Skip to contentSkip to conversation

NAT updates for nf_tables

PATCH Superseded Arturo Borrero Gonzalez 3 participants 10 messages Jul 1, 2014 – Jul 25, 2014

Subsystems (2)

v3 of 4 Superseded Latest is v4 →

lore
Share & tools
lore
#0 lore raw

v3 cover NAT updates for nf_tables

The following series implements some updates for NAT in nf_tables.

First of all, I add a new flag attribute to allow clients of nft_nat to specify additional config flags. This enables implementing port randomization and persistence to be set from nft.

Two patches split the masquerade code from ip[6]t_MASQUERADE.c to generic modules, so we can use this NAT type from nft_nat.

Then, the nft_nat code is splitted in AF specific parts, so we avoid potential dependencies regarding AF specific symbols in the last patch.

The last patch finally implements masquerade for nft_nat.

The v2 series included some fixes and additionals checks, as requested by Florian Westphal.

This v3 series includes changes requested by Pablo Neira.

Comments are welcomed.

---

Arturo Borrero Gonzalez (5):
netfilter: nft_nat: include a flag attribute
netfilter: nf_nat_masquerade_ipv4: code factorization
netfilter: nf_nat_masquerade_ipv6: code factorization
netfilter: nft_nat: split code in AF parts
netfilter: nft_nat: add masquerade support

 .../net/netfilter/ipv4/nf_nat_masquerade_ipv4.h    |   14 ++
 .../net/netfilter/ipv6/nf_nat_masquerade_ipv6.h    |   10 +
 include/net/netfilter/nft_nat.h                    |   22 +++
 include/uapi/linux/netfilter/nf_nat.h              |    5 +
 include/uapi/linux/netfilter/nf_tables.h           |   10 +
 net/ipv4/netfilter/Kconfig                         |   14 ++
 net/ipv4/netfilter/Makefile                        |    2 
 net/ipv4/netfilter/ipt_MASQUERADE.c                |  108 +-------------
 net/ipv4/netfilter/nf_nat_masquerade_ipv4.c        |  155 ++++++++++++++++++++
 net/ipv4/netfilter/nft_nat_ipv4.c                  |  133 +++++++++++++++++
 net/ipv6/netfilter/Kconfig                         |   14 ++
 net/ipv6/netfilter/Makefile                        |    2 
 net/ipv6/netfilter/ip6t_MASQUERADE.c               |   76 +---------
 net/ipv6/netfilter/nf_nat_masquerade_ipv6.c        |  121 ++++++++++++++++
 net/ipv6/netfilter/nft_nat_ipv6.c                  |  132 +++++++++++++++++
 net/netfilter/nft_nat.c                            |  156 ++++++--------------
 16 files changed, 688 insertions(+), 286 deletions(-)
 create mode 100644 include/net/netfilter/ipv4/nf_nat_masquerade_ipv4.h
 create mode 100644 include/net/netfilter/ipv6/nf_nat_masquerade_ipv6.h
 create mode 100644 include/net/netfilter/nft_nat.h
 create mode 100644 net/ipv4/netfilter/nf_nat_masquerade_ipv4.c
 create mode 100644 net/ipv4/netfilter/nft_nat_ipv4.c
 create mode 100644 net/ipv6/netfilter/nf_nat_masquerade_ipv6.c
 create mode 100644 net/ipv6/netfilter/nft_nat_ipv6.c
Signature · 1 line
Arturo Borrero Gonzalez
#1 lore raw

PATCH v3 1/5 netfilter: nft_nat: include a flag attribute

Both SNAT and DNAT (and the upcoming masquerade) can have additional configuration parameters, such as port randomization or NAT addressing persistence.
We can cover these scenarios by simply adding a flag attribute for userspace to fill when needed.

The flags to use are defined in include/uapi/linux/netfilter/nf_nat.h,
NF_NAT_RANGE_MAP_IPS
NF_NAT_RANGE_PROTO_SPECIFIED
NF_NAT_RANGE_PROTO_RANDOM
NF_NAT_RANGE_PERSISTENT
NF_NAT_RANGE_PROTO_RANDOM_FULLY NF_NAT_RANGE_PROTO_RANDOM_ALL

The caller must take care of not messing up with the flags, as they are added unconditionally to the final resulting nf_nat_range.

---
v2: address Florian Westphal's comments: check all flag bits to be known.
v3: style cleanup requested by Pablo Neira. Mask name shortened.

 include/uapi/linux/netfilter/nf_nat.h    |    5 +++++
 include/uapi/linux/netfilter/nf_tables.h |    2 ++
 net/netfilter/nft_nat.c                  |   16 ++++++++++++++++
 3 files changed, 23 insertions(+)
include/uapi/linux/netfilter/nf_nat.h +5 -0
index 1ad3659..898db2d 100644
@@ -13,6 +13,11 @@
 #define NF_NAT_RANGE_PROTO_RANDOM_ALL		\
 	(NF_NAT_RANGE_PROTO_RANDOM | NF_NAT_RANGE_PROTO_RANDOM_FULLY)
 
+#define NF_NAT_RANGE_MASK					\
+	(NF_NAT_RANGE_MAP_IPS|NF_NAT_RANGE_PROTO_SPECIFIED	\
+	 |NF_NAT_RANGE_PROTO_RANDOM|NF_NAT_RANGE_PERSISTENT	\
+	 |NF_NAT_RANGE_PROTO_RANDOM_FULLY)
+
 struct nf_nat_ipv4_range {
 	unsigned int			flags;
 	__be32				min_ip;
include/uapi/linux/netfilter/nf_tables.h +2 -0
index 2a88f64..92c211b 100644
@@ -773,6 +773,7 @@ enum nft_nat_types {
  * @NFTA_NAT_REG_ADDR_MAX: source register of address range end (NLA_U32: nft_registers)
  * @NFTA_NAT_REG_PROTO_MIN: source register of proto range start (NLA_U32: nft_registers)
  * @NFTA_NAT_REG_PROTO_MAX: source register of proto range end (NLA_U32: nft_registers)
+ * @NFTA_NAT_FLAGS: additional NAT configuration (NF_NAT_RANGE_*) (NLA_U32)
  */
 enum nft_nat_attributes {
 	NFTA_NAT_UNSPEC,
@@ -782,6 +783,7 @@ enum nft_nat_attributes {
 	NFTA_NAT_REG_ADDR_MAX,
 	NFTA_NAT_REG_PROTO_MIN,
 	NFTA_NAT_REG_PROTO_MAX,
+	NFTA_NAT_FLAGS,
 	__NFTA_NAT_MAX
 };
 #define NFTA_NAT_MAX		(__NFTA_NAT_MAX - 1)
net/netfilter/nft_nat.c +16 -0
index 79ff58c..799550b 100644
@@ -33,6 +33,7 @@ struct nft_nat {
 	enum nft_registers      sreg_proto_max:8;
 	enum nf_nat_manip_type  type:8;
 	u8			family;
+	u16			flags;
 };
 
 static void nft_nat_eval(const struct nft_expr *expr,
@@ -71,6 +72,8 @@ static void nft_nat_eval(const struct nft_expr *expr,
 		range.flags |= NF_NAT_RANGE_PROTO_SPECIFIED;
 	}
 
+	range.flags |= priv->flags;
+
 	data[NFT_REG_VERDICT].verdict =
 		nf_nat_setup_info(ct, &range, priv->type);
 }
@@ -82,6 +85,7 @@ static const struct nla_policy nft_nat_policy[NFTA_NAT_MAX + 1] = {
 	[NFTA_NAT_REG_ADDR_MAX]	 = { .type = NLA_U32 },
 	[NFTA_NAT_REG_PROTO_MIN] = { .type = NLA_U32 },
 	[NFTA_NAT_REG_PROTO_MAX] = { .type = NLA_U32 },
+	[NFTA_NAT_FLAGS]	 = { .type = NLA_U32 },
 };
 
 static int nft_nat_init(const struct nft_ctx *ctx, const struct nft_expr *expr,
@@ -149,6 +153,12 @@ static int nft_nat_init(const struct nft_ctx *ctx, const struct nft_expr *expr,
 	} else
 		priv->sreg_proto_max = priv->sreg_proto_min;
 
+	if (tb[NFTA_NAT_FLAGS]) {
+		priv->flags = ntohl(nla_get_be32(tb[NFTA_NAT_FLAGS]));
+		if (priv->flags & ~NF_NAT_RANGE_MASK)
+			return -EINVAL;
+	}
+
 	return 0;
 }
 
@@ -183,6 +193,12 @@ static int nft_nat_dump(struct sk_buff *skb, const struct nft_expr *expr)
 				 htonl(priv->sreg_proto_max)))
 			goto nla_put_failure;
 	}
+
+	if (priv->flags != 0) {
+		if (nla_put_be32(skb, NFTA_NAT_FLAGS, htonl(priv->flags)))
+			goto nla_put_failure;
+	}
+
 	return 0;
 
 nla_put_failure:
#2 lore raw

PATCH v3 2/5 netfilter: nf_nat_masquerade_ipv4: code factorization

In reply to Arturo Borrero Gonzalez

Let's refactor the code so we can reach the masquerade functionality from outside the xt context (ie, nftables).

The patch includes adding an atomic counter to the masquerade notifier: the stuff to be done by the notifier is the same in any case, and agnostic about who called it. Only one notification handler is needed.

This factorization only involves IPv4; a similar patch will follow to handle IPv6.

---
v2: no changes.
v3: no changes.

 .../net/netfilter/ipv4/nf_nat_masquerade_ipv4.h    |   14 ++
 net/ipv4/netfilter/Kconfig                         |    7 +
 net/ipv4/netfilter/Makefile                        |    1 
 net/ipv4/netfilter/ipt_MASQUERADE.c                |  108 +-------------
 net/ipv4/netfilter/nf_nat_masquerade_ipv4.c        |  155 ++++++++++++++++++++
 5 files changed, 186 insertions(+), 99 deletions(-)
 create mode 100644 include/net/netfilter/ipv4/nf_nat_masquerade_ipv4.h
 create mode 100644 net/ipv4/netfilter/nf_nat_masquerade_ipv4.c
include/net/netfilter/ipv4/nf_nat_masquerade_ipv4.h new file +14 -0
new file mode 100644
index 0000000..a9c001c
@@ -0,0 +1,14 @@
+#ifndef _NF_NAT_MASQUERADE_IPV4_H_
+#define _NF_NAT_MASQUERADE_IPV4_H_
+
+#include <net/netfilter/nf_nat.h>
+
+unsigned int
+nf_nat_masquerade_ipv4(struct sk_buff *skb, unsigned int hooknum,
+		       const struct nf_nat_range *range,
+		       const struct net_device *out);
+
+void nf_nat_masquerade_ipv4_register_notifier(void);
+void nf_nat_masquerade_ipv4_unregister_notifier(void);
+
+#endif /*_NF_NAT_MASQUERADE_IPV4_H_ */
net/ipv4/netfilter/Kconfig +7 -0
index a26ce03..f2d2202 100644
@@ -193,8 +193,15 @@ config NF_NAT_IPV4
 
 if NF_NAT_IPV4
 
+config NF_NAT_MASQUERADE_IPV4
+	tristate "IPv4 masquerade support"
+	help
+	This is the kernel functionality to provide NAT in the masquerade
+	flavour (automatic source address selection).
+
 config IP_NF_TARGET_MASQUERADE
 	tristate "MASQUERADE target support"
+	select NF_NAT_MASQUERADE_IPV4
 	default m if NETFILTER_ADVANCED=n
 	help
 	  Masquerading is a special case of NAT: all outgoing connections are
net/ipv4/netfilter/Makefile +1 -0
index 90b8240..a7bfa0a 100644
@@ -23,6 +23,7 @@ obj-$(CONFIG_NF_DEFRAG_IPV4) += nf_defrag_ipv4.o
 obj-$(CONFIG_NF_NAT_H323) += nf_nat_h323.o
 obj-$(CONFIG_NF_NAT_PPTP) += nf_nat_pptp.o
 obj-$(CONFIG_NF_NAT_SNMP_BASIC) += nf_nat_snmp_basic.o
+obj-$(CONFIG_NF_NAT_MASQUERADE_IPV4) += nf_nat_masquerade_ipv4.o
 
 # NAT protocols (nf_nat)
 obj-$(CONFIG_NF_NAT_PROTO_GRE) += nf_nat_proto_gre.o
net/ipv4/netfilter/ipt_MASQUERADE.c +9 -99

138 lines not on this page · raw message

net/ipv4/netfilter/nf_nat_masquerade_ipv4.c new file +155 -0

155 lines not on this page · raw message

#3 lore raw

Re: [linux PATCH v3 2/5] netfilter: nf_nat_masquerade_ipv4: code factorization

Quoted 4 lines · Arturo Borrero Gonzalez

Let's refactor the code so we can reach the masquerade functionality from
outside the xt context (ie, nftables).

Quoted 8 lines

The patch includes adding an atomic counter to the masquerade notifier:
the
stuff to be done by the notifier is the same in any case, and agnostic about who called it. Only one notification handler is needed.

This factorization only involves IPv4; a similar patch will follow to handle
IPv6.

Just a suggestion, the NAT support is parameterizable at runtime. An alternative would be an expression to load the local address.

Signature · 1 line
Diese Nachricht wurde von meinem Android-Mobiltelefon mit K-9 Mail gesendet.
#4 lore raw

Re: [linux PATCH v3 2/5] netfilter: nf_nat_masquerade_ipv4: code factorization

Hi Patrick,

Quoted 10 lines · Patrick McHardy
Quoted 9 lines · Arturo Borrero Gonzalez

Let's refactor the code so we can reach the masquerade functionality from
outside the xt context (ie, nftables).

The patch includes adding an atomic counter to the masquerade notifier:
the
stuff to be done by the notifier is the same in any case, and agnostic about who called it. Only one notification handler is needed.

Quoted 6 lines
Quoted 3 lines

This factorization only involves IPv4; a similar patch will follow to handle
IPv6.

Just a suggestion, the NAT support is parameterizable at runtime. An alternative would be an expression to load the local address.

That seems quite natural way to make it without requiring kernel changes, I like it. The only problem that I see is that I don't come up with a way to handle the conntrack cleanup case that needs to happen if the interface is brought down with this approach.

#5 lore raw

PATCH v3 3/5 netfilter: nf_nat_masquerade_ipv6: code factorization

In reply to Arturo Borrero Gonzalez

Let's refactor the code so we can reach the masquerade functionality from outside the xt context (ie, nftables).

The patch includes adding an atomic counter to the masquerade notifier: the stuff to be done by the notifier is the same in any case, and agnostic about who called it. Only one notification handler is needed.

This factorization only involves IPv6; a similar patch exists to handle IPv4.

---
v2: no changes.
v3: no changes.

 .../net/netfilter/ipv6/nf_nat_masquerade_ipv6.h    |   10 ++
 net/ipv6/netfilter/Kconfig                         |    7 +
 net/ipv6/netfilter/Makefile                        |    1 
 net/ipv6/netfilter/ip6t_MASQUERADE.c               |   76 +------------
 net/ipv6/netfilter/nf_nat_masquerade_ipv6.c        |  121 ++++++++++++++++++++
 5 files changed, 144 insertions(+), 71 deletions(-)
 create mode 100644 include/net/netfilter/ipv6/nf_nat_masquerade_ipv6.h
 create mode 100644 net/ipv6/netfilter/nf_nat_masquerade_ipv6.c
include/net/netfilter/ipv6/nf_nat_masquerade_ipv6.h new file +10 -0
new file mode 100644
index 0000000..0a13396
@@ -0,0 +1,10 @@
+#ifndef _NF_NAT_MASQUERADE_IPV6_H_
+#define _NF_NAT_MASQUERADE_IPV6_H_
+
+unsigned int
+nf_nat_masquerade_ipv6(struct sk_buff *skb, const struct nf_nat_range *range,
+		       const struct net_device *out);
+void nf_nat_masquerade_ipv6_register_notifier(void);
+void nf_nat_masquerade_ipv6_unregister_notifier(void);
+
+#endif /* _NF_NAT_MASQUERADE_IPV6_H_ */
net/ipv6/netfilter/Kconfig +7 -0
index 4bff1f2..86d24f9 100644
@@ -241,8 +241,15 @@ config NF_NAT_IPV6
 
 if NF_NAT_IPV6
 
+config NF_NAT_MASQUERADE_IPV6
+	tristate "IPv6 masquerade support"
+	help
+	 This is the kernel functionality to provide NAT in the masquerade
+	 flavour (automatic source address selection) for IPv6.
+
 config IP6_NF_TARGET_MASQUERADE
 	tristate "MASQUERADE target support"
+	select NF_NAT_MASQUERADE_IPV6
 	help
 	  Masquerading is a special case of NAT: all outgoing connections are
 	  changed to seem to come from a particular interface's address, and
net/ipv6/netfilter/Makefile +1 -0
index 70d3dd6..9d85d60 100644
@@ -18,6 +18,7 @@ obj-$(CONFIG_NF_CONNTRACK_IPV6) += nf_conntrack_ipv6.o
 
 nf_nat_ipv6-y		:= nf_nat_l3proto_ipv6.o nf_nat_proto_icmpv6.o
 obj-$(CONFIG_NF_NAT_IPV6) += nf_nat_ipv6.o
+obj-$(CONFIG_NF_NAT_MASQUERADE_IPV6) += nf_nat_masquerade_ipv6.o
 
 # defrag
 nf_defrag_ipv6-y := nf_defrag_ipv6_hooks.o nf_conntrack_reasm.o
net/ipv6/netfilter/ip6t_MASQUERADE.c +5 -71

103 lines not on this page · raw message

net/ipv6/netfilter/nf_nat_masquerade_ipv6.c new file +121 -0

121 lines not on this page · raw message

#6 lore raw

PATCH v3 4/5 netfilter: nft_nat: split code in AF parts

In reply to Arturo Borrero Gonzalez

This patch refactorices the nft_nat code into AF specific parts, allowing further work in the AF specific zones, like adding masquerade support.

While at it, fix coding style in several places.

---
v2: no changes.
v3: flags is u16. This also applies to the patch 1/5. Common _dump() function.

 include/net/netfilter/nft_nat.h   |   22 ++++++
 net/ipv4/netfilter/Kconfig        |    7 ++
 net/ipv4/netfilter/Makefile       |    1 
 net/ipv4/netfilter/nft_nat_ipv4.c |   99 ++++++++++++++++++++++++++++
 net/ipv6/netfilter/Kconfig        |    7 ++
 net/ipv6/netfilter/Makefile       |    1 
 net/ipv6/netfilter/nft_nat_ipv6.c |   99 ++++++++++++++++++++++++++++
 net/netfilter/nft_nat.c           |  130 ++++++++-----------------------------
 8 files changed, 263 insertions(+), 103 deletions(-)
 create mode 100644 include/net/netfilter/nft_nat.h
 create mode 100644 net/ipv4/netfilter/nft_nat_ipv4.c
 create mode 100644 net/ipv6/netfilter/nft_nat_ipv6.c
include/net/netfilter/nft_nat.h new file +22 -0
new file mode 100644
index 0000000..0ac1546
@@ -0,0 +1,22 @@
+#ifndef _NFT_NAT_H_
+#define _NFT_NAT_H_
+
+struct nft_nat {
+	enum nft_registers      sreg_addr_min:8;
+	enum nft_registers      sreg_addr_max:8;
+	enum nft_registers      sreg_proto_min:8;
+	enum nft_registers      sreg_proto_max:8;
+	enum nf_nat_manip_type  type:8;
+	u8			family;
+	u16			flags;
+};
+
+extern const struct nla_policy nft_nat_policy[];
+
+int nft_nat_init(const struct nft_ctx *ctx,
+		 const struct nft_expr *expr,
+		 const struct nlattr * const tb[]);
+
+int nft_nat_dump(struct sk_buff *skb, const struct nft_expr *expr);
+
+#endif /* _NFT_NAT_H_ */
net/ipv4/netfilter/Kconfig +7 -0
index f2d2202..6d7355c 100644
@@ -199,6 +199,13 @@ config NF_NAT_MASQUERADE_IPV4
 	This is the kernel functionality to provide NAT in the masquerade
 	flavour (automatic source address selection).
 
+config NFT_NAT_IPV4
+	tristate "nft_nat IPv4 support"
+	depends on NFT_NAT
+	select NF_NAT_MASQUERADE_IPV4
+	help
+	  This is the nftables expression that handles NAT in IPv4.
+
 config IP_NF_TARGET_MASQUERADE
 	tristate "MASQUERADE target support"
 	select NF_NAT_MASQUERADE_IPV4
net/ipv4/netfilter/Makefile +1 -0
index a7bfa0a..1c76c34 100644
@@ -32,6 +32,7 @@ obj-$(CONFIG_NF_TABLES_IPV4) += nf_tables_ipv4.o
 obj-$(CONFIG_NFT_CHAIN_ROUTE_IPV4) += nft_chain_route_ipv4.o
 obj-$(CONFIG_NFT_CHAIN_NAT_IPV4) += nft_chain_nat_ipv4.o
 obj-$(CONFIG_NFT_REJECT_IPV4) += nft_reject_ipv4.o
+obj-$(CONFIG_NFT_NAT_IPV4) += nft_nat_ipv4.o
 obj-$(CONFIG_NF_TABLES_ARP) += nf_tables_arp.o
 
 # generic IP tables 
net/ipv4/netfilter/nft_nat_ipv4.c new file +99 -0

99 lines not on this page · raw message

net/ipv6/netfilter/Kconfig +7 -0
index 86d24f9..9438bd4 100644
@@ -247,6 +247,13 @@ config NF_NAT_MASQUERADE_IPV6
 	 This is the kernel functionality to provide NAT in the masquerade
 	 flavour (automatic source address selection) for IPv6.
 
+config NFT_NAT_IPV6
+	tristate "nft_nat IPv6 support"
+	depends on NFT_NAT
+	select NF_NAT_MASQUERADE_IPV6
+	help
+	 This is the nftables expression that handles NAT in IPv6.
+
 config IP6_NF_TARGET_MASQUERADE
 	tristate "MASQUERADE target support"
 	select NF_NAT_MASQUERADE_IPV6
net/ipv6/netfilter/Makefile +1 -0
index 9d85d60..cd48175 100644
@@ -29,6 +29,7 @@ obj-$(CONFIG_NF_TABLES_IPV6) += nf_tables_ipv6.o
 obj-$(CONFIG_NFT_CHAIN_ROUTE_IPV6) += nft_chain_route_ipv6.o
 obj-$(CONFIG_NFT_CHAIN_NAT_IPV6) += nft_chain_nat_ipv6.o
 obj-$(CONFIG_NFT_REJECT_IPV6) += nft_reject_ipv6.o
+obj-$(CONFIG_NFT_NAT_IPV6) += nft_nat_ipv6.o
 
 # matches
 obj-$(CONFIG_IP6_NF_MATCH_AH) += ip6t_ah.o
net/ipv6/netfilter/nft_nat_ipv6.c new file +99 -0

99 lines not on this page · raw message

net/netfilter/nft_nat.c +27 -103

193 lines not on this page · raw message

#7 lore raw

PATCH v3 5/5 netfilter: nft_nat: add masquerade support

In reply to Arturo Borrero Gonzalez

This patch adds masquerade support to nft_nat.

Note that enum nf_nat_manip_type is replaced by enum nft_nat_types in order to support masquerade.

---
v2: fix missing htonl() conversion for NFTA_NAT_TYPE in _ipv6_dump().
v3: no changes.

 include/net/netfilter/nft_nat.h          |    2 +-
 include/uapi/linux/netfilter/nf_tables.h |    8 ++++--
 net/ipv4/netfilter/nft_nat_ipv4.c        |   40 ++++++++++++++++++++++++++++--
 net/ipv6/netfilter/nft_nat_ipv6.c        |   39 +++++++++++++++++++++++++++--
 net/netfilter/nft_nat.c                  |   20 ++++-----------
 5 files changed, 84 insertions(+), 25 deletions(-)
include/net/netfilter/nft_nat.h +1 -1
index 0ac1546..f5e108c 100644
@@ -6,7 +6,7 @@ struct nft_nat {
 	enum nft_registers      sreg_addr_max:8;
 	enum nft_registers      sreg_proto_min:8;
 	enum nft_registers      sreg_proto_max:8;
-	enum nf_nat_manip_type  type:8;
+	enum nft_nat_types	type:8;
 	u8			family;
 	u16			flags;
 };
include/uapi/linux/netfilter/nf_tables.h +5 -3
index 92c211b..ce75cb7 100644
@@ -754,14 +754,16 @@ enum nft_reject_attributes {
 #define NFTA_REJECT_MAX		(__NFTA_REJECT_MAX - 1)
 
 /**
- * enum nft_nat_types - nf_tables nat expression NAT types
+ * enum nft_nat_types - nf_tables nat expression NAT types.
  *
  * @NFT_NAT_SNAT: source NAT
  * @NFT_NAT_DNAT: destination NAT
+ * @NFT_NAT_MASQUERADE: masquerade NAT
  */
 enum nft_nat_types {
-	NFT_NAT_SNAT,
-	NFT_NAT_DNAT,
+	NFT_NAT_SNAT,		/* NF_NAT_MANIP_SRC */
+	NFT_NAT_DNAT,		/* NF_NAT_MANIP_DST */
+	NFT_NAT_MASQUERADE,
 };
 
 /**
net/ipv4/netfilter/nft_nat_ipv4.c +37 -3
index eea3cd9..d246eb3 100644
@@ -37,6 +37,7 @@ static void nft_nat_ipv4_eval(const struct nft_expr *expr,
 	enum ip_conntrack_info ctinfo;
 	struct nf_conn *ct = nf_ct_get(pkt->skb, &ctinfo);
 	struct nf_nat_range range;
+	unsigned int verdict;
 
 	memset(&range, 0, sizeof(range));
 	if (priv->sreg_addr_min) {
@@ -59,8 +60,40 @@ static void nft_nat_ipv4_eval(const struct nft_expr *expr,
 
 	range.flags |= priv->flags;
 
-	data[NFT_REG_VERDICT].verdict =
-				nf_nat_setup_info(ct, &range, priv->type);
+	if (priv->type == NFT_NAT_MASQUERADE) {
+		verdict = nf_nat_masquerade_ipv4(pkt->skb, pkt->ops->hooknum,
+						 &range, pkt->out);
+	} else {
+		verdict = nf_nat_setup_info(ct, &range, priv->type);
+	}
+
+	data[NFT_REG_VERDICT].verdict = verdict;
+}
+
+static int nft_nat_ipv4_init(const struct nft_ctx *ctx,
+			     const struct nft_expr *expr,
+			     const struct nlattr * const tb[])
+{
+	int ret;
+	struct nft_nat *priv = nft_expr_priv(expr);
+
+	ret = nft_nat_init(ctx, expr, tb);
+	if (ret < 0)
+		goto out;
+
+	if (priv->type == NFT_NAT_MASQUERADE)
+		nf_nat_masquerade_ipv4_register_notifier();
+out:
+	return ret;
+}
+
+static void nft_nat_ipv4_destroy(const struct nft_ctx *ctx,
+				 const struct nft_expr *expr)
+{
+	struct nft_nat *priv = nft_expr_priv(expr);
+
+	if (priv->type == NFT_NAT_MASQUERADE)
+		nf_nat_masquerade_ipv4_unregister_notifier();
 }
 
 static struct nft_expr_type nft_nat_ipv4_type;
@@ -68,7 +101,8 @@ static const struct nft_expr_ops nft_nat_ipv4_ops = {
 	.type           = &nft_nat_ipv4_type,
 	.size           = NFT_EXPR_SIZE(sizeof(struct nft_nat)),
 	.eval           = nft_nat_ipv4_eval,
-	.init           = nft_nat_init,
+	.init           = nft_nat_ipv4_init,
+	.destroy	= nft_nat_ipv4_destroy,
 	.dump           = nft_nat_dump,
 };
 
net/ipv6/netfilter/nft_nat_ipv6.c +36 -3
index bf1498a..0e4f277 100644
@@ -27,6 +27,7 @@
 #include <net/netfilter/nf_nat_l3proto.h>
 #include <net/ip.h>
 #include <net/netfilter/nft_nat.h>
+#include <net/netfilter/ipv6/nf_nat_masquerade_ipv6.h>
 
 static void nft_nat_ipv6_eval(const struct nft_expr *expr,
 			      struct nft_data data[NFT_REG_MAX + 1],
@@ -36,6 +37,7 @@ static void nft_nat_ipv6_eval(const struct nft_expr *expr,
 	enum ip_conntrack_info ctinfo;
 	struct nf_conn *ct = nf_ct_get(pkt->skb, &ctinfo);
 	struct nf_nat_range range;
+	unsigned int verdict;
 
 	memset(&range, 0, sizeof(range));
 	if (priv->sreg_addr_min) {
@@ -59,8 +61,38 @@ static void nft_nat_ipv6_eval(const struct nft_expr *expr,
 
 	range.flags |= priv->flags;
 
-	data[NFT_REG_VERDICT].verdict =
-				nf_nat_setup_info(ct, &range, priv->type);
+	if (priv->type == NFT_NAT_MASQUERADE)
+		verdict = nf_nat_masquerade_ipv6(pkt->skb, &range, pkt->out);
+	else
+		verdict = nf_nat_setup_info(ct, &range, priv->type);
+
+	data[NFT_REG_VERDICT].verdict = verdict;
+}
+
+static int nft_nat_ipv6_init(const struct nft_ctx *ctx,
+			     const struct nft_expr *expr,
+			     const struct nlattr * const tb[])
+{
+	int ret;
+	struct nft_nat *priv = nft_expr_priv(expr);
+
+	ret = nft_nat_init(ctx, expr, tb);
+	if (ret < 0)
+		goto out;
+
+	if (priv->type == NFT_NAT_MASQUERADE)
+		nf_nat_masquerade_ipv6_register_notifier();
+out:
+	return ret;
+}
+
+static void nft_nat_ipv6_destroy(const struct nft_ctx *ctx,
+				 const struct nft_expr *expr)
+{
+	struct nft_nat *priv = nft_expr_priv(expr);
+
+	if (priv->type == NFT_NAT_MASQUERADE)
+		nf_nat_masquerade_ipv6_unregister_notifier();
 }
 
 static struct nft_expr_type nft_nat_ipv6_type;
@@ -68,7 +100,8 @@ static const struct nft_expr_ops nft_nat_ipv6_ops = {
 	.type           = &nft_nat_ipv6_type,
 	.size           = NFT_EXPR_SIZE(sizeof(struct nft_nat)),
 	.eval           = nft_nat_ipv6_eval,
-	.init           = nft_nat_init,
+	.init           = nft_nat_ipv6_init,
+	.destroy	= nft_nat_ipv6_destroy,
 	.dump           = nft_nat_dump,
 };
 
net/netfilter/nft_nat.c +5 -15
index ec9c283..d603acb 100644
@@ -49,12 +49,11 @@ int nft_nat_init(const struct nft_ctx *ctx, const struct nft_expr *expr,
 	if (tb[NFTA_NAT_TYPE] == NULL)
 		return -EINVAL;
 
-	switch (ntohl(nla_get_be32(tb[NFTA_NAT_TYPE]))) {
+	priv->type = ntohl(nla_get_be32(tb[NFTA_NAT_TYPE]));
+	switch (priv->type) {
 	case NFT_NAT_SNAT:
-		priv->type = NF_NAT_MANIP_SRC;
-		break;
 	case NFT_NAT_DNAT:
-		priv->type = NF_NAT_MANIP_DST;
+	case NFT_NAT_MASQUERADE:
 		break;
 	default:
 		return -EINVAL;
@@ -122,17 +121,8 @@ int nft_nat_dump(struct sk_buff *skb, const struct nft_expr *expr)
 {
 	const struct nft_nat *priv = nft_expr_priv(expr);
 
-	switch (priv->type) {
-	case NF_NAT_MANIP_SRC:
-		if (nla_put_be32(skb, NFTA_NAT_TYPE, htonl(NFT_NAT_SNAT)))
-			goto nla_put_failure;
-		break;
-	case NF_NAT_MANIP_DST:
-		if (nla_put_be32(skb, NFTA_NAT_TYPE, htonl(NFT_NAT_DNAT)))
-			goto nla_put_failure;
-		break;
-	}
-
+	if (nla_put_be32(skb, NFTA_NAT_TYPE, htonl(priv->type)))
+		goto nla_put_failure;
 	if (nla_put_be32(skb, NFTA_NAT_FAMILY, htonl(priv->family)))
 		goto nla_put_failure;
 	if (nla_put_be32(skb,
#8 lore raw
In reply to Arturo Borrero Gonzalez

Hi Patrick,

Would you be OK if we push this patchset into mainstream? I think we can investigate the fetch interface address and store in register approach that you proposed to implement masquerading later on. The missing bits are the conntrack cleanup routine, I think that needs some "scratchpad" area to store the last address/interface that have been used. We can probably revisit this later once that generic state infrastructure for nf_tables (to support stateful expressions in some generic way) is in place?

If you don't like the idea, please let me know, and I'll defer this masquerading patchset.

Thanks!

Quoted 58 lines · Arturo Borrero Gonzalez

The following series implements some updates for NAT in nf_tables.

First of all, I add a new flag attribute to allow clients of nft_nat to specify additional config flags. This enables implementing port randomization and persistence to be set from nft.

Two patches split the masquerade code from ip[6]t_MASQUERADE.c to generic modules, so we can use this NAT type from nft_nat.

Then, the nft_nat code is splitted in AF specific parts, so we avoid potential dependencies regarding AF specific symbols in the last patch.

The last patch finally implements masquerade for nft_nat.

The v2 series included some fixes and additionals checks, as requested by Florian Westphal.

This v3 series includes changes requested by Pablo Neira.

Comments are welcomed.

---

Arturo Borrero Gonzalez (5):
netfilter: nft_nat: include a flag attribute
netfilter: nf_nat_masquerade_ipv4: code factorization
netfilter: nf_nat_masquerade_ipv6: code factorization
netfilter: nft_nat: split code in AF parts
netfilter: nft_nat: add masquerade support

 .../net/netfilter/ipv4/nf_nat_masquerade_ipv4.h    |   14 ++
 .../net/netfilter/ipv6/nf_nat_masquerade_ipv6.h    |   10 +
 include/net/netfilter/nft_nat.h                    |   22 +++
 include/uapi/linux/netfilter/nf_nat.h              |    5 +
 include/uapi/linux/netfilter/nf_tables.h           |   10 +
 net/ipv4/netfilter/Kconfig                         |   14 ++
 net/ipv4/netfilter/Makefile                        |    2 
 net/ipv4/netfilter/ipt_MASQUERADE.c                |  108 +-------------
 net/ipv4/netfilter/nf_nat_masquerade_ipv4.c        |  155 ++++++++++++++++++++
 net/ipv4/netfilter/nft_nat_ipv4.c                  |  133 +++++++++++++++++
 net/ipv6/netfilter/Kconfig                         |   14 ++
 net/ipv6/netfilter/Makefile                        |    2 
 net/ipv6/netfilter/ip6t_MASQUERADE.c               |   76 +---------
 net/ipv6/netfilter/nf_nat_masquerade_ipv6.c        |  121 ++++++++++++++++
 net/ipv6/netfilter/nft_nat_ipv6.c                  |  132 +++++++++++++++++
 net/netfilter/nft_nat.c                            |  156 ++++++--------------
 16 files changed, 688 insertions(+), 286 deletions(-)
 create mode 100644 include/net/netfilter/ipv4/nf_nat_masquerade_ipv4.h
 create mode 100644 include/net/netfilter/ipv6/nf_nat_masquerade_ipv6.h
 create mode 100644 include/net/netfilter/nft_nat.h
 create mode 100644 net/ipv4/netfilter/nf_nat_masquerade_ipv4.c
 create mode 100644 net/ipv4/netfilter/nft_nat_ipv4.c
 create mode 100644 net/ipv6/netfilter/nf_nat_masquerade_ipv6.c
 create mode 100644 net/ipv6/netfilter/nft_nat_ipv6.c
Signature · 1 line
Arturo Borrero Gonzalez
#9 lore raw
Quoted 10 lines · Pablo Neira Ayuso

Hi Patrick,

Would you be OK if we push this patchset into mainstream? I think we can investigate the fetch interface address and store in register approach that you proposed to implement masquerading later on. The missing bits are the conntrack cleanup routine, I think that needs some "scratchpad" area to store the last address/interface that have been used. We can probably revisit this later once that generic state infrastructure for nf_tables (to support stateful expressions in some generic way) is in place?

Not sure right now about the specifics, I'm out of order until my notebook has been repaired.

Quoted 4 lines

If you don't like the idea, please let me know, and I'll defer this masquerading patchset.

Thanks!

Sure, please go ahead.