netfilter: nft: add queue module

This patch adds a new nft module named "nft_queue" which provides
a new nftables expression that allows you to enqueue packets to
userspace via the nfnetlink_queue subsystem. It provides the same
level of functionality as NFQUEUE and it shares some code with it.

Signed-off-by: Eric Leblond <eric@regit.org>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>

include/uapi/linux/netfilter/nf_tables.h

@@ -657,6 +657,26 @@ enum nft_log_attributes {
 };
 #define NFTA_LOG_MAX		(__NFTA_LOG_MAX - 1)
 
+/**
+ * enum nft_queue_attributes - nf_tables queue expression netlink attributes
+ *
+ * @NFTA_QUEUE_NUM: netlink queue to send messages to (NLA_U16)
+ * @NFTA_QUEUE_TOTAL: number of queues to load balance packets on (NLA_U16)
+ * @NFTA_QUEUE_FLAGS: various flags (NLA_U16)
+ */
+enum nft_queue_attributes {
+	NFTA_QUEUE_UNSPEC,
+	NFTA_QUEUE_NUM,
+	NFTA_QUEUE_TOTAL,
+	NFTA_QUEUE_FLAGS,
+	__NFTA_QUEUE_MAX
+};
+#define NFTA_QUEUE_MAX		(__NFTA_QUEUE_MAX - 1)
+
+#define NFT_QUEUE_FLAG_BYPASS		0x01 /* for compatibility with v2 */
+#define NFT_QUEUE_FLAG_CPU_FANOUT	0x02 /* use current CPU (no hashing) */
+#define NFT_QUEUE_FLAG_MASK		0x03
+
 /**
  * enum nft_reject_types - nf_tables reject expression reject types
  *

net/netfilter/Kconfig

@@ -456,6 +456,15 @@ config NFT_NAT
 	depends on NF_NAT
 	tristate "Netfilter nf_tables nat module"
 
+config NFT_QUEUE
+	depends on NF_TABLES
+	depends on NETFILTER_XTABLES
+	depends on NETFILTER_NETLINK_QUEUE
+	tristate "Netfilter nf_tables queue module"
+	help
+	  This is required if you intend to use the userspace queueing
+	  infrastructure (also known as NFQUEUE) from nftables.
+
 config NFT_COMPAT
 	depends on NF_TABLES
 	depends on NETFILTER_XTABLES

net/netfilter/Makefile

@@ -76,6 +76,7 @@ obj-$(CONFIG_NFT_META)		+= nft_meta.o
 obj-$(CONFIG_NFT_CT)		+= nft_ct.o
 obj-$(CONFIG_NFT_LIMIT)		+= nft_limit.o
 obj-$(CONFIG_NFT_NAT)		+= nft_nat.o
+obj-$(CONFIG_NFT_QUEUE)		+= nft_queue.o
 #nf_tables-objs			+= nft_meta_target.o
 obj-$(CONFIG_NFT_RBTREE)	+= nft_rbtree.o
 obj-$(CONFIG_NFT_HASH)		+= nft_hash.o

net/netfilter/nft_queue.c

@@ -0,0 +1,134 @@
+/*
+ * Copyright (c) 2013 Eric Leblond <eric@regit.org>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * Development of this code partly funded by OISF
+ * (http://www.openinfosecfoundation.org/)
+ */
+
+#include <linux/kernel.h>
+#include <linux/init.h>
+#include <linux/module.h>
+#include <linux/netlink.h>
+#include <linux/jhash.h>
+#include <linux/netfilter.h>
+#include <linux/netfilter/nf_tables.h>
+#include <net/netfilter/nf_tables.h>
+#include <net/netfilter/nf_queue.h>
+
+static u32 jhash_initval __read_mostly;
+
+struct nft_queue {
+	u16	queuenum;
+	u16	queues_total;
+	u16	flags;
+	u8	family;
+};
+
+static void nft_queue_eval(const struct nft_expr *expr,
+			   struct nft_data data[NFT_REG_MAX + 1],
+			   const struct nft_pktinfo *pkt)
+{
+	struct nft_queue *priv = nft_expr_priv(expr);
+	u32 queue = priv->queuenum;
+	u32 ret;
+
+	if (priv->queues_total > 1) {
+		if (priv->flags & NFT_QUEUE_FLAG_CPU_FANOUT) {
+			int cpu = smp_processor_id();
+
+			queue = priv->queuenum + cpu % priv->queues_total;
+		} else {
+			queue = nfqueue_hash(pkt->skb, queue,
+					     priv->queues_total, priv->family,
+					     jhash_initval);
+		}
+	}
+
+	ret = NF_QUEUE_NR(queue);
+	if (priv->flags & NFT_QUEUE_FLAG_BYPASS)
+		ret |= NF_VERDICT_FLAG_QUEUE_BYPASS;
+
+	data[NFT_REG_VERDICT].verdict = ret;
+}
+
+static const struct nla_policy nft_queue_policy[NFTA_QUEUE_MAX + 1] = {
+	[NFTA_QUEUE_NUM]	= { .type = NLA_U16 },
+	[NFTA_QUEUE_TOTAL]	= { .type = NLA_U16 },
+	[NFTA_QUEUE_FLAGS]	= { .type = NLA_U16 },
+};
+
+static int nft_queue_init(const struct nft_ctx *ctx,
+			   const struct nft_expr *expr,
+			   const struct nlattr * const tb[])
+{
+	struct nft_queue *priv = nft_expr_priv(expr);
+
+	if (tb[NFTA_QUEUE_NUM] == NULL)
+		return -EINVAL;
+
+	init_hashrandom(&jhash_initval);
+	priv->family = ctx->afi->family;
+	priv->queuenum = ntohs(nla_get_be16(tb[NFTA_QUEUE_NUM]));
+
+	if (tb[NFTA_QUEUE_TOTAL] != NULL)
+		priv->queues_total = ntohs(nla_get_be16(tb[NFTA_QUEUE_TOTAL]));
+	if (tb[NFTA_QUEUE_FLAGS] != NULL) {
+		priv->flags = ntohs(nla_get_be16(tb[NFTA_QUEUE_FLAGS]));
+		if (priv->flags & ~NFT_QUEUE_FLAG_MASK)
+			return -EINVAL;
+	}
+	return 0;
+}
+
+static int nft_queue_dump(struct sk_buff *skb, const struct nft_expr *expr)
+{
+	const struct nft_queue *priv = nft_expr_priv(expr);
+
+	if (nla_put_be16(skb, NFTA_QUEUE_NUM, htons(priv->queuenum)) ||
+	    nla_put_be16(skb, NFTA_QUEUE_TOTAL, htons(priv->queues_total)) ||
+	    nla_put_be16(skb, NFTA_QUEUE_FLAGS, htons(priv->flags)))
+		goto nla_put_failure;
+
+	return 0;
+
+nla_put_failure:
+	return -1;
+}
+
+static struct nft_expr_type nft_queue_type;
+static const struct nft_expr_ops nft_queue_ops = {
+	.type		= &nft_queue_type,
+	.size		= NFT_EXPR_SIZE(sizeof(struct nft_queue)),
+	.eval		= nft_queue_eval,
+	.init		= nft_queue_init,
+	.dump		= nft_queue_dump,
+};
+
+static struct nft_expr_type nft_queue_type __read_mostly = {
+	.name		= "queue",
+	.ops		= &nft_queue_ops,
+	.policy		= nft_queue_policy,
+	.maxattr	= NFTA_QUEUE_MAX,
+	.owner		= THIS_MODULE,
+};
+
+static int __init nft_queue_module_init(void)
+{
+	return nft_register_expr(&nft_queue_type);
+}
+
+static void __exit nft_queue_module_exit(void)
+{
+	nft_unregister_expr(&nft_queue_type);
+}
+
+module_init(nft_queue_module_init);
+module_exit(nft_queue_module_exit);
+
+MODULE_LICENSE("GPL");
+MODULE_AUTHOR("Eric Leblond <eric@regit.org>");
+MODULE_ALIAS_NFT_EXPR("queue");