netfilter: conntrack: remove central spinlock nf_conntrack_lock

nf_conntrack_lock is a monolithic lock and suffers from huge contention
on current generation servers (8 or more core/threads).

Perf locking congestion is clear on base kernel:

-  72.56%  ksoftirqd/6  [kernel.kallsyms]    [k] _raw_spin_lock_bh
   - _raw_spin_lock_bh
      + 25.33% init_conntrack
      + 24.86% nf_ct_delete_from_lists
      + 24.62% __nf_conntrack_confirm
      + 24.38% destroy_conntrack
      + 0.70% tcp_packet
+   2.21%  ksoftirqd/6  [kernel.kallsyms]    [k] fib_table_lookup
+   1.15%  ksoftirqd/6  [kernel.kallsyms]    [k] __slab_free
+   0.77%  ksoftirqd/6  [kernel.kallsyms]    [k] inet_getpeer
+   0.70%  ksoftirqd/6  [nf_conntrack]       [k] nf_ct_delete
+   0.55%  ksoftirqd/6  [ip_tables]          [k] ipt_do_table

This patch change conntrack locking and provides a huge performance
improvement.  SYN-flood attack tested on a 24-core E5-2695v2(ES) with
10Gbit/s ixgbe (with tool trafgen):

 Base kernel:   810.405 new conntrack/sec
 After patch: 2.233.876 new conntrack/sec

Notice other floods attack (SYN+ACK or ACK) can easily be deflected using:
 # iptables -A INPUT -m state --state INVALID -j DROP
 # sysctl -w net/netfilter/nf_conntrack_tcp_loose=0

Use an array of hashed spinlocks to protect insertions/deletions of
conntracks into the hash table. 1024 spinlocks seem to give good
results, at minimal cost (4KB memory). Due to lockdep max depth,
1024 becomes 8 if CONFIG_LOCKDEP=y

The hash resize is a bit tricky, because we need to take all locks in
the array. A seqcount_t is used to synchronize the hash table users
with the resizing process.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: Jesper Dangaard Brouer <brouer@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Reviewed-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>

include/net/netfilter/nf_conntrack_core.h

@@ -77,7 +77,12 @@ print_tuple(struct seq_file *s, const struct nf_conntrack_tuple *tuple,
             const struct nf_conntrack_l3proto *l3proto,
             const struct nf_conntrack_l4proto *proto);
 
-extern spinlock_t nf_conntrack_lock ;
+#ifdef CONFIG_LOCKDEP
+# define CONNTRACK_LOCKS 8
+#else
+# define CONNTRACK_LOCKS 1024
+#endif
+extern spinlock_t nf_conntrack_locks[CONNTRACK_LOCKS];
 
 extern spinlock_t nf_conntrack_expect_lock;
 

include/net/netns/conntrack.h

@@ -5,6 +5,7 @@
 #include <linux/list_nulls.h>
 #include <linux/atomic.h>
 #include <linux/netfilter/nf_conntrack_tcp.h>
+#include <linux/seqlock.h>
 
 struct ctl_table_header;
 struct nf_conntrack_ecache;
@@ -90,6 +91,7 @@ struct netns_ct {
 	int			sysctl_checksum;
 
 	unsigned int		htable_size;
+	seqcount_t		generation;
 	struct kmem_cache	*nf_conntrack_cachep;
 	struct hlist_nulls_head	*hash;
 	struct hlist_head	*expect_hash;

net/netfilter/nf_conntrack_helper.c

@@ -423,12 +423,16 @@ static void __nf_conntrack_helper_unregister(struct nf_conntrack_helper *me,
 			unhelp(h, me);
 		spin_unlock_bh(&pcpu->lock);
 	}
-	spin_lock_bh(&nf_conntrack_lock);
+	local_bh_disable();
 	for (i = 0; i < net->ct.htable_size; i++) {
-		hlist_nulls_for_each_entry(h, nn, &net->ct.hash[i], hnnode)
-			unhelp(h, me);
+		spin_lock(&nf_conntrack_locks[i % CONNTRACK_LOCKS]);
+		if (i < net->ct.htable_size) {
+			hlist_nulls_for_each_entry(h, nn, &net->ct.hash[i], hnnode)
+				unhelp(h, me);
+		}
+		spin_unlock(&nf_conntrack_locks[i % CONNTRACK_LOCKS]);
 	}
-	spin_unlock_bh(&nf_conntrack_lock);
+	local_bh_enable();
 }
 
 void nf_conntrack_helper_unregister(struct nf_conntrack_helper *me)

net/netfilter/nf_conntrack_netlink.c

@@ -764,14 +764,23 @@ ctnetlink_dump_table(struct sk_buff *skb, struct netlink_callback *cb)
 	struct nfgenmsg *nfmsg = nlmsg_data(cb->nlh);
 	u_int8_t l3proto = nfmsg->nfgen_family;
 	int res;
+	spinlock_t *lockp;
+
 #ifdef CONFIG_NF_CONNTRACK_MARK
 	const struct ctnetlink_dump_filter *filter = cb->data;
 #endif
 
-	spin_lock_bh(&nf_conntrack_lock);
 	last = (struct nf_conn *)cb->args[1];
+
+	local_bh_disable();
 	for (; cb->args[0] < net->ct.htable_size; cb->args[0]++) {
 restart:
+		lockp = &nf_conntrack_locks[cb->args[0] % CONNTRACK_LOCKS];
+		spin_lock(lockp);
+		if (cb->args[0] >= net->ct.htable_size) {
+			spin_unlock(lockp);
+			goto out;
+		}
 		hlist_nulls_for_each_entry(h, n, &net->ct.hash[cb->args[0]],
 					 hnnode) {
 			if (NF_CT_DIRECTION(h) != IP_CT_DIR_ORIGINAL)
@@ -803,16 +812,18 @@ ctnetlink_dump_table(struct sk_buff *skb, struct netlink_callback *cb)
 			if (res < 0) {
 				nf_conntrack_get(&ct->ct_general);
 				cb->args[1] = (unsigned long)ct;
+				spin_unlock(lockp);
 				goto out;
 			}
 		}
+		spin_unlock(lockp);
 		if (cb->args[1]) {
 			cb->args[1] = 0;
 			goto restart;
 		}
 	}
 out:
-	spin_unlock_bh(&nf_conntrack_lock);
+	local_bh_enable();
 	if (last)
 		nf_ct_put(last);