netfilter: accounting rework: ct_extend + 64bit counters (v4)

Initially netfilter has had 64bit counters for conntrack-based accounting, but it was changed in 2.6.14 to save memory. Unfortunately in-kernel 64bit counters are still required, for example for "connbytes" extension. However, 64bit counters waste a lot of memory and it was not possible to enable/disable it runtime. This patch: - reimplements accounting with respect to the extension infrastructure, - makes one global version of seq_print_acct() instead of two seq_print_counters(), - makes it possible to enable it at boot time (for CONFIG_SYSCTL/CONFIG_SYSFS=n), - makes it possible to enable/disable it at runtime by sysctl or sysfs, - extends counters from 32bit to 64bit, - renames ip_conntrack_counter -> nf_conn_counter, - enables accounting code unconditionally (no longer depends on CONFIG_NF_CT_ACCT), - set initial accounting enable state based on CONFIG_NF_CT_ACCT - removes buggy IPCT_COUNTER_FILLING event handling. If accounting is enabled newly created connections get additional acct extend. Old connections are not changed as it is not possible to add a ct_extend area to confirmed conntrack. Accounting is performed for all connections with acct extend regardless of a current state of "net.netfilter.nf_conntrack_acct". Signed-off-by: Krzysztof Piotr Oledzki <[email protected]> Signed-off-by: Patrick McHardy <[email protected]> Signed-off-by: David S. Miller <[email protected]>
kishon · Jul 21, 2008 · 5840157 · 5840157
1 parent 07a7c10
commit 5840157
Show file tree

Hide file tree

Showing 15 changed files with 248 additions and 86 deletions.
diff --git a/Documentation/feature-removal-schedule.txt b/Documentation/feature-removal-schedule.txt
@@ -336,3 +336,13 @@ When:	After the only user (hal) has seen a release with the patches
 Why:	Over 1K .text/.data size reduction, data is available in other
 	ways (ioctls)
 Who:	Johannes Berg <[email protected]>
+
+---------------------------
+
+What: CONFIG_NF_CT_ACCT
+When: 2.6.29
+Why:  Accounting can now be enabled/disabled without kernel recompilation.
+      Currently used only to set a default value for a feature that is also
+      controlled by a kernel/module/sysfs/sysctl parameter.
+Who:  Krzysztof Piotr Oledzki <[email protected]>
+
diff --git a/Documentation/kernel-parameters.txt b/Documentation/kernel-parameters.txt
@@ -1279,6 +1279,13 @@ and is between 256 and 4096 characters. It is defined in the file
 			This usage is only documented in each driver source
 			file if at all.
 
+	nf_conntrack.acct=
+			[NETFILTER] Enable connection tracking flow accounting
+			0 to disable accounting
+			1 to enable accounting
+			Default value depends on CONFIG_NF_CT_ACCT that is
+			going to be removed in 2.6.29.
+
 	nfsaddrs=	[NFS]
 			See Documentation/filesystems/nfsroot.txt.
 

diff --git a/include/linux/netfilter/nf_conntrack_common.h b/include/linux/netfilter/nf_conntrack_common.h
@@ -122,7 +122,7 @@ enum ip_conntrack_events
 	IPCT_NATINFO_BIT = 10,
 	IPCT_NATINFO = (1 << IPCT_NATINFO_BIT),
 
-	/* Counter highest bit has been set */
+	/* Counter highest bit has been set, unused */
 	IPCT_COUNTER_FILLING_BIT = 11,
 	IPCT_COUNTER_FILLING = (1 << IPCT_COUNTER_FILLING_BIT),
 
@@ -145,12 +145,6 @@ enum ip_conntrack_expect_events {
 };
 
 #ifdef __KERNEL__
-struct ip_conntrack_counter
-{
-	u_int32_t packets;
-	u_int32_t bytes;
-};
-
 struct ip_conntrack_stat
 {
 	unsigned int searched;

diff --git a/include/linux/netfilter/nfnetlink_conntrack.h b/include/linux/netfilter/nfnetlink_conntrack.h
@@ -115,10 +115,10 @@ enum ctattr_protoinfo_sctp {
 
 enum ctattr_counters {
 	CTA_COUNTERS_UNSPEC,
-	CTA_COUNTERS_PACKETS,		/* old 64bit counters */
-	CTA_COUNTERS_BYTES,		/* old 64bit counters */
-	CTA_COUNTERS32_PACKETS,
-	CTA_COUNTERS32_BYTES,
+	CTA_COUNTERS_PACKETS,		/* 64bit counters */
+	CTA_COUNTERS_BYTES,		/* 64bit counters */
+	CTA_COUNTERS32_PACKETS,		/* old 32bit counters, unused */
+	CTA_COUNTERS32_BYTES,		/* old 32bit counters, unused */
 	__CTA_COUNTERS_MAX
 };
 #define CTA_COUNTERS_MAX (__CTA_COUNTERS_MAX - 1)

diff --git a/include/net/netfilter/nf_conntrack.h b/include/net/netfilter/nf_conntrack.h
@@ -88,7 +88,6 @@ struct nf_conn_help {
 	u8 expecting[NF_CT_MAX_EXPECT_CLASSES];
 };
 
-
 #include <net/netfilter/ipv4/nf_conntrack_ipv4.h>
 #include <net/netfilter/ipv6/nf_conntrack_ipv6.h>
 
@@ -111,11 +110,6 @@ struct nf_conn
 	/* Timer function; drops refcnt when it goes off. */
 	struct timer_list timeout;
 
-#ifdef CONFIG_NF_CT_ACCT
-	/* Accounting Information (same cache line as other written members) */
-	struct ip_conntrack_counter counters[IP_CT_DIR_MAX];
-#endif
-
 #if defined(CONFIG_NF_CONNTRACK_MARK)
 	u_int32_t mark;
 #endif

diff --git a/include/net/netfilter/nf_conntrack_acct.h b/include/net/netfilter/nf_conntrack_acct.h
@@ -0,0 +1,51 @@
+/*
+ * (C) 2008 Krzysztof Piotr Oledzki <[email protected]>
+ *
+ * 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.
+ */
+
+#ifndef _NF_CONNTRACK_ACCT_H
+#define _NF_CONNTRACK_ACCT_H
+#include <linux/netfilter/nf_conntrack_common.h>
+#include <linux/netfilter/nf_conntrack_tuple_common.h>
+#include <net/netfilter/nf_conntrack.h>
+#include <net/netfilter/nf_conntrack_extend.h>
+
+struct nf_conn_counter {
+	u_int64_t packets;
+	u_int64_t bytes;
+};
+
+extern int nf_ct_acct;
+
+static inline
+struct nf_conn_counter *nf_conn_acct_find(const struct nf_conn *ct)
+{
+	return nf_ct_ext_find(ct, NF_CT_EXT_ACCT);
+}
+
+static inline
+struct nf_conn_counter *nf_ct_acct_ext_add(struct nf_conn *ct, gfp_t gfp)
+{
+	struct nf_conn_counter *acct;
+
+	if (!nf_ct_acct)
+		return NULL;
+
+	acct = nf_ct_ext_add(ct, NF_CT_EXT_ACCT, gfp);
+	if (!acct)
+		pr_debug("failed to add accounting extension area");
+
+
+	return acct;
+};
+
+extern unsigned int
+seq_print_acct(struct seq_file *s, const struct nf_conn *ct, int dir);
+
+extern int nf_conntrack_acct_init(void);
+extern void nf_conntrack_acct_fini(void);
+
+#endif /* _NF_CONNTRACK_ACCT_H */
diff --git a/include/net/netfilter/nf_conntrack_extend.h b/include/net/netfilter/nf_conntrack_extend.h
@@ -7,11 +7,13 @@ enum nf_ct_ext_id
 {
 	NF_CT_EXT_HELPER,
 	NF_CT_EXT_NAT,
+	NF_CT_EXT_ACCT,
 	NF_CT_EXT_NUM,
 };
 
 #define NF_CT_EXT_HELPER_TYPE struct nf_conn_help
 #define NF_CT_EXT_NAT_TYPE struct nf_conn_nat
+#define NF_CT_EXT_ACCT_TYPE struct nf_conn_counter
 
 /* Extensions: optional stuff which isn't permanently in struct. */
 struct nf_ct_ext {

diff --git a/net/ipv4/netfilter/nf_conntrack_l3proto_ipv4_compat.c b/net/ipv4/netfilter/nf_conntrack_l3proto_ipv4_compat.c
@@ -18,19 +18,7 @@
 #include <net/netfilter/nf_conntrack_l3proto.h>
 #include <net/netfilter/nf_conntrack_l4proto.h>
 #include <net/netfilter/nf_conntrack_expect.h>
-
-#ifdef CONFIG_NF_CT_ACCT
-static unsigned int
-seq_print_counters(struct seq_file *s,
-		   const struct ip_conntrack_counter *counter)
-{
-	return seq_printf(s, "packets=%llu bytes=%llu ",
-			  (unsigned long long)counter->packets,
-			  (unsigned long long)counter->bytes);
-}
-#else
-#define seq_print_counters(x, y)	0
-#endif
+#include <net/netfilter/nf_conntrack_acct.h>
 
 struct ct_iter_state {
 	unsigned int bucket;
@@ -127,7 +115,7 @@ static int ct_seq_show(struct seq_file *s, void *v)
 			l3proto, l4proto))
 		return -ENOSPC;
 
-	if (seq_print_counters(s, &ct->counters[IP_CT_DIR_ORIGINAL]))
+	if (seq_print_acct(s, ct, IP_CT_DIR_ORIGINAL))
 		return -ENOSPC;
 
 	if (!(test_bit(IPS_SEEN_REPLY_BIT, &ct->status)))
@@ -138,7 +126,7 @@ static int ct_seq_show(struct seq_file *s, void *v)
 			l3proto, l4proto))
 		return -ENOSPC;
 
-	if (seq_print_counters(s, &ct->counters[IP_CT_DIR_REPLY]))
+	if (seq_print_acct(s, ct, IP_CT_DIR_REPLY))
 		return -ENOSPC;
 
 	if (test_bit(IPS_ASSURED_BIT, &ct->status))

diff --git a/net/netfilter/Kconfig b/net/netfilter/Kconfig
@@ -49,6 +49,15 @@ config NF_CT_ACCT
 	  Those counters can be used for flow-based accounting or the
 	  `connbytes' match.
 
+	  Please note that currently this option only sets a default state.
+	  You may change it at boot time with nf_conntrack.acct=0/1 kernel
+	  paramater or by loading the nf_conntrack module with acct=0/1.
+
+	  You may also disable/enable it on a running system with:
+	   sysctl net.netfilter.nf_conntrack_acct=0/1
+
+	  This option will be removed in 2.6.29.
+
 	  If unsure, say `N'.
 
 config NF_CONNTRACK_MARK

diff --git a/net/netfilter/Makefile b/net/netfilter/Makefile
@@ -1,6 +1,6 @@
 netfilter-objs := core.o nf_log.o nf_queue.o nf_sockopt.o
 
-nf_conntrack-y	:= nf_conntrack_core.o nf_conntrack_standalone.o nf_conntrack_expect.o nf_conntrack_helper.o nf_conntrack_proto.o nf_conntrack_l3proto_generic.o nf_conntrack_proto_generic.o nf_conntrack_proto_tcp.o nf_conntrack_proto_udp.o nf_conntrack_extend.o
+nf_conntrack-y	:= nf_conntrack_core.o nf_conntrack_standalone.o nf_conntrack_expect.o nf_conntrack_helper.o nf_conntrack_proto.o nf_conntrack_l3proto_generic.o nf_conntrack_proto_generic.o nf_conntrack_proto_tcp.o nf_conntrack_proto_udp.o nf_conntrack_extend.o nf_conntrack_acct.o
 nf_conntrack-$(CONFIG_NF_CONNTRACK_EVENTS) += nf_conntrack_ecache.o
 
 obj-$(CONFIG_NETFILTER) = netfilter.o

diff --git a/net/netfilter/nf_conntrack_acct.c b/net/netfilter/nf_conntrack_acct.c
@@ -0,0 +1,104 @@
+/* Accouting handling for netfilter. */
+
+/*
+ * (C) 2008 Krzysztof Piotr Oledzki <[email protected]>
+ *
+ * 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.
+ */
+
+#include <linux/netfilter.h>
+#include <linux/kernel.h>
+#include <linux/moduleparam.h>
+
+#include <net/netfilter/nf_conntrack.h>
+#include <net/netfilter/nf_conntrack_extend.h>
+#include <net/netfilter/nf_conntrack_acct.h>
+
+#ifdef CONFIG_NF_CT_ACCT
+#define NF_CT_ACCT_DEFAULT 1
+#else
+#define NF_CT_ACCT_DEFAULT 0
+#endif
+
+int nf_ct_acct __read_mostly = NF_CT_ACCT_DEFAULT;
+EXPORT_SYMBOL_GPL(nf_ct_acct);
+
+module_param_named(acct, nf_ct_acct, bool, 0644);
+MODULE_PARM_DESC(acct, "Enable connection tracking flow accounting.");
+
+#ifdef CONFIG_SYSCTL
+static struct ctl_table_header *acct_sysctl_header;
+static struct ctl_table acct_sysctl_table[] = {
+	{
+		.ctl_name	= CTL_UNNUMBERED,
+		.procname	= "nf_conntrack_acct",
+		.data		= &nf_ct_acct,
+		.maxlen		= sizeof(unsigned int),
+		.mode		= 0644,
+		.proc_handler	= &proc_dointvec,
+	},
+	{}
+};
+#endif /* CONFIG_SYSCTL */
+
+unsigned int
+seq_print_acct(struct seq_file *s, const struct nf_conn *ct, int dir)
+{
+	struct nf_conn_counter *acct;
+
+	acct = nf_conn_acct_find(ct);
+	if (!acct)
+		return 0;
+
+	return seq_printf(s, "packets=%llu bytes=%llu ",
+			  (unsigned long long)acct[dir].packets,
+			  (unsigned long long)acct[dir].bytes);
+};
+EXPORT_SYMBOL_GPL(seq_print_acct);
+
+static struct nf_ct_ext_type acct_extend __read_mostly = {
+	.len	= sizeof(struct nf_conn_counter[IP_CT_DIR_MAX]),
+	.align	= __alignof__(struct nf_conn_counter[IP_CT_DIR_MAX]),
+	.id	= NF_CT_EXT_ACCT,
+};
+
+int nf_conntrack_acct_init(void)
+{
+	int ret;
+
+#ifdef CONFIG_NF_CT_ACCT
+	printk(KERN_WARNING "CONFIG_NF_CT_ACCT is deprecated and will be removed soon. Plase use\n");
+	printk(KERN_WARNING "nf_conntrack.acct=1 kernel paramater, acct=1 nf_conntrack module option or\n");
+	printk(KERN_WARNING "sysctl net.netfilter.nf_conntrack_acct=1 to enable it.\n");
+#endif
+
+	ret = nf_ct_extend_register(&acct_extend);
+	if (ret < 0) {
+		printk(KERN_ERR "nf_conntrack_acct: Unable to register extension\n");
+		return ret;
+	}
+
+#ifdef CONFIG_SYSCTL
+	acct_sysctl_header = register_sysctl_paths(nf_net_netfilter_sysctl_path,
+				acct_sysctl_table);
+
+	if (!acct_sysctl_header) {
+		nf_ct_extend_unregister(&acct_extend);
+
+		printk(KERN_ERR "nf_conntrack_acct: can't register to sysctl.\n");
+		return -ENOMEM;
+	}
+#endif
+
+	return 0;
+}
+
+void nf_conntrack_acct_fini(void)
+{
+#ifdef CONFIG_SYSCTL
+	unregister_sysctl_table(acct_sysctl_header);
+#endif
+	nf_ct_extend_unregister(&acct_extend);
+}