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  bridge: Update outdated comment on promiscuous mode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, 58 insertions, 13 deletions

diff --git a/include/net/ipv6.h b/include/net/ipv6.h
index 574337fe72dd..a2db816e8461 100644
--- a/include/net/ipv6.h
+++ b/include/net/ipv6.h
@@ -19,6 +19,7 @@
 #include <net/if_inet6.h>
 #include <net/ndisc.h>
 #include <net/flow.h>
+#include <net/flow_keys.h>
 #include <net/snmp.h>
 
 #define SIN6_LEN_RFC2133	24
@@ -298,11 +299,6 @@ static inline bool ipv6_accept_ra(struct inet6_dev *idev)
 }
 
 #if IS_ENABLED(CONFIG_IPV6)
-static inline int ip6_frag_nqueues(struct net *net)
-{
-	return net->ipv6.frags.nqueues;
-}
-
 static inline int ip6_frag_mem(struct net *net)
 {
 	return sum_frag_mem_limit(&net->ipv6.frags);
@@ -495,8 +491,8 @@ struct ip6_create_arg {
 	u8 ecn;
 };
 
-void ip6_frag_init(struct inet_frag_queue *q, void *a);
-bool ip6_frag_match(struct inet_frag_queue *q, void *a);
+void ip6_frag_init(struct inet_frag_queue *q, const void *a);
+bool ip6_frag_match(const struct inet_frag_queue *q, const void *a);
 
 /*
  *	Equivalent of ipv4 struct ip
@@ -557,24 +553,29 @@ static inline u32 __ipv6_addr_jhash(const struct in6_addr *a, const u32 initval)
 static inline bool ipv6_addr_loopback(const struct in6_addr *a)
 {
 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
-	const unsigned long *ul = (const unsigned long *)a;
+	const __be64 *be = (const __be64 *)a;
 
-	return (ul[0] | (ul[1] ^ cpu_to_be64(1))) == 0UL;
+	return (be[0] | (be[1] ^ cpu_to_be64(1))) == 0UL;
 #else
 	return (a->s6_addr32[0] | a->s6_addr32[1] |
-		a->s6_addr32[2] | (a->s6_addr32[3] ^ htonl(1))) == 0;
+		a->s6_addr32[2] | (a->s6_addr32[3] ^ cpu_to_be32(1))) == 0;
 #endif
 }
 
+/*
+ * Note that we must __force cast these to unsigned long to make sparse happy,
+ * since all of the endian-annotated types are fixed size regardless of arch.
+ */
 static inline bool ipv6_addr_v4mapped(const struct in6_addr *a)
 {
 	return (
 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
-		*(__be64 *)a |
+		*(unsigned long *)a |
 #else
-		(a->s6_addr32[0] | a->s6_addr32[1]) |
+		(__force unsigned long)(a->s6_addr32[0] | a->s6_addr32[1]) |
 #endif
-		(a->s6_addr32[2] ^ htonl(0x0000ffff))) == 0UL;
+		(__force unsigned long)(a->s6_addr32[2] ^
+					cpu_to_be32(0x0000ffff))) == 0UL;
 }
 
 /*
@@ -684,6 +685,50 @@ static inline int ip6_sk_dst_hoplimit(struct ipv6_pinfo *np, struct flowi6 *fl6,
 	return hlimit;
 }
 
+#if IS_ENABLED(CONFIG_IPV6)
+static inline void ip6_set_txhash(struct sock *sk)
+{
+	struct inet_sock *inet = inet_sk(sk);
+	struct ipv6_pinfo *np = inet6_sk(sk);
+	struct flow_keys keys;
+
+	keys.src = (__force __be32)ipv6_addr_hash(&np->saddr);
+	keys.dst = (__force __be32)ipv6_addr_hash(&sk->sk_v6_daddr);
+	keys.port16[0] = inet->inet_sport;
+	keys.port16[1] = inet->inet_dport;
+
+	sk->sk_txhash = flow_hash_from_keys(&keys);
+}
+
+static inline __be32 ip6_make_flowlabel(struct net *net, struct sk_buff *skb,
+					__be32 flowlabel, bool autolabel)
+{
+	if (!flowlabel && (autolabel || net->ipv6.sysctl.auto_flowlabels)) {
+		__be32 hash;
+
+		hash = skb_get_hash(skb);
+
+		/* Since this is being sent on the wire obfuscate hash a bit
+		 * to minimize possbility that any useful information to an
+		 * attacker is leaked. Only lower 20 bits are relevant.
+		 */
+		hash ^= hash >> 12;
+
+		flowlabel = hash & IPV6_FLOWLABEL_MASK;
+	}
+
+	return flowlabel;
+}
+#else
+static inline void ip6_set_txhash(struct sock *sk) { }
+static inline __be32 ip6_make_flowlabel(struct net *net, struct sk_buff *skb,
+					__be32 flowlabel, bool autolabel)
+{
+	return flowlabel;
+}
+#endif
+
+
 /*
  *	Header manipulation
  */