[PATCH nf-next v2 00/11] netfilter: nf_tables_offload: support more expr and obj offload

[PATCH nf-next v2 00/11] netfilter: nf_tables_offload: support more expr and obj offload

[wenxu]

From: wenxu <wenxu@ucloud.cn>

This series patch support more expr and obj offload: 
fw_nedev, set payload, tunnel encap/decap action,
tunnel meta match, objref offload
Keep the action data in reg through immedidate offload

The follwing is the test sample:

# nft add table netdev firewall
# nft add tunnel netdev firewall encap tunid 1000 tundst 0xf198a8ac tunsrc 0x4b98a8ac tunrelease 0
# nft add tunnel netdev firewall decap tunid 0 tundst 0 tunsrc 0  tunrelease 1
# nft add chain netdev firewall aclout { type filter hook ingress device mlx_pf0vf0 priority - 300 \; }
# nft --debug=netlink add rule netdev firewall aclout ip daddr 10.0.1.7  @ll,0,48 set 0x00002e9ca06e2596 @ll,48,48 set 0xfaffffffffff tunnel name encap fwd to gretap
  [ meta load protocol => reg 1 ]
  [ cmp eq reg 1 0x00000008 ]
  [ payload load 4b @ network header + 16 => reg 1 ]
  [ cmp eq reg 1 0x0701000a ]
  [ immediate reg 1 0x6ea09c2e 0x00009625 ]
  [ payload write reg 1 => 6b @ link header + 0 csum_type 0 csum_off 0 csum_flags 0x0 ]
  [ immediate reg 1 0xfffffffa 0x0000ffff ]
  [ payload write reg 1 => 6b @ link header + 6 csum_type 0 csum_off 0 csum_flags 0x0 ]
  [ objref type 6 name encap ]
  [ immediate reg 1 0x00000019 ]
  [ fwd sreg_dev 1 ]


# nft add chain netdev firewall aclin { type filter hook ingress device gretap priority - 300 \; }
# nft --debug=netlink add rule netdev firewall aclin ip daddr 10.0.0.7 tunnel tunid 1000 tunnel tundst 172.168.152.75 tunnel tunsrc 172.168.152.241 tunnel name decap @ll,0,48 set 0x0000525400001275 @ll,48,48 set 0xfaffffffffff fwd to mlx_pf0vf0
  [ meta load protocol => reg 1 ]
  [ cmp eq reg 1 0x00000008 ]
  [ payload load 4b @ network header + 16 => reg 1 ]
  [ cmp eq reg 1 0x0700000a ]
  [ tunnel load id => reg 1 ]
  [ cmp eq reg 1 0x000003e8 ]
  [ tunnel load tun_dst => reg 1 ]
  [ cmp eq reg 1 0xaca8984b ]
  [ tunnel load tun_src => reg 1 ]
  [ cmp eq reg 1 0xaca898f1 ]
  [ objref type 6 name decap ]
  [ immediate reg 1 0x00005452 0x00007512 ]
  [ payload write reg 1 => 6b @ link header + 0 csum_type 0 csum_off 0 csum_flags 0x0 ]
  [ immediate reg 1 0xfffffffa 0x0000ffff ]
  [ payload write reg 1 => 6b @ link header + 6 csum_type 0 csum_off 0 csum_flags 0x0 ]
  [ immediate reg 1 0x0000000f ]
  [ fwd sreg_dev 1 ]

wenxu (11):
  netfilter: nf_flow_offload: add net in offload_ctx
  netfilter: nf_tables_offload: add offload_actions callback
  netfilter: nf_tables_offload: split nft_offload_reg to match and
    action type
  netfilter: nft_immediate: add offload support for actions
  netfilter: nft_fwd_netdev: add fw_netdev action support
  netfilter: nft_payload: add nft_set_payload offload support
  netfilter: nft_tunnel: support NFT_TUNNEL_SRC/DST_IP match
  netfilter: nft_tunnel: support tunnel meta match offload
  netfilter: nft_tunnel: add NFTA_TUNNEL_KEY_RELEASE action
  netfilter: nft_objref: add nft_objref_type offload
  netfilter: nft_tunnel: support nft_tunnel_obj offload

 include/net/netfilter/nf_tables.h         |  10 ++-
 include/net/netfilter/nf_tables_offload.h |  27 ++++++-
 include/uapi/linux/netfilter/nf_tables.h  |   3 +
 net/netfilter/nf_tables_api.c             |   2 +-
 net/netfilter/nf_tables_offload.c         |   7 +-
 net/netfilter/nft_cmp.c                   |  10 +--
 net/netfilter/nft_fwd_netdev.c            |  30 ++++++++
 net/netfilter/nft_immediate.c             |  47 +++++++-----
 net/netfilter/nft_meta.c                  |   6 +-
 net/netfilter/nft_objref.c                |  15 ++++
 net/netfilter/nft_payload.c               |  90 +++++++++++++++++++---
 net/netfilter/nft_tunnel.c                | 123 ++++++++++++++++++++++++++----
 12 files changed, 312 insertions(+), 58 deletions(-)

-- 
1.8.3.1

[PATCH nf-next v2 01/11] netfilter: nf_flow_offload: add net in offload_ctx

[wenxu]

From: wenxu <wenxu@ucloud.cn>

In the offload_ctx, the net can be used for other actions
such as fwd netdev

Signed-off-by: wenxu <wenxu@ucloud.cn>
---
 include/net/netfilter/nf_tables_offload.h | 3 ++-
 net/netfilter/nf_tables_api.c             | 2 +-
 net/netfilter/nf_tables_offload.c         | 3 ++-
 3 files changed, 5 insertions(+), 3 deletions(-)

diff --git a/include/net/netfilter/nf_tables_offload.h b/include/net/netfilter/nf_tables_offload.h
index 3196663..ad61958 100644
--- a/include/net/netfilter/nf_tables_offload.h
+++ b/include/net/netfilter/nf_tables_offload.h
@@ -24,6 +24,7 @@ struct nft_offload_ctx {
 		__be16				l3num;
 		u8				protonum;
 	} dep;
+	struct net *net;
 	unsigned int				num_actions;
 	struct nft_offload_reg			regs[NFT_REG32_15 + 1];
 };
@@ -60,7 +61,7 @@ struct nft_flow_rule {
 #define NFT_OFFLOAD_F_ACTION	(1 << 0)
 
 struct nft_rule;
-struct nft_flow_rule *nft_flow_rule_create(const struct nft_rule *rule);
+struct nft_flow_rule *nft_flow_rule_create(struct net *net, const struct nft_rule *rule);
 void nft_flow_rule_destroy(struct nft_flow_rule *flow);
 int nft_flow_rule_offload_commit(struct net *net);
 
diff --git a/net/netfilter/nf_tables_api.c b/net/netfilter/nf_tables_api.c
index 605a7cf..c6dc173 100644
--- a/net/netfilter/nf_tables_api.c
+++ b/net/netfilter/nf_tables_api.c
@@ -2844,7 +2844,7 @@ static int nf_tables_newrule(struct net *net, struct sock *nlsk,
 		return nft_table_validate(net, table);
 
 	if (chain->flags & NFT_CHAIN_HW_OFFLOAD) {
-		flow = nft_flow_rule_create(rule);
+		flow = nft_flow_rule_create(net, rule);
 		if (IS_ERR(flow))
 			return PTR_ERR(flow);
 
diff --git a/net/netfilter/nf_tables_offload.c b/net/netfilter/nf_tables_offload.c
index 64f5fd5..5c1fef7 100644
--- a/net/netfilter/nf_tables_offload.c
+++ b/net/netfilter/nf_tables_offload.c
@@ -28,12 +28,13 @@ static struct nft_flow_rule *nft_flow_rule_alloc(int num_actions)
 	return flow;
 }
 
-struct nft_flow_rule *nft_flow_rule_create(const struct nft_rule *rule)
+struct nft_flow_rule *nft_flow_rule_create(struct net *net, const struct nft_rule *rule)
 {
 	struct nft_offload_ctx ctx = {
 		.dep	= {
 			.type	= NFT_OFFLOAD_DEP_UNSPEC,
 		},
+		.net = net,
 	};
 	struct nft_flow_rule *flow;
 	int num_actions = 0, err;
-- 
1.8.3.1

[PATCH nf-next v2 02/11] netfilter: nf_tables_offload: add offload_actions callback

[wenxu]

From: wenxu <wenxu@ucloud.cn>

There will be zero one or serval actions for some expr. such as
payload set and immediate. The one payload set action maybe split
to several offload entries. And only NFT_REG_VERDICT immediate
action has 1 offload entry 

Signed-off-by: wenxu <wenxu@ucloud.cn>
---
 include/net/netfilter/nf_tables.h         | 7 ++++++-
 include/net/netfilter/nf_tables_offload.h | 2 --
 net/netfilter/nf_tables_offload.c         | 4 ++--
 net/netfilter/nft_immediate.c             | 2 +-
 4 files changed, 9 insertions(+), 6 deletions(-)

diff --git a/include/net/netfilter/nf_tables.h b/include/net/netfilter/nf_tables.h
index 9b62456..9285df2 100644
--- a/include/net/netfilter/nf_tables.h
+++ b/include/net/netfilter/nf_tables.h
@@ -785,7 +785,7 @@ struct nft_expr_ops {
 	int				(*offload)(struct nft_offload_ctx *ctx,
 						   struct nft_flow_rule *flow,
 						   const struct nft_expr *expr);
-	u32				offload_flags;
+	int				(*offload_actions)(const struct nft_expr *expr);
 	const struct nft_expr_type	*type;
 	void				*data;
 };
@@ -794,6 +794,11 @@ struct nft_expr_ops {
 #define NFT_EXPR_SIZE(size)		(sizeof(struct nft_expr) + \
 					 ALIGN(size, __alignof__(struct nft_expr)))
 
+static inline int nft_offload_action(const struct nft_expr *expr)
+{
+	return 1;
+}
+
 /**
  *	struct nft_expr - nf_tables expression
  *
diff --git a/include/net/netfilter/nf_tables_offload.h b/include/net/netfilter/nf_tables_offload.h
index ad61958..275d014 100644
--- a/include/net/netfilter/nf_tables_offload.h
+++ b/include/net/netfilter/nf_tables_offload.h
@@ -58,8 +58,6 @@ struct nft_flow_rule {
 	struct flow_rule	*rule;
 };
 
-#define NFT_OFFLOAD_F_ACTION	(1 << 0)
-
 struct nft_rule;
 struct nft_flow_rule *nft_flow_rule_create(struct net *net, const struct nft_rule *rule);
 void nft_flow_rule_destroy(struct nft_flow_rule *flow);
diff --git a/net/netfilter/nf_tables_offload.c b/net/netfilter/nf_tables_offload.c
index 5c1fef7..33543f5 100644
--- a/net/netfilter/nf_tables_offload.c
+++ b/net/netfilter/nf_tables_offload.c
@@ -42,8 +42,8 @@ struct nft_flow_rule *nft_flow_rule_create(struct net *net, const struct nft_rul
 
 	expr = nft_expr_first(rule);
 	while (expr->ops && expr != nft_expr_last(rule)) {
-		if (expr->ops->offload_flags & NFT_OFFLOAD_F_ACTION)
-			num_actions++;
+		if (expr->ops->offload_actions)
+			num_actions += expr->ops->offload_actions(expr);
 
 		expr = nft_expr_next(expr);
 	}
diff --git a/net/netfilter/nft_immediate.c b/net/netfilter/nft_immediate.c
index ca2ae4b..391f699 100644
--- a/net/netfilter/nft_immediate.c
+++ b/net/netfilter/nft_immediate.c
@@ -163,7 +163,7 @@ static int nft_immediate_offload(struct nft_offload_ctx *ctx,
 	.dump		= nft_immediate_dump,
 	.validate	= nft_immediate_validate,
 	.offload	= nft_immediate_offload,
-	.offload_flags	= NFT_OFFLOAD_F_ACTION,
+	.offload_actions = nft_offload_action,
 };
 
 struct nft_expr_type nft_imm_type __read_mostly = {
-- 
1.8.3.1

[PATCH nf-next v2 03/11] netfilter: nf_tables_offload: split nft_offload_reg to match and action type

[wenxu]

From: wenxu <wenxu@ucloud.cn>

Currently the nft_offload_reg is only can be used for match condition.
Can not be used for action. Add nft_offload_reg_type to make nft_offload_reg
can be used for action also.

Signed-off-by: wenxu <wenxu@ucloud.cn>
---
 include/net/netfilter/nf_tables_offload.h | 20 +++++++++++++++++-
 net/netfilter/nft_cmp.c                   | 10 ++++-----
 net/netfilter/nft_meta.c                  |  6 ++++--
 net/netfilter/nft_payload.c               | 34 ++++++++++++++++++++-----------
 4 files changed, 50 insertions(+), 20 deletions(-)

diff --git a/include/net/netfilter/nf_tables_offload.h b/include/net/netfilter/nf_tables_offload.h
index 275d014..82e3936 100644
--- a/include/net/netfilter/nf_tables_offload.h
+++ b/include/net/netfilter/nf_tables_offload.h
@@ -4,7 +4,13 @@
 #include <net/flow_offload.h>
 #include <net/netfilter/nf_tables.h>
 
-struct nft_offload_reg {
+enum nft_offload_reg_type {
+	NFT_OFFLOAD_REG_UNSPEC	= 0,
+	NFT_OFFLOAD_REG_MATCH,
+	NFT_OFFLOAD_REG_ACTION,
+};
+
+struct nft_offload_match {
 	u32		key;
 	u32		len;
 	u32		base_offset;
@@ -12,6 +18,18 @@ struct nft_offload_reg {
 	struct nft_data	mask;
 };
 
+struct nft_offload_action {
+	struct nft_data	data;
+};
+
+struct nft_offload_reg {
+	enum nft_offload_reg_type type;
+	union {
+		struct nft_offload_match match;
+		struct nft_offload_action action;
+	};
+};
+
 enum nft_offload_dep_type {
 	NFT_OFFLOAD_DEP_UNSPEC	= 0,
 	NFT_OFFLOAD_DEP_NETWORK,
diff --git a/net/netfilter/nft_cmp.c b/net/netfilter/nft_cmp.c
index bd173b1..ee38cba 100644
--- a/net/netfilter/nft_cmp.c
+++ b/net/netfilter/nft_cmp.c
@@ -116,14 +116,14 @@ static int __nft_cmp_offload(struct nft_offload_ctx *ctx,
 	u8 *mask = (u8 *)&flow->match.mask;
 	u8 *key = (u8 *)&flow->match.key;
 
-	if (priv->op != NFT_CMP_EQ)
+	if (priv->op != NFT_CMP_EQ || reg->type != NFT_OFFLOAD_REG_MATCH)
 		return -EOPNOTSUPP;
 
-	memcpy(key + reg->offset, &priv->data, priv->len);
-	memcpy(mask + reg->offset, &reg->mask, priv->len);
+	memcpy(key + reg->match.offset, &priv->data, priv->len);
+	memcpy(mask + reg->match.offset, &reg->match.mask, priv->len);
 
-	flow->match.dissector.used_keys |= BIT(reg->key);
-	flow->match.dissector.offset[reg->key] = reg->base_offset;
+	flow->match.dissector.used_keys |= BIT(reg->match.key);
+	flow->match.dissector.offset[reg->match.key] = reg->match.base_offset;
 
 	nft_offload_update_dependency(ctx, &priv->data, priv->len);
 
diff --git a/net/netfilter/nft_meta.c b/net/netfilter/nft_meta.c
index f1b1d94..6bb5ba6 100644
--- a/net/netfilter/nft_meta.c
+++ b/net/netfilter/nft_meta.c
@@ -498,15 +498,17 @@ static int nft_meta_get_offload(struct nft_offload_ctx *ctx,
 	const struct nft_meta *priv = nft_expr_priv(expr);
 	struct nft_offload_reg *reg = &ctx->regs[priv->dreg];
 
+	reg->type = NFT_OFFLOAD_REG_MATCH;
+
 	switch (priv->key) {
 	case NFT_META_PROTOCOL:
 		NFT_OFFLOAD_MATCH(FLOW_DISSECTOR_KEY_BASIC, basic, n_proto,
-				  sizeof(__u16), reg);
+				  sizeof(__u16), &reg->match);
 		nft_offload_set_dependency(ctx, NFT_OFFLOAD_DEP_NETWORK);
 		break;
 	case NFT_META_L4PROTO:
 		NFT_OFFLOAD_MATCH(FLOW_DISSECTOR_KEY_BASIC, basic, ip_proto,
-				  sizeof(__u8), reg);
+				  sizeof(__u8), &reg->match);
 		nft_offload_set_dependency(ctx, NFT_OFFLOAD_DEP_TRANSPORT);
 		break;
 	default:
diff --git a/net/netfilter/nft_payload.c b/net/netfilter/nft_payload.c
index 22a80eb..36efa1c 100644
--- a/net/netfilter/nft_payload.c
+++ b/net/netfilter/nft_payload.c
@@ -159,14 +159,16 @@ static int nft_payload_offload_ll(struct nft_offload_ctx *ctx,
 {
 	struct nft_offload_reg *reg = &ctx->regs[priv->dreg];
 
+	reg->type = NFT_OFFLOAD_REG_MATCH;
+
 	switch (priv->offset) {
 	case offsetof(struct ethhdr, h_source):
 		NFT_OFFLOAD_MATCH(FLOW_DISSECTOR_KEY_ETH_ADDRS, eth_addrs,
-				  src, ETH_ALEN, reg);
+				  src, ETH_ALEN, &reg->match);
 		break;
 	case offsetof(struct ethhdr, h_dest):
 		NFT_OFFLOAD_MATCH(FLOW_DISSECTOR_KEY_ETH_ADDRS, eth_addrs,
-				  dst, ETH_ALEN, reg);
+				  dst, ETH_ALEN, &reg->match);
 		break;
 	}
 
@@ -179,18 +181,20 @@ static int nft_payload_offload_ip(struct nft_offload_ctx *ctx,
 {
 	struct nft_offload_reg *reg = &ctx->regs[priv->dreg];
 
+	reg->type = NFT_OFFLOAD_REG_MATCH;
+
 	switch (priv->offset) {
 	case offsetof(struct iphdr, saddr):
 		NFT_OFFLOAD_MATCH(FLOW_DISSECTOR_KEY_IPV4_ADDRS, ipv4, src,
-				  sizeof(struct in_addr), reg);
+				  sizeof(struct in_addr), &reg->match);
 		break;
 	case offsetof(struct iphdr, daddr):
 		NFT_OFFLOAD_MATCH(FLOW_DISSECTOR_KEY_IPV4_ADDRS, ipv4, dst,
-				  sizeof(struct in_addr), reg);
+				  sizeof(struct in_addr), &reg->match);
 		break;
 	case offsetof(struct iphdr, protocol):
 		NFT_OFFLOAD_MATCH(FLOW_DISSECTOR_KEY_BASIC, basic, ip_proto,
-				  sizeof(__u8), reg);
+				  sizeof(__u8), &reg->match);
 		nft_offload_set_dependency(ctx, NFT_OFFLOAD_DEP_TRANSPORT);
 		break;
 	default:
@@ -206,18 +210,20 @@ static int nft_payload_offload_ip6(struct nft_offload_ctx *ctx,
 {
 	struct nft_offload_reg *reg = &ctx->regs[priv->dreg];
 
+	reg->type = NFT_OFFLOAD_REG_MATCH;
+
 	switch (priv->offset) {
 	case offsetof(struct ipv6hdr, saddr):
 		NFT_OFFLOAD_MATCH(FLOW_DISSECTOR_KEY_IPV6_ADDRS, ipv6, src,
-				  sizeof(struct in6_addr), reg);
+				  sizeof(struct in6_addr), &reg->match);
 		break;
 	case offsetof(struct ipv6hdr, daddr):
 		NFT_OFFLOAD_MATCH(FLOW_DISSECTOR_KEY_IPV6_ADDRS, ipv6, dst,
-				  sizeof(struct in6_addr), reg);
+				  sizeof(struct in6_addr), &reg->match);
 		break;
 	case offsetof(struct ipv6hdr, nexthdr):
 		NFT_OFFLOAD_MATCH(FLOW_DISSECTOR_KEY_BASIC, basic, ip_proto,
-				  sizeof(__u8), reg);
+				  sizeof(__u8), &reg->match);
 		nft_offload_set_dependency(ctx, NFT_OFFLOAD_DEP_TRANSPORT);
 		break;
 	default:
@@ -253,14 +259,16 @@ static int nft_payload_offload_tcp(struct nft_offload_ctx *ctx,
 {
 	struct nft_offload_reg *reg = &ctx->regs[priv->dreg];
 
+	reg->type = NFT_OFFLOAD_REG_MATCH;
+
 	switch (priv->offset) {
 	case offsetof(struct tcphdr, source):
 		NFT_OFFLOAD_MATCH(FLOW_DISSECTOR_KEY_PORTS, tp, src,
-				  sizeof(__be16), reg);
+				  sizeof(__be16), &reg->match);
 		break;
 	case offsetof(struct tcphdr, dest):
 		NFT_OFFLOAD_MATCH(FLOW_DISSECTOR_KEY_PORTS, tp, dst,
-				  sizeof(__be16), reg);
+				  sizeof(__be16), &reg->match);
 		break;
 	default:
 		return -EOPNOTSUPP;
@@ -275,14 +283,16 @@ static int nft_payload_offload_udp(struct nft_offload_ctx *ctx,
 {
 	struct nft_offload_reg *reg = &ctx->regs[priv->dreg];
 
+	reg->type = NFT_OFFLOAD_REG_MATCH;
+
 	switch (priv->offset) {
 	case offsetof(struct udphdr, source):
 		NFT_OFFLOAD_MATCH(FLOW_DISSECTOR_KEY_PORTS, tp, src,
-				  sizeof(__be16), reg);
+				  sizeof(__be16), &reg->match);
 		break;
 	case offsetof(struct udphdr, dest):
 		NFT_OFFLOAD_MATCH(FLOW_DISSECTOR_KEY_PORTS, tp, dst,
-				  sizeof(__be16), reg);
+				  sizeof(__be16), &reg->match);
 		break;
 	default:
 		return -EOPNOTSUPP;
-- 
1.8.3.1

[PATCH nf-next v2 04/11] netfilter: nft_immediate: add offload support for actions

[wenxu]

From: wenxu <wenxu@ucloud.cn>

Immediate offload support for other action to handle the offload_reg

Signed-off-by: wenxu <wenxu@ucloud.cn>
---
 net/netfilter/nft_immediate.c | 47 +++++++++++++++++++++++++++----------------
 1 file changed, 30 insertions(+), 17 deletions(-)

diff --git a/net/netfilter/nft_immediate.c b/net/netfilter/nft_immediate.c
index 391f699..34facc3 100644
--- a/net/netfilter/nft_immediate.c
+++ b/net/netfilter/nft_immediate.c
@@ -130,29 +130,42 @@ static int nft_immediate_offload(struct nft_offload_ctx *ctx,
 				 const struct nft_expr *expr)
 {
 	const struct nft_immediate_expr *priv = nft_expr_priv(expr);
+	const struct nft_data *data = &priv->data;
 	struct flow_action_entry *entry;
-	const struct nft_data *data;
-
-	if (priv->dreg != NFT_REG_VERDICT)
-		return -EOPNOTSUPP;
-
-	entry = &flow->rule->action.entries[ctx->num_actions++];
 
-	data = &priv->data;
-	switch (data->verdict.code) {
-	case NF_ACCEPT:
-		entry->id = FLOW_ACTION_ACCEPT;
-		break;
-	case NF_DROP:
-		entry->id = FLOW_ACTION_DROP;
-		break;
-	default:
-		return -EOPNOTSUPP;
+	if (priv->dreg == NFT_REG_VERDICT) {
+		entry = &flow->rule->action.entries[ctx->num_actions++];
+
+		switch (data->verdict.code) {
+		case NF_ACCEPT:
+			entry->id = FLOW_ACTION_ACCEPT;
+			break;
+		case NF_DROP:
+			entry->id = FLOW_ACTION_DROP;
+			break;
+		default:
+			return -EOPNOTSUPP;
+		}
+	} else {
+		struct nft_offload_reg *reg = &ctx->regs[priv->dreg];
+
+		reg->type = NFT_OFFLOAD_REG_ACTION;
+		memcpy(&reg->action.data, data, sizeof(*data));
 	}
 
 	return 0;
 }
 
+static int nft_immediate_offload_actions(const struct nft_expr *expr)
+{
+	const struct nft_immediate_expr *priv = nft_expr_priv(expr);
+
+	if (priv->dreg == NFT_REG_VERDICT)
+		return 1;
+	else
+		return 0;
+}
+
 static const struct nft_expr_ops nft_imm_ops = {
 	.type		= &nft_imm_type,
 	.size		= NFT_EXPR_SIZE(sizeof(struct nft_immediate_expr)),
@@ -163,7 +176,7 @@ static int nft_immediate_offload(struct nft_offload_ctx *ctx,
 	.dump		= nft_immediate_dump,
 	.validate	= nft_immediate_validate,
 	.offload	= nft_immediate_offload,
-	.offload_actions = nft_offload_action,
+	.offload_actions = nft_immediate_offload_actions,
 };
 
 struct nft_expr_type nft_imm_type __read_mostly = {
-- 
1.8.3.1

[PATCH nf-next v2 05/11] netfilter: nft_fwd_netdev: add fw_netdev action support

[wenxu]

From: wenxu <wenxu@ucloud.cn>

fwd_netdev action offload:
nft --debug=netlink add rule netdev firewall aclout ip daddr 10.0.1.7 fwd to eth0

Signed-off-by: wenxu <wenxu@ucloud.cn>
---
 net/netfilter/nft_fwd_netdev.c | 30 ++++++++++++++++++++++++++++++
 1 file changed, 30 insertions(+)

diff --git a/net/netfilter/nft_fwd_netdev.c b/net/netfilter/nft_fwd_netdev.c
index 61b7f93..06dbd98 100644
--- a/net/netfilter/nft_fwd_netdev.c
+++ b/net/netfilter/nft_fwd_netdev.c
@@ -15,6 +15,7 @@
 #include <net/netfilter/nf_dup_netdev.h>
 #include <net/neighbour.h>
 #include <net/ip.h>
+#include <net/netfilter/nf_tables_offload.h>
 
 struct nft_fwd_netdev {
 	enum nft_registers	sreg_dev:8;
@@ -63,6 +64,33 @@ static int nft_fwd_netdev_dump(struct sk_buff *skb, const struct nft_expr *expr)
 	return -1;
 }
 
+static int nft_fwd_netdev_offload(struct nft_offload_ctx *ctx,
+				  struct nft_flow_rule *flow,
+				  const struct nft_expr *expr)
+{
+	const struct nft_fwd_netdev *priv = nft_expr_priv(expr);
+	struct nft_offload_reg *reg = &ctx->regs[priv->sreg_dev];
+	const struct nft_data *data = &reg->action.data;
+	struct flow_action_entry *entry;
+	struct net_device *dev;
+	int oif = -1;
+
+	if (reg->type != NFT_OFFLOAD_REG_ACTION)
+		return -EOPNOTSUPP;
+
+	entry = &flow->rule->action.entries[ctx->num_actions++];
+
+	memcpy(&oif, data->data, sizeof(oif));
+	dev = __dev_get_by_index(ctx->net, oif);
+	if (!dev)
+		return -EOPNOTSUPP;
+
+	entry->id = FLOW_ACTION_REDIRECT;
+	entry->dev = dev;
+
+	return 0;
+}
+
 struct nft_fwd_neigh {
 	enum nft_registers	sreg_dev:8;
 	enum nft_registers	sreg_addr:8;
@@ -194,6 +222,8 @@ static int nft_fwd_neigh_dump(struct sk_buff *skb, const struct nft_expr *expr)
 	.eval		= nft_fwd_netdev_eval,
 	.init		= nft_fwd_netdev_init,
 	.dump		= nft_fwd_netdev_dump,
+	.offload	= nft_fwd_netdev_offload,
+	.offload_actions = nft_offload_action,
 };
 
 static const struct nft_expr_ops *
-- 
1.8.3.1

[PATCH nf-next v2 06/11] netfilter: nft_payload: add nft_set_payload offload support

[wenxu]

From: wenxu <wenxu@ucloud.cn>

currently payload set only support ll header

nft --debug=netlink add rule netdev firewall aclout ip daddr 10.0.1.7 @ll,0,48
set 0x00002e9ca06e2596 @ll,48,48 set 0xfaffffffffff fwd to eth0

Signed-off-by: wenxu <wenxu@ucloud.cn>
---
 net/netfilter/nft_payload.c | 56 +++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 56 insertions(+)

diff --git a/net/netfilter/nft_payload.c b/net/netfilter/nft_payload.c
index 36efa1c..544fc40 100644
--- a/net/netfilter/nft_payload.c
+++ b/net/netfilter/nft_payload.c
@@ -572,12 +572,68 @@ static int nft_payload_set_dump(struct sk_buff *skb, const struct nft_expr *expr
 	return -1;
 }
 
+static int nft_payload_set_offload(struct nft_offload_ctx *ctx,
+				   struct nft_flow_rule *flow,
+				   const struct nft_expr *expr)
+{
+	const struct nft_payload_set *priv = nft_expr_priv(expr);
+	struct nft_offload_reg *reg = &ctx->regs[priv->sreg];
+	const struct nft_data *data = &reg->action.data;
+	struct flow_action_entry *entry;
+	u32 len = priv->len;
+	u32 offset, last;
+	int n_actions, i;
+
+	if (priv->base != NFT_PAYLOAD_LL_HEADER || len > 16)
+		return -EOPNOTSUPP;
+
+	offset = priv->offset;
+	n_actions = len >> 2;
+	last = len & 0x3;
+
+	for (i = 0; i < n_actions; i++) {
+		entry = &flow->rule->action.entries[ctx->num_actions++];
+
+		entry->id = FLOW_ACTION_MANGLE;
+		entry->mangle.htype = FLOW_ACT_MANGLE_HDR_TYPE_ETH;
+		entry->mangle.mask = 0;
+		entry->mangle.val = data->data[i];
+		entry->mangle.offset = offset;
+		offset = offset + 4;
+	}
+
+	if (last) {
+		entry = &flow->rule->action.entries[ctx->num_actions++];
+
+		entry->id = FLOW_ACTION_MANGLE;
+		entry->mangle.htype = FLOW_ACT_MANGLE_HDR_TYPE_ETH;
+		entry->mangle.mask = ~((1 << (last * 8)) - 1);
+		entry->mangle.val = data->data[i];
+		entry->mangle.offset = offset;
+	}
+
+	return 0;
+}
+
+static int nft_payload_set_offload_actions(const struct nft_expr *expr)
+{
+	const struct nft_payload_set *priv = nft_expr_priv(expr);
+	u32 len = priv->len;
+
+	if (priv->base != NFT_PAYLOAD_LL_HEADER || len > 16)
+		return 0;
+
+	return (len >> 2) + !!(len & 3);
+}
+
 static const struct nft_expr_ops nft_payload_set_ops = {
 	.type		= &nft_payload_type,
 	.size		= NFT_EXPR_SIZE(sizeof(struct nft_payload_set)),
 	.eval		= nft_payload_set_eval,
 	.init		= nft_payload_set_init,
 	.dump		= nft_payload_set_dump,
+	.offload	= nft_payload_set_offload,
+	.offload_actions = nft_payload_set_offload_actions,
 };
 
 static const struct nft_expr_ops *
-- 
1.8.3.1

[PATCH nf-next v2 07/11] netfilter: nft_tunnel: support NFT_TUNNEL_SRC/DST_IP match

[wenxu]

From: wenxu <wenxu@ucloud.cn>

Add new two NFT_TUNNEL_SRC/DST_IP match in nft_tunnel

Signed-off-by: wenxu <wenxu@ucloud.cn>
---
 include/uapi/linux/netfilter/nf_tables.h |  2 ++
 net/netfilter/nft_tunnel.c               | 46 +++++++++++++++++++++++++-------
 2 files changed, 38 insertions(+), 10 deletions(-)

diff --git a/include/uapi/linux/netfilter/nf_tables.h b/include/uapi/linux/netfilter/nf_tables.h
index 82abaa1..173690a 100644
--- a/include/uapi/linux/netfilter/nf_tables.h
+++ b/include/uapi/linux/netfilter/nf_tables.h
@@ -1765,6 +1765,8 @@ enum nft_tunnel_key_attributes {
 enum nft_tunnel_keys {
 	NFT_TUNNEL_PATH,
 	NFT_TUNNEL_ID,
+	NFT_TUNNEL_SRC_IP,
+	NFT_TUNNEL_DST_IP,
 	__NFT_TUNNEL_MAX
 };
 #define NFT_TUNNEL_MAX	(__NFT_TUNNEL_MAX - 1)
diff --git a/net/netfilter/nft_tunnel.c b/net/netfilter/nft_tunnel.c
index 3d4c2ae..e218163 100644
--- a/net/netfilter/nft_tunnel.c
+++ b/net/netfilter/nft_tunnel.c
@@ -18,6 +18,18 @@ struct nft_tunnel {
 	enum nft_tunnel_mode	mode:8;
 };
 
+bool nft_tunnel_mode_validate(enum nft_tunnel_mode priv_mode, u8 tun_mode)
+{
+	if (priv_mode == NFT_TUNNEL_MODE_NONE ||
+	    (priv_mode == NFT_TUNNEL_MODE_RX &&
+	     !(tun_mode & IP_TUNNEL_INFO_TX)) ||
+	    (priv_mode == NFT_TUNNEL_MODE_TX &&
+	     (tun_mode & IP_TUNNEL_INFO_TX)))
+		return true;
+
+	return false;
+}
+
 static void nft_tunnel_get_eval(const struct nft_expr *expr,
 				struct nft_regs *regs,
 				const struct nft_pktinfo *pkt)
@@ -34,11 +46,7 @@ static void nft_tunnel_get_eval(const struct nft_expr *expr,
 			nft_reg_store8(dest, false);
 			return;
 		}
-		if (priv->mode == NFT_TUNNEL_MODE_NONE ||
-		    (priv->mode == NFT_TUNNEL_MODE_RX &&
-		     !(tun_info->mode & IP_TUNNEL_INFO_TX)) ||
-		    (priv->mode == NFT_TUNNEL_MODE_TX &&
-		     (tun_info->mode & IP_TUNNEL_INFO_TX)))
+		if (nft_tunnel_mode_validate(priv->mode, tun_info->mode))
 			nft_reg_store8(dest, true);
 		else
 			nft_reg_store8(dest, false);
@@ -48,15 +56,31 @@ static void nft_tunnel_get_eval(const struct nft_expr *expr,
 			regs->verdict.code = NFT_BREAK;
 			return;
 		}
-		if (priv->mode == NFT_TUNNEL_MODE_NONE ||
-		    (priv->mode == NFT_TUNNEL_MODE_RX &&
-		     !(tun_info->mode & IP_TUNNEL_INFO_TX)) ||
-		    (priv->mode == NFT_TUNNEL_MODE_TX &&
-		     (tun_info->mode & IP_TUNNEL_INFO_TX)))
+		if (nft_tunnel_mode_validate(priv->mode, tun_info->mode))
 			*dest = ntohl(tunnel_id_to_key32(tun_info->key.tun_id));
 		else
 			regs->verdict.code = NFT_BREAK;
 		break;
+	case NFT_TUNNEL_SRC_IP:
+		if (!tun_info) {
+			regs->verdict.code = NFT_BREAK;
+			return;
+		}
+		if (nft_tunnel_mode_validate(priv->mode, tun_info->mode))
+			*dest = ntohl(tun_info->key.u.ipv4.src);
+		else
+			regs->verdict.code = NFT_BREAK;
+		break;
+	case NFT_TUNNEL_DST_IP:
+		if (!tun_info) {
+			regs->verdict.code = NFT_BREAK;
+			return;
+		}
+		if (nft_tunnel_mode_validate(priv->mode, tun_info->mode))
+			*dest = ntohl(tun_info->key.u.ipv4.dst);
+		else
+			regs->verdict.code = NFT_BREAK;
+		break;
 	default:
 		WARN_ON(1);
 		regs->verdict.code = NFT_BREAK;
@@ -86,6 +110,8 @@ static int nft_tunnel_get_init(const struct nft_ctx *ctx,
 		len = sizeof(u8);
 		break;
 	case NFT_TUNNEL_ID:
+	case NFT_TUNNEL_SRC_IP:
+	case NFT_TUNNEL_DST_IP:
 		len = sizeof(u32);
 		break;
 	default:
-- 
1.8.3.1

[PATCH nf-next v2 08/11] netfilter: nft_tunnel: support tunnel meta match offload

[wenxu]

From: wenxu <wenxu@ucloud.cn>

Add tunnel meta match offload. Currently support for NFT_TUNNEL_ID
NFT_TUNNEL_SRC_IP and NFT_TUNNEL_DST_IP

Signed-off-by: wenxu <wenxu@ucloud.cn>
---
 include/net/netfilter/nf_tables_offload.h |  2 ++
 net/netfilter/nft_tunnel.c                | 33 +++++++++++++++++++++++++++++++
 2 files changed, 35 insertions(+)

diff --git a/include/net/netfilter/nf_tables_offload.h b/include/net/netfilter/nf_tables_offload.h
index 82e3936..4405b16 100644
--- a/include/net/netfilter/nf_tables_offload.h
+++ b/include/net/netfilter/nf_tables_offload.h
@@ -62,6 +62,8 @@ struct nft_flow_key {
 	struct flow_dissector_key_ip			ip;
 	struct flow_dissector_key_vlan			vlan;
 	struct flow_dissector_key_eth_addrs		eth_addrs;
+	struct flow_dissector_key_keyid         enc_key_id;
+	struct flow_dissector_key_ipv4_addrs	enc_ipv4;
 } __aligned(BITS_PER_LONG / 8); /* Ensure that we can do comparisons as longs. */
 
 struct nft_flow_match {
diff --git a/net/netfilter/nft_tunnel.c b/net/netfilter/nft_tunnel.c
index e218163..900c94f 100644
--- a/net/netfilter/nft_tunnel.c
+++ b/net/netfilter/nft_tunnel.c
@@ -11,6 +11,7 @@
 #include <net/ip_tunnels.h>
 #include <net/vxlan.h>
 #include <net/erspan.h>
+#include <net/netfilter/nf_tables_offload.h>
 
 struct nft_tunnel {
 	enum nft_tunnel_keys	key:8;
@@ -149,6 +150,37 @@ static int nft_tunnel_get_dump(struct sk_buff *skb,
 	return -1;
 }
 
+static int nft_tunnel_get_offload(struct nft_offload_ctx *ctx,
+				  struct nft_flow_rule *flow,
+				  const struct nft_expr *expr)
+{
+	const struct nft_tunnel *priv = nft_expr_priv(expr);
+	struct nft_offload_reg *reg = &ctx->regs[priv->dreg];
+
+	if (priv->mode == NFT_TUNNEL_MODE_TX)
+		return -EOPNOTSUPP;
+
+	switch (priv->key) {
+	case NFT_TUNNEL_ID:
+		NFT_OFFLOAD_MATCH(FLOW_DISSECTOR_KEY_ENC_KEYID, enc_key_id, keyid,
+				  sizeof(__u32), &reg->match);
+		break;
+	case NFT_TUNNEL_SRC_IP:
+		NFT_OFFLOAD_MATCH(FLOW_DISSECTOR_KEY_ENC_IPV4_ADDRS, enc_ipv4, src,
+				  sizeof(__u32), &reg->match);
+		break;
+	case NFT_TUNNEL_DST_IP:
+		NFT_OFFLOAD_MATCH(FLOW_DISSECTOR_KEY_ENC_IPV4_ADDRS, enc_ipv4, dst,
+				  sizeof(__u32), &reg->match);
+		break;
+	case NFT_TUNNEL_PATH:
+	default:
+		return -EOPNOTSUPP;
+	}
+
+	return 0;
+}
+
 static struct nft_expr_type nft_tunnel_type;
 static const struct nft_expr_ops nft_tunnel_get_ops = {
 	.type		= &nft_tunnel_type,
@@ -156,6 +188,7 @@ static int nft_tunnel_get_dump(struct sk_buff *skb,
 	.eval		= nft_tunnel_get_eval,
 	.init		= nft_tunnel_get_init,
 	.dump		= nft_tunnel_get_dump,
+	.offload	= nft_tunnel_get_offload,
 };
 
 static struct nft_expr_type nft_tunnel_type __read_mostly = {
-- 
1.8.3.1

[PATCH nf-next v2 09/11] netfilter: nft_tunnel: add NFTA_TUNNEL_KEY_RELEASE action

[wenxu]

From: wenxu <wenxu@ucloud.cn>

Add new NFTA_TUNNEL_KEY_RELEASE action for future offload
feature

Signed-off-by: wenxu <wenxu@ucloud.cn>
---
 include/uapi/linux/netfilter/nf_tables.h |  1 +
 net/netfilter/nft_tunnel.c               | 24 +++++++++++++++++++++---
 2 files changed, 22 insertions(+), 3 deletions(-)

diff --git a/include/uapi/linux/netfilter/nf_tables.h b/include/uapi/linux/netfilter/nf_tables.h
index 173690a..4489b66 100644
--- a/include/uapi/linux/netfilter/nf_tables.h
+++ b/include/uapi/linux/netfilter/nf_tables.h
@@ -1758,6 +1758,7 @@ enum nft_tunnel_key_attributes {
 	NFTA_TUNNEL_KEY_SPORT,
 	NFTA_TUNNEL_KEY_DPORT,
 	NFTA_TUNNEL_KEY_OPTS,
+	NFTA_TUNNEL_KEY_RELEASE,
 	__NFTA_TUNNEL_KEY_MAX
 };
 #define NFTA_TUNNEL_KEY_MAX	(__NFTA_TUNNEL_KEY_MAX - 1)
diff --git a/net/netfilter/nft_tunnel.c b/net/netfilter/nft_tunnel.c
index 900c94f..0e0a34d 100644
--- a/net/netfilter/nft_tunnel.c
+++ b/net/netfilter/nft_tunnel.c
@@ -211,6 +211,7 @@ struct nft_tunnel_opts {
 struct nft_tunnel_obj {
 	struct metadata_dst	*md;
 	struct nft_tunnel_opts	opts;
+	bool tunnel_key_release;
 };
 
 static const struct nla_policy nft_tunnel_ip_policy[NFTA_TUNNEL_KEY_IP_MAX + 1] = {
@@ -395,6 +396,7 @@ static int nft_tunnel_obj_opts_init(const struct nft_ctx *ctx,
 	[NFTA_TUNNEL_KEY_TOS]	= { .type = NLA_U8, },
 	[NFTA_TUNNEL_KEY_TTL]	= { .type = NLA_U8, },
 	[NFTA_TUNNEL_KEY_OPTS]	= { .type = NLA_NESTED, },
+	[NFTA_TUNNEL_KEY_RELEASE]	= { .type = NLA_U8, },
 };
 
 static int nft_tunnel_obj_init(const struct nft_ctx *ctx,
@@ -406,6 +408,12 @@ static int nft_tunnel_obj_init(const struct nft_ctx *ctx,
 	struct metadata_dst *md;
 	int err;
 
+	if (tb[NFTA_TUNNEL_KEY_RELEASE]) {
+		priv->tunnel_key_release = !!nla_get_u8(tb[NFTA_TUNNEL_KEY_RELEASE]);
+		if (priv->tunnel_key_release)
+			return 0;
+	}
+
 	if (!tb[NFTA_TUNNEL_KEY_ID])
 		return -EINVAL;
 
@@ -488,8 +496,11 @@ static inline void nft_tunnel_obj_eval(struct nft_object *obj,
 	struct sk_buff *skb = pkt->skb;
 
 	skb_dst_drop(skb);
-	dst_hold((struct dst_entry *) priv->md);
-	skb_dst_set(skb, (struct dst_entry *) priv->md);
+
+	if (!priv->tunnel_key_release) {
+		dst_hold((struct dst_entry *)priv->md);
+		skb_dst_set(skb, (struct dst_entry *)priv->md);
+	}
 }
 
 static int nft_tunnel_ip_dump(struct sk_buff *skb, struct ip_tunnel_info *info)
@@ -591,6 +602,12 @@ static int nft_tunnel_obj_dump(struct sk_buff *skb,
 	struct nft_tunnel_obj *priv = nft_obj_data(obj);
 	struct ip_tunnel_info *info = &priv->md->u.tun_info;
 
+	if (priv->tunnel_key_release) {
+		if (nla_put_u8(skb, NFTA_TUNNEL_KEY_RELEASE, 1))
+			goto nla_put_failure;
+		return 0;
+	}
+
 	if (nla_put_be32(skb, NFTA_TUNNEL_KEY_ID,
 			 tunnel_id_to_key32(info->key.tun_id)) ||
 	    nft_tunnel_ip_dump(skb, info) < 0 ||
@@ -612,7 +629,8 @@ static void nft_tunnel_obj_destroy(const struct nft_ctx *ctx,
 {
 	struct nft_tunnel_obj *priv = nft_obj_data(obj);
 
-	metadata_dst_free(priv->md);
+	if (!priv->tunnel_key_release)
+		metadata_dst_free(priv->md);
 }
 
 static struct nft_object_type nft_tunnel_obj_type;
-- 
1.8.3.1

Re: [PATCH nf-next v2 03/11] netfilter: nf_tables_offload: split nft_offload_reg to match and action type

[Pablo Neira Ayuso]

On Thu, Jul 25, 2019 at 09:33:58PM +0800, wenxu@ucloud.cn wrote:
> From: wenxu <wenxu@ucloud.cn>
> 
> Currently the nft_offload_reg is only can be used for match condition.
> Can not be used for action. Add nft_offload_reg_type to make nft_offload_reg
> can be used for action also.

I think this patch provides what you need:

https://patchwork.ozlabs.org/patch/1140431/

that is access to the data that the immediate stores in the registers.

Re: [PATCH nf-next v2 03/11] netfilter: nf_tables_offload: split nft_offload_reg to match and action type

[wenxu]

I will repost the series based on this patch

在 2019/8/1 20:15, Pablo Neira Ayuso 写道:
> On Thu, Jul 25, 2019 at 09:33:58PM +0800, wenxu@ucloud.cn wrote:
>> From: wenxu <wenxu@ucloud.cn>
>>
>> Currently the nft_offload_reg is only can be used for match condition.
>> Can not be used for action. Add nft_offload_reg_type to make nft_offload_reg
>> can be used for action also.
> I think this patch provides what you need:
>
> https://patchwork.ozlabs.org/patch/1140431/
>
> that is access to the data that the immediate stores in the registers.
>