[RFC PATCH 0/2] nft: improve reject support

[RFC PATCH 0/2] nft: improve reject support

[Eric Leblond]

Hello,

Here's a small patchset implementing IPv6 reject as well as TCP RST based
reject. It uses the same factorisation technique as the one used for
nft_queue. I've tested ICMP reject and still need to fix nftables to be
able to test TCP reset. This explains in part the RFC status of the patch.

Patchset statistics:
 include/net/netfilter/nf_reject.h    | 297 +++++++++++++++++++++++++++++++++++
 net/ipv4/netfilter/Kconfig           |   4 -
 net/ipv4/netfilter/Makefile          |   1 -
 net/ipv4/netfilter/ipt_REJECT.c      | 124 +--------------
 net/ipv4/netfilter/nft_reject_ipv4.c | 123 ---------------
 net/ipv6/netfilter/ip6t_REJECT.c     | 177 +--------------------
 net/netfilter/Kconfig                |   4 +
 net/netfilter/Makefile               |   1 +
 net/netfilter/nft_reject.c           | 143 +++++++++++++++++
 9 files changed, 455 insertions(+), 419 deletions(-)

BR,
--
Eric

[PATCH 1/2] netfilter: REJECT: separate reusable code

[Eric Leblond]

This patch prepares the addition of TCP reset support in
the nft_reject module by moving reusable code into a header
file.

Signed-off-by: Eric Leblond <eric@regit.org>
---
 include/net/netfilter/nf_reject.h | 297 ++++++++++++++++++++++++++++++++++++++
 net/ipv4/netfilter/ipt_REJECT.c   | 124 +---------------
 net/ipv6/netfilter/ip6t_REJECT.c  | 177 +----------------------
 3 files changed, 307 insertions(+), 291 deletions(-)
 create mode 100644 include/net/netfilter/nf_reject.h

diff --git a/include/net/netfilter/nf_reject.h b/include/net/netfilter/nf_reject.h
new file mode 100644
index 0000000..0bb089a
--- /dev/null
+++ b/include/net/netfilter/nf_reject.h
@@ -0,0 +1,297 @@
+#ifndef _NF_REJECT_H
+#define _NF_REJECT_H
+
+#include <net/ip.h>
+#include <net/tcp.h>
+#include <net/route.h>
+#include <net/dst.h>
+
+#if IS_ENABLED(CONFIG_IP6_NF_IPTABLES)
+#include <net/ipv6.h>
+#include <net/ip6_route.h>
+#include <linux/netfilter_ipv6.h>
+#endif
+
+static inline void send_unreach(struct sk_buff *skb_in, int code)
+{
+	icmp_send(skb_in, ICMP_DEST_UNREACH, code, 0);
+}
+
+/* Send RST reply */
+static inline void send_reset(struct sk_buff *oldskb, int hook)
+{
+	struct sk_buff *nskb;
+	const struct iphdr *oiph;
+	struct iphdr *niph;
+	const struct tcphdr *oth;
+	struct tcphdr _otcph, *tcph;
+
+	/* IP header checks: fragment. */
+	if (ip_hdr(oldskb)->frag_off & htons(IP_OFFSET))
+		return;
+
+	oth = skb_header_pointer(oldskb, ip_hdrlen(oldskb),
+				 sizeof(_otcph), &_otcph);
+	if (oth == NULL)
+		return;
+
+	/* No RST for RST. */
+	if (oth->rst)
+		return;
+
+	if (skb_rtable(oldskb)->rt_flags & (RTCF_BROADCAST | RTCF_MULTICAST))
+		return;
+
+	/* Check checksum */
+	if (nf_ip_checksum(oldskb, hook, ip_hdrlen(oldskb), IPPROTO_TCP))
+		return;
+	oiph = ip_hdr(oldskb);
+
+	nskb = alloc_skb(sizeof(struct iphdr) + sizeof(struct tcphdr) +
+			 LL_MAX_HEADER, GFP_ATOMIC);
+	if (!nskb)
+		return;
+
+	skb_reserve(nskb, LL_MAX_HEADER);
+
+	skb_reset_network_header(nskb);
+	niph = (struct iphdr *)skb_put(nskb, sizeof(struct iphdr));
+	niph->version	= 4;
+	niph->ihl	= sizeof(struct iphdr) / 4;
+	niph->tos	= 0;
+	niph->id	= 0;
+	niph->frag_off	= htons(IP_DF);
+	niph->protocol	= IPPROTO_TCP;
+	niph->check	= 0;
+	niph->saddr	= oiph->daddr;
+	niph->daddr	= oiph->saddr;
+
+	skb_reset_transport_header(nskb);
+	tcph = (struct tcphdr *)skb_put(nskb, sizeof(struct tcphdr));
+	memset(tcph, 0, sizeof(*tcph));
+	tcph->source	= oth->dest;
+	tcph->dest	= oth->source;
+	tcph->doff	= sizeof(struct tcphdr) / 4;
+
+	if (oth->ack)
+		tcph->seq = oth->ack_seq;
+	else {
+		tcph->ack_seq = htonl(ntohl(oth->seq) + oth->syn + oth->fin +
+				      oldskb->len - ip_hdrlen(oldskb) -
+				      (oth->doff << 2));
+		tcph->ack = 1;
+	}
+
+	tcph->rst	= 1;
+	tcph->check = ~tcp_v4_check(sizeof(struct tcphdr), niph->saddr,
+				    niph->daddr, 0);
+	nskb->ip_summed = CHECKSUM_PARTIAL;
+	nskb->csum_start = (unsigned char *)tcph - nskb->head;
+	nskb->csum_offset = offsetof(struct tcphdr, check);
+
+	/* ip_route_me_harder expects skb->dst to be set */
+	skb_dst_set_noref(nskb, skb_dst(oldskb));
+
+	nskb->protocol = htons(ETH_P_IP);
+	if (ip_route_me_harder(nskb, RTN_UNSPEC))
+		goto free_nskb;
+
+	niph->ttl	= ip4_dst_hoplimit(skb_dst(nskb));
+
+	/* "Never happens" */
+	if (nskb->len > dst_mtu(skb_dst(nskb)))
+		goto free_nskb;
+
+	nf_ct_attach(nskb, oldskb);
+
+#ifdef CONFIG_BRIDGE_NETFILTER
+	/* If we use ip_local_out for bridged traffic, the MAC source on
+	 * the RST will be ours, instead of the destination's.  This confuses
+	 * some routers/firewalls, and they drop the packet.  So we need to
+	 * build the eth header using the original destination's MAC as the
+	 * source, and send the RST packet directly.
+	 */
+	if (oldskb->nf_bridge) {
+		struct ethhdr *oeth = eth_hdr(oldskb);
+		nskb->dev = oldskb->nf_bridge->physindev;
+		niph->tot_len = htons(nskb->len);
+		ip_send_check(niph);
+		if (dev_hard_header(nskb, nskb->dev, ntohs(nskb->protocol),
+				    oeth->h_source, oeth->h_dest, nskb->len) < 0)
+			goto free_nskb;
+		dev_queue_xmit(nskb);
+	} else
+#endif
+		ip_local_out(nskb);
+
+	return;
+
+ free_nskb:
+	kfree_skb(nskb);
+}
+
+#if IS_ENABLED(CONFIG_IP6_NF_IPTABLES)
+
+static inline void
+send_unreach6(struct net *net, struct sk_buff *skb_in, unsigned char code,
+	     unsigned int hooknum)
+{
+	if (hooknum == NF_INET_LOCAL_OUT && skb_in->dev == NULL)
+		skb_in->dev = net->loopback_dev;
+
+	icmpv6_send(skb_in, ICMPV6_DEST_UNREACH, code, 0);
+}
+
+/* Send RST reply */
+static inline void send_reset6(struct net *net, struct sk_buff *oldskb, int hook)
+{
+	struct sk_buff *nskb;
+	struct tcphdr otcph, *tcph;
+	unsigned int otcplen, hh_len;
+	int tcphoff, needs_ack;
+	const struct ipv6hdr *oip6h = ipv6_hdr(oldskb);
+	struct ipv6hdr *ip6h;
+#define DEFAULT_TOS_VALUE	0x0U
+	const __u8 tclass = DEFAULT_TOS_VALUE;
+	struct dst_entry *dst = NULL;
+	u8 proto;
+	__be16 frag_off;
+	struct flowi6 fl6;
+
+	if ((!(ipv6_addr_type(&oip6h->saddr) & IPV6_ADDR_UNICAST)) ||
+	    (!(ipv6_addr_type(&oip6h->daddr) & IPV6_ADDR_UNICAST))) {
+		pr_debug("addr is not unicast.\n");
+		return;
+	}
+
+	proto = oip6h->nexthdr;
+	tcphoff = ipv6_skip_exthdr(oldskb, ((u8*)(oip6h+1) - oldskb->data), &proto, &frag_off);
+
+	if ((tcphoff < 0) || (tcphoff > oldskb->len)) {
+		pr_debug("Cannot get TCP header.\n");
+		return;
+	}
+
+	otcplen = oldskb->len - tcphoff;
+
+	/* IP header checks: fragment, too short. */
+	if (proto != IPPROTO_TCP || otcplen < sizeof(struct tcphdr)) {
+		pr_debug("proto(%d) != IPPROTO_TCP, "
+			 "or too short. otcplen = %d\n",
+			 proto, otcplen);
+		return;
+	}
+
+	if (skb_copy_bits(oldskb, tcphoff, &otcph, sizeof(struct tcphdr)))
+		BUG();
+
+	/* No RST for RST. */
+	if (otcph.rst) {
+		pr_debug("RST is set\n");
+		return;
+	}
+
+	/* Check checksum. */
+	if (nf_ip6_checksum(oldskb, hook, tcphoff, IPPROTO_TCP)) {
+		pr_debug("TCP checksum is invalid\n");
+		return;
+	}
+
+	memset(&fl6, 0, sizeof(fl6));
+	fl6.flowi6_proto = IPPROTO_TCP;
+	fl6.saddr = oip6h->daddr;
+	fl6.daddr = oip6h->saddr;
+	fl6.fl6_sport = otcph.dest;
+	fl6.fl6_dport = otcph.source;
+	security_skb_classify_flow(oldskb, flowi6_to_flowi(&fl6));
+	dst = ip6_route_output(net, NULL, &fl6);
+	if (dst == NULL || dst->error) {
+		dst_release(dst);
+		return;
+	}
+	dst = xfrm_lookup(net, dst, flowi6_to_flowi(&fl6), NULL, 0);
+	if (IS_ERR(dst))
+		return;
+
+	hh_len = (dst->dev->hard_header_len + 15)&~15;
+	nskb = alloc_skb(hh_len + 15 + dst->header_len + sizeof(struct ipv6hdr)
+			 + sizeof(struct tcphdr) + dst->trailer_len,
+			 GFP_ATOMIC);
+
+	if (!nskb) {
+		net_dbg_ratelimited("cannot alloc skb\n");
+		dst_release(dst);
+		return;
+	}
+
+	skb_dst_set(nskb, dst);
+
+	skb_reserve(nskb, hh_len + dst->header_len);
+
+	skb_put(nskb, sizeof(struct ipv6hdr));
+	skb_reset_network_header(nskb);
+	ip6h = ipv6_hdr(nskb);
+	ip6_flow_hdr(ip6h, tclass, 0);
+	ip6h->hop_limit = ip6_dst_hoplimit(dst);
+	ip6h->nexthdr = IPPROTO_TCP;
+	ip6h->saddr = oip6h->daddr;
+	ip6h->daddr = oip6h->saddr;
+
+	skb_reset_transport_header(nskb);
+	tcph = (struct tcphdr *)skb_put(nskb, sizeof(struct tcphdr));
+	/* Truncate to length (no data) */
+	tcph->doff = sizeof(struct tcphdr)/4;
+	tcph->source = otcph.dest;
+	tcph->dest = otcph.source;
+
+	if (otcph.ack) {
+		needs_ack = 0;
+		tcph->seq = otcph.ack_seq;
+		tcph->ack_seq = 0;
+	} else {
+		needs_ack = 1;
+		tcph->ack_seq = htonl(ntohl(otcph.seq) + otcph.syn + otcph.fin
+				      + otcplen - (otcph.doff<<2));
+		tcph->seq = 0;
+	}
+
+	/* Reset flags */
+	((u_int8_t *)tcph)[13] = 0;
+	tcph->rst = 1;
+	tcph->ack = needs_ack;
+	tcph->window = 0;
+	tcph->urg_ptr = 0;
+	tcph->check = 0;
+
+	/* Adjust TCP checksum */
+	tcph->check = csum_ipv6_magic(&ipv6_hdr(nskb)->saddr,
+				      &ipv6_hdr(nskb)->daddr,
+				      sizeof(struct tcphdr), IPPROTO_TCP,
+				      csum_partial(tcph,
+						   sizeof(struct tcphdr), 0));
+
+	nf_ct_attach(nskb, oldskb);
+
+#ifdef CONFIG_BRIDGE_NETFILTER
+	/* If we use ip6_local_out for bridged traffic, the MAC source on
+	 * the RST will be ours, instead of the destination's.  This confuses
+	 * some routers/firewalls, and they drop the packet.  So we need to
+	 * build the eth header using the original destination's MAC as the
+	 * source, and send the RST packet directly.
+	 */
+	if (oldskb->nf_bridge) {
+		struct ethhdr *oeth = eth_hdr(oldskb);
+		nskb->dev = oldskb->nf_bridge->physindev;
+		nskb->protocol = htons(ETH_P_IPV6);
+		ip6h->payload_len = htons(sizeof(struct tcphdr));
+		if (dev_hard_header(nskb, nskb->dev, ntohs(nskb->protocol),
+				    oeth->h_source, oeth->h_dest, nskb->len) < 0)
+			return;
+		dev_queue_xmit(nskb);
+	} else
+#endif
+		ip6_local_out(nskb);
+}
+
+#endif /* CONFIG_IP6_NF_IPTABLES */
+#endif /* _NF_REJECT_H */
diff --git a/net/ipv4/netfilter/ipt_REJECT.c b/net/ipv4/netfilter/ipt_REJECT.c
index b969131..f40305f 100644
--- a/net/ipv4/netfilter/ipt_REJECT.c
+++ b/net/ipv4/netfilter/ipt_REJECT.c
@@ -17,10 +17,6 @@
 #include <linux/udp.h>
 #include <linux/icmp.h>
 #include <net/icmp.h>
-#include <net/ip.h>
-#include <net/tcp.h>
-#include <net/route.h>
-#include <net/dst.h>
 #include <linux/netfilter/x_tables.h>
 #include <linux/netfilter_ipv4/ip_tables.h>
 #include <linux/netfilter_ipv4/ipt_REJECT.h>
@@ -28,128 +24,12 @@
 #include <linux/netfilter_bridge.h>
 #endif
 
+#include <net/netfilter/nf_reject.h>
+
 MODULE_LICENSE("GPL");
 MODULE_AUTHOR("Netfilter Core Team <coreteam@netfilter.org>");
 MODULE_DESCRIPTION("Xtables: packet \"rejection\" target for IPv4");
 
-/* Send RST reply */
-static void send_reset(struct sk_buff *oldskb, int hook)
-{
-	struct sk_buff *nskb;
-	const struct iphdr *oiph;
-	struct iphdr *niph;
-	const struct tcphdr *oth;
-	struct tcphdr _otcph, *tcph;
-
-	/* IP header checks: fragment. */
-	if (ip_hdr(oldskb)->frag_off & htons(IP_OFFSET))
-		return;
-
-	oth = skb_header_pointer(oldskb, ip_hdrlen(oldskb),
-				 sizeof(_otcph), &_otcph);
-	if (oth == NULL)
-		return;
-
-	/* No RST for RST. */
-	if (oth->rst)
-		return;
-
-	if (skb_rtable(oldskb)->rt_flags & (RTCF_BROADCAST | RTCF_MULTICAST))
-		return;
-
-	/* Check checksum */
-	if (nf_ip_checksum(oldskb, hook, ip_hdrlen(oldskb), IPPROTO_TCP))
-		return;
-	oiph = ip_hdr(oldskb);
-
-	nskb = alloc_skb(sizeof(struct iphdr) + sizeof(struct tcphdr) +
-			 LL_MAX_HEADER, GFP_ATOMIC);
-	if (!nskb)
-		return;
-
-	skb_reserve(nskb, LL_MAX_HEADER);
-
-	skb_reset_network_header(nskb);
-	niph = (struct iphdr *)skb_put(nskb, sizeof(struct iphdr));
-	niph->version	= 4;
-	niph->ihl	= sizeof(struct iphdr) / 4;
-	niph->tos	= 0;
-	niph->id	= 0;
-	niph->frag_off	= htons(IP_DF);
-	niph->protocol	= IPPROTO_TCP;
-	niph->check	= 0;
-	niph->saddr	= oiph->daddr;
-	niph->daddr	= oiph->saddr;
-
-	skb_reset_transport_header(nskb);
-	tcph = (struct tcphdr *)skb_put(nskb, sizeof(struct tcphdr));
-	memset(tcph, 0, sizeof(*tcph));
-	tcph->source	= oth->dest;
-	tcph->dest	= oth->source;
-	tcph->doff	= sizeof(struct tcphdr) / 4;
-
-	if (oth->ack)
-		tcph->seq = oth->ack_seq;
-	else {
-		tcph->ack_seq = htonl(ntohl(oth->seq) + oth->syn + oth->fin +
-				      oldskb->len - ip_hdrlen(oldskb) -
-				      (oth->doff << 2));
-		tcph->ack = 1;
-	}
-
-	tcph->rst	= 1;
-	tcph->check = ~tcp_v4_check(sizeof(struct tcphdr), niph->saddr,
-				    niph->daddr, 0);
-	nskb->ip_summed = CHECKSUM_PARTIAL;
-	nskb->csum_start = (unsigned char *)tcph - nskb->head;
-	nskb->csum_offset = offsetof(struct tcphdr, check);
-
-	/* ip_route_me_harder expects skb->dst to be set */
-	skb_dst_set_noref(nskb, skb_dst(oldskb));
-
-	nskb->protocol = htons(ETH_P_IP);
-	if (ip_route_me_harder(nskb, RTN_UNSPEC))
-		goto free_nskb;
-
-	niph->ttl	= ip4_dst_hoplimit(skb_dst(nskb));
-
-	/* "Never happens" */
-	if (nskb->len > dst_mtu(skb_dst(nskb)))
-		goto free_nskb;
-
-	nf_ct_attach(nskb, oldskb);
-
-#ifdef CONFIG_BRIDGE_NETFILTER
-	/* If we use ip_local_out for bridged traffic, the MAC source on
-	 * the RST will be ours, instead of the destination's.  This confuses
-	 * some routers/firewalls, and they drop the packet.  So we need to
-	 * build the eth header using the original destination's MAC as the
-	 * source, and send the RST packet directly.
-	 */
-	if (oldskb->nf_bridge) {
-		struct ethhdr *oeth = eth_hdr(oldskb);
-		nskb->dev = oldskb->nf_bridge->physindev;
-		niph->tot_len = htons(nskb->len);
-		ip_send_check(niph);
-		if (dev_hard_header(nskb, nskb->dev, ntohs(nskb->protocol),
-				    oeth->h_source, oeth->h_dest, nskb->len) < 0)
-			goto free_nskb;
-		dev_queue_xmit(nskb);
-	} else
-#endif
-		ip_local_out(nskb);
-
-	return;
-
- free_nskb:
-	kfree_skb(nskb);
-}
-
-static inline void send_unreach(struct sk_buff *skb_in, int code)
-{
-	icmp_send(skb_in, ICMP_DEST_UNREACH, code, 0);
-}
-
 static unsigned int
 reject_tg(struct sk_buff *skb, const struct xt_action_param *par)
 {
diff --git a/net/ipv6/netfilter/ip6t_REJECT.c b/net/ipv6/netfilter/ip6t_REJECT.c
index da00a2e..3461c3c 100644
--- a/net/ipv6/netfilter/ip6t_REJECT.c
+++ b/net/ipv6/netfilter/ip6t_REJECT.c
@@ -23,181 +23,20 @@
 #include <linux/skbuff.h>
 #include <linux/icmpv6.h>
 #include <linux/netdevice.h>
-#include <net/ipv6.h>
-#include <net/tcp.h>
 #include <net/icmp.h>
 #include <net/ip6_checksum.h>
 #include <net/ip6_fib.h>
-#include <net/ip6_route.h>
 #include <net/flow.h>
 #include <linux/netfilter/x_tables.h>
 #include <linux/netfilter_ipv6/ip6_tables.h>
 #include <linux/netfilter_ipv6/ip6t_REJECT.h>
 
+#include <net/netfilter/nf_reject.h>
+
 MODULE_AUTHOR("Yasuyuki KOZAKAI <yasuyuki.kozakai@toshiba.co.jp>");
 MODULE_DESCRIPTION("Xtables: packet \"rejection\" target for IPv6");
 MODULE_LICENSE("GPL");
 
-/* Send RST reply */
-static void send_reset(struct net *net, struct sk_buff *oldskb, int hook)
-{
-	struct sk_buff *nskb;
-	struct tcphdr otcph, *tcph;
-	unsigned int otcplen, hh_len;
-	int tcphoff, needs_ack;
-	const struct ipv6hdr *oip6h = ipv6_hdr(oldskb);
-	struct ipv6hdr *ip6h;
-#define DEFAULT_TOS_VALUE	0x0U
-	const __u8 tclass = DEFAULT_TOS_VALUE;
-	struct dst_entry *dst = NULL;
-	u8 proto;
-	__be16 frag_off;
-	struct flowi6 fl6;
-
-	if ((!(ipv6_addr_type(&oip6h->saddr) & IPV6_ADDR_UNICAST)) ||
-	    (!(ipv6_addr_type(&oip6h->daddr) & IPV6_ADDR_UNICAST))) {
-		pr_debug("addr is not unicast.\n");
-		return;
-	}
-
-	proto = oip6h->nexthdr;
-	tcphoff = ipv6_skip_exthdr(oldskb, ((u8*)(oip6h+1) - oldskb->data), &proto, &frag_off);
-
-	if ((tcphoff < 0) || (tcphoff > oldskb->len)) {
-		pr_debug("Cannot get TCP header.\n");
-		return;
-	}
-
-	otcplen = oldskb->len - tcphoff;
-
-	/* IP header checks: fragment, too short. */
-	if (proto != IPPROTO_TCP || otcplen < sizeof(struct tcphdr)) {
-		pr_debug("proto(%d) != IPPROTO_TCP, "
-			 "or too short. otcplen = %d\n",
-			 proto, otcplen);
-		return;
-	}
-
-	if (skb_copy_bits(oldskb, tcphoff, &otcph, sizeof(struct tcphdr)))
-		BUG();
-
-	/* No RST for RST. */
-	if (otcph.rst) {
-		pr_debug("RST is set\n");
-		return;
-	}
-
-	/* Check checksum. */
-	if (nf_ip6_checksum(oldskb, hook, tcphoff, IPPROTO_TCP)) {
-		pr_debug("TCP checksum is invalid\n");
-		return;
-	}
-
-	memset(&fl6, 0, sizeof(fl6));
-	fl6.flowi6_proto = IPPROTO_TCP;
-	fl6.saddr = oip6h->daddr;
-	fl6.daddr = oip6h->saddr;
-	fl6.fl6_sport = otcph.dest;
-	fl6.fl6_dport = otcph.source;
-	security_skb_classify_flow(oldskb, flowi6_to_flowi(&fl6));
-	dst = ip6_route_output(net, NULL, &fl6);
-	if (dst == NULL || dst->error) {
-		dst_release(dst);
-		return;
-	}
-	dst = xfrm_lookup(net, dst, flowi6_to_flowi(&fl6), NULL, 0);
-	if (IS_ERR(dst))
-		return;
-
-	hh_len = (dst->dev->hard_header_len + 15)&~15;
-	nskb = alloc_skb(hh_len + 15 + dst->header_len + sizeof(struct ipv6hdr)
-			 + sizeof(struct tcphdr) + dst->trailer_len,
-			 GFP_ATOMIC);
-
-	if (!nskb) {
-		net_dbg_ratelimited("cannot alloc skb\n");
-		dst_release(dst);
-		return;
-	}
-
-	skb_dst_set(nskb, dst);
-
-	skb_reserve(nskb, hh_len + dst->header_len);
-
-	skb_put(nskb, sizeof(struct ipv6hdr));
-	skb_reset_network_header(nskb);
-	ip6h = ipv6_hdr(nskb);
-	ip6_flow_hdr(ip6h, tclass, 0);
-	ip6h->hop_limit = ip6_dst_hoplimit(dst);
-	ip6h->nexthdr = IPPROTO_TCP;
-	ip6h->saddr = oip6h->daddr;
-	ip6h->daddr = oip6h->saddr;
-
-	skb_reset_transport_header(nskb);
-	tcph = (struct tcphdr *)skb_put(nskb, sizeof(struct tcphdr));
-	/* Truncate to length (no data) */
-	tcph->doff = sizeof(struct tcphdr)/4;
-	tcph->source = otcph.dest;
-	tcph->dest = otcph.source;
-
-	if (otcph.ack) {
-		needs_ack = 0;
-		tcph->seq = otcph.ack_seq;
-		tcph->ack_seq = 0;
-	} else {
-		needs_ack = 1;
-		tcph->ack_seq = htonl(ntohl(otcph.seq) + otcph.syn + otcph.fin
-				      + otcplen - (otcph.doff<<2));
-		tcph->seq = 0;
-	}
-
-	/* Reset flags */
-	((u_int8_t *)tcph)[13] = 0;
-	tcph->rst = 1;
-	tcph->ack = needs_ack;
-	tcph->window = 0;
-	tcph->urg_ptr = 0;
-	tcph->check = 0;
-
-	/* Adjust TCP checksum */
-	tcph->check = csum_ipv6_magic(&ipv6_hdr(nskb)->saddr,
-				      &ipv6_hdr(nskb)->daddr,
-				      sizeof(struct tcphdr), IPPROTO_TCP,
-				      csum_partial(tcph,
-						   sizeof(struct tcphdr), 0));
-
-	nf_ct_attach(nskb, oldskb);
-
-#ifdef CONFIG_BRIDGE_NETFILTER
-	/* If we use ip6_local_out for bridged traffic, the MAC source on
-	 * the RST will be ours, instead of the destination's.  This confuses
-	 * some routers/firewalls, and they drop the packet.  So we need to
-	 * build the eth header using the original destination's MAC as the
-	 * source, and send the RST packet directly.
-	 */
-	if (oldskb->nf_bridge) {
-		struct ethhdr *oeth = eth_hdr(oldskb);
-		nskb->dev = oldskb->nf_bridge->physindev;
-		nskb->protocol = htons(ETH_P_IPV6);
-		ip6h->payload_len = htons(sizeof(struct tcphdr));
-		if (dev_hard_header(nskb, nskb->dev, ntohs(nskb->protocol),
-				    oeth->h_source, oeth->h_dest, nskb->len) < 0)
-			return;
-		dev_queue_xmit(nskb);
-	} else
-#endif
-		ip6_local_out(nskb);
-}
-
-static inline void
-send_unreach(struct net *net, struct sk_buff *skb_in, unsigned char code,
-	     unsigned int hooknum)
-{
-	if (hooknum == NF_INET_LOCAL_OUT && skb_in->dev == NULL)
-		skb_in->dev = net->loopback_dev;
-
-	icmpv6_send(skb_in, ICMPV6_DEST_UNREACH, code, 0);
-}
 
 static unsigned int
 reject_tg6(struct sk_buff *skb, const struct xt_action_param *par)
@@ -208,25 +47,25 @@ reject_tg6(struct sk_buff *skb, const struct xt_action_param *par)
 	pr_debug("%s: medium point\n", __func__);
 	switch (reject->with) {
 	case IP6T_ICMP6_NO_ROUTE:
-		send_unreach(net, skb, ICMPV6_NOROUTE, par->hooknum);
+		send_unreach6(net, skb, ICMPV6_NOROUTE, par->hooknum);
 		break;
 	case IP6T_ICMP6_ADM_PROHIBITED:
-		send_unreach(net, skb, ICMPV6_ADM_PROHIBITED, par->hooknum);
+		send_unreach6(net, skb, ICMPV6_ADM_PROHIBITED, par->hooknum);
 		break;
 	case IP6T_ICMP6_NOT_NEIGHBOUR:
-		send_unreach(net, skb, ICMPV6_NOT_NEIGHBOUR, par->hooknum);
+		send_unreach6(net, skb, ICMPV6_NOT_NEIGHBOUR, par->hooknum);
 		break;
 	case IP6T_ICMP6_ADDR_UNREACH:
-		send_unreach(net, skb, ICMPV6_ADDR_UNREACH, par->hooknum);
+		send_unreach6(net, skb, ICMPV6_ADDR_UNREACH, par->hooknum);
 		break;
 	case IP6T_ICMP6_PORT_UNREACH:
-		send_unreach(net, skb, ICMPV6_PORT_UNREACH, par->hooknum);
+		send_unreach6(net, skb, ICMPV6_PORT_UNREACH, par->hooknum);
 		break;
 	case IP6T_ICMP6_ECHOREPLY:
 		/* Do nothing */
 		break;
 	case IP6T_TCP_RESET:
-		send_reset(net, skb, par->hooknum);
+		send_reset6(net, skb, par->hooknum);
 		break;
 	default:
 		net_info_ratelimited("case %u not handled yet\n", reject->with);
-- 
1.8.5.1

[PATCH 2/2] netfilter: nft: reject support for IPv6 and TCP reset

[Eric Leblond]

This patch moves nft_reject_ipv4 to nft_reject and adds support
for IPv6 protocol. This patch uses functions included in nf_reject.h
to implement reject by TCP reset.

Signed-off-by: Eric Leblond <eric@regit.org>
---
 net/ipv4/netfilter/Kconfig           |   4 -
 net/ipv4/netfilter/Makefile          |   1 -
 net/ipv4/netfilter/nft_reject_ipv4.c | 123 ------------------------------
 net/netfilter/Kconfig                |   4 +
 net/netfilter/Makefile               |   1 +
 net/netfilter/nft_reject.c           | 143 +++++++++++++++++++++++++++++++++++
 6 files changed, 148 insertions(+), 128 deletions(-)
 delete mode 100644 net/ipv4/netfilter/nft_reject_ipv4.c
 create mode 100644 net/netfilter/nft_reject.c

diff --git a/net/ipv4/netfilter/Kconfig b/net/ipv4/netfilter/Kconfig
index 40d5607..9d3d69a 100644
--- a/net/ipv4/netfilter/Kconfig
+++ b/net/ipv4/netfilter/Kconfig
@@ -40,10 +40,6 @@ config NF_TABLES_IPV4
 	depends on NF_TABLES
 	tristate "IPv4 nf_tables support"
 
-config NFT_REJECT_IPV4
-	depends on NF_TABLES_IPV4
-	tristate "nf_tables IPv4 reject support"
-
 config NFT_CHAIN_ROUTE_IPV4
 	depends on NF_TABLES_IPV4
 	tristate "IPv4 nf_tables route chain support"
diff --git a/net/ipv4/netfilter/Makefile b/net/ipv4/netfilter/Makefile
index 19df72b..c16be9d 100644
--- a/net/ipv4/netfilter/Makefile
+++ b/net/ipv4/netfilter/Makefile
@@ -28,7 +28,6 @@ obj-$(CONFIG_NF_NAT_SNMP_BASIC) += nf_nat_snmp_basic.o
 obj-$(CONFIG_NF_NAT_PROTO_GRE) += nf_nat_proto_gre.o
 
 obj-$(CONFIG_NF_TABLES_IPV4) += nf_tables_ipv4.o
-obj-$(CONFIG_NFT_REJECT_IPV4) += nft_reject_ipv4.o
 obj-$(CONFIG_NFT_CHAIN_ROUTE_IPV4) += nft_chain_route_ipv4.o
 obj-$(CONFIG_NFT_CHAIN_NAT_IPV4) += nft_chain_nat_ipv4.o
 obj-$(CONFIG_NF_TABLES_ARP) += nf_tables_arp.o
diff --git a/net/ipv4/netfilter/nft_reject_ipv4.c b/net/ipv4/netfilter/nft_reject_ipv4.c
deleted file mode 100644
index fff5ba1..0000000
--- a/net/ipv4/netfilter/nft_reject_ipv4.c
+++ /dev/null
@@ -1,123 +0,0 @@
-/*
- * Copyright (c) 2008-2009 Patrick McHardy <kaber@trash.net>
- *
- * This program is free software; you can redistribute it and/or modify
- * it under the terms of the GNU General Public License version 2 as
- * published by the Free Software Foundation.
- *
- * Development of this code funded by Astaro AG (http://www.astaro.com/)
- */
-
-#include <linux/kernel.h>
-#include <linux/init.h>
-#include <linux/module.h>
-#include <linux/netlink.h>
-#include <linux/netfilter.h>
-#include <linux/netfilter/nf_tables.h>
-#include <net/netfilter/nf_tables.h>
-#include <net/icmp.h>
-
-struct nft_reject {
-	enum nft_reject_types	type:8;
-	u8			icmp_code;
-};
-
-static void nft_reject_eval(const struct nft_expr *expr,
-			      struct nft_data data[NFT_REG_MAX + 1],
-			      const struct nft_pktinfo *pkt)
-{
-	struct nft_reject *priv = nft_expr_priv(expr);
-
-	switch (priv->type) {
-	case NFT_REJECT_ICMP_UNREACH:
-		icmp_send(pkt->skb, ICMP_DEST_UNREACH, priv->icmp_code, 0);
-		break;
-	case NFT_REJECT_TCP_RST:
-		break;
-	}
-
-	data[NFT_REG_VERDICT].verdict = NF_DROP;
-}
-
-static const struct nla_policy nft_reject_policy[NFTA_REJECT_MAX + 1] = {
-	[NFTA_REJECT_TYPE]		= { .type = NLA_U32 },
-	[NFTA_REJECT_ICMP_CODE]		= { .type = NLA_U8 },
-};
-
-static int nft_reject_init(const struct nft_ctx *ctx,
-			   const struct nft_expr *expr,
-			   const struct nlattr * const tb[])
-{
-	struct nft_reject *priv = nft_expr_priv(expr);
-
-	if (tb[NFTA_REJECT_TYPE] == NULL)
-		return -EINVAL;
-
-	priv->type = ntohl(nla_get_be32(tb[NFTA_REJECT_TYPE]));
-	switch (priv->type) {
-	case NFT_REJECT_ICMP_UNREACH:
-		if (tb[NFTA_REJECT_ICMP_CODE] == NULL)
-			return -EINVAL;
-		priv->icmp_code = nla_get_u8(tb[NFTA_REJECT_ICMP_CODE]);
-	case NFT_REJECT_TCP_RST:
-		break;
-	default:
-		return -EINVAL;
-	}
-
-	return 0;
-}
-
-static int nft_reject_dump(struct sk_buff *skb, const struct nft_expr *expr)
-{
-	const struct nft_reject *priv = nft_expr_priv(expr);
-
-	if (nla_put_be32(skb, NFTA_REJECT_TYPE, priv->type))
-		goto nla_put_failure;
-
-	switch (priv->type) {
-	case NFT_REJECT_ICMP_UNREACH:
-		if (nla_put_u8(skb, NFTA_REJECT_ICMP_CODE, priv->icmp_code))
-			goto nla_put_failure;
-		break;
-	}
-
-	return 0;
-
-nla_put_failure:
-	return -1;
-}
-
-static struct nft_expr_type nft_reject_type;
-static const struct nft_expr_ops nft_reject_ops = {
-	.type		= &nft_reject_type,
-	.size		= NFT_EXPR_SIZE(sizeof(struct nft_reject)),
-	.eval		= nft_reject_eval,
-	.init		= nft_reject_init,
-	.dump		= nft_reject_dump,
-};
-
-static struct nft_expr_type nft_reject_type __read_mostly = {
-	.name		= "reject",
-	.ops		= &nft_reject_ops,
-	.policy		= nft_reject_policy,
-	.maxattr	= NFTA_REJECT_MAX,
-	.owner		= THIS_MODULE,
-};
-
-static int __init nft_reject_module_init(void)
-{
-	return nft_register_expr(&nft_reject_type);
-}
-
-static void __exit nft_reject_module_exit(void)
-{
-	nft_unregister_expr(&nft_reject_type);
-}
-
-module_init(nft_reject_module_init);
-module_exit(nft_reject_module_exit);
-
-MODULE_LICENSE("GPL");
-MODULE_AUTHOR("Patrick McHardy <kaber@trash.net>");
-MODULE_ALIAS_NFT_EXPR("reject");
diff --git a/net/netfilter/Kconfig b/net/netfilter/Kconfig
index 5263605..9482822 100644
--- a/net/netfilter/Kconfig
+++ b/net/netfilter/Kconfig
@@ -465,6 +465,10 @@ config NFT_QUEUE
 	  This is required if you intend to use the userspace queueing
 	  infrastructure (also known as NFQUEUE) from nftables.
 
+config NFT_REJECT
+	depends on NF_TABLES
+	tristate "nf_tables reject support"
+
 config NFT_COMPAT
 	depends on NF_TABLES
 	depends on NETFILTER_XTABLES
diff --git a/net/netfilter/Makefile b/net/netfilter/Makefile
index e763746..ba578f6 100644
--- a/net/netfilter/Makefile
+++ b/net/netfilter/Makefile
@@ -77,6 +77,7 @@ obj-$(CONFIG_NFT_CT)		+= nft_ct.o
 obj-$(CONFIG_NFT_LIMIT)		+= nft_limit.o
 obj-$(CONFIG_NFT_NAT)		+= nft_nat.o
 obj-$(CONFIG_NFT_QUEUE)		+= nft_queue.o
+obj-$(CONFIG_NFT_REJECT) 	+= nft_reject.o
 #nf_tables-objs			+= nft_meta_target.o
 obj-$(CONFIG_NFT_RBTREE)	+= nft_rbtree.o
 obj-$(CONFIG_NFT_HASH)		+= nft_hash.o
diff --git a/net/netfilter/nft_reject.c b/net/netfilter/nft_reject.c
new file mode 100644
index 0000000..002143c
--- /dev/null
+++ b/net/netfilter/nft_reject.c
@@ -0,0 +1,143 @@
+/*
+ * Copyright (c) 2008-2009 Patrick McHardy <kaber@trash.net>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * Development of this code funded by Astaro AG (http://www.astaro.com/)
+ */
+
+#include <linux/kernel.h>
+#include <linux/init.h>
+#include <linux/module.h>
+#include <linux/netlink.h>
+#include <linux/netfilter.h>
+#include <linux/netfilter/nf_tables.h>
+#include <net/netfilter/nf_tables.h>
+#include <net/icmp.h>
+#if IS_ENABLED(CONFIG_IP6_NF_IPTABLES)
+#include <net/ipv6.h>
+#endif
+
+#include <net/netfilter/nf_reject.h>
+
+struct nft_reject {
+	enum nft_reject_types	type:8;
+	u8			icmp_code;
+	u8			family;
+};
+
+static void nft_reject_eval(const struct nft_expr *expr,
+			      struct nft_data data[NFT_REG_MAX + 1],
+			      const struct nft_pktinfo *pkt)
+{
+	struct nft_reject *priv = nft_expr_priv(expr);
+	struct net *net = dev_net((pkt->in != NULL) ? pkt->in : pkt->out);
+
+	switch (priv->type) {
+	case NFT_REJECT_ICMP_UNREACH:
+		if (priv->family == NFPROTO_IPV4)
+			send_unreach(pkt->skb, priv->icmp_code);
+#if IS_ENABLED(CONFIG_IP6_NF_IPTABLES)
+		else if (priv->family == NFPROTO_IPV6)
+			send_unreach6(net, pkt->skb, priv->icmp_code,
+				      pkt->hooknum);
+#endif
+		break;
+	case NFT_REJECT_TCP_RST:
+		if (priv->family == NFPROTO_IPV4)
+			send_reset(pkt->skb, pkt->hooknum);
+#if IS_ENABLED(CONFIG_IP6_NF_IPTABLES)
+		else if (priv->family == NFPROTO_IPV6)
+			send_reset6(net, pkt->skb, pkt->hooknum);
+#endif
+		break;
+	}
+
+	data[NFT_REG_VERDICT].verdict = NF_DROP;
+}
+
+static const struct nla_policy nft_reject_policy[NFTA_REJECT_MAX + 1] = {
+	[NFTA_REJECT_TYPE]		= { .type = NLA_U32 },
+	[NFTA_REJECT_ICMP_CODE]		= { .type = NLA_U8 },
+};
+
+static int nft_reject_init(const struct nft_ctx *ctx,
+			   const struct nft_expr *expr,
+			   const struct nlattr * const tb[])
+{
+	struct nft_reject *priv = nft_expr_priv(expr);
+
+	if (tb[NFTA_REJECT_TYPE] == NULL)
+		return -EINVAL;
+
+	priv->family = ctx->afi->family;
+	priv->type = ntohl(nla_get_be32(tb[NFTA_REJECT_TYPE]));
+	switch (priv->type) {
+	case NFT_REJECT_ICMP_UNREACH:
+		if (tb[NFTA_REJECT_ICMP_CODE] == NULL)
+			return -EINVAL;
+		priv->icmp_code = nla_get_u8(tb[NFTA_REJECT_ICMP_CODE]);
+	case NFT_REJECT_TCP_RST:
+		break;
+	default:
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static int nft_reject_dump(struct sk_buff *skb, const struct nft_expr *expr)
+{
+	const struct nft_reject *priv = nft_expr_priv(expr);
+
+	if (nla_put_be32(skb, NFTA_REJECT_TYPE, priv->type))
+		goto nla_put_failure;
+
+	switch (priv->type) {
+	case NFT_REJECT_ICMP_UNREACH:
+		if (nla_put_u8(skb, NFTA_REJECT_ICMP_CODE, priv->icmp_code))
+			goto nla_put_failure;
+		break;
+	}
+
+	return 0;
+
+nla_put_failure:
+	return -1;
+}
+
+static struct nft_expr_type nft_reject_type;
+static const struct nft_expr_ops nft_reject_ops = {
+	.type		= &nft_reject_type,
+	.size		= NFT_EXPR_SIZE(sizeof(struct nft_reject)),
+	.eval		= nft_reject_eval,
+	.init		= nft_reject_init,
+	.dump		= nft_reject_dump,
+};
+
+static struct nft_expr_type nft_reject_type __read_mostly = {
+	.name		= "reject",
+	.ops		= &nft_reject_ops,
+	.policy		= nft_reject_policy,
+	.maxattr	= NFTA_REJECT_MAX,
+	.owner		= THIS_MODULE,
+};
+
+static int __init nft_reject_module_init(void)
+{
+	return nft_register_expr(&nft_reject_type);
+}
+
+static void __exit nft_reject_module_exit(void)
+{
+	nft_unregister_expr(&nft_reject_type);
+}
+
+module_init(nft_reject_module_init);
+module_exit(nft_reject_module_exit);
+
+MODULE_LICENSE("GPL");
+MODULE_AUTHOR("Patrick McHardy <kaber@trash.net>");
+MODULE_ALIAS_NFT_EXPR("reject");
-- 
1.8.5.1

[RFC PATCH 0/2] nft: finish reject support

[Eric Leblond]

Hello,

Here's an update of yesterday patchset. There is only one single change
fixing endianess in the dump function.

BR,
--
Eric

[PATCH 1/2] netfilter: REJECT: separate reusable code

[Eric Leblond]

This patch prepares the addition of TCP reset support in
the nft_reject module by moving reusable code into a header
file.

Signed-off-by: Eric Leblond <eric@regit.org>
---
 include/net/netfilter/nf_reject.h | 297 ++++++++++++++++++++++++++++++++++++++
 net/ipv4/netfilter/ipt_REJECT.c   | 124 +---------------
 net/ipv6/netfilter/ip6t_REJECT.c  | 177 +----------------------
 3 files changed, 307 insertions(+), 291 deletions(-)
 create mode 100644 include/net/netfilter/nf_reject.h

diff --git a/include/net/netfilter/nf_reject.h b/include/net/netfilter/nf_reject.h
new file mode 100644
index 0000000..0bb089a
--- /dev/null
+++ b/include/net/netfilter/nf_reject.h
@@ -0,0 +1,297 @@
+#ifndef _NF_REJECT_H
+#define _NF_REJECT_H
+
+#include <net/ip.h>
+#include <net/tcp.h>
+#include <net/route.h>
+#include <net/dst.h>
+
+#if IS_ENABLED(CONFIG_IP6_NF_IPTABLES)
+#include <net/ipv6.h>
+#include <net/ip6_route.h>
+#include <linux/netfilter_ipv6.h>
+#endif
+
+static inline void send_unreach(struct sk_buff *skb_in, int code)
+{
+	icmp_send(skb_in, ICMP_DEST_UNREACH, code, 0);
+}
+
+/* Send RST reply */
+static inline void send_reset(struct sk_buff *oldskb, int hook)
+{
+	struct sk_buff *nskb;
+	const struct iphdr *oiph;
+	struct iphdr *niph;
+	const struct tcphdr *oth;
+	struct tcphdr _otcph, *tcph;
+
+	/* IP header checks: fragment. */
+	if (ip_hdr(oldskb)->frag_off & htons(IP_OFFSET))
+		return;
+
+	oth = skb_header_pointer(oldskb, ip_hdrlen(oldskb),
+				 sizeof(_otcph), &_otcph);
+	if (oth == NULL)
+		return;
+
+	/* No RST for RST. */
+	if (oth->rst)
+		return;
+
+	if (skb_rtable(oldskb)->rt_flags & (RTCF_BROADCAST | RTCF_MULTICAST))
+		return;
+
+	/* Check checksum */
+	if (nf_ip_checksum(oldskb, hook, ip_hdrlen(oldskb), IPPROTO_TCP))
+		return;
+	oiph = ip_hdr(oldskb);
+
+	nskb = alloc_skb(sizeof(struct iphdr) + sizeof(struct tcphdr) +
+			 LL_MAX_HEADER, GFP_ATOMIC);
+	if (!nskb)
+		return;
+
+	skb_reserve(nskb, LL_MAX_HEADER);
+
+	skb_reset_network_header(nskb);
+	niph = (struct iphdr *)skb_put(nskb, sizeof(struct iphdr));
+	niph->version	= 4;
+	niph->ihl	= sizeof(struct iphdr) / 4;
+	niph->tos	= 0;
+	niph->id	= 0;
+	niph->frag_off	= htons(IP_DF);
+	niph->protocol	= IPPROTO_TCP;
+	niph->check	= 0;
+	niph->saddr	= oiph->daddr;
+	niph->daddr	= oiph->saddr;
+
+	skb_reset_transport_header(nskb);
+	tcph = (struct tcphdr *)skb_put(nskb, sizeof(struct tcphdr));
+	memset(tcph, 0, sizeof(*tcph));
+	tcph->source	= oth->dest;
+	tcph->dest	= oth->source;
+	tcph->doff	= sizeof(struct tcphdr) / 4;
+
+	if (oth->ack)
+		tcph->seq = oth->ack_seq;
+	else {
+		tcph->ack_seq = htonl(ntohl(oth->seq) + oth->syn + oth->fin +
+				      oldskb->len - ip_hdrlen(oldskb) -
+				      (oth->doff << 2));
+		tcph->ack = 1;
+	}
+
+	tcph->rst	= 1;
+	tcph->check = ~tcp_v4_check(sizeof(struct tcphdr), niph->saddr,
+				    niph->daddr, 0);
+	nskb->ip_summed = CHECKSUM_PARTIAL;
+	nskb->csum_start = (unsigned char *)tcph - nskb->head;
+	nskb->csum_offset = offsetof(struct tcphdr, check);
+
+	/* ip_route_me_harder expects skb->dst to be set */
+	skb_dst_set_noref(nskb, skb_dst(oldskb));
+
+	nskb->protocol = htons(ETH_P_IP);
+	if (ip_route_me_harder(nskb, RTN_UNSPEC))
+		goto free_nskb;
+
+	niph->ttl	= ip4_dst_hoplimit(skb_dst(nskb));
+
+	/* "Never happens" */
+	if (nskb->len > dst_mtu(skb_dst(nskb)))
+		goto free_nskb;
+
+	nf_ct_attach(nskb, oldskb);
+
+#ifdef CONFIG_BRIDGE_NETFILTER
+	/* If we use ip_local_out for bridged traffic, the MAC source on
+	 * the RST will be ours, instead of the destination's.  This confuses
+	 * some routers/firewalls, and they drop the packet.  So we need to
+	 * build the eth header using the original destination's MAC as the
+	 * source, and send the RST packet directly.
+	 */
+	if (oldskb->nf_bridge) {
+		struct ethhdr *oeth = eth_hdr(oldskb);
+		nskb->dev = oldskb->nf_bridge->physindev;
+		niph->tot_len = htons(nskb->len);
+		ip_send_check(niph);
+		if (dev_hard_header(nskb, nskb->dev, ntohs(nskb->protocol),
+				    oeth->h_source, oeth->h_dest, nskb->len) < 0)
+			goto free_nskb;
+		dev_queue_xmit(nskb);
+	} else
+#endif
+		ip_local_out(nskb);
+
+	return;
+
+ free_nskb:
+	kfree_skb(nskb);
+}
+
+#if IS_ENABLED(CONFIG_IP6_NF_IPTABLES)
+
+static inline void
+send_unreach6(struct net *net, struct sk_buff *skb_in, unsigned char code,
+	     unsigned int hooknum)
+{
+	if (hooknum == NF_INET_LOCAL_OUT && skb_in->dev == NULL)
+		skb_in->dev = net->loopback_dev;
+
+	icmpv6_send(skb_in, ICMPV6_DEST_UNREACH, code, 0);
+}
+
+/* Send RST reply */
+static inline void send_reset6(struct net *net, struct sk_buff *oldskb, int hook)
+{
+	struct sk_buff *nskb;
+	struct tcphdr otcph, *tcph;
+	unsigned int otcplen, hh_len;
+	int tcphoff, needs_ack;
+	const struct ipv6hdr *oip6h = ipv6_hdr(oldskb);
+	struct ipv6hdr *ip6h;
+#define DEFAULT_TOS_VALUE	0x0U
+	const __u8 tclass = DEFAULT_TOS_VALUE;
+	struct dst_entry *dst = NULL;
+	u8 proto;
+	__be16 frag_off;
+	struct flowi6 fl6;
+
+	if ((!(ipv6_addr_type(&oip6h->saddr) & IPV6_ADDR_UNICAST)) ||
+	    (!(ipv6_addr_type(&oip6h->daddr) & IPV6_ADDR_UNICAST))) {
+		pr_debug("addr is not unicast.\n");
+		return;
+	}
+
+	proto = oip6h->nexthdr;
+	tcphoff = ipv6_skip_exthdr(oldskb, ((u8*)(oip6h+1) - oldskb->data), &proto, &frag_off);
+
+	if ((tcphoff < 0) || (tcphoff > oldskb->len)) {
+		pr_debug("Cannot get TCP header.\n");
+		return;
+	}
+
+	otcplen = oldskb->len - tcphoff;
+
+	/* IP header checks: fragment, too short. */
+	if (proto != IPPROTO_TCP || otcplen < sizeof(struct tcphdr)) {
+		pr_debug("proto(%d) != IPPROTO_TCP, "
+			 "or too short. otcplen = %d\n",
+			 proto, otcplen);
+		return;
+	}
+
+	if (skb_copy_bits(oldskb, tcphoff, &otcph, sizeof(struct tcphdr)))
+		BUG();
+
+	/* No RST for RST. */
+	if (otcph.rst) {
+		pr_debug("RST is set\n");
+		return;
+	}
+
+	/* Check checksum. */
+	if (nf_ip6_checksum(oldskb, hook, tcphoff, IPPROTO_TCP)) {
+		pr_debug("TCP checksum is invalid\n");
+		return;
+	}
+
+	memset(&fl6, 0, sizeof(fl6));
+	fl6.flowi6_proto = IPPROTO_TCP;
+	fl6.saddr = oip6h->daddr;
+	fl6.daddr = oip6h->saddr;
+	fl6.fl6_sport = otcph.dest;
+	fl6.fl6_dport = otcph.source;
+	security_skb_classify_flow(oldskb, flowi6_to_flowi(&fl6));
+	dst = ip6_route_output(net, NULL, &fl6);
+	if (dst == NULL || dst->error) {
+		dst_release(dst);
+		return;
+	}
+	dst = xfrm_lookup(net, dst, flowi6_to_flowi(&fl6), NULL, 0);
+	if (IS_ERR(dst))
+		return;
+
+	hh_len = (dst->dev->hard_header_len + 15)&~15;
+	nskb = alloc_skb(hh_len + 15 + dst->header_len + sizeof(struct ipv6hdr)
+			 + sizeof(struct tcphdr) + dst->trailer_len,
+			 GFP_ATOMIC);
+
+	if (!nskb) {
+		net_dbg_ratelimited("cannot alloc skb\n");
+		dst_release(dst);
+		return;
+	}
+
+	skb_dst_set(nskb, dst);
+
+	skb_reserve(nskb, hh_len + dst->header_len);
+
+	skb_put(nskb, sizeof(struct ipv6hdr));
+	skb_reset_network_header(nskb);
+	ip6h = ipv6_hdr(nskb);
+	ip6_flow_hdr(ip6h, tclass, 0);
+	ip6h->hop_limit = ip6_dst_hoplimit(dst);
+	ip6h->nexthdr = IPPROTO_TCP;
+	ip6h->saddr = oip6h->daddr;
+	ip6h->daddr = oip6h->saddr;
+
+	skb_reset_transport_header(nskb);
+	tcph = (struct tcphdr *)skb_put(nskb, sizeof(struct tcphdr));
+	/* Truncate to length (no data) */
+	tcph->doff = sizeof(struct tcphdr)/4;
+	tcph->source = otcph.dest;
+	tcph->dest = otcph.source;
+
+	if (otcph.ack) {
+		needs_ack = 0;
+		tcph->seq = otcph.ack_seq;
+		tcph->ack_seq = 0;
+	} else {
+		needs_ack = 1;
+		tcph->ack_seq = htonl(ntohl(otcph.seq) + otcph.syn + otcph.fin
+				      + otcplen - (otcph.doff<<2));
+		tcph->seq = 0;
+	}
+
+	/* Reset flags */
+	((u_int8_t *)tcph)[13] = 0;
+	tcph->rst = 1;
+	tcph->ack = needs_ack;
+	tcph->window = 0;
+	tcph->urg_ptr = 0;
+	tcph->check = 0;
+
+	/* Adjust TCP checksum */
+	tcph->check = csum_ipv6_magic(&ipv6_hdr(nskb)->saddr,
+				      &ipv6_hdr(nskb)->daddr,
+				      sizeof(struct tcphdr), IPPROTO_TCP,
+				      csum_partial(tcph,
+						   sizeof(struct tcphdr), 0));
+
+	nf_ct_attach(nskb, oldskb);
+
+#ifdef CONFIG_BRIDGE_NETFILTER
+	/* If we use ip6_local_out for bridged traffic, the MAC source on
+	 * the RST will be ours, instead of the destination's.  This confuses
+	 * some routers/firewalls, and they drop the packet.  So we need to
+	 * build the eth header using the original destination's MAC as the
+	 * source, and send the RST packet directly.
+	 */
+	if (oldskb->nf_bridge) {
+		struct ethhdr *oeth = eth_hdr(oldskb);
+		nskb->dev = oldskb->nf_bridge->physindev;
+		nskb->protocol = htons(ETH_P_IPV6);
+		ip6h->payload_len = htons(sizeof(struct tcphdr));
+		if (dev_hard_header(nskb, nskb->dev, ntohs(nskb->protocol),
+				    oeth->h_source, oeth->h_dest, nskb->len) < 0)
+			return;
+		dev_queue_xmit(nskb);
+	} else
+#endif
+		ip6_local_out(nskb);
+}
+
+#endif /* CONFIG_IP6_NF_IPTABLES */
+#endif /* _NF_REJECT_H */
diff --git a/net/ipv4/netfilter/ipt_REJECT.c b/net/ipv4/netfilter/ipt_REJECT.c
index b969131..f40305f 100644
--- a/net/ipv4/netfilter/ipt_REJECT.c
+++ b/net/ipv4/netfilter/ipt_REJECT.c
@@ -17,10 +17,6 @@
 #include <linux/udp.h>
 #include <linux/icmp.h>
 #include <net/icmp.h>
-#include <net/ip.h>
-#include <net/tcp.h>
-#include <net/route.h>
-#include <net/dst.h>
 #include <linux/netfilter/x_tables.h>
 #include <linux/netfilter_ipv4/ip_tables.h>
 #include <linux/netfilter_ipv4/ipt_REJECT.h>
@@ -28,128 +24,12 @@
 #include <linux/netfilter_bridge.h>
 #endif
 
+#include <net/netfilter/nf_reject.h>
+
 MODULE_LICENSE("GPL");
 MODULE_AUTHOR("Netfilter Core Team <coreteam@netfilter.org>");
 MODULE_DESCRIPTION("Xtables: packet \"rejection\" target for IPv4");
 
-/* Send RST reply */
-static void send_reset(struct sk_buff *oldskb, int hook)
-{
-	struct sk_buff *nskb;
-	const struct iphdr *oiph;
-	struct iphdr *niph;
-	const struct tcphdr *oth;
-	struct tcphdr _otcph, *tcph;
-
-	/* IP header checks: fragment. */
-	if (ip_hdr(oldskb)->frag_off & htons(IP_OFFSET))
-		return;
-
-	oth = skb_header_pointer(oldskb, ip_hdrlen(oldskb),
-				 sizeof(_otcph), &_otcph);
-	if (oth == NULL)
-		return;
-
-	/* No RST for RST. */
-	if (oth->rst)
-		return;
-
-	if (skb_rtable(oldskb)->rt_flags & (RTCF_BROADCAST | RTCF_MULTICAST))
-		return;
-
-	/* Check checksum */
-	if (nf_ip_checksum(oldskb, hook, ip_hdrlen(oldskb), IPPROTO_TCP))
-		return;
-	oiph = ip_hdr(oldskb);
-
-	nskb = alloc_skb(sizeof(struct iphdr) + sizeof(struct tcphdr) +
-			 LL_MAX_HEADER, GFP_ATOMIC);
-	if (!nskb)
-		return;
-
-	skb_reserve(nskb, LL_MAX_HEADER);
-
-	skb_reset_network_header(nskb);
-	niph = (struct iphdr *)skb_put(nskb, sizeof(struct iphdr));
-	niph->version	= 4;
-	niph->ihl	= sizeof(struct iphdr) / 4;
-	niph->tos	= 0;
-	niph->id	= 0;
-	niph->frag_off	= htons(IP_DF);
-	niph->protocol	= IPPROTO_TCP;
-	niph->check	= 0;
-	niph->saddr	= oiph->daddr;
-	niph->daddr	= oiph->saddr;
-
-	skb_reset_transport_header(nskb);
-	tcph = (struct tcphdr *)skb_put(nskb, sizeof(struct tcphdr));
-	memset(tcph, 0, sizeof(*tcph));
-	tcph->source	= oth->dest;
-	tcph->dest	= oth->source;
-	tcph->doff	= sizeof(struct tcphdr) / 4;
-
-	if (oth->ack)
-		tcph->seq = oth->ack_seq;
-	else {
-		tcph->ack_seq = htonl(ntohl(oth->seq) + oth->syn + oth->fin +
-				      oldskb->len - ip_hdrlen(oldskb) -
-				      (oth->doff << 2));
-		tcph->ack = 1;
-	}
-
-	tcph->rst	= 1;
-	tcph->check = ~tcp_v4_check(sizeof(struct tcphdr), niph->saddr,
-				    niph->daddr, 0);
-	nskb->ip_summed = CHECKSUM_PARTIAL;
-	nskb->csum_start = (unsigned char *)tcph - nskb->head;
-	nskb->csum_offset = offsetof(struct tcphdr, check);
-
-	/* ip_route_me_harder expects skb->dst to be set */
-	skb_dst_set_noref(nskb, skb_dst(oldskb));
-
-	nskb->protocol = htons(ETH_P_IP);
-	if (ip_route_me_harder(nskb, RTN_UNSPEC))
-		goto free_nskb;
-
-	niph->ttl	= ip4_dst_hoplimit(skb_dst(nskb));
-
-	/* "Never happens" */
-	if (nskb->len > dst_mtu(skb_dst(nskb)))
-		goto free_nskb;
-
-	nf_ct_attach(nskb, oldskb);
-
-#ifdef CONFIG_BRIDGE_NETFILTER
-	/* If we use ip_local_out for bridged traffic, the MAC source on
-	 * the RST will be ours, instead of the destination's.  This confuses
-	 * some routers/firewalls, and they drop the packet.  So we need to
-	 * build the eth header using the original destination's MAC as the
-	 * source, and send the RST packet directly.
-	 */
-	if (oldskb->nf_bridge) {
-		struct ethhdr *oeth = eth_hdr(oldskb);
-		nskb->dev = oldskb->nf_bridge->physindev;
-		niph->tot_len = htons(nskb->len);
-		ip_send_check(niph);
-		if (dev_hard_header(nskb, nskb->dev, ntohs(nskb->protocol),
-				    oeth->h_source, oeth->h_dest, nskb->len) < 0)
-			goto free_nskb;
-		dev_queue_xmit(nskb);
-	} else
-#endif
-		ip_local_out(nskb);
-
-	return;
-
- free_nskb:
-	kfree_skb(nskb);
-}
-
-static inline void send_unreach(struct sk_buff *skb_in, int code)
-{
-	icmp_send(skb_in, ICMP_DEST_UNREACH, code, 0);
-}
-
 static unsigned int
 reject_tg(struct sk_buff *skb, const struct xt_action_param *par)
 {
diff --git a/net/ipv6/netfilter/ip6t_REJECT.c b/net/ipv6/netfilter/ip6t_REJECT.c
index da00a2e..3461c3c 100644
--- a/net/ipv6/netfilter/ip6t_REJECT.c
+++ b/net/ipv6/netfilter/ip6t_REJECT.c
@@ -23,181 +23,20 @@
 #include <linux/skbuff.h>
 #include <linux/icmpv6.h>
 #include <linux/netdevice.h>
-#include <net/ipv6.h>
-#include <net/tcp.h>
 #include <net/icmp.h>
 #include <net/ip6_checksum.h>
 #include <net/ip6_fib.h>
-#include <net/ip6_route.h>
 #include <net/flow.h>
 #include <linux/netfilter/x_tables.h>
 #include <linux/netfilter_ipv6/ip6_tables.h>
 #include <linux/netfilter_ipv6/ip6t_REJECT.h>
 
+#include <net/netfilter/nf_reject.h>
+
 MODULE_AUTHOR("Yasuyuki KOZAKAI <yasuyuki.kozakai@toshiba.co.jp>");
 MODULE_DESCRIPTION("Xtables: packet \"rejection\" target for IPv6");
 MODULE_LICENSE("GPL");
 
-/* Send RST reply */
-static void send_reset(struct net *net, struct sk_buff *oldskb, int hook)
-{
-	struct sk_buff *nskb;
-	struct tcphdr otcph, *tcph;
-	unsigned int otcplen, hh_len;
-	int tcphoff, needs_ack;
-	const struct ipv6hdr *oip6h = ipv6_hdr(oldskb);
-	struct ipv6hdr *ip6h;
-#define DEFAULT_TOS_VALUE	0x0U
-	const __u8 tclass = DEFAULT_TOS_VALUE;
-	struct dst_entry *dst = NULL;
-	u8 proto;
-	__be16 frag_off;
-	struct flowi6 fl6;
-
-	if ((!(ipv6_addr_type(&oip6h->saddr) & IPV6_ADDR_UNICAST)) ||
-	    (!(ipv6_addr_type(&oip6h->daddr) & IPV6_ADDR_UNICAST))) {
-		pr_debug("addr is not unicast.\n");
-		return;
-	}
-
-	proto = oip6h->nexthdr;
-	tcphoff = ipv6_skip_exthdr(oldskb, ((u8*)(oip6h+1) - oldskb->data), &proto, &frag_off);
-
-	if ((tcphoff < 0) || (tcphoff > oldskb->len)) {
-		pr_debug("Cannot get TCP header.\n");
-		return;
-	}
-
-	otcplen = oldskb->len - tcphoff;
-
-	/* IP header checks: fragment, too short. */
-	if (proto != IPPROTO_TCP || otcplen < sizeof(struct tcphdr)) {
-		pr_debug("proto(%d) != IPPROTO_TCP, "
-			 "or too short. otcplen = %d\n",
-			 proto, otcplen);
-		return;
-	}
-
-	if (skb_copy_bits(oldskb, tcphoff, &otcph, sizeof(struct tcphdr)))
-		BUG();
-
-	/* No RST for RST. */
-	if (otcph.rst) {
-		pr_debug("RST is set\n");
-		return;
-	}
-
-	/* Check checksum. */
-	if (nf_ip6_checksum(oldskb, hook, tcphoff, IPPROTO_TCP)) {
-		pr_debug("TCP checksum is invalid\n");
-		return;
-	}
-
-	memset(&fl6, 0, sizeof(fl6));
-	fl6.flowi6_proto = IPPROTO_TCP;
-	fl6.saddr = oip6h->daddr;
-	fl6.daddr = oip6h->saddr;
-	fl6.fl6_sport = otcph.dest;
-	fl6.fl6_dport = otcph.source;
-	security_skb_classify_flow(oldskb, flowi6_to_flowi(&fl6));
-	dst = ip6_route_output(net, NULL, &fl6);
-	if (dst == NULL || dst->error) {
-		dst_release(dst);
-		return;
-	}
-	dst = xfrm_lookup(net, dst, flowi6_to_flowi(&fl6), NULL, 0);
-	if (IS_ERR(dst))
-		return;
-
-	hh_len = (dst->dev->hard_header_len + 15)&~15;
-	nskb = alloc_skb(hh_len + 15 + dst->header_len + sizeof(struct ipv6hdr)
-			 + sizeof(struct tcphdr) + dst->trailer_len,
-			 GFP_ATOMIC);
-
-	if (!nskb) {
-		net_dbg_ratelimited("cannot alloc skb\n");
-		dst_release(dst);
-		return;
-	}
-
-	skb_dst_set(nskb, dst);
-
-	skb_reserve(nskb, hh_len + dst->header_len);
-
-	skb_put(nskb, sizeof(struct ipv6hdr));
-	skb_reset_network_header(nskb);
-	ip6h = ipv6_hdr(nskb);
-	ip6_flow_hdr(ip6h, tclass, 0);
-	ip6h->hop_limit = ip6_dst_hoplimit(dst);
-	ip6h->nexthdr = IPPROTO_TCP;
-	ip6h->saddr = oip6h->daddr;
-	ip6h->daddr = oip6h->saddr;
-
-	skb_reset_transport_header(nskb);
-	tcph = (struct tcphdr *)skb_put(nskb, sizeof(struct tcphdr));
-	/* Truncate to length (no data) */
-	tcph->doff = sizeof(struct tcphdr)/4;
-	tcph->source = otcph.dest;
-	tcph->dest = otcph.source;
-
-	if (otcph.ack) {
-		needs_ack = 0;
-		tcph->seq = otcph.ack_seq;
-		tcph->ack_seq = 0;
-	} else {
-		needs_ack = 1;
-		tcph->ack_seq = htonl(ntohl(otcph.seq) + otcph.syn + otcph.fin
-				      + otcplen - (otcph.doff<<2));
-		tcph->seq = 0;
-	}
-
-	/* Reset flags */
-	((u_int8_t *)tcph)[13] = 0;
-	tcph->rst = 1;
-	tcph->ack = needs_ack;
-	tcph->window = 0;
-	tcph->urg_ptr = 0;
-	tcph->check = 0;
-
-	/* Adjust TCP checksum */
-	tcph->check = csum_ipv6_magic(&ipv6_hdr(nskb)->saddr,
-				      &ipv6_hdr(nskb)->daddr,
-				      sizeof(struct tcphdr), IPPROTO_TCP,
-				      csum_partial(tcph,
-						   sizeof(struct tcphdr), 0));
-
-	nf_ct_attach(nskb, oldskb);
-
-#ifdef CONFIG_BRIDGE_NETFILTER
-	/* If we use ip6_local_out for bridged traffic, the MAC source on
-	 * the RST will be ours, instead of the destination's.  This confuses
-	 * some routers/firewalls, and they drop the packet.  So we need to
-	 * build the eth header using the original destination's MAC as the
-	 * source, and send the RST packet directly.
-	 */
-	if (oldskb->nf_bridge) {
-		struct ethhdr *oeth = eth_hdr(oldskb);
-		nskb->dev = oldskb->nf_bridge->physindev;
-		nskb->protocol = htons(ETH_P_IPV6);
-		ip6h->payload_len = htons(sizeof(struct tcphdr));
-		if (dev_hard_header(nskb, nskb->dev, ntohs(nskb->protocol),
-				    oeth->h_source, oeth->h_dest, nskb->len) < 0)
-			return;
-		dev_queue_xmit(nskb);
-	} else
-#endif
-		ip6_local_out(nskb);
-}
-
-static inline void
-send_unreach(struct net *net, struct sk_buff *skb_in, unsigned char code,
-	     unsigned int hooknum)
-{
-	if (hooknum == NF_INET_LOCAL_OUT && skb_in->dev == NULL)
-		skb_in->dev = net->loopback_dev;
-
-	icmpv6_send(skb_in, ICMPV6_DEST_UNREACH, code, 0);
-}
 
 static unsigned int
 reject_tg6(struct sk_buff *skb, const struct xt_action_param *par)
@@ -208,25 +47,25 @@ reject_tg6(struct sk_buff *skb, const struct xt_action_param *par)
 	pr_debug("%s: medium point\n", __func__);
 	switch (reject->with) {
 	case IP6T_ICMP6_NO_ROUTE:
-		send_unreach(net, skb, ICMPV6_NOROUTE, par->hooknum);
+		send_unreach6(net, skb, ICMPV6_NOROUTE, par->hooknum);
 		break;
 	case IP6T_ICMP6_ADM_PROHIBITED:
-		send_unreach(net, skb, ICMPV6_ADM_PROHIBITED, par->hooknum);
+		send_unreach6(net, skb, ICMPV6_ADM_PROHIBITED, par->hooknum);
 		break;
 	case IP6T_ICMP6_NOT_NEIGHBOUR:
-		send_unreach(net, skb, ICMPV6_NOT_NEIGHBOUR, par->hooknum);
+		send_unreach6(net, skb, ICMPV6_NOT_NEIGHBOUR, par->hooknum);
 		break;
 	case IP6T_ICMP6_ADDR_UNREACH:
-		send_unreach(net, skb, ICMPV6_ADDR_UNREACH, par->hooknum);
+		send_unreach6(net, skb, ICMPV6_ADDR_UNREACH, par->hooknum);
 		break;
 	case IP6T_ICMP6_PORT_UNREACH:
-		send_unreach(net, skb, ICMPV6_PORT_UNREACH, par->hooknum);
+		send_unreach6(net, skb, ICMPV6_PORT_UNREACH, par->hooknum);
 		break;
 	case IP6T_ICMP6_ECHOREPLY:
 		/* Do nothing */
 		break;
 	case IP6T_TCP_RESET:
-		send_reset(net, skb, par->hooknum);
+		send_reset6(net, skb, par->hooknum);
 		break;
 	default:
 		net_info_ratelimited("case %u not handled yet\n", reject->with);
-- 
1.8.5.1

Re: [PATCH 1/2] netfilter: REJECT: separate reusable code

[Pablo Neira Ayuso]

Hi Eric,

Some comments on this patch.

On Thu, Dec 12, 2013 at 08:44:40AM +0100, Eric Leblond wrote:
> This patch prepares the addition of TCP reset support in
> the nft_reject module by moving reusable code into a header
> file.
> 
> Signed-off-by: Eric Leblond <eric@regit.org>
> ---
>  include/net/netfilter/nf_reject.h | 297 ++++++++++++++++++++++++++++++++++++++
>  net/ipv4/netfilter/ipt_REJECT.c   | 124 +---------------
>  net/ipv6/netfilter/ip6t_REJECT.c  | 177 +----------------------
>  3 files changed, 307 insertions(+), 291 deletions(-)
>  create mode 100644 include/net/netfilter/nf_reject.h
> 
> diff --git a/include/net/netfilter/nf_reject.h b/include/net/netfilter/nf_reject.h
> new file mode 100644
> index 0000000..0bb089a
> --- /dev/null
> +++ b/include/net/netfilter/nf_reject.h
> @@ -0,0 +1,297 @@
> +#ifndef _NF_REJECT_H
> +#define _NF_REJECT_H
> +
> +#include <net/ip.h>
> +#include <net/tcp.h>
> +#include <net/route.h>
> +#include <net/dst.h>
> +
> +#if IS_ENABLED(CONFIG_IP6_NF_IPTABLES)
> +#include <net/ipv6.h>
> +#include <net/ip6_route.h>
> +#include <linux/netfilter_ipv6.h>
> +#endif
> +
> +static inline void send_unreach(struct sk_buff *skb_in, int code)
> +{
> +	icmp_send(skb_in, ICMP_DEST_UNREACH, code, 0);
> +}
> +
> +/* Send RST reply */
> +static inline void send_reset(struct sk_buff *oldskb, int hook)

I think that this function is too large to use the inline approach
that we have used in nft_queue. I've been considering moving this
common code to a different module and export these symbols but I
don't find a nice way to make it.

So my suggestion is to have two different include files:

include/net/netfilter/ipv4/nf_reject.h, that contains:

        static inline void send_unreach(...
        static void send_reset(...

that you can include from ipt_REJECT and nft_reject.

And include/net/netfilter/ipv6/nf_reject.h, that contains:

        static inline void send_unreach6(...
        static void send_reset6(...

that you can include from ip6t_REJECT and nft_reject.

While at it, it is probably good to prepend the prefix nf_ to those
functions.

Let me know if you have any concern with this approach.

Re: [PATCH 1/2] netfilter: REJECT: separate reusable code

[Eric Leblond]

Hello Pablo,

On Fri, 2013-12-20 at 10:50 +0100, Pablo Neira Ayuso wrote:
> Hi Eric,
> 
> Some comments on this patch.
> 
> On Thu, Dec 12, 2013 at 08:44:40AM +0100, Eric Leblond wrote:
> > This patch prepares the addition of TCP reset support in
> > the nft_reject module by moving reusable code into a header
> > file.
> > 
> > Signed-off-by: Eric Leblond <eric@regit.org>
> > ---
> >  include/net/netfilter/nf_reject.h | 297 ++++++++++++++++++++++++++++++++++++++
> >  net/ipv4/netfilter/ipt_REJECT.c   | 124 +---------------
> >  net/ipv6/netfilter/ip6t_REJECT.c  | 177 +----------------------
> >  3 files changed, 307 insertions(+), 291 deletions(-)
> >  create mode 100644 include/net/netfilter/nf_reject.h
> > 
> > diff --git a/include/net/netfilter/nf_reject.h b/include/net/netfilter/nf_reject.h
> > new file mode 100644
> > index 0000000..0bb089a
> > --- /dev/null
> > +++ b/include/net/netfilter/nf_reject.h
> > @@ -0,0 +1,297 @@
> > +#ifndef _NF_REJECT_H
> > +#define _NF_REJECT_H
> > +
> > +#include <net/ip.h>
> > +#include <net/tcp.h>
> > +#include <net/route.h>
> > +#include <net/dst.h>
> > +
> > +#if IS_ENABLED(CONFIG_IP6_NF_IPTABLES)
> > +#include <net/ipv6.h>
> > +#include <net/ip6_route.h>
> > +#include <linux/netfilter_ipv6.h>
> > +#endif
> > +
> > +static inline void send_unreach(struct sk_buff *skb_in, int code)
> > +{
> > +	icmp_send(skb_in, ICMP_DEST_UNREACH, code, 0);
> > +}
> > +
> > +/* Send RST reply */
> > +static inline void send_reset(struct sk_buff *oldskb, int hook)
> 
> I think that this function is too large to use the inline approach
> that we have used in nft_queue. I've been considering moving this
> common code to a different module and export these symbols but I
> don't find a nice way to make it.
> 
> So my suggestion is to have two different include files:
> 
> include/net/netfilter/ipv4/nf_reject.h, that contains:
> 
>         static inline void send_unreach(...
>         static void send_reset(...
> 
> that you can include from ipt_REJECT and nft_reject.
> 
> And include/net/netfilter/ipv6/nf_reject.h, that contains:
> 
>         static inline void send_unreach6(...
>         static void send_reset6(...
> 
> that you can include from ip6t_REJECT and nft_reject.
> 
> While at it, it is probably good to prepend the prefix nf_ to those
> functions.
> 
> Let me know if you have any concern with this approach.

I've implemented this but there is a corner case that I don't like: if
IPv6 is build as module then nft reject should be build as module too.
Putting a dependency on IPv6 to nft reject is not working as IPv4 only
setup will not be possible.
So I only see two solutions:
     1. We make IPv6 to be dependent of nft reject IPv6 (I don"t think
        this is a good choice)
     2. We keep two separate versions of the nft reject module (IPv4 and
        IPv6) to avoid this problem (not satisfying solution).

Do you have a better idea ?

I'm in favor of 2. as I think that it is a bit too much to act on IPv6
configuration value.

FYI the build message is:
net/built-in.o: In function `nft_reject_eval':
:(.text+0x2774a): undefined reference to `nf_ip6_checksum'
:(.text+0x27792): undefined reference to `ip6_route_output'


BR,
-- 
Eric Leblond <eric@regit.org>

[RFC PATCHv3 finish reject support]

[Eric Leblond]

Hello,

Here's a rework of previous patchset. No change on first patch if we
omit rebase on current code. Second patch now uses separate includes
for IPv6 and IPv4. It also fixes the dependency issue with IPv6 code
by adding a depends in Kconfig.

BR,
--
Eric Leblond

[RFC PATCHv3 1/2] netfilter: REJECT: separate reusable code

[Eric Leblond]

This patch prepares the addition of TCP reset support in
the nft_reject module by moving reusable code into a header
file.

Signed-off-by: Eric Leblond <eric@regit.org>
---
 include/net/netfilter/ipv4/nf_reject.h | 128 +++++++++++++++++++++++
 include/net/netfilter/ipv6/nf_reject.h | 171 +++++++++++++++++++++++++++++++
 net/ipv4/netfilter/ipt_REJECT.c        | 140 ++------------------------
 net/ipv6/netfilter/ip6t_REJECT.c       | 179 ++-------------------------------
 4 files changed, 317 insertions(+), 301 deletions(-)
 create mode 100644 include/net/netfilter/ipv4/nf_reject.h
 create mode 100644 include/net/netfilter/ipv6/nf_reject.h

diff --git a/include/net/netfilter/ipv4/nf_reject.h b/include/net/netfilter/ipv4/nf_reject.h
new file mode 100644
index 0000000..931fbf8
--- /dev/null
+++ b/include/net/netfilter/ipv4/nf_reject.h
@@ -0,0 +1,128 @@
+#ifndef _IPV4_NF_REJECT_H
+#define _IPV4_NF_REJECT_H
+
+#include <net/ip.h>
+#include <net/tcp.h>
+#include <net/route.h>
+#include <net/dst.h>
+
+static inline void nf_send_unreach(struct sk_buff *skb_in, int code)
+{
+	icmp_send(skb_in, ICMP_DEST_UNREACH, code, 0);
+}
+
+/* Send RST reply */
+static void nf_send_reset(struct sk_buff *oldskb, int hook)
+{
+	struct sk_buff *nskb;
+	const struct iphdr *oiph;
+	struct iphdr *niph;
+	const struct tcphdr *oth;
+	struct tcphdr _otcph, *tcph;
+
+	/* IP header checks: fragment. */
+	if (ip_hdr(oldskb)->frag_off & htons(IP_OFFSET))
+		return;
+
+	oth = skb_header_pointer(oldskb, ip_hdrlen(oldskb),
+				 sizeof(_otcph), &_otcph);
+	if (oth == NULL)
+		return;
+
+	/* No RST for RST. */
+	if (oth->rst)
+		return;
+
+	if (skb_rtable(oldskb)->rt_flags & (RTCF_BROADCAST | RTCF_MULTICAST))
+		return;
+
+	/* Check checksum */
+	if (nf_ip_checksum(oldskb, hook, ip_hdrlen(oldskb), IPPROTO_TCP))
+		return;
+	oiph = ip_hdr(oldskb);
+
+	nskb = alloc_skb(sizeof(struct iphdr) + sizeof(struct tcphdr) +
+			 LL_MAX_HEADER, GFP_ATOMIC);
+	if (!nskb)
+		return;
+
+	skb_reserve(nskb, LL_MAX_HEADER);
+
+	skb_reset_network_header(nskb);
+	niph = (struct iphdr *)skb_put(nskb, sizeof(struct iphdr));
+	niph->version	= 4;
+	niph->ihl	= sizeof(struct iphdr) / 4;
+	niph->tos	= 0;
+	niph->id	= 0;
+	niph->frag_off	= htons(IP_DF);
+	niph->protocol	= IPPROTO_TCP;
+	niph->check	= 0;
+	niph->saddr	= oiph->daddr;
+	niph->daddr	= oiph->saddr;
+
+	skb_reset_transport_header(nskb);
+	tcph = (struct tcphdr *)skb_put(nskb, sizeof(struct tcphdr));
+	memset(tcph, 0, sizeof(*tcph));
+	tcph->source	= oth->dest;
+	tcph->dest	= oth->source;
+	tcph->doff	= sizeof(struct tcphdr) / 4;
+
+	if (oth->ack)
+		tcph->seq = oth->ack_seq;
+	else {
+		tcph->ack_seq = htonl(ntohl(oth->seq) + oth->syn + oth->fin +
+				      oldskb->len - ip_hdrlen(oldskb) -
+				      (oth->doff << 2));
+		tcph->ack = 1;
+	}
+
+	tcph->rst	= 1;
+	tcph->check = ~tcp_v4_check(sizeof(struct tcphdr), niph->saddr,
+				    niph->daddr, 0);
+	nskb->ip_summed = CHECKSUM_PARTIAL;
+	nskb->csum_start = (unsigned char *)tcph - nskb->head;
+	nskb->csum_offset = offsetof(struct tcphdr, check);
+
+	/* ip_route_me_harder expects skb->dst to be set */
+	skb_dst_set_noref(nskb, skb_dst(oldskb));
+
+	nskb->protocol = htons(ETH_P_IP);
+	if (ip_route_me_harder(nskb, RTN_UNSPEC))
+		goto free_nskb;
+
+	niph->ttl	= ip4_dst_hoplimit(skb_dst(nskb));
+
+	/* "Never happens" */
+	if (nskb->len > dst_mtu(skb_dst(nskb)))
+		goto free_nskb;
+
+	nf_ct_attach(nskb, oldskb);
+
+#ifdef CONFIG_BRIDGE_NETFILTER
+	/* If we use ip_local_out for bridged traffic, the MAC source on
+	 * the RST will be ours, instead of the destination's.  This confuses
+	 * some routers/firewalls, and they drop the packet.  So we need to
+	 * build the eth header using the original destination's MAC as the
+	 * source, and send the RST packet directly.
+	 */
+	if (oldskb->nf_bridge) {
+		struct ethhdr *oeth = eth_hdr(oldskb);
+		nskb->dev = oldskb->nf_bridge->physindev;
+		niph->tot_len = htons(nskb->len);
+		ip_send_check(niph);
+		if (dev_hard_header(nskb, nskb->dev, ntohs(nskb->protocol),
+				    oeth->h_source, oeth->h_dest, nskb->len) < 0)
+			goto free_nskb;
+		dev_queue_xmit(nskb);
+	} else
+#endif
+		ip_local_out(nskb);
+
+	return;
+
+ free_nskb:
+	kfree_skb(nskb);
+}
+
+
+#endif /* _IPV4_NF_REJECT_H */
diff --git a/include/net/netfilter/ipv6/nf_reject.h b/include/net/netfilter/ipv6/nf_reject.h
new file mode 100644
index 0000000..710d17e
--- /dev/null
+++ b/include/net/netfilter/ipv6/nf_reject.h
@@ -0,0 +1,171 @@
+#ifndef _IPV6_NF_REJECT_H
+#define _IPV6_NF_REJECT_H
+
+#include <net/ipv6.h>
+#include <net/ip6_route.h>
+#include <net/ip6_fib.h>
+#include <net/ip6_checksum.h>
+#include <linux/netfilter_ipv6.h>
+
+static inline void
+nf_send_unreach6(struct net *net, struct sk_buff *skb_in, unsigned char code,
+	     unsigned int hooknum)
+{
+	if (hooknum == NF_INET_LOCAL_OUT && skb_in->dev == NULL)
+		skb_in->dev = net->loopback_dev;
+
+	icmpv6_send(skb_in, ICMPV6_DEST_UNREACH, code, 0);
+}
+
+/* Send RST reply */
+static void nf_send_reset6(struct net *net, struct sk_buff *oldskb, int hook)
+{
+	struct sk_buff *nskb;
+	struct tcphdr otcph, *tcph;
+	unsigned int otcplen, hh_len;
+	int tcphoff, needs_ack;
+	const struct ipv6hdr *oip6h = ipv6_hdr(oldskb);
+	struct ipv6hdr *ip6h;
+#define DEFAULT_TOS_VALUE	0x0U
+	const __u8 tclass = DEFAULT_TOS_VALUE;
+	struct dst_entry *dst = NULL;
+	u8 proto;
+	__be16 frag_off;
+	struct flowi6 fl6;
+
+	if ((!(ipv6_addr_type(&oip6h->saddr) & IPV6_ADDR_UNICAST)) ||
+	    (!(ipv6_addr_type(&oip6h->daddr) & IPV6_ADDR_UNICAST))) {
+		pr_debug("addr is not unicast.\n");
+		return;
+	}
+
+	proto = oip6h->nexthdr;
+	tcphoff = ipv6_skip_exthdr(oldskb, ((u8*)(oip6h+1) - oldskb->data), &proto, &frag_off);
+
+	if ((tcphoff < 0) || (tcphoff > oldskb->len)) {
+		pr_debug("Cannot get TCP header.\n");
+		return;
+	}
+
+	otcplen = oldskb->len - tcphoff;
+
+	/* IP header checks: fragment, too short. */
+	if (proto != IPPROTO_TCP || otcplen < sizeof(struct tcphdr)) {
+		pr_debug("proto(%d) != IPPROTO_TCP, "
+			 "or too short. otcplen = %d\n",
+			 proto, otcplen);
+		return;
+	}
+
+	if (skb_copy_bits(oldskb, tcphoff, &otcph, sizeof(struct tcphdr)))
+		BUG();
+
+	/* No RST for RST. */
+	if (otcph.rst) {
+		pr_debug("RST is set\n");
+		return;
+	}
+
+	/* Check checksum. */
+	if (nf_ip6_checksum(oldskb, hook, tcphoff, IPPROTO_TCP)) {
+		pr_debug("TCP checksum is invalid\n");
+		return;
+	}
+
+	memset(&fl6, 0, sizeof(fl6));
+	fl6.flowi6_proto = IPPROTO_TCP;
+	fl6.saddr = oip6h->daddr;
+	fl6.daddr = oip6h->saddr;
+	fl6.fl6_sport = otcph.dest;
+	fl6.fl6_dport = otcph.source;
+	security_skb_classify_flow(oldskb, flowi6_to_flowi(&fl6));
+	dst = ip6_route_output(net, NULL, &fl6);
+	if (dst == NULL || dst->error) {
+		dst_release(dst);
+		return;
+	}
+	dst = xfrm_lookup(net, dst, flowi6_to_flowi(&fl6), NULL, 0);
+	if (IS_ERR(dst))
+		return;
+
+	hh_len = (dst->dev->hard_header_len + 15)&~15;
+	nskb = alloc_skb(hh_len + 15 + dst->header_len + sizeof(struct ipv6hdr)
+			 + sizeof(struct tcphdr) + dst->trailer_len,
+			 GFP_ATOMIC);
+
+	if (!nskb) {
+		net_dbg_ratelimited("cannot alloc skb\n");
+		dst_release(dst);
+		return;
+	}
+
+	skb_dst_set(nskb, dst);
+
+	skb_reserve(nskb, hh_len + dst->header_len);
+
+	skb_put(nskb, sizeof(struct ipv6hdr));
+	skb_reset_network_header(nskb);
+	ip6h = ipv6_hdr(nskb);
+	ip6_flow_hdr(ip6h, tclass, 0);
+	ip6h->hop_limit = ip6_dst_hoplimit(dst);
+	ip6h->nexthdr = IPPROTO_TCP;
+	ip6h->saddr = oip6h->daddr;
+	ip6h->daddr = oip6h->saddr;
+
+	skb_reset_transport_header(nskb);
+	tcph = (struct tcphdr *)skb_put(nskb, sizeof(struct tcphdr));
+	/* Truncate to length (no data) */
+	tcph->doff = sizeof(struct tcphdr)/4;
+	tcph->source = otcph.dest;
+	tcph->dest = otcph.source;
+
+	if (otcph.ack) {
+		needs_ack = 0;
+		tcph->seq = otcph.ack_seq;
+		tcph->ack_seq = 0;
+	} else {
+		needs_ack = 1;
+		tcph->ack_seq = htonl(ntohl(otcph.seq) + otcph.syn + otcph.fin
+				      + otcplen - (otcph.doff<<2));
+		tcph->seq = 0;
+	}
+
+	/* Reset flags */
+	((u_int8_t *)tcph)[13] = 0;
+	tcph->rst = 1;
+	tcph->ack = needs_ack;
+	tcph->window = 0;
+	tcph->urg_ptr = 0;
+	tcph->check = 0;
+
+	/* Adjust TCP checksum */
+	tcph->check = csum_ipv6_magic(&ipv6_hdr(nskb)->saddr,
+				      &ipv6_hdr(nskb)->daddr,
+				      sizeof(struct tcphdr), IPPROTO_TCP,
+				      csum_partial(tcph,
+						   sizeof(struct tcphdr), 0));
+
+	nf_ct_attach(nskb, oldskb);
+
+#ifdef CONFIG_BRIDGE_NETFILTER
+	/* If we use ip6_local_out for bridged traffic, the MAC source on
+	 * the RST will be ours, instead of the destination's.  This confuses
+	 * some routers/firewalls, and they drop the packet.  So we need to
+	 * build the eth header using the original destination's MAC as the
+	 * source, and send the RST packet directly.
+	 */
+	if (oldskb->nf_bridge) {
+		struct ethhdr *oeth = eth_hdr(oldskb);
+		nskb->dev = oldskb->nf_bridge->physindev;
+		nskb->protocol = htons(ETH_P_IPV6);
+		ip6h->payload_len = htons(sizeof(struct tcphdr));
+		if (dev_hard_header(nskb, nskb->dev, ntohs(nskb->protocol),
+				    oeth->h_source, oeth->h_dest, nskb->len) < 0)
+			return;
+		dev_queue_xmit(nskb);
+	} else
+#endif
+		ip6_local_out(nskb);
+}
+
+#endif /* _IPV6_NF_REJECT_H */
diff --git a/net/ipv4/netfilter/ipt_REJECT.c b/net/ipv4/netfilter/ipt_REJECT.c
index b969131..5b6e0df 100644
--- a/net/ipv4/netfilter/ipt_REJECT.c
+++ b/net/ipv4/netfilter/ipt_REJECT.c
@@ -17,10 +17,6 @@
 #include <linux/udp.h>
 #include <linux/icmp.h>
 #include <net/icmp.h>
-#include <net/ip.h>
-#include <net/tcp.h>
-#include <net/route.h>
-#include <net/dst.h>
 #include <linux/netfilter/x_tables.h>
 #include <linux/netfilter_ipv4/ip_tables.h>
 #include <linux/netfilter_ipv4/ipt_REJECT.h>
@@ -28,128 +24,12 @@
 #include <linux/netfilter_bridge.h>
 #endif
 
+#include <net/netfilter/ipv4/nf_reject.h>
+
 MODULE_LICENSE("GPL");
 MODULE_AUTHOR("Netfilter Core Team <coreteam@netfilter.org>");
 MODULE_DESCRIPTION("Xtables: packet \"rejection\" target for IPv4");
 
-/* Send RST reply */
-static void send_reset(struct sk_buff *oldskb, int hook)
-{
-	struct sk_buff *nskb;
-	const struct iphdr *oiph;
-	struct iphdr *niph;
-	const struct tcphdr *oth;
-	struct tcphdr _otcph, *tcph;
-
-	/* IP header checks: fragment. */
-	if (ip_hdr(oldskb)->frag_off & htons(IP_OFFSET))
-		return;
-
-	oth = skb_header_pointer(oldskb, ip_hdrlen(oldskb),
-				 sizeof(_otcph), &_otcph);
-	if (oth == NULL)
-		return;
-
-	/* No RST for RST. */
-	if (oth->rst)
-		return;
-
-	if (skb_rtable(oldskb)->rt_flags & (RTCF_BROADCAST | RTCF_MULTICAST))
-		return;
-
-	/* Check checksum */
-	if (nf_ip_checksum(oldskb, hook, ip_hdrlen(oldskb), IPPROTO_TCP))
-		return;
-	oiph = ip_hdr(oldskb);
-
-	nskb = alloc_skb(sizeof(struct iphdr) + sizeof(struct tcphdr) +
-			 LL_MAX_HEADER, GFP_ATOMIC);
-	if (!nskb)
-		return;
-
-	skb_reserve(nskb, LL_MAX_HEADER);
-
-	skb_reset_network_header(nskb);
-	niph = (struct iphdr *)skb_put(nskb, sizeof(struct iphdr));
-	niph->version	= 4;
-	niph->ihl	= sizeof(struct iphdr) / 4;
-	niph->tos	= 0;
-	niph->id	= 0;
-	niph->frag_off	= htons(IP_DF);
-	niph->protocol	= IPPROTO_TCP;
-	niph->check	= 0;
-	niph->saddr	= oiph->daddr;
-	niph->daddr	= oiph->saddr;
-
-	skb_reset_transport_header(nskb);
-	tcph = (struct tcphdr *)skb_put(nskb, sizeof(struct tcphdr));
-	memset(tcph, 0, sizeof(*tcph));
-	tcph->source	= oth->dest;
-	tcph->dest	= oth->source;
-	tcph->doff	= sizeof(struct tcphdr) / 4;
-
-	if (oth->ack)
-		tcph->seq = oth->ack_seq;
-	else {
-		tcph->ack_seq = htonl(ntohl(oth->seq) + oth->syn + oth->fin +
-				      oldskb->len - ip_hdrlen(oldskb) -
-				      (oth->doff << 2));
-		tcph->ack = 1;
-	}
-
-	tcph->rst	= 1;
-	tcph->check = ~tcp_v4_check(sizeof(struct tcphdr), niph->saddr,
-				    niph->daddr, 0);
-	nskb->ip_summed = CHECKSUM_PARTIAL;
-	nskb->csum_start = (unsigned char *)tcph - nskb->head;
-	nskb->csum_offset = offsetof(struct tcphdr, check);
-
-	/* ip_route_me_harder expects skb->dst to be set */
-	skb_dst_set_noref(nskb, skb_dst(oldskb));
-
-	nskb->protocol = htons(ETH_P_IP);
-	if (ip_route_me_harder(nskb, RTN_UNSPEC))
-		goto free_nskb;
-
-	niph->ttl	= ip4_dst_hoplimit(skb_dst(nskb));
-
-	/* "Never happens" */
-	if (nskb->len > dst_mtu(skb_dst(nskb)))
-		goto free_nskb;
-
-	nf_ct_attach(nskb, oldskb);
-
-#ifdef CONFIG_BRIDGE_NETFILTER
-	/* If we use ip_local_out for bridged traffic, the MAC source on
-	 * the RST will be ours, instead of the destination's.  This confuses
-	 * some routers/firewalls, and they drop the packet.  So we need to
-	 * build the eth header using the original destination's MAC as the
-	 * source, and send the RST packet directly.
-	 */
-	if (oldskb->nf_bridge) {
-		struct ethhdr *oeth = eth_hdr(oldskb);
-		nskb->dev = oldskb->nf_bridge->physindev;
-		niph->tot_len = htons(nskb->len);
-		ip_send_check(niph);
-		if (dev_hard_header(nskb, nskb->dev, ntohs(nskb->protocol),
-				    oeth->h_source, oeth->h_dest, nskb->len) < 0)
-			goto free_nskb;
-		dev_queue_xmit(nskb);
-	} else
-#endif
-		ip_local_out(nskb);
-
-	return;
-
- free_nskb:
-	kfree_skb(nskb);
-}
-
-static inline void send_unreach(struct sk_buff *skb_in, int code)
-{
-	icmp_send(skb_in, ICMP_DEST_UNREACH, code, 0);
-}
-
 static unsigned int
 reject_tg(struct sk_buff *skb, const struct xt_action_param *par)
 {
@@ -157,28 +37,28 @@ reject_tg(struct sk_buff *skb, const struct xt_action_param *par)
 
 	switch (reject->with) {
 	case IPT_ICMP_NET_UNREACHABLE:
-		send_unreach(skb, ICMP_NET_UNREACH);
+		nf_send_unreach(skb, ICMP_NET_UNREACH);
 		break;
 	case IPT_ICMP_HOST_UNREACHABLE:
-		send_unreach(skb, ICMP_HOST_UNREACH);
+		nf_send_unreach(skb, ICMP_HOST_UNREACH);
 		break;
 	case IPT_ICMP_PROT_UNREACHABLE:
-		send_unreach(skb, ICMP_PROT_UNREACH);
+		nf_send_unreach(skb, ICMP_PROT_UNREACH);
 		break;
 	case IPT_ICMP_PORT_UNREACHABLE:
-		send_unreach(skb, ICMP_PORT_UNREACH);
+		nf_send_unreach(skb, ICMP_PORT_UNREACH);
 		break;
 	case IPT_ICMP_NET_PROHIBITED:
-		send_unreach(skb, ICMP_NET_ANO);
+		nf_send_unreach(skb, ICMP_NET_ANO);
 		break;
 	case IPT_ICMP_HOST_PROHIBITED:
-		send_unreach(skb, ICMP_HOST_ANO);
+		nf_send_unreach(skb, ICMP_HOST_ANO);
 		break;
 	case IPT_ICMP_ADMIN_PROHIBITED:
-		send_unreach(skb, ICMP_PKT_FILTERED);
+		nf_send_unreach(skb, ICMP_PKT_FILTERED);
 		break;
 	case IPT_TCP_RESET:
-		send_reset(skb, par->hooknum);
+		nf_send_reset(skb, par->hooknum);
 	case IPT_ICMP_ECHOREPLY:
 		/* Doesn't happen. */
 		break;
diff --git a/net/ipv6/netfilter/ip6t_REJECT.c b/net/ipv6/netfilter/ip6t_REJECT.c
index da00a2e..544b0a9 100644
--- a/net/ipv6/netfilter/ip6t_REJECT.c
+++ b/net/ipv6/netfilter/ip6t_REJECT.c
@@ -23,181 +23,18 @@
 #include <linux/skbuff.h>
 #include <linux/icmpv6.h>
 #include <linux/netdevice.h>
-#include <net/ipv6.h>
-#include <net/tcp.h>
 #include <net/icmp.h>
-#include <net/ip6_checksum.h>
-#include <net/ip6_fib.h>
-#include <net/ip6_route.h>
 #include <net/flow.h>
 #include <linux/netfilter/x_tables.h>
 #include <linux/netfilter_ipv6/ip6_tables.h>
 #include <linux/netfilter_ipv6/ip6t_REJECT.h>
 
+#include <net/netfilter/ipv6/nf_reject.h>
+
 MODULE_AUTHOR("Yasuyuki KOZAKAI <yasuyuki.kozakai@toshiba.co.jp>");
 MODULE_DESCRIPTION("Xtables: packet \"rejection\" target for IPv6");
 MODULE_LICENSE("GPL");
 
-/* Send RST reply */
-static void send_reset(struct net *net, struct sk_buff *oldskb, int hook)
-{
-	struct sk_buff *nskb;
-	struct tcphdr otcph, *tcph;
-	unsigned int otcplen, hh_len;
-	int tcphoff, needs_ack;
-	const struct ipv6hdr *oip6h = ipv6_hdr(oldskb);
-	struct ipv6hdr *ip6h;
-#define DEFAULT_TOS_VALUE	0x0U
-	const __u8 tclass = DEFAULT_TOS_VALUE;
-	struct dst_entry *dst = NULL;
-	u8 proto;
-	__be16 frag_off;
-	struct flowi6 fl6;
-
-	if ((!(ipv6_addr_type(&oip6h->saddr) & IPV6_ADDR_UNICAST)) ||
-	    (!(ipv6_addr_type(&oip6h->daddr) & IPV6_ADDR_UNICAST))) {
-		pr_debug("addr is not unicast.\n");
-		return;
-	}
-
-	proto = oip6h->nexthdr;
-	tcphoff = ipv6_skip_exthdr(oldskb, ((u8*)(oip6h+1) - oldskb->data), &proto, &frag_off);
-
-	if ((tcphoff < 0) || (tcphoff > oldskb->len)) {
-		pr_debug("Cannot get TCP header.\n");
-		return;
-	}
-
-	otcplen = oldskb->len - tcphoff;
-
-	/* IP header checks: fragment, too short. */
-	if (proto != IPPROTO_TCP || otcplen < sizeof(struct tcphdr)) {
-		pr_debug("proto(%d) != IPPROTO_TCP, "
-			 "or too short. otcplen = %d\n",
-			 proto, otcplen);
-		return;
-	}
-
-	if (skb_copy_bits(oldskb, tcphoff, &otcph, sizeof(struct tcphdr)))
-		BUG();
-
-	/* No RST for RST. */
-	if (otcph.rst) {
-		pr_debug("RST is set\n");
-		return;
-	}
-
-	/* Check checksum. */
-	if (nf_ip6_checksum(oldskb, hook, tcphoff, IPPROTO_TCP)) {
-		pr_debug("TCP checksum is invalid\n");
-		return;
-	}
-
-	memset(&fl6, 0, sizeof(fl6));
-	fl6.flowi6_proto = IPPROTO_TCP;
-	fl6.saddr = oip6h->daddr;
-	fl6.daddr = oip6h->saddr;
-	fl6.fl6_sport = otcph.dest;
-	fl6.fl6_dport = otcph.source;
-	security_skb_classify_flow(oldskb, flowi6_to_flowi(&fl6));
-	dst = ip6_route_output(net, NULL, &fl6);
-	if (dst == NULL || dst->error) {
-		dst_release(dst);
-		return;
-	}
-	dst = xfrm_lookup(net, dst, flowi6_to_flowi(&fl6), NULL, 0);
-	if (IS_ERR(dst))
-		return;
-
-	hh_len = (dst->dev->hard_header_len + 15)&~15;
-	nskb = alloc_skb(hh_len + 15 + dst->header_len + sizeof(struct ipv6hdr)
-			 + sizeof(struct tcphdr) + dst->trailer_len,
-			 GFP_ATOMIC);
-
-	if (!nskb) {
-		net_dbg_ratelimited("cannot alloc skb\n");
-		dst_release(dst);
-		return;
-	}
-
-	skb_dst_set(nskb, dst);
-
-	skb_reserve(nskb, hh_len + dst->header_len);
-
-	skb_put(nskb, sizeof(struct ipv6hdr));
-	skb_reset_network_header(nskb);
-	ip6h = ipv6_hdr(nskb);
-	ip6_flow_hdr(ip6h, tclass, 0);
-	ip6h->hop_limit = ip6_dst_hoplimit(dst);
-	ip6h->nexthdr = IPPROTO_TCP;
-	ip6h->saddr = oip6h->daddr;
-	ip6h->daddr = oip6h->saddr;
-
-	skb_reset_transport_header(nskb);
-	tcph = (struct tcphdr *)skb_put(nskb, sizeof(struct tcphdr));
-	/* Truncate to length (no data) */
-	tcph->doff = sizeof(struct tcphdr)/4;
-	tcph->source = otcph.dest;
-	tcph->dest = otcph.source;
-
-	if (otcph.ack) {
-		needs_ack = 0;
-		tcph->seq = otcph.ack_seq;
-		tcph->ack_seq = 0;
-	} else {
-		needs_ack = 1;
-		tcph->ack_seq = htonl(ntohl(otcph.seq) + otcph.syn + otcph.fin
-				      + otcplen - (otcph.doff<<2));
-		tcph->seq = 0;
-	}
-
-	/* Reset flags */
-	((u_int8_t *)tcph)[13] = 0;
-	tcph->rst = 1;
-	tcph->ack = needs_ack;
-	tcph->window = 0;
-	tcph->urg_ptr = 0;
-	tcph->check = 0;
-
-	/* Adjust TCP checksum */
-	tcph->check = csum_ipv6_magic(&ipv6_hdr(nskb)->saddr,
-				      &ipv6_hdr(nskb)->daddr,
-				      sizeof(struct tcphdr), IPPROTO_TCP,
-				      csum_partial(tcph,
-						   sizeof(struct tcphdr), 0));
-
-	nf_ct_attach(nskb, oldskb);
-
-#ifdef CONFIG_BRIDGE_NETFILTER
-	/* If we use ip6_local_out for bridged traffic, the MAC source on
-	 * the RST will be ours, instead of the destination's.  This confuses
-	 * some routers/firewalls, and they drop the packet.  So we need to
-	 * build the eth header using the original destination's MAC as the
-	 * source, and send the RST packet directly.
-	 */
-	if (oldskb->nf_bridge) {
-		struct ethhdr *oeth = eth_hdr(oldskb);
-		nskb->dev = oldskb->nf_bridge->physindev;
-		nskb->protocol = htons(ETH_P_IPV6);
-		ip6h->payload_len = htons(sizeof(struct tcphdr));
-		if (dev_hard_header(nskb, nskb->dev, ntohs(nskb->protocol),
-				    oeth->h_source, oeth->h_dest, nskb->len) < 0)
-			return;
-		dev_queue_xmit(nskb);
-	} else
-#endif
-		ip6_local_out(nskb);
-}
-
-static inline void
-send_unreach(struct net *net, struct sk_buff *skb_in, unsigned char code,
-	     unsigned int hooknum)
-{
-	if (hooknum == NF_INET_LOCAL_OUT && skb_in->dev == NULL)
-		skb_in->dev = net->loopback_dev;
-
-	icmpv6_send(skb_in, ICMPV6_DEST_UNREACH, code, 0);
-}
 
 static unsigned int
 reject_tg6(struct sk_buff *skb, const struct xt_action_param *par)
@@ -208,25 +45,25 @@ reject_tg6(struct sk_buff *skb, const struct xt_action_param *par)
 	pr_debug("%s: medium point\n", __func__);
 	switch (reject->with) {
 	case IP6T_ICMP6_NO_ROUTE:
-		send_unreach(net, skb, ICMPV6_NOROUTE, par->hooknum);
+		nf_send_unreach6(net, skb, ICMPV6_NOROUTE, par->hooknum);
 		break;
 	case IP6T_ICMP6_ADM_PROHIBITED:
-		send_unreach(net, skb, ICMPV6_ADM_PROHIBITED, par->hooknum);
+		nf_send_unreach6(net, skb, ICMPV6_ADM_PROHIBITED, par->hooknum);
 		break;
 	case IP6T_ICMP6_NOT_NEIGHBOUR:
-		send_unreach(net, skb, ICMPV6_NOT_NEIGHBOUR, par->hooknum);
+		nf_send_unreach6(net, skb, ICMPV6_NOT_NEIGHBOUR, par->hooknum);
 		break;
 	case IP6T_ICMP6_ADDR_UNREACH:
-		send_unreach(net, skb, ICMPV6_ADDR_UNREACH, par->hooknum);
+		nf_send_unreach6(net, skb, ICMPV6_ADDR_UNREACH, par->hooknum);
 		break;
 	case IP6T_ICMP6_PORT_UNREACH:
-		send_unreach(net, skb, ICMPV6_PORT_UNREACH, par->hooknum);
+		nf_send_unreach6(net, skb, ICMPV6_PORT_UNREACH, par->hooknum);
 		break;
 	case IP6T_ICMP6_ECHOREPLY:
 		/* Do nothing */
 		break;
 	case IP6T_TCP_RESET:
-		send_reset(net, skb, par->hooknum);
+		nf_send_reset6(net, skb, par->hooknum);
 		break;
 	default:
 		net_info_ratelimited("case %u not handled yet\n", reject->with);
-- 
1.8.5.2

Re: [RFC PATCHv3 1/2] netfilter: REJECT: separate reusable code

[Pablo Neira Ayuso]

On Sun, Dec 29, 2013 at 12:28:13PM +0100, Eric Leblond wrote:
> This patch prepares the addition of TCP reset support in
> the nft_reject module by moving reusable code into a header
> file.

Applied, thanks Eric.

[RFC PATCHv3 2/2] netfilter: nft: reject support for IPv6 and TCP reset

[Eric Leblond]

This patch moves nft_reject_ipv4 to nft_reject and adds support
for IPv6 protocol. This patch uses functions included in nf_reject.h
to implement reject by TCP reset.

The code has to be build as a module if NF_TABLES_IPV6 is also a
module to avoid compilation error due to usage of IPv6 functions.
This has been done in Kconfig by using the construct:
 depends on NF_TABLES_IPV6 || !NF_TABLES_IPV6
This seems a bit weird in term of syntax but works perfectly.

Signed-off-by: Eric Leblond <eric@regit.org>
---
 net/ipv4/netfilter/Kconfig           |   4 -
 net/ipv4/netfilter/Makefile          |   1 -
 net/ipv4/netfilter/nft_reject_ipv4.c | 123 ------------------------------
 net/netfilter/Kconfig                |   6 ++
 net/netfilter/Makefile               |   1 +
 net/netfilter/nft_reject.c           | 144 +++++++++++++++++++++++++++++++++++
 6 files changed, 151 insertions(+), 128 deletions(-)
 delete mode 100644 net/ipv4/netfilter/nft_reject_ipv4.c
 create mode 100644 net/netfilter/nft_reject.c

diff --git a/net/ipv4/netfilter/Kconfig b/net/ipv4/netfilter/Kconfig
index 40d5607..9d3d69a 100644
--- a/net/ipv4/netfilter/Kconfig
+++ b/net/ipv4/netfilter/Kconfig
@@ -40,10 +40,6 @@ config NF_TABLES_IPV4
 	depends on NF_TABLES
 	tristate "IPv4 nf_tables support"
 
-config NFT_REJECT_IPV4
-	depends on NF_TABLES_IPV4
-	tristate "nf_tables IPv4 reject support"
-
 config NFT_CHAIN_ROUTE_IPV4
 	depends on NF_TABLES_IPV4
 	tristate "IPv4 nf_tables route chain support"
diff --git a/net/ipv4/netfilter/Makefile b/net/ipv4/netfilter/Makefile
index 19df72b..c16be9d 100644
--- a/net/ipv4/netfilter/Makefile
+++ b/net/ipv4/netfilter/Makefile
@@ -28,7 +28,6 @@ obj-$(CONFIG_NF_NAT_SNMP_BASIC) += nf_nat_snmp_basic.o
 obj-$(CONFIG_NF_NAT_PROTO_GRE) += nf_nat_proto_gre.o
 
 obj-$(CONFIG_NF_TABLES_IPV4) += nf_tables_ipv4.o
-obj-$(CONFIG_NFT_REJECT_IPV4) += nft_reject_ipv4.o
 obj-$(CONFIG_NFT_CHAIN_ROUTE_IPV4) += nft_chain_route_ipv4.o
 obj-$(CONFIG_NFT_CHAIN_NAT_IPV4) += nft_chain_nat_ipv4.o
 obj-$(CONFIG_NF_TABLES_ARP) += nf_tables_arp.o
diff --git a/net/ipv4/netfilter/nft_reject_ipv4.c b/net/ipv4/netfilter/nft_reject_ipv4.c
deleted file mode 100644
index fff5ba1..0000000
--- a/net/ipv4/netfilter/nft_reject_ipv4.c
+++ /dev/null
@@ -1,123 +0,0 @@
-/*
- * Copyright (c) 2008-2009 Patrick McHardy <kaber@trash.net>
- *
- * This program is free software; you can redistribute it and/or modify
- * it under the terms of the GNU General Public License version 2 as
- * published by the Free Software Foundation.
- *
- * Development of this code funded by Astaro AG (http://www.astaro.com/)
- */
-
-#include <linux/kernel.h>
-#include <linux/init.h>
-#include <linux/module.h>
-#include <linux/netlink.h>
-#include <linux/netfilter.h>
-#include <linux/netfilter/nf_tables.h>
-#include <net/netfilter/nf_tables.h>
-#include <net/icmp.h>
-
-struct nft_reject {
-	enum nft_reject_types	type:8;
-	u8			icmp_code;
-};
-
-static void nft_reject_eval(const struct nft_expr *expr,
-			      struct nft_data data[NFT_REG_MAX + 1],
-			      const struct nft_pktinfo *pkt)
-{
-	struct nft_reject *priv = nft_expr_priv(expr);
-
-	switch (priv->type) {
-	case NFT_REJECT_ICMP_UNREACH:
-		icmp_send(pkt->skb, ICMP_DEST_UNREACH, priv->icmp_code, 0);
-		break;
-	case NFT_REJECT_TCP_RST:
-		break;
-	}
-
-	data[NFT_REG_VERDICT].verdict = NF_DROP;
-}
-
-static const struct nla_policy nft_reject_policy[NFTA_REJECT_MAX + 1] = {
-	[NFTA_REJECT_TYPE]		= { .type = NLA_U32 },
-	[NFTA_REJECT_ICMP_CODE]		= { .type = NLA_U8 },
-};
-
-static int nft_reject_init(const struct nft_ctx *ctx,
-			   const struct nft_expr *expr,
-			   const struct nlattr * const tb[])
-{
-	struct nft_reject *priv = nft_expr_priv(expr);
-
-	if (tb[NFTA_REJECT_TYPE] == NULL)
-		return -EINVAL;
-
-	priv->type = ntohl(nla_get_be32(tb[NFTA_REJECT_TYPE]));
-	switch (priv->type) {
-	case NFT_REJECT_ICMP_UNREACH:
-		if (tb[NFTA_REJECT_ICMP_CODE] == NULL)
-			return -EINVAL;
-		priv->icmp_code = nla_get_u8(tb[NFTA_REJECT_ICMP_CODE]);
-	case NFT_REJECT_TCP_RST:
-		break;
-	default:
-		return -EINVAL;
-	}
-
-	return 0;
-}
-
-static int nft_reject_dump(struct sk_buff *skb, const struct nft_expr *expr)
-{
-	const struct nft_reject *priv = nft_expr_priv(expr);
-
-	if (nla_put_be32(skb, NFTA_REJECT_TYPE, priv->type))
-		goto nla_put_failure;
-
-	switch (priv->type) {
-	case NFT_REJECT_ICMP_UNREACH:
-		if (nla_put_u8(skb, NFTA_REJECT_ICMP_CODE, priv->icmp_code))
-			goto nla_put_failure;
-		break;
-	}
-
-	return 0;
-
-nla_put_failure:
-	return -1;
-}
-
-static struct nft_expr_type nft_reject_type;
-static const struct nft_expr_ops nft_reject_ops = {
-	.type		= &nft_reject_type,
-	.size		= NFT_EXPR_SIZE(sizeof(struct nft_reject)),
-	.eval		= nft_reject_eval,
-	.init		= nft_reject_init,
-	.dump		= nft_reject_dump,
-};
-
-static struct nft_expr_type nft_reject_type __read_mostly = {
-	.name		= "reject",
-	.ops		= &nft_reject_ops,
-	.policy		= nft_reject_policy,
-	.maxattr	= NFTA_REJECT_MAX,
-	.owner		= THIS_MODULE,
-};
-
-static int __init nft_reject_module_init(void)
-{
-	return nft_register_expr(&nft_reject_type);
-}
-
-static void __exit nft_reject_module_exit(void)
-{
-	nft_unregister_expr(&nft_reject_type);
-}
-
-module_init(nft_reject_module_init);
-module_exit(nft_reject_module_exit);
-
-MODULE_LICENSE("GPL");
-MODULE_AUTHOR("Patrick McHardy <kaber@trash.net>");
-MODULE_ALIAS_NFT_EXPR("reject");
diff --git a/net/netfilter/Kconfig b/net/netfilter/Kconfig
index 01f9f64..a1dec61 100644
--- a/net/netfilter/Kconfig
+++ b/net/netfilter/Kconfig
@@ -465,6 +465,12 @@ config NFT_QUEUE
 	  This is required if you intend to use the userspace queueing
 	  infrastructure (also known as NFQUEUE) from nftables.
 
+config NFT_REJECT
+	depends on NF_TABLES
+	depends on NF_TABLES_IPV6 || !NF_TABLES_IPV6
+	default m if NETFILTER_ADVANCED=n
+	tristate "Netfilter nf_tables reject support"
+
 config NFT_COMPAT
 	depends on NF_TABLES
 	depends on NETFILTER_XTABLES
diff --git a/net/netfilter/Makefile b/net/netfilter/Makefile
index dcc818a..39e4a7b 100644
--- a/net/netfilter/Makefile
+++ b/net/netfilter/Makefile
@@ -77,6 +77,7 @@ obj-$(CONFIG_NFT_CT)		+= nft_ct.o
 obj-$(CONFIG_NFT_LIMIT)		+= nft_limit.o
 obj-$(CONFIG_NFT_NAT)		+= nft_nat.o
 obj-$(CONFIG_NFT_QUEUE)		+= nft_queue.o
+obj-$(CONFIG_NFT_REJECT) 	+= nft_reject.o
 obj-$(CONFIG_NFT_RBTREE)	+= nft_rbtree.o
 obj-$(CONFIG_NFT_HASH)		+= nft_hash.o
 obj-$(CONFIG_NFT_COUNTER)	+= nft_counter.o
diff --git a/net/netfilter/nft_reject.c b/net/netfilter/nft_reject.c
new file mode 100644
index 0000000..0d690d4
--- /dev/null
+++ b/net/netfilter/nft_reject.c
@@ -0,0 +1,144 @@
+/*
+ * Copyright (c) 2008-2009 Patrick McHardy <kaber@trash.net>
+ * Copyright (c) 2013 Eric Leblond <eric@regit.org>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * Development of this code funded by Astaro AG (http://www.astaro.com/)
+ */
+
+#include <linux/kernel.h>
+#include <linux/init.h>
+#include <linux/module.h>
+#include <linux/netlink.h>
+#include <linux/netfilter.h>
+#include <linux/netfilter/nf_tables.h>
+#include <net/netfilter/nf_tables.h>
+#include <net/icmp.h>
+#include <net/netfilter/ipv4/nf_reject.h>
+
+#if IS_ENABLED(CONFIG_NF_TABLES_IPV6)
+#include <net/netfilter/ipv6/nf_reject.h>
+#endif
+
+struct nft_reject {
+	enum nft_reject_types	type:8;
+	u8			icmp_code;
+	u8			family;
+};
+
+static void nft_reject_eval(const struct nft_expr *expr,
+			      struct nft_data data[NFT_REG_MAX + 1],
+			      const struct nft_pktinfo *pkt)
+{
+	struct nft_reject *priv = nft_expr_priv(expr);
+	struct net *net = dev_net((pkt->in != NULL) ? pkt->in : pkt->out);
+
+	switch (priv->type) {
+	case NFT_REJECT_ICMP_UNREACH:
+		if (priv->family == NFPROTO_IPV4)
+			nf_send_unreach(pkt->skb, priv->icmp_code);
+#if IS_ENABLED(CONFIG_NF_TABLES_IPV6)
+		else if (priv->family == NFPROTO_IPV6)
+			nf_send_unreach6(net, pkt->skb, priv->icmp_code,
+				      pkt->hooknum);
+#endif
+		break;
+	case NFT_REJECT_TCP_RST:
+		if (priv->family == NFPROTO_IPV4)
+			nf_send_reset(pkt->skb, pkt->hooknum);
+#if IS_ENABLED(CONFIG_NF_TABLES_IPV6)
+		else if (priv->family == NFPROTO_IPV6)
+			nf_send_reset6(net, pkt->skb, pkt->hooknum);
+#endif
+		break;
+	}
+
+	data[NFT_REG_VERDICT].verdict = NF_DROP;
+}
+
+static const struct nla_policy nft_reject_policy[NFTA_REJECT_MAX + 1] = {
+	[NFTA_REJECT_TYPE]		= { .type = NLA_U32 },
+	[NFTA_REJECT_ICMP_CODE]		= { .type = NLA_U8 },
+};
+
+static int nft_reject_init(const struct nft_ctx *ctx,
+			   const struct nft_expr *expr,
+			   const struct nlattr * const tb[])
+{
+	struct nft_reject *priv = nft_expr_priv(expr);
+
+	if (tb[NFTA_REJECT_TYPE] == NULL)
+		return -EINVAL;
+
+	priv->family = ctx->afi->family;
+	priv->type = ntohl(nla_get_be32(tb[NFTA_REJECT_TYPE]));
+	switch (priv->type) {
+	case NFT_REJECT_ICMP_UNREACH:
+		if (tb[NFTA_REJECT_ICMP_CODE] == NULL)
+			return -EINVAL;
+		priv->icmp_code = nla_get_u8(tb[NFTA_REJECT_ICMP_CODE]);
+	case NFT_REJECT_TCP_RST:
+		break;
+	default:
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static int nft_reject_dump(struct sk_buff *skb, const struct nft_expr *expr)
+{
+	const struct nft_reject *priv = nft_expr_priv(expr);
+
+	if (nla_put_be32(skb, NFTA_REJECT_TYPE, htonl(priv->type)))
+		goto nla_put_failure;
+
+	switch (priv->type) {
+	case NFT_REJECT_ICMP_UNREACH:
+		if (nla_put_u8(skb, NFTA_REJECT_ICMP_CODE, priv->icmp_code))
+			goto nla_put_failure;
+		break;
+	}
+
+	return 0;
+
+nla_put_failure:
+	return -1;
+}
+
+static struct nft_expr_type nft_reject_type;
+static const struct nft_expr_ops nft_reject_ops = {
+	.type		= &nft_reject_type,
+	.size		= NFT_EXPR_SIZE(sizeof(struct nft_reject)),
+	.eval		= nft_reject_eval,
+	.init		= nft_reject_init,
+	.dump		= nft_reject_dump,
+};
+
+static struct nft_expr_type nft_reject_type __read_mostly = {
+	.name		= "reject",
+	.ops		= &nft_reject_ops,
+	.policy		= nft_reject_policy,
+	.maxattr	= NFTA_REJECT_MAX,
+	.owner		= THIS_MODULE,
+};
+
+static int __init nft_reject_module_init(void)
+{
+	return nft_register_expr(&nft_reject_type);
+}
+
+static void __exit nft_reject_module_exit(void)
+{
+	nft_unregister_expr(&nft_reject_type);
+}
+
+module_init(nft_reject_module_init);
+module_exit(nft_reject_module_exit);
+
+MODULE_LICENSE("GPL");
+MODULE_AUTHOR("Patrick McHardy <kaber@trash.net>");
+MODULE_ALIAS_NFT_EXPR("reject");
-- 
1.8.5.2

Re: [RFC PATCHv3 2/2] netfilter: nft: reject support for IPv6 and TCP reset

[Pablo Neira Ayuso]

On Sun, Dec 29, 2013 at 12:28:14PM +0100, Eric Leblond wrote:
> This patch moves nft_reject_ipv4 to nft_reject and adds support
> for IPv6 protocol. This patch uses functions included in nf_reject.h
> to implement reject by TCP reset.
> 
> The code has to be build as a module if NF_TABLES_IPV6 is also a
> module to avoid compilation error due to usage of IPv6 functions.
> This has been done in Kconfig by using the construct:
>  depends on NF_TABLES_IPV6 || !NF_TABLES_IPV6
> This seems a bit weird in term of syntax but works perfectly.

Also applied, thanks.

[PATCH 2/2] netfilter: nft: reject support for IPv6 and TCP reset

[Eric Leblond]

This patch moves nft_reject_ipv4 to nft_reject and adds support
for IPv6 protocol. This patch uses functions included in nf_reject.h
to implement reject by TCP reset.

Signed-off-by: Eric Leblond <eric@regit.org>
---
 net/ipv4/netfilter/Kconfig           |   4 -
 net/ipv4/netfilter/Makefile          |   1 -
 net/ipv4/netfilter/nft_reject_ipv4.c | 123 ------------------------------
 net/netfilter/Kconfig                |   4 +
 net/netfilter/Makefile               |   1 +
 net/netfilter/nft_reject.c           | 143 +++++++++++++++++++++++++++++++++++
 6 files changed, 148 insertions(+), 128 deletions(-)
 delete mode 100644 net/ipv4/netfilter/nft_reject_ipv4.c
 create mode 100644 net/netfilter/nft_reject.c

diff --git a/net/ipv4/netfilter/Kconfig b/net/ipv4/netfilter/Kconfig
index 40d5607..9d3d69a 100644
--- a/net/ipv4/netfilter/Kconfig
+++ b/net/ipv4/netfilter/Kconfig
@@ -40,10 +40,6 @@ config NF_TABLES_IPV4
 	depends on NF_TABLES
 	tristate "IPv4 nf_tables support"
 
-config NFT_REJECT_IPV4
-	depends on NF_TABLES_IPV4
-	tristate "nf_tables IPv4 reject support"
-
 config NFT_CHAIN_ROUTE_IPV4
 	depends on NF_TABLES_IPV4
 	tristate "IPv4 nf_tables route chain support"
diff --git a/net/ipv4/netfilter/Makefile b/net/ipv4/netfilter/Makefile
index 19df72b..c16be9d 100644
--- a/net/ipv4/netfilter/Makefile
+++ b/net/ipv4/netfilter/Makefile
@@ -28,7 +28,6 @@ obj-$(CONFIG_NF_NAT_SNMP_BASIC) += nf_nat_snmp_basic.o
 obj-$(CONFIG_NF_NAT_PROTO_GRE) += nf_nat_proto_gre.o
 
 obj-$(CONFIG_NF_TABLES_IPV4) += nf_tables_ipv4.o
-obj-$(CONFIG_NFT_REJECT_IPV4) += nft_reject_ipv4.o
 obj-$(CONFIG_NFT_CHAIN_ROUTE_IPV4) += nft_chain_route_ipv4.o
 obj-$(CONFIG_NFT_CHAIN_NAT_IPV4) += nft_chain_nat_ipv4.o
 obj-$(CONFIG_NF_TABLES_ARP) += nf_tables_arp.o
diff --git a/net/ipv4/netfilter/nft_reject_ipv4.c b/net/ipv4/netfilter/nft_reject_ipv4.c
deleted file mode 100644
index fff5ba1..0000000
--- a/net/ipv4/netfilter/nft_reject_ipv4.c
+++ /dev/null
@@ -1,123 +0,0 @@
-/*
- * Copyright (c) 2008-2009 Patrick McHardy <kaber@trash.net>
- *
- * This program is free software; you can redistribute it and/or modify
- * it under the terms of the GNU General Public License version 2 as
- * published by the Free Software Foundation.
- *
- * Development of this code funded by Astaro AG (http://www.astaro.com/)
- */
-
-#include <linux/kernel.h>
-#include <linux/init.h>
-#include <linux/module.h>
-#include <linux/netlink.h>
-#include <linux/netfilter.h>
-#include <linux/netfilter/nf_tables.h>
-#include <net/netfilter/nf_tables.h>
-#include <net/icmp.h>
-
-struct nft_reject {
-	enum nft_reject_types	type:8;
-	u8			icmp_code;
-};
-
-static void nft_reject_eval(const struct nft_expr *expr,
-			      struct nft_data data[NFT_REG_MAX + 1],
-			      const struct nft_pktinfo *pkt)
-{
-	struct nft_reject *priv = nft_expr_priv(expr);
-
-	switch (priv->type) {
-	case NFT_REJECT_ICMP_UNREACH:
-		icmp_send(pkt->skb, ICMP_DEST_UNREACH, priv->icmp_code, 0);
-		break;
-	case NFT_REJECT_TCP_RST:
-		break;
-	}
-
-	data[NFT_REG_VERDICT].verdict = NF_DROP;
-}
-
-static const struct nla_policy nft_reject_policy[NFTA_REJECT_MAX + 1] = {
-	[NFTA_REJECT_TYPE]		= { .type = NLA_U32 },
-	[NFTA_REJECT_ICMP_CODE]		= { .type = NLA_U8 },
-};
-
-static int nft_reject_init(const struct nft_ctx *ctx,
-			   const struct nft_expr *expr,
-			   const struct nlattr * const tb[])
-{
-	struct nft_reject *priv = nft_expr_priv(expr);
-
-	if (tb[NFTA_REJECT_TYPE] == NULL)
-		return -EINVAL;
-
-	priv->type = ntohl(nla_get_be32(tb[NFTA_REJECT_TYPE]));
-	switch (priv->type) {
-	case NFT_REJECT_ICMP_UNREACH:
-		if (tb[NFTA_REJECT_ICMP_CODE] == NULL)
-			return -EINVAL;
-		priv->icmp_code = nla_get_u8(tb[NFTA_REJECT_ICMP_CODE]);
-	case NFT_REJECT_TCP_RST:
-		break;
-	default:
-		return -EINVAL;
-	}
-
-	return 0;
-}
-
-static int nft_reject_dump(struct sk_buff *skb, const struct nft_expr *expr)
-{
-	const struct nft_reject *priv = nft_expr_priv(expr);
-
-	if (nla_put_be32(skb, NFTA_REJECT_TYPE, priv->type))
-		goto nla_put_failure;
-
-	switch (priv->type) {
-	case NFT_REJECT_ICMP_UNREACH:
-		if (nla_put_u8(skb, NFTA_REJECT_ICMP_CODE, priv->icmp_code))
-			goto nla_put_failure;
-		break;
-	}
-
-	return 0;
-
-nla_put_failure:
-	return -1;
-}
-
-static struct nft_expr_type nft_reject_type;
-static const struct nft_expr_ops nft_reject_ops = {
-	.type		= &nft_reject_type,
-	.size		= NFT_EXPR_SIZE(sizeof(struct nft_reject)),
-	.eval		= nft_reject_eval,
-	.init		= nft_reject_init,
-	.dump		= nft_reject_dump,
-};
-
-static struct nft_expr_type nft_reject_type __read_mostly = {
-	.name		= "reject",
-	.ops		= &nft_reject_ops,
-	.policy		= nft_reject_policy,
-	.maxattr	= NFTA_REJECT_MAX,
-	.owner		= THIS_MODULE,
-};
-
-static int __init nft_reject_module_init(void)
-{
-	return nft_register_expr(&nft_reject_type);
-}
-
-static void __exit nft_reject_module_exit(void)
-{
-	nft_unregister_expr(&nft_reject_type);
-}
-
-module_init(nft_reject_module_init);
-module_exit(nft_reject_module_exit);
-
-MODULE_LICENSE("GPL");
-MODULE_AUTHOR("Patrick McHardy <kaber@trash.net>");
-MODULE_ALIAS_NFT_EXPR("reject");
diff --git a/net/netfilter/Kconfig b/net/netfilter/Kconfig
index 5263605..9482822 100644
--- a/net/netfilter/Kconfig
+++ b/net/netfilter/Kconfig
@@ -465,6 +465,10 @@ config NFT_QUEUE
 	  This is required if you intend to use the userspace queueing
 	  infrastructure (also known as NFQUEUE) from nftables.
 
+config NFT_REJECT
+	depends on NF_TABLES
+	tristate "nf_tables reject support"
+
 config NFT_COMPAT
 	depends on NF_TABLES
 	depends on NETFILTER_XTABLES
diff --git a/net/netfilter/Makefile b/net/netfilter/Makefile
index e763746..ba578f6 100644
--- a/net/netfilter/Makefile
+++ b/net/netfilter/Makefile
@@ -77,6 +77,7 @@ obj-$(CONFIG_NFT_CT)		+= nft_ct.o
 obj-$(CONFIG_NFT_LIMIT)		+= nft_limit.o
 obj-$(CONFIG_NFT_NAT)		+= nft_nat.o
 obj-$(CONFIG_NFT_QUEUE)		+= nft_queue.o
+obj-$(CONFIG_NFT_REJECT) 	+= nft_reject.o
 #nf_tables-objs			+= nft_meta_target.o
 obj-$(CONFIG_NFT_RBTREE)	+= nft_rbtree.o
 obj-$(CONFIG_NFT_HASH)		+= nft_hash.o
diff --git a/net/netfilter/nft_reject.c b/net/netfilter/nft_reject.c
new file mode 100644
index 0000000..f7f30a1
--- /dev/null
+++ b/net/netfilter/nft_reject.c
@@ -0,0 +1,143 @@
+/*
+ * Copyright (c) 2008-2009 Patrick McHardy <kaber@trash.net>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * Development of this code funded by Astaro AG (http://www.astaro.com/)
+ */
+
+#include <linux/kernel.h>
+#include <linux/init.h>
+#include <linux/module.h>
+#include <linux/netlink.h>
+#include <linux/netfilter.h>
+#include <linux/netfilter/nf_tables.h>
+#include <net/netfilter/nf_tables.h>
+#include <net/icmp.h>
+#if IS_ENABLED(CONFIG_IP6_NF_IPTABLES)
+#include <net/ipv6.h>
+#endif
+
+#include <net/netfilter/nf_reject.h>
+
+struct nft_reject {
+	enum nft_reject_types	type:8;
+	u8			icmp_code;
+	u8			family;
+};
+
+static void nft_reject_eval(const struct nft_expr *expr,
+			      struct nft_data data[NFT_REG_MAX + 1],
+			      const struct nft_pktinfo *pkt)
+{
+	struct nft_reject *priv = nft_expr_priv(expr);
+	struct net *net = dev_net((pkt->in != NULL) ? pkt->in : pkt->out);
+
+	switch (priv->type) {
+	case NFT_REJECT_ICMP_UNREACH:
+		if (priv->family == NFPROTO_IPV4)
+			send_unreach(pkt->skb, priv->icmp_code);
+#if IS_ENABLED(CONFIG_IP6_NF_IPTABLES)
+		else if (priv->family == NFPROTO_IPV6)
+			send_unreach6(net, pkt->skb, priv->icmp_code,
+				      pkt->hooknum);
+#endif
+		break;
+	case NFT_REJECT_TCP_RST:
+		if (priv->family == NFPROTO_IPV4)
+			send_reset(pkt->skb, pkt->hooknum);
+#if IS_ENABLED(CONFIG_IP6_NF_IPTABLES)
+		else if (priv->family == NFPROTO_IPV6)
+			send_reset6(net, pkt->skb, pkt->hooknum);
+#endif
+		break;
+	}
+
+	data[NFT_REG_VERDICT].verdict = NF_DROP;
+}
+
+static const struct nla_policy nft_reject_policy[NFTA_REJECT_MAX + 1] = {
+	[NFTA_REJECT_TYPE]		= { .type = NLA_U32 },
+	[NFTA_REJECT_ICMP_CODE]		= { .type = NLA_U8 },
+};
+
+static int nft_reject_init(const struct nft_ctx *ctx,
+			   const struct nft_expr *expr,
+			   const struct nlattr * const tb[])
+{
+	struct nft_reject *priv = nft_expr_priv(expr);
+
+	if (tb[NFTA_REJECT_TYPE] == NULL)
+		return -EINVAL;
+
+	priv->family = ctx->afi->family;
+	priv->type = ntohl(nla_get_be32(tb[NFTA_REJECT_TYPE]));
+	switch (priv->type) {
+	case NFT_REJECT_ICMP_UNREACH:
+		if (tb[NFTA_REJECT_ICMP_CODE] == NULL)
+			return -EINVAL;
+		priv->icmp_code = nla_get_u8(tb[NFTA_REJECT_ICMP_CODE]);
+	case NFT_REJECT_TCP_RST:
+		break;
+	default:
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static int nft_reject_dump(struct sk_buff *skb, const struct nft_expr *expr)
+{
+	const struct nft_reject *priv = nft_expr_priv(expr);
+
+	if (nla_put_be32(skb, NFTA_REJECT_TYPE, htonl(priv->type)))
+		goto nla_put_failure;
+
+	switch (priv->type) {
+	case NFT_REJECT_ICMP_UNREACH:
+		if (nla_put_u8(skb, NFTA_REJECT_ICMP_CODE, priv->icmp_code))
+			goto nla_put_failure;
+		break;
+	}
+
+	return 0;
+
+nla_put_failure:
+	return -1;
+}
+
+static struct nft_expr_type nft_reject_type;
+static const struct nft_expr_ops nft_reject_ops = {
+	.type		= &nft_reject_type,
+	.size		= NFT_EXPR_SIZE(sizeof(struct nft_reject)),
+	.eval		= nft_reject_eval,
+	.init		= nft_reject_init,
+	.dump		= nft_reject_dump,
+};
+
+static struct nft_expr_type nft_reject_type __read_mostly = {
+	.name		= "reject",
+	.ops		= &nft_reject_ops,
+	.policy		= nft_reject_policy,
+	.maxattr	= NFTA_REJECT_MAX,
+	.owner		= THIS_MODULE,
+};
+
+static int __init nft_reject_module_init(void)
+{
+	return nft_register_expr(&nft_reject_type);
+}
+
+static void __exit nft_reject_module_exit(void)
+{
+	nft_unregister_expr(&nft_reject_type);
+}
+
+module_init(nft_reject_module_init);
+module_exit(nft_reject_module_exit);
+
+MODULE_LICENSE("GPL");
+MODULE_AUTHOR("Patrick McHardy <kaber@trash.net>");
+MODULE_ALIAS_NFT_EXPR("reject");
-- 
1.8.5.1

Re: [PATCH 2/2] netfilter: nft: reject support for IPv6 and TCP reset

[Tomasz Bursztyka]

Hi Eric,

> +#if IS_ENABLED(CONFIG_IP6_NF_IPTABLES)

Isn't it NF_TABLES_IPV6 instead?


Tomasz

Re: [PATCH 2/2] netfilter: nft: reject support for IPv6 and TCP reset

[Tomasz Bursztyka]

>> +#if IS_ENABLED(CONFIG_IP6_NF_IPTABLES)
>
> Isn't it NF_TABLES_IPV6 instead?
>

I of course meant CONFIG_NF_TABLES_IPV6, sorry

Re: [PATCH 2/2] netfilter: nft: reject support for IPv6 and TCP reset

[Eric Leblond]

Hi,

On Thu, 2013-12-12 at 09:57 +0200, Tomasz Bursztyka wrote:
> >> +#if IS_ENABLED(CONFIG_IP6_NF_IPTABLES)
> >
> > Isn't it NF_TABLES_IPV6 instead?
> >
> 
> I of course meant CONFIG_NF_TABLES_IPV6, sorry

Good point and this also means we have a issue in nf_reject.h where we
use CONFIG_IP6_NF_IPTABLES which is out-of-scope in nft_reject.

I'll try to fix that.

++
-- 
Eric Leblond <eric@regit.org>