nft: improve reject support
Subsystems (2)
cover nft: improve reject support
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
This patch prepares the addition of TCP reset support in the nft_reject module by moving reusable code into a header file.
--- 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
include/net/netfilter/nf_reject.h new file +297 -0
@@ -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 */
net/ipv4/netfilter/ipt_REJECT.c +2 -122
140 lines not on this page · raw message
net/ipv6/netfilter/ip6t_REJECT.c +8 -169
214 lines not on this page · raw message
Signature · 1 line
1.8.5.1
PATCH 2/2 netfilter: nft: reject support for IPv6 and TCP reset
In reply to Eric LeblondThis 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.
--- 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
net/ipv4/netfilter/Kconfig +0 -4
@@ -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"
net/ipv4/netfilter/Makefile +0 -1
@@ -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-$(CONFIG_NF_TABLES_ARP) += nf_tables_arp.o
net/ipv4/netfilter/nft_reject_ipv4.c deleted +0 -123
123 lines not on this page · raw message
net/netfilter/Kconfig +4 -0
@@ -465,6 +465,10 @@ config NFT_QUEUE This is required if you intend to use the userspace queueing infrastructure (also known as NFQUEUE) from nftables. +config NFT_REJECT + depends on NF_TABLES + tristate "nf_tables reject support" + config NFT_COMPAT depends on NF_TABLES depends on NETFILTER_XTABLES
net/netfilter/Makefile +1 -0
@@ -77,6 +77,7 @@ obj-$(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
net/netfilter/nft_reject.c new file +143 -0
143 lines not on this page · raw message
Signature · 1 line
1.8.5.1
cover nft: finish reject support
In reply to Eric LeblondHello,
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
This patch prepares the addition of TCP reset support in the nft_reject module by moving reusable code into a header file.
--- 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
include/net/netfilter/nf_reject.h new file +297 -0
@@ -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 */
net/ipv4/netfilter/ipt_REJECT.c +2 -122
140 lines not on this page · raw message
net/ipv6/netfilter/ip6t_REJECT.c +8 -169
214 lines not on this page · raw message
Signature · 1 line
1.8.5.1
Pablo Neira Ayuso
maintainerRe: [PATCH 1/2] netfilter: REJECT: separate reusable code
Hi Eric,
Some comments on this patch.
Quoted 29 lines · Eric Leblond
This patch prepares the addition of TCP reset support in the
nft_rejectmodule by moving reusable code into a header file.Signed-off-by Eric Leblond <[email protected]>--- 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.hinclude/net/netfilter/nf_reject.h new file +11
-0@@ -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/net/netfilter/nf_reject.h line 21
Quoted 10 lines
include/net/netfilter/nf_reject.h new file +10
-0+#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
Hello Pablo,
Quoted 64 lines · Pablo Neira Ayuso
Hi Eric,
Some comments on this patch.
Quoted 29 lines · Eric Leblond
This patch prepares the addition of TCP reset support in the
nft_rejectmodule by moving reusable code into a header file.Signed-off-by Eric Leblond <[email protected]>--- 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.hinclude/net/netfilter/nf_reject.h new file +11
-0@@ -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/net/netfilter/nf_reject.h line 21
Quoted 10 lines
include/net/netfilter/nf_reject.h new file +10
-0+#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_REJECTandnft_reject.And
include/net/netfilter/ipv6/nf_reject.h, that contains:static inline void send_unreach6(...
static void send_reset6(...
Quoted 6 lines
that you can include from
ip6t_REJECTandnft_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,
Signature · 1 line
Eric Leblond <[email protected]>
RFC PATCH v3 1/1 [RFC PATCHv3 finish reject support]
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 PATCH v3 1/2 netfilter: REJECT: separate reusable code
This patch prepares the addition of TCP reset support in the nft_reject module by moving reusable code into a header file.
--- 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
include/net/netfilter/ipv4/nf_reject.h new file +128 -0
@@ -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 */
include/net/netfilter/ipv6/nf_reject.h new file +171 -0
171 lines not on this page · raw message
net/ipv4/netfilter/ipt_REJECT.c +10 -130
176 lines not on this page · raw message
net/ipv6/netfilter/ip6t_REJECT.c +8 -171
214 lines not on this page · raw message
Signature · 1 line
1.8.5.2
Re: [RFC PATCHv3 1/2] netfilter: REJECT: separate reusable code
Quoted 3 lines · Eric Leblond
This patch prepares the addition of TCP reset support in the
nft_rejectmodule by moving reusable code into a header file.
Applied, thanks Eric.
RFC PATCH v3 2/2 netfilter: nft: reject support for IPv6 and TCP reset
In reply to Eric LeblondThis 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.
--- 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
net/ipv4/netfilter/Kconfig +0 -4
@@ -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"
net/ipv4/netfilter/Makefile +0 -1
@@ -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-$(CONFIG_NF_TABLES_ARP) += nf_tables_arp.o
net/ipv4/netfilter/nft_reject_ipv4.c deleted +0 -123
123 lines not on this page · raw message
net/netfilter/Kconfig +6 -0
@@ -465,6 +465,12 @@ config NFT_QUEUE This is required if you intend to use the userspace queueing infrastructure (also known as NFQUEUE) from nftables. +config NFT_REJECT + depends on NF_TABLES + depends on NF_TABLES_IPV6 || !NF_TABLES_IPV6 + default m if NETFILTER_ADVANCED=n + tristate "Netfilter nf_tables reject support" + config NFT_COMPAT depends on NF_TABLES depends on NETFILTER_XTABLES
net/netfilter/Makefile +1 -0
@@ -77,6 +77,7 @@ obj-$(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-$(CONFIG_NFT_COUNTER) += nft_counter.o
net/netfilter/nft_reject.c new file +144 -0
144 lines not on this page · raw message
Signature · 1 line
1.8.5.2
Pablo Neira Ayuso
maintainerRe: [RFC PATCHv3 2/2] netfilter: nft: reject support for IPv6 and TCP reset
Quoted 9 lines · Eric Leblond
This patch moves
nft_reject_ipv4tonft_rejectand adds support for IPv6 protocol. This patch uses functions included innf_reject.hto implement reject by TCP reset.The code has to be build as a module if
NF_TABLES_IPV6is also a module to avoid compilation error due to usage of IPv6 functions. This has been done inKconfigby using the construct:
depends onNF_TABLES_IPV6|| !NF_TABLES_IPV6
This seems a bit weird in term of syntax but works perfectly.
Also applied, thanks.
PATCH 2/2 netfilter: nft: reject support for IPv6 and TCP reset
In reply to Eric LeblondThis patch moves nft_reject_ipv4 to nft_reject and adds support for IPv6 protocol. This patch uses functions included in nf_reject.h to implement reject by TCP reset.
--- net/ipv4/netfilter/Kconfig | 4 - net/ipv4/netfilter/Makefile | 1 - net/ipv4/netfilter/nft_reject_ipv4.c | 123 ------------------------------ net/netfilter/Kconfig | 4 + net/netfilter/Makefile | 1 + net/netfilter/nft_reject.c | 143 +++++++++++++++++++++++++++++++++++ 6 files changed, 148 insertions(+), 128 deletions(-) delete mode 100644 net/ipv4/netfilter/nft_reject_ipv4.c create mode 100644 net/netfilter/nft_reject.c
net/ipv4/netfilter/Kconfig +0 -4
@@ -40,10 +40,6 @@ config NF_TABLES_IPV4 depends on NF_TABLES tristate "IPv4 nf_tables support" -config NFT_REJECT_IPV4 - depends on NF_TABLES_IPV4 - tristate "nf_tables IPv4 reject support" - config NFT_CHAIN_ROUTE_IPV4 depends on NF_TABLES_IPV4 tristate "IPv4 nf_tables route chain support"
net/ipv4/netfilter/Makefile +0 -1
@@ -28,7 +28,6 @@ obj-$(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
net/ipv4/netfilter/nft_reject_ipv4.c deleted +0 -123
123 lines not on this page · raw message
net/netfilter/Kconfig +4 -0
@@ -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
net/netfilter/Makefile +1 -0
@@ -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
net/netfilter/nft_reject.c new file +143 -0
143 lines not on this page · raw message
Signature · 1 line
1.8.5.1
Re: [PATCH 2/2] netfilter: nft: reject support for IPv6 and TCP reset
Hi Eric,
Quoted 1 line
+#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
Quoted 5 lines
Quoted 2 lines
+#if
IS_ENABLED(CONFIG_IP6_NF_IPTABLES)Isn't it
NF_TABLES_IPV6instead?
I of course meant CONFIG_NF_TABLES_IPV6, sorry
Re: [PATCH 2/2] netfilter: nft: reject support for IPv6 and TCP reset
Hi,
Quoted 6 lines · Tomasz Bursztyka
Quoted 4 lines
Quoted 1 line
+#if
IS_ENABLED(CONFIG_IP6_NF_IPTABLES)Isn't it
NF_TABLES_IPV6instead?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.
++
Signature · 1 line
Eric Leblond <[email protected]>