netfilter: add extended accounting infrastructure over nfnetlink

We currently have two ways to account traffic in netfilter:

- iptables chain and rule counters:

 # iptables -L -n -v
Chain INPUT (policy DROP 3 packets, 867 bytes)
 pkts bytes target     prot opt in     out     source               destination
    8  1104 ACCEPT     all  --  lo     *       0.0.0.0/0            0.0.0.0/0

- use flow-based accounting provided by ctnetlink:

 # conntrack -L
tcp      6 431999 ESTABLISHED src=192.168.1.130 dst=212.106.219.168 sport=58152 dport=80 packets=47 bytes=7654 src=212.106.219.168 dst=192.168.1.130 sport=80 dport=58152 packets=49 bytes=66340 [ASSURED] mark=0 use=1

While trying to display real-time accounting statistics, we require
to pool the kernel periodically to obtain this information. This is
OK if the number of flows is relatively low. However, in case that
the number of flows is huge, we can spend a considerable amount of
cycles to iterate over the list of flows that have been obtained.

Moreover, if we want to obtain the sum of the flow accounting results
that match some criteria, we have to iterate over the whole list of
existing flows, look for matchings and update the counters.

This patch adds the extended accounting infrastructure for
nfnetlink which aims to allow displaying real-time traffic accounting
without the need of complicated and resource-consuming implementation
in user-space. Basically, this new infrastructure allows you to create
accounting objects. One accounting object is composed of packet and
byte counters.

In order to manipulate create accounting objects, you require the
new libnetfilter_acct library. It contains several examples of use:

libnetfilter_acct/examples# ./nfacct-add http-traffic
libnetfilter_acct/examples# ./nfacct-get
http-traffic = { pkts = 000000000000,   bytes = 000000000000 };

Then, you can use one of this accounting objects in several iptables
rules using the new nfacct match (which comes in a follow-up patch):

 # iptables -I INPUT -p tcp --sport 80 -m nfacct --nfacct-name http-traffic
 # iptables -I OUTPUT -p tcp --dport 80 -m nfacct --nfacct-name http-traffic

The idea is simple: if one packet matches the rule, the nfacct match
updates the counters.

Thanks to Patrick McHardy, Eric Dumazet, Changli Gao for reviewing and
providing feedback for this contribution.

Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>

6 files changed, 400 insertions, 1 deletions

diff --git a/include/linux/netfilter/Kbuild b/include/linux/netfilter/Kbuild
index d81f7719b01..6785246e6e6 100644
--- a/include/linux/netfilter/Kbuild
+++ b/include/linux/netfilter/Kbuild
@@ -7,6 +7,7 @@ header-y += nf_conntrack_tcp.h
 header-y += nf_conntrack_tuple_common.h
 header-y += nf_nat.h
 header-y += nfnetlink.h
+header-y += nfnetlink_acct.h
 header-y += nfnetlink_compat.h
 header-y += nfnetlink_conntrack.h
 header-y += nfnetlink_log.h
diff --git a/include/linux/netfilter/nfnetlink.h b/include/linux/netfilter/nfnetlink.h
index 74d33861473..b64454c2f79 100644
--- a/include/linux/netfilter/nfnetlink.h
+++ b/include/linux/netfilter/nfnetlink.h
@@ -48,7 +48,8 @@ struct nfgenmsg {
 #define NFNL_SUBSYS_ULOG		4
 #define NFNL_SUBSYS_OSF			5
 #define NFNL_SUBSYS_IPSET		6
-#define NFNL_SUBSYS_COUNT		7
+#define NFNL_SUBSYS_ACCT		7
+#define NFNL_SUBSYS_COUNT		8
 
 #ifdef __KERNEL__
 
diff --git a/include/linux/netfilter/nfnetlink_acct.h b/include/linux/netfilter/nfnetlink_acct.h
new file mode 100644
index 00000000000..7c4279b4ae7
--- /dev/null
+++ b/include/linux/netfilter/nfnetlink_acct.h
@@ -0,0 +1,36 @@
+#ifndef _NFNL_ACCT_H_
+#define _NFNL_ACCT_H_
+
+#ifndef NFACCT_NAME_MAX
+#define NFACCT_NAME_MAX		32
+#endif
+
+enum nfnl_acct_msg_types {
+	NFNL_MSG_ACCT_NEW,
+	NFNL_MSG_ACCT_GET,
+	NFNL_MSG_ACCT_GET_CTRZERO,
+	NFNL_MSG_ACCT_DEL,
+	NFNL_MSG_ACCT_MAX
+};
+
+enum nfnl_acct_type {
+	NFACCT_UNSPEC,
+	NFACCT_NAME,
+	NFACCT_PKTS,
+	NFACCT_BYTES,
+	NFACCT_USE,
+	__NFACCT_MAX
+};
+#define NFACCT_MAX (__NFACCT_MAX - 1)
+
+#ifdef __KERNEL__
+
+struct nf_acct;
+
+extern struct nf_acct *nfnl_acct_find_get(const char *filter_name);
+extern void nfnl_acct_put(struct nf_acct *acct);
+extern void nfnl_acct_update(const struct sk_buff *skb, struct nf_acct *nfacct);
+
+#endif /* __KERNEL__ */
+
+#endif /* _NFNL_ACCT_H */
diff --git a/net/netfilter/Kconfig b/net/netfilter/Kconfig
index d5597b759ba..77326acd1f5 100644
--- a/net/netfilter/Kconfig
+++ b/net/netfilter/Kconfig
@@ -4,6 +4,14 @@ menu "Core Netfilter Configuration"
 config NETFILTER_NETLINK
 	tristate
 
+config NETFILTER_NETLINK_ACCT
+tristate "Netfilter NFACCT over NFNETLINK interface"
+	depends on NETFILTER_ADVANCED
+	select NETFILTER_NETLINK
+	help
+	  If this option is enabled, the kernel will include support
+	  for extended accounting via NFNETLINK.
+
 config NETFILTER_NETLINK_QUEUE
 	tristate "Netfilter NFQUEUE over NFNETLINK interface"
 	depends on NETFILTER_ADVANCED
diff --git a/net/netfilter/Makefile b/net/netfilter/Makefile
index 1a02853df86..4da1c879644 100644
--- a/net/netfilter/Makefile
+++ b/net/netfilter/Makefile
@@ -7,6 +7,7 @@ nf_conntrack-$(CONFIG_NF_CONNTRACK_EVENTS) += nf_conntrack_ecache.o
 obj-$(CONFIG_NETFILTER) = netfilter.o
 
 obj-$(CONFIG_NETFILTER_NETLINK) += nfnetlink.o
+obj-$(CONFIG_NETFILTER_NETLINK_ACCT) += nfnetlink_acct.o
 obj-$(CONFIG_NETFILTER_NETLINK_QUEUE) += nfnetlink_queue.o
 obj-$(CONFIG_NETFILTER_NETLINK_LOG) += nfnetlink_log.o
 
diff --git a/net/netfilter/nfnetlink_acct.c b/net/netfilter/nfnetlink_acct.c
new file mode 100644
index 00000000000..362ab6ca3dc
--- /dev/null
+++ b/net/netfilter/nfnetlink_acct.c
@@ -0,0 +1,352 @@
+/*
+ * (C) 2011 Pablo Neira Ayuso <pablo@netfilter.org>
+ * (C) 2011 Intra2net AG <http://www.intra2net.com>
+ *
+ * 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 (or any later at your option).
+ */
+#include <linux/init.h>
+#include <linux/module.h>
+#include <linux/kernel.h>
+#include <linux/skbuff.h>
+#include <linux/netlink.h>
+#include <linux/rculist.h>
+#include <linux/slab.h>
+#include <linux/types.h>
+#include <linux/errno.h>
+#include <net/netlink.h>
+#include <net/sock.h>
+#include <asm/atomic.h>
+
+#include <linux/netfilter.h>
+#include <linux/netfilter/nfnetlink.h>
+#include <linux/netfilter/nfnetlink_acct.h>
+
+MODULE_LICENSE("GPL");
+MODULE_AUTHOR("Pablo Neira Ayuso <pablo@netfilter.org>");
+MODULE_DESCRIPTION("nfacct: Extended Netfilter accounting infrastructure");
+
+static LIST_HEAD(nfnl_acct_list);
+
+struct nf_acct {
+	atomic64_t		pkts;
+	atomic64_t		bytes;
+	struct list_head	head;
+	atomic_t		refcnt;
+	char			name[NFACCT_NAME_MAX];
+	struct rcu_head		rcu_head;
+};
+
+static int
+nfnl_acct_new(struct sock *nfnl, struct sk_buff *skb,
+	     const struct nlmsghdr *nlh, const struct nlattr * const tb[])
+{
+	struct nf_acct *nfacct, *matching = NULL;
+	char *acct_name;
+
+	if (!tb[NFACCT_NAME])
+		return -EINVAL;
+
+	acct_name = nla_data(tb[NFACCT_NAME]);
+
+	list_for_each_entry(nfacct, &nfnl_acct_list, head) {
+		if (strncmp(nfacct->name, acct_name, NFACCT_NAME_MAX) != 0)
+			continue;
+
+                if (nlh->nlmsg_flags & NLM_F_EXCL)
+			return -EEXIST;
+
+		matching = nfacct;
+		break;
+        }
+
+	if (matching) {
+		if (nlh->nlmsg_flags & NLM_F_REPLACE) {
+			/* reset counters if you request a replacement. */
+			atomic64_set(&matching->pkts, 0);
+			atomic64_set(&matching->bytes, 0);
+			return 0;
+		}
+		return -EBUSY;
+	}
+
+	nfacct = kzalloc(sizeof(struct nf_acct), GFP_KERNEL);
+	if (nfacct == NULL)
+		return -ENOMEM;
+
+	strncpy(nfacct->name, nla_data(tb[NFACCT_NAME]), NFACCT_NAME_MAX);
+
+	if (tb[NFACCT_BYTES]) {
+		atomic64_set(&nfacct->bytes,
+			     be64_to_cpu(nla_get_u64(tb[NFACCT_BYTES])));
+	}
+	if (tb[NFACCT_PKTS]) {
+		atomic64_set(&nfacct->pkts,
+			     be64_to_cpu(nla_get_u64(tb[NFACCT_PKTS])));
+	}
+	atomic_set(&nfacct->refcnt, 1);
+	list_add_tail_rcu(&nfacct->head, &nfnl_acct_list);
+	return 0;
+}
+
+static int
+nfnl_acct_fill_info(struct sk_buff *skb, u32 pid, u32 seq, u32 type,
+		   int event, struct nf_acct *acct)
+{
+	struct nlmsghdr *nlh;
+	struct nfgenmsg *nfmsg;
+	unsigned int flags = pid ? NLM_F_MULTI : 0;
+	u64 pkts, bytes;
+
+	event |= NFNL_SUBSYS_ACCT << 8;
+	nlh = nlmsg_put(skb, pid, seq, event, sizeof(*nfmsg), flags);
+	if (nlh == NULL)
+		goto nlmsg_failure;
+
+	nfmsg = nlmsg_data(nlh);
+	nfmsg->nfgen_family = AF_UNSPEC;
+	nfmsg->version = NFNETLINK_V0;
+	nfmsg->res_id = 0;
+
+	NLA_PUT_STRING(skb, NFACCT_NAME, acct->name);
+
+	if (type == NFNL_MSG_ACCT_GET_CTRZERO) {
+		pkts = atomic64_xchg(&acct->pkts, 0);
+		bytes = atomic64_xchg(&acct->bytes, 0);
+	} else {
+		pkts = atomic64_read(&acct->pkts);
+		bytes = atomic64_read(&acct->bytes);
+	}
+	NLA_PUT_BE64(skb, NFACCT_PKTS, cpu_to_be64(pkts));
+	NLA_PUT_BE64(skb, NFACCT_BYTES, cpu_to_be64(bytes));
+	NLA_PUT_BE32(skb, NFACCT_USE, htonl(atomic_read(&acct->refcnt)));
+
+	nlmsg_end(skb, nlh);
+	return skb->len;
+
+nlmsg_failure:
+nla_put_failure:
+	nlmsg_cancel(skb, nlh);
+	return -1;
+}
+
+static int
+nfnl_acct_dump(struct sk_buff *skb, struct netlink_callback *cb)
+{
+	struct nf_acct *cur, *last;
+
+	if (cb->args[2])
+		return 0;
+
+	last = (struct nf_acct *)cb->args[1];
+	if (cb->args[1])
+		cb->args[1] = 0;
+
+	rcu_read_lock();
+	list_for_each_entry_rcu(cur, &nfnl_acct_list, head) {
+		if (last && cur != last)
+			continue;
+
+		if (nfnl_acct_fill_info(skb, NETLINK_CB(cb->skb).pid,
+				       cb->nlh->nlmsg_seq,
+				       NFNL_MSG_TYPE(cb->nlh->nlmsg_type),
+				       NFNL_MSG_ACCT_NEW, cur) < 0) {
+			cb->args[1] = (unsigned long)cur;
+			break;
+		}
+	}
+	if (!cb->args[1])
+		cb->args[2] = 1;
+	rcu_read_unlock();
+	return skb->len;
+}
+
+static int
+nfnl_acct_get(struct sock *nfnl, struct sk_buff *skb,
+	     const struct nlmsghdr *nlh, const struct nlattr * const tb[])
+{
+	int ret = 0;
+	struct nf_acct *cur;
+	char *acct_name;
+
+	if (nlh->nlmsg_flags & NLM_F_DUMP) {
+		return netlink_dump_start(nfnl, skb, nlh, nfnl_acct_dump,
+					  NULL, 0);
+	}
+
+	if (!tb[NFACCT_NAME])
+		return -EINVAL;
+	acct_name = nla_data(tb[NFACCT_NAME]);
+
+	list_for_each_entry(cur, &nfnl_acct_list, head) {
+		struct sk_buff *skb2;
+
+		if (strncmp(cur->name, acct_name, NFACCT_NAME_MAX)!= 0)
+			continue;
+
+		skb2 = nlmsg_new(NLMSG_DEFAULT_SIZE, GFP_KERNEL);
+		if (skb2 == NULL)
+			break;
+
+		ret = nfnl_acct_fill_info(skb2, NETLINK_CB(skb).pid,
+					 nlh->nlmsg_seq,
+					 NFNL_MSG_TYPE(nlh->nlmsg_type),
+					 NFNL_MSG_ACCT_NEW, cur);
+		if (ret <= 0)
+			kfree_skb(skb2);
+
+		break;
+	}
+	return ret;
+}
+
+/* try to delete object, fail if it is still in use. */
+static int nfnl_acct_try_del(struct nf_acct *cur)
+{
+	int ret = 0;
+
+	/* we want to avoid races with nfnl_acct_find_get. */
+	if (atomic_dec_and_test(&cur->refcnt)) {
+		/* We are protected by nfnl mutex. */
+		list_del_rcu(&cur->head);
+		kfree_rcu(cur, rcu_head);
+	} else {
+		/* still in use, restore reference counter. */
+		atomic_inc(&cur->refcnt);
+		ret = -EBUSY;
+	}
+	return ret;
+}
+
+static int
+nfnl_acct_del(struct sock *nfnl, struct sk_buff *skb,
+	     const struct nlmsghdr *nlh, const struct nlattr * const tb[])
+{
+	char *acct_name;
+	struct nf_acct *cur;
+	int ret = -ENOENT;
+
+	if (!tb[NFACCT_NAME]) {
+		list_for_each_entry(cur, &nfnl_acct_list, head)
+			nfnl_acct_try_del(cur);
+
+		return 0;
+	}
+	acct_name = nla_data(tb[NFACCT_NAME]);
+
+	list_for_each_entry(cur, &nfnl_acct_list, head) {
+		if (strncmp(cur->name, acct_name, NFACCT_NAME_MAX) != 0)
+			continue;
+
+		ret = nfnl_acct_try_del(cur);
+		if (ret < 0)
+			return ret;
+
+		break;
+	}
+	return ret;
+}
+
+static const struct nla_policy nfnl_acct_policy[NFACCT_MAX+1] = {
+	[NFACCT_NAME] = { .type = NLA_NUL_STRING, .len = NFACCT_NAME_MAX-1 },
+	[NFACCT_BYTES] = { .type = NLA_U64 },
+	[NFACCT_PKTS] = { .type = NLA_U64 },
+};
+
+static const struct nfnl_callback nfnl_acct_cb[NFNL_MSG_ACCT_MAX] = {
+	[NFNL_MSG_ACCT_NEW]		= { .call = nfnl_acct_new,
+					    .attr_count = NFACCT_MAX,
+					    .policy = nfnl_acct_policy },
+	[NFNL_MSG_ACCT_GET] 		= { .call = nfnl_acct_get,
+					    .attr_count = NFACCT_MAX,
+					    .policy = nfnl_acct_policy },
+	[NFNL_MSG_ACCT_GET_CTRZERO] 	= { .call = nfnl_acct_get,
+					    .attr_count = NFACCT_MAX,
+					    .policy = nfnl_acct_policy },
+	[NFNL_MSG_ACCT_DEL]		= { .call = nfnl_acct_del,
+					    .attr_count = NFACCT_MAX,
+					    .policy = nfnl_acct_policy },
+};
+
+static const struct nfnetlink_subsystem nfnl_acct_subsys = {
+	.name				= "acct",
+	.subsys_id			= NFNL_SUBSYS_ACCT,
+	.cb_count			= NFNL_MSG_ACCT_MAX,
+	.cb				= nfnl_acct_cb,
+};
+
+MODULE_ALIAS_NFNL_SUBSYS(NFNL_SUBSYS_ACCT);
+
+struct nf_acct *nfnl_acct_find_get(const char *acct_name)
+{
+	struct nf_acct *cur, *acct = NULL;
+
+	rcu_read_lock();
+	list_for_each_entry_rcu(cur, &nfnl_acct_list, head) {
+		if (strncmp(cur->name, acct_name, NFACCT_NAME_MAX)!= 0)
+			continue;
+
+		if (!try_module_get(THIS_MODULE))
+			goto err;
+
+		if (!atomic_inc_not_zero(&cur->refcnt)) {
+			module_put(THIS_MODULE);
+			goto err;
+		}
+
+		acct = cur;
+		break;
+	}
+err:
+	rcu_read_unlock();
+	return acct;
+}
+EXPORT_SYMBOL_GPL(nfnl_acct_find_get);
+
+void nfnl_acct_put(struct nf_acct *acct)
+{
+	atomic_dec(&acct->refcnt);
+	module_put(THIS_MODULE);
+}
+EXPORT_SYMBOL_GPL(nfnl_acct_put);
+
+void nfnl_acct_update(const struct sk_buff *skb, struct nf_acct *nfacct)
+{
+	atomic64_inc(&nfacct->pkts);
+	atomic64_add(skb->len, &nfacct->bytes);
+}
+EXPORT_SYMBOL_GPL(nfnl_acct_update);
+
+static int __init nfnl_acct_init(void)
+{
+	int ret;
+
+	pr_info("nfnl_acct: registering with nfnetlink.\n");
+	ret = nfnetlink_subsys_register(&nfnl_acct_subsys);
+	if (ret < 0) {
+		pr_err("nfnl_acct_init: cannot register with nfnetlink.\n");
+		goto err_out;
+	}
+	return 0;
+err_out:
+	return ret;
+}
+
+static void __exit nfnl_acct_exit(void)
+{
+	struct nf_acct *cur, *tmp;
+
+	pr_info("nfnl_acct: unregistering from nfnetlink.\n");
+	nfnetlink_subsys_unregister(&nfnl_acct_subsys);
+
+	list_for_each_entry_safe(cur, tmp, &nfnl_acct_list, head) {
+		list_del_rcu(&cur->head);
+		/* We are sure that our objects have no clients at this point,
+		 * it's safe to release them all without checking refcnt. */
+		kfree_rcu(cur, rcu_head);
+	}
+}
+
+module_init(nfnl_acct_init);
+module_exit(nfnl_acct_exit);