Merge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-next

Pull networking changes from David S Miller:

 1) Remove the ipv4 routing cache.  Now lookups go directly into the FIB
    trie and use prebuilt routes cached there.

    No more garbage collection, no more rDOS attacks on the routing
    cache.  Instead we now get predictable and consistent performance,
    no matter what the pattern of traffic we service.

    This has been almost 2 years in the making.  Special thanks to
    Julian Anastasov, Eric Dumazet, Steffen Klassert, and others who
    have helped along the way.

    I'm sure that with a change of this magnitude there will be some
    kind of fallout, but such things ought the be simple to fix at this
    point.  Luckily I'm not European so I'll be around all of August to
    fix things :-)

    The major stages of this work here are each fronted by a forced
    merge commit whose commit message contains a top-level description
    of the motivations and implementation issues.

 2) Pre-demux of established ipv4 TCP sockets, saves a route demux on
    input.

 3) TCP SYN/ACK performance tweaks from Eric Dumazet.

 4) Add namespace support for netfilter L4 conntrack helpers, from Gao
    Feng.

 5) Add config mechanism for Energy Efficient Ethernet to ethtool, from
    Yuval Mintz.

 6) Remove quadratic behavior from /proc/net/unix, from Eric Dumazet.

 7) Support for connection tracker helpers in userspace, from Pablo
    Neira Ayuso.

 8) Allow userspace driven TX load balancing functions in TEAM driver,
    from Jiri Pirko.

 9) Kill off NLMSG_PUT and RTA_PUT macros, more gross stuff with
    embedded gotos.

10) TCP Small Queues, essentially minimize the amount of TCP data queued
    up in the packet scheduler layer.  Whereas the existing BQL (Byte
    Queue Limits) limits the pkt_sched --> netdevice queuing levels,
    this controls the TCP --> pkt_sched queueing levels.

    From Eric Dumazet.

11) Reduce the number of get_page/put_page ops done on SKB fragments,
    from Alexander Duyck.

12) Implement protection against blind resets in TCP (RFC 5961), from
    Eric Dumazet.

13) Support the client side of TCP Fast Open, basically the ability to
    send data in the SYN exchange, from Yuchung Cheng.

    Basically, the sender queues up data with a sendmsg() call using
    MSG_FASTOPEN, then they do the connect() which emits the queued up
    fastopen data.

14) Avoid all the problems we get into in TCP when timers or PMTU events
    hit a locked socket.  The TCP Small Queues changes added a
    tcp_release_cb() that allows us to queue work up to the
    release_sock() caller, and that's what we use here too.  From Eric
    Dumazet.

15) Zero copy on TX support for TUN driver, from Michael S. Tsirkin.

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-next: (1870 commits)
  genetlink: define lockdep_genl_is_held() when CONFIG_LOCKDEP
  r8169: revert "add byte queue limit support".
  ipv4: Change rt->rt_iif encoding.
  net: Make skb->skb_iif always track skb->dev
  ipv4: Prepare for change of rt->rt_iif encoding.
  ipv4: Remove all RTCF_DIRECTSRC handliing.
  ipv4: Really ignore ICMP address requests/replies.
  decnet: Don't set RTCF_DIRECTSRC.
  net/ipv4/ip_vti.c: Fix __rcu warnings detected by sparse.
  ipv4: Remove redundant assignment
  rds: set correct msg_namelen
  openvswitch: potential NULL deref in sample()
  tcp: dont drop MTU reduction indications
  bnx2x: Add new 57840 device IDs
  tcp: avoid oops in tcp_metrics and reset tcpm_stamp
  niu: Change niu_rbr_fill() to use unlikely() to check niu_rbr_add_page() return value
  niu: Fix to check for dma mapping errors.
  net: Fix references to out-of-scope variables in put_cmsg_compat()
  net: ethernet: davinci_emac: add pm_runtime support
  net: ethernet: davinci_emac: Remove unnecessary #include
  ...

1 files changed, 50 insertions, 43 deletions

diff --git a/net/ipv4/ip_output.c b/net/ipv4/ip_output.c
index 451f97c42eb4..ba39a52d18c1 100644
--- a/net/ipv4/ip_output.c
+++ b/net/ipv4/ip_output.c
@@ -113,19 +113,6 @@ int ip_local_out(struct sk_buff *skb)
 }
 EXPORT_SYMBOL_GPL(ip_local_out);
 
-/* dev_loopback_xmit for use with netfilter. */
-static int ip_dev_loopback_xmit(struct sk_buff *newskb)
-{
-	skb_reset_mac_header(newskb);
-	__skb_pull(newskb, skb_network_offset(newskb));
-	newskb->pkt_type = PACKET_LOOPBACK;
-	newskb->ip_summed = CHECKSUM_UNNECESSARY;
-	WARN_ON(!skb_dst(newskb));
-	skb_dst_force(newskb);
-	netif_rx_ni(newskb);
-	return 0;
-}
-
 static inline int ip_select_ttl(struct inet_sock *inet, struct dst_entry *dst)
 {
 	int ttl = inet->uc_ttl;
@@ -183,6 +170,7 @@ static inline int ip_finish_output2(struct sk_buff *skb)
 	struct net_device *dev = dst->dev;
 	unsigned int hh_len = LL_RESERVED_SPACE(dev);
 	struct neighbour *neigh;
+	u32 nexthop;
 
 	if (rt->rt_type == RTN_MULTICAST) {
 		IP_UPD_PO_STATS(dev_net(dev), IPSTATS_MIB_OUTMCAST, skb->len);
@@ -200,19 +188,22 @@ static inline int ip_finish_output2(struct sk_buff *skb)
 		}
 		if (skb->sk)
 			skb_set_owner_w(skb2, skb->sk);
-		kfree_skb(skb);
+		consume_skb(skb);
 		skb = skb2;
 	}
 
-	rcu_read_lock();
-	neigh = dst_get_neighbour_noref(dst);
+	rcu_read_lock_bh();
+	nexthop = rt->rt_gateway ? rt->rt_gateway : ip_hdr(skb)->daddr;
+	neigh = __ipv4_neigh_lookup_noref(dev, nexthop);
+	if (unlikely(!neigh))
+		neigh = __neigh_create(&arp_tbl, &nexthop, dev, false);
 	if (neigh) {
-		int res = neigh_output(neigh, skb);
+		int res = dst_neigh_output(dst, neigh, skb);
 
-		rcu_read_unlock();
+		rcu_read_unlock_bh();
 		return res;
 	}
-	rcu_read_unlock();
+	rcu_read_unlock_bh();
 
 	net_dbg_ratelimited("%s: No header cache and no neighbour!\n",
 			    __func__);
@@ -281,7 +272,7 @@ int ip_mc_output(struct sk_buff *skb)
 			if (newskb)
 				NF_HOOK(NFPROTO_IPV4, NF_INET_POST_ROUTING,
 					newskb, NULL, newskb->dev,
-					ip_dev_loopback_xmit);
+					dev_loopback_xmit);
 		}
 
 		/* Multicasts with ttl 0 must not go beyond the host */
@@ -296,7 +287,7 @@ int ip_mc_output(struct sk_buff *skb)
 		struct sk_buff *newskb = skb_clone(skb, GFP_ATOMIC);
 		if (newskb)
 			NF_HOOK(NFPROTO_IPV4, NF_INET_POST_ROUTING, newskb,
-				NULL, newskb->dev, ip_dev_loopback_xmit);
+				NULL, newskb->dev, dev_loopback_xmit);
 	}
 
 	return NF_HOOK_COND(NFPROTO_IPV4, NF_INET_POST_ROUTING, skb, NULL,
@@ -380,7 +371,7 @@ int ip_queue_xmit(struct sk_buff *skb, struct flowi *fl)
 	skb_dst_set_noref(skb, &rt->dst);
 
 packet_routed:
-	if (inet_opt && inet_opt->opt.is_strictroute && fl4->daddr != rt->rt_gateway)
+	if (inet_opt && inet_opt->opt.is_strictroute && rt->rt_gateway)
 		goto no_route;
 
 	/* OK, we know where to send it, allocate and build IP header. */
@@ -709,7 +700,7 @@ slow_path:
 
 		IP_INC_STATS(dev_net(dev), IPSTATS_MIB_FRAGCREATES);
 	}
-	kfree_skb(skb);
+	consume_skb(skb);
 	IP_INC_STATS(dev_net(dev), IPSTATS_MIB_FRAGOKS);
 	return err;
 
@@ -1472,19 +1463,34 @@ static int ip_reply_glue_bits(void *dptr, char *to, int offset,
 
 /*
  *	Generic function to send a packet as reply to another packet.
- *	Used to send TCP resets so far. ICMP should use this function too.
+ *	Used to send some TCP resets/acks so far.
  *
- *	Should run single threaded per socket because it uses the sock
- *     	structure to pass arguments.
+ *	Use a fake percpu inet socket to avoid false sharing and contention.
  */
-void ip_send_reply(struct sock *sk, struct sk_buff *skb, __be32 daddr,
-		   const struct ip_reply_arg *arg, unsigned int len)
+static DEFINE_PER_CPU(struct inet_sock, unicast_sock) = {
+	.sk = {
+		.__sk_common = {
+			.skc_refcnt = ATOMIC_INIT(1),
+		},
+		.sk_wmem_alloc	= ATOMIC_INIT(1),
+		.sk_allocation	= GFP_ATOMIC,
+		.sk_flags	= (1UL << SOCK_USE_WRITE_QUEUE),
+	},
+	.pmtudisc	= IP_PMTUDISC_WANT,
+	.uc_ttl		= -1,
+};
+
+void ip_send_unicast_reply(struct net *net, struct sk_buff *skb, __be32 daddr,
+			   __be32 saddr, const struct ip_reply_arg *arg,
+			   unsigned int len)
 {
-	struct inet_sock *inet = inet_sk(sk);
 	struct ip_options_data replyopts;
 	struct ipcm_cookie ipc;
 	struct flowi4 fl4;
 	struct rtable *rt = skb_rtable(skb);
+	struct sk_buff *nskb;
+	struct sock *sk;
+	struct inet_sock *inet;
 
 	if (ip_options_echo(&replyopts.opt.opt, skb))
 		return;
@@ -1502,38 +1508,39 @@ void ip_send_reply(struct sock *sk, struct sk_buff *skb, __be32 daddr,
 
 	flowi4_init_output(&fl4, arg->bound_dev_if, 0,
 			   RT_TOS(arg->tos),
-			   RT_SCOPE_UNIVERSE, sk->sk_protocol,
+			   RT_SCOPE_UNIVERSE, ip_hdr(skb)->protocol,
 			   ip_reply_arg_flowi_flags(arg),
-			   daddr, rt->rt_spec_dst,
+			   daddr, saddr,
 			   tcp_hdr(skb)->source, tcp_hdr(skb)->dest);
 	security_skb_classify_flow(skb, flowi4_to_flowi(&fl4));
-	rt = ip_route_output_key(sock_net(sk), &fl4);
+	rt = ip_route_output_key(net, &fl4);
 	if (IS_ERR(rt))
 		return;
 
-	/* And let IP do all the hard work.
+	inet = &get_cpu_var(unicast_sock);
 
-	   This chunk is not reenterable, hence spinlock.
-	   Note that it uses the fact, that this function is called
-	   with locally disabled BH and that sk cannot be already spinlocked.
-	 */
-	bh_lock_sock(sk);
 	inet->tos = arg->tos;
+	sk = &inet->sk;
 	sk->sk_priority = skb->priority;
 	sk->sk_protocol = ip_hdr(skb)->protocol;
 	sk->sk_bound_dev_if = arg->bound_dev_if;
+	sock_net_set(sk, net);
+	__skb_queue_head_init(&sk->sk_write_queue);
+	sk->sk_sndbuf = sysctl_wmem_default;
 	ip_append_data(sk, &fl4, ip_reply_glue_bits, arg->iov->iov_base, len, 0,
 		       &ipc, &rt, MSG_DONTWAIT);
-	if ((skb = skb_peek(&sk->sk_write_queue)) != NULL) {
+	nskb = skb_peek(&sk->sk_write_queue);
+	if (nskb) {
 		if (arg->csumoffset >= 0)
-			*((__sum16 *)skb_transport_header(skb) +
-			  arg->csumoffset) = csum_fold(csum_add(skb->csum,
+			*((__sum16 *)skb_transport_header(nskb) +
+			  arg->csumoffset) = csum_fold(csum_add(nskb->csum,
 								arg->csum));
-		skb->ip_summed = CHECKSUM_NONE;
+		nskb->ip_summed = CHECKSUM_NONE;
+		skb_set_queue_mapping(nskb, skb_get_queue_mapping(skb));
 		ip_push_pending_frames(sk, &fl4);
 	}
 
-	bh_unlock_sock(sk);
+	put_cpu_var(unicast_sock);
 
 	ip_rt_put(rt);
 }