			break;
 		case RTN_THROW:
 		case RTN_UNREACHABLE:
 		default:
@@ -2485,6 +2507,17 @@ static int ip6_pkt_prohibit_out(struct n
 	return ip6_pkt_drop(skb, ICMPV6_ADM_PROHIBITED, IPSTATS_MIB_OUTNOROUTES);
 }
 
+static int ip6_pkt_policy_failed(struct sk_buff *skb)
+{
+	return ip6_pkt_drop(skb, ICMPV6_POLICY_FAIL, IPSTATS_MIB_INNOROUTES);
+}
+
+static int ip6_pkt_policy_failed_out(struct net *net, struct sock *sk, struct sk_buff *skb)
+{
+	skb->dev = skb_dst(skb)->dev;
+	return ip6_pkt_drop(skb, ICMPV6_POLICY_FAIL, IPSTATS_MIB_OUTNOROUTES);
+}
+
 /*
  *	Allocate a dst for local (unicast / anycast) address.
  */
@@ -2727,7 +2760,8 @@ static int rtm_to_fib6_config(struct sk_
 	if (rtm->rtm_type == RTN_UNREACHABLE ||
 	    rtm->rtm_type == RTN_BLACKHOLE ||
 	    rtm->rtm_type == RTN_PROHIBIT ||
-	    rtm->rtm_type == RTN_THROW)
+	    rtm->rtm_type == RTN_THROW ||
+	    rtm->rtm_type == RTN_POLICY_FAILED)
 		cfg->fc_flags |= RTF_REJECT;
 
 	if (rtm->rtm_type == RTN_LOCAL)
@@ -3086,6 +3120,9 @@ static int rt6_fill_node(struct net *net
 		case -EACCES:
 			rtm->rtm_type = RTN_PROHIBIT;
 			break;
+		case -EPERM:
+			rtm->rtm_type = RTN_POLICY_FAILED;
+			break;
 		case -EAGAIN:
 			rtm->rtm_type = RTN_THROW;
 			break;
@@ -3359,6 +3396,8 @@ static int ip6_route_dev_notify(struct n
 #ifdef CONFIG_IPV6_MULTIPLE_TABLES
 		net->ipv6.ip6_prohibit_entry->dst.dev = dev;
 		net->ipv6.ip6_prohibit_entry->rt6i_idev = in6_dev_get(dev);
+		net->ipv6.ip6_policy_failed_entry->dst.dev = dev;
+		net->ipv6.ip6_policy_failed_entry->rt6i_idev = in6_dev_get(dev);
 		net->ipv6.ip6_blk_hole_entry->dst.dev = dev;
 		net->ipv6.ip6_blk_hole_entry->rt6i_idev = in6_dev_get(dev);
 #endif
@@ -3387,6 +3425,7 @@
 		in6_dev_put(net->ipv6.ip6_null_entry->rt6i_idev);
 #ifdef CONFIG_IPV6_MULTIPLE_TABLES
 		in6_dev_put(net->ipv6.ip6_prohibit_entry->rt6i_idev);
+		in6_dev_put(net->ipv6.ip6_policy_failed_entry->rt6i_idev);
 		in6_dev_put(net->ipv6.ip6_blk_hole_entry->rt6i_idev);
 #endif
 	}
@@ -3575,6 +3614,17 @@ static int __net_init ip6_route_net_init
 	net->ipv6.ip6_blk_hole_entry->dst.ops = &net->ipv6.ip6_dst_ops;
 	dst_init_metrics(&net->ipv6.ip6_blk_hole_entry->dst,
 			 ip6_template_metrics, true);
+
+	net->ipv6.ip6_policy_failed_entry =
+		kmemdup(&ip6_policy_failed_entry_template,
+			sizeof(*net->ipv6.ip6_policy_failed_entry), GFP_KERNEL);
+	if (!net->ipv6.ip6_policy_failed_entry)
+		goto out_ip6_blk_hole_entry;
+	net->ipv6.ip6_policy_failed_entry->dst.path =
+		(struct dst_entry *)net->ipv6.ip6_policy_failed_entry;
+	net->ipv6.ip6_policy_failed_entry->dst.ops = &net->ipv6.ip6_dst_ops;
+	dst_init_metrics(&net->ipv6.ip6_policy_failed_entry->dst,
+			 ip6_template_metrics, true);
 #endif
 
 	net->ipv6.sysctl.flush_delay = 0;
@@ -3593,6 +3643,8 @@ out:
 	return ret;
 
 #ifdef CONFIG_IPV6_MULTIPLE_TABLES
+out_ip6_blk_hole_entry:
+	kfree(net->ipv6.ip6_blk_hole_entry);
 out_ip6_prohibit_entry:
 	kfree(net->ipv6.ip6_prohibit_entry);
 out_ip6_null_entry:
@@ -3610,6 +3662,7 @@ static void __net_exit ip6_route_net_exi
 #ifdef CONFIG_IPV6_MULTIPLE_TABLES
 	kfree(net->ipv6.ip6_prohibit_entry);
 	kfree(net->ipv6.ip6_blk_hole_entry);
+	kfree(net->ipv6.ip6_policy_failed_entry);
 #endif
 	dst_entries_destroy(&net->ipv6.ip6_dst_ops);
 }
@@ -3707,6 +3760,9 @@ int __init ip6_route_init(void)
 	init_net.ipv6.ip6_prohibit_entry->rt6i_idev = in6_dev_get(init_net.loopback_dev);
 	init_net.ipv6.ip6_blk_hole_entry->dst.dev = init_net.loopback_dev;
 	init_net.ipv6.ip6_blk_hole_entry->rt6i_idev = in6_dev_get(init_net.loopback_dev);
+	init_net.ipv6.ip6_policy_failed_entry->dst.dev = init_net.loopback_dev;
+	init_net.ipv6.ip6_policy_failed_entry->rt6i_idev =
+		in6_dev_get(init_net.loopback_dev);
   #endif
 	ret = fib6_init();
 	if (ret)

Hi there! I nearly gave up on getting LXC to run reliably on a recent version of OpenWrt until i stumbled upon your bugfix for this patch via the Turris OS forum. Since this fix doens't seem to have made it back into the OpenWrt repository yet, i took the liberty of filing a bug report with OpenWrt so it can be integrated with the official OpenWrt sources. I already had the strong suspicion that this problem was introduced by an OpenWrt itself as all other reports of "waiting for lo to become free"-related issues had a hard time finding reliable reproducers while on OpenWrt it happened almost immediately every time a network namespace was destroyed.

Thanks for your good work in finding this, it saved me a lot of headaches trying to circumvent it (wit persistent namespaces and whatnot)!