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

Pull networking updates from David Miller:
 "Highlights:

   1) Steady transitioning of the BPF instructure to a generic spot so
      all kernel subsystems can make use of it, from Alexei Starovoitov.

   2) SFC driver supports busy polling, from Alexandre Rames.

   3) Take advantage of hash table in UDP multicast delivery, from David
      Held.

   4) Lighten locking, in particular by getting rid of the LRU lists, in
      inet frag handling.  From Florian Westphal.

   5) Add support for various RFC6458 control messages in SCTP, from
      Geir Ola Vaagland.

   6) Allow to filter bridge forwarding database dumps by device, from
      Jamal Hadi Salim.

   7) virtio-net also now supports busy polling, from Jason Wang.

   8) Some low level optimization tweaks in pktgen from Jesper Dangaard
      Brouer.

   9) Add support for ipv6 address generation modes, so that userland
      can have some input into the process.  From Jiri Pirko.

  10) Consolidate common TCP connection request code in ipv4 and ipv6,
      from Octavian Purdila.

  11) New ARP packet logger in netfilter, from Pablo Neira Ayuso.

  12) Generic resizable RCU hash table, with intial users in netlink and
      nftables.  From Thomas Graf.

  13) Maintain a name assignment type so that userspace can see where a
      network device name came from (enumerated by kernel, assigned
      explicitly by userspace, etc.) From Tom Gundersen.

  14) Automatic flow label generation on transmit in ipv6, from Tom
      Herbert.

  15) New packet timestamping facilities from Willem de Bruijn, meant to
      assist in measuring latencies going into/out-of the packet
      scheduler, latency from TCP data transmission to ACK, etc"

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-next: (1536 commits)
  cxgb4 : Disable recursive mailbox commands when enabling vi
  net: reduce USB network driver config options.
  tg3: Modify tg3_tso_bug() to handle multiple TX rings
  amd-xgbe: Perform phy connect/disconnect at dev open/stop
  amd-xgbe: Use dma_set_mask_and_coherent to set DMA mask
  net: sun4i-emac: fix memory leak on bad packet
  sctp: fix possible seqlock seadlock in sctp_packet_transmit()
  Revert "net: phy: Set the driver when registering an MDIO bus device"
  cxgb4vf: Turn off SGE RX/TX Callback Timers and interrupts in PCI shutdown routine
  team: Simplify return path of team_newlink
  bridge: Update outdated comment on promiscuous mode
  net-timestamp: ACK timestamp for bytestreams
  net-timestamp: TCP timestamping
  net-timestamp: SCHED timestamp on entering packet scheduler
  net-timestamp: add key to disambiguate concurrent datagrams
  net-timestamp: move timestamp flags out of sk_flags
  net-timestamp: extend SCM_TIMESTAMPING ancillary data struct
  cxgb4i : Move stray CPL definitions to cxgb4 driver
  tcp: reduce spurious retransmits due to transient SACK reneging
  qlcnic: Initialize dcbnl_ops before register_netdev
  ...

1 files changed, 49 insertions, 36 deletions

diff --git a/net/core/flow_dissector.c b/net/core/flow_dissector.c
index 107ed12a5323..5f362c1d0332 100644
--- a/net/core/flow_dissector.c
+++ b/net/core/flow_dissector.c
@@ -80,6 +80,8 @@ ip:
 	case htons(ETH_P_IPV6): {
 		const struct ipv6hdr *iph;
 		struct ipv6hdr _iph;
+		__be32 flow_label;
+
 ipv6:
 		iph = skb_header_pointer(skb, nhoff, sizeof(_iph), &_iph);
 		if (!iph)
@@ -89,6 +91,21 @@ ipv6:
 		flow->src = (__force __be32)ipv6_addr_hash(&iph->saddr);
 		flow->dst = (__force __be32)ipv6_addr_hash(&iph->daddr);
 		nhoff += sizeof(struct ipv6hdr);
+
+		flow_label = ip6_flowlabel(iph);
+		if (flow_label) {
+			/* Awesome, IPv6 packet has a flow label so we can
+			 * use that to represent the ports without any
+			 * further dissection.
+			 */
+			flow->n_proto = proto;
+			flow->ip_proto = ip_proto;
+			flow->ports = flow_label;
+			flow->thoff = (u16)nhoff;
+
+			return true;
+		}
+
 		break;
 	}
 	case htons(ETH_P_8021AD):
@@ -175,6 +192,7 @@ ipv6:
 		break;
 	}
 
+	flow->n_proto = proto;
 	flow->ip_proto = ip_proto;
 	flow->ports = skb_flow_get_ports(skb, nhoff, ip_proto);
 	flow->thoff = (u16) nhoff;
@@ -195,12 +213,33 @@ static __always_inline u32 __flow_hash_3words(u32 a, u32 b, u32 c)
 	return jhash_3words(a, b, c, hashrnd);
 }
 
-static __always_inline u32 __flow_hash_1word(u32 a)
+static inline u32 __flow_hash_from_keys(struct flow_keys *keys)
 {
-	__flow_hash_secret_init();
-	return jhash_1word(a, hashrnd);
+	u32 hash;
+
+	/* get a consistent hash (same value on both flow directions) */
+	if (((__force u32)keys->dst < (__force u32)keys->src) ||
+	    (((__force u32)keys->dst == (__force u32)keys->src) &&
+	     ((__force u16)keys->port16[1] < (__force u16)keys->port16[0]))) {
+		swap(keys->dst, keys->src);
+		swap(keys->port16[0], keys->port16[1]);
+	}
+
+	hash = __flow_hash_3words((__force u32)keys->dst,
+				  (__force u32)keys->src,
+				  (__force u32)keys->ports);
+	if (!hash)
+		hash = 1;
+
+	return hash;
 }
 
+u32 flow_hash_from_keys(struct flow_keys *keys)
+{
+	return __flow_hash_from_keys(keys);
+}
+EXPORT_SYMBOL(flow_hash_from_keys);
+
 /*
  * __skb_get_hash: calculate a flow hash based on src/dst addresses
  * and src/dst port numbers.  Sets hash in skb to non-zero hash value
@@ -210,7 +249,6 @@ static __always_inline u32 __flow_hash_1word(u32 a)
 void __skb_get_hash(struct sk_buff *skb)
 {
 	struct flow_keys keys;
-	u32 hash;
 
 	if (!skb_flow_dissect(skb, &keys))
 		return;
@@ -218,21 +256,9 @@ void __skb_get_hash(struct sk_buff *skb)
 	if (keys.ports)
 		skb->l4_hash = 1;
 
-	/* get a consistent hash (same value on both flow directions) */
-	if (((__force u32)keys.dst < (__force u32)keys.src) ||
-	    (((__force u32)keys.dst == (__force u32)keys.src) &&
-	     ((__force u16)keys.port16[1] < (__force u16)keys.port16[0]))) {
-		swap(keys.dst, keys.src);
-		swap(keys.port16[0], keys.port16[1]);
-	}
-
-	hash = __flow_hash_3words((__force u32)keys.dst,
-				  (__force u32)keys.src,
-				  (__force u32)keys.ports);
-	if (!hash)
-		hash = 1;
+	skb->sw_hash = 1;
 
-	skb->hash = hash;
+	skb->hash = __flow_hash_from_keys(&keys);
 }
 EXPORT_SYMBOL(__skb_get_hash);
 
@@ -240,7 +266,7 @@ EXPORT_SYMBOL(__skb_get_hash);
  * Returns a Tx hash based on the given packet descriptor a Tx queues' number
  * to be used as a distribution range.
  */
-u16 __skb_tx_hash(const struct net_device *dev, const struct sk_buff *skb,
+u16 __skb_tx_hash(const struct net_device *dev, struct sk_buff *skb,
 		  unsigned int num_tx_queues)
 {
 	u32 hash;
@@ -260,13 +286,7 @@ u16 __skb_tx_hash(const struct net_device *dev, const struct sk_buff *skb,
 		qcount = dev->tc_to_txq[tc].count;
 	}
 
-	if (skb->sk && skb->sk->sk_hash)
-		hash = skb->sk->sk_hash;
-	else
-		hash = (__force u16) skb->protocol;
-	hash = __flow_hash_1word(hash);
-
-	return (u16) (((u64) hash * qcount) >> 32) + qoffset;
+	return (u16) (((u64)skb_get_hash(skb) * qcount) >> 32) + qoffset;
 }
 EXPORT_SYMBOL(__skb_tx_hash);
 
@@ -338,17 +358,10 @@ static inline int get_xps_queue(struct net_device *dev, struct sk_buff *skb)
 		if (map) {
 			if (map->len == 1)
 				queue_index = map->queues[0];
-			else {
-				u32 hash;
-				if (skb->sk && skb->sk->sk_hash)
-					hash = skb->sk->sk_hash;
-				else
-					hash = (__force u16) skb->protocol ^
-					    skb->hash;
-				hash = __flow_hash_1word(hash);
+			else
 				queue_index = map->queues[
-				    ((u64)hash * map->len) >> 32];
-			}
+				    ((u64)skb_get_hash(skb) * map->len) >> 32];
+
 			if (unlikely(queue_index >= dev->real_num_tx_queues))
 				queue_index = -1;
 		}