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

Pull networking updates from Davic Miller:

 1) Support busy polling generically, for all NAPI drivers.  From Eric
    Dumazet.

 2) Add byte/packet counter support to nft_ct, from Floriani Westphal.

 3) Add RSS/XPS support to mvneta driver, from Gregory Clement.

 4) Implement IPV6_HDRINCL socket option for raw sockets, from Hannes
    Frederic Sowa.

 5) Add support for T6 adapter to cxgb4 driver, from Hariprasad Shenai.

 6) Add support for VLAN device bridging to mlxsw switch driver, from
    Ido Schimmel.

 7) Add driver for Netronome NFP4000/NFP6000, from Jakub Kicinski.

 8) Provide hwmon interface to mlxsw switch driver, from Jiri Pirko.

 9) Reorganize wireless drivers into per-vendor directories just like we
    do for ethernet drivers.  From Kalle Valo.

10) Provide a way for administrators "destroy" connected sockets via the
    SOCK_DESTROY socket netlink diag operation.  From Lorenzo Colitti.

11) Add support to add/remove multicast routes via netlink, from Nikolay
    Aleksandrov.

12) Make TCP keepalive settings per-namespace, from Nikolay Borisov.

13) Add forwarding and packet duplication facilities to nf_tables, from
    Pablo Neira Ayuso.

14) Dead route support in MPLS, from Roopa Prabhu.

15) TSO support for thunderx chips, from Sunil Goutham.

16) Add driver for IBM's System i/p VNIC protocol, from Thomas Falcon.

17) Rationalize, consolidate, and more completely document the checksum
    offloading facilities in the networking stack.  From Tom Herbert.

18) Support aborting an ongoing scan in mac80211/cfg80211, from
    Vidyullatha Kanchanapally.

19) Use per-bucket spinlock for bpf hash facility, from Tom Leiming.

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-next: (1375 commits)
  net: bnxt: always return values from _bnxt_get_max_rings
  net: bpf: reject invalid shifts
  phonet: properly unshare skbs in phonet_rcv()
  dwc_eth_qos: Fix dma address for multi-fragment skbs
  phy: remove an unneeded condition
  mdio: remove an unneed condition
  mdio_bus: NULL dereference on allocation error
  net: Fix typo in netdev_intersect_features
  net: freescale: mac-fec: Fix build error from phy_device API change
  net: freescale: ucc_geth: Fix build error from phy_device API change
  bonding: Prevent IPv6 link local address on enslaved devices
  IB/mlx5: Add flow steering support
  net/mlx5_core: Export flow steering API
  net/mlx5_core: Make ipv4/ipv6 location more clear
  net/mlx5_core: Enable flow steering support for the IB driver
  net/mlx5_core: Initialize namespaces only when supported by device
  net/mlx5_core: Set priority attributes
  net/mlx5_core: Connect flow tables
  net/mlx5_core: Introduce modify flow table command
  net/mlx5_core: Managing root flow table
  ...

1 files changed, 251 insertions, 0 deletions

diff --git a/net/core/sock_reuseport.c b/net/core/sock_reuseport.c
new file mode 100644
index 000000000000..1df98c557440
--- /dev/null
+++ b/net/core/sock_reuseport.c
@@ -0,0 +1,251 @@
+/*
+ * To speed up listener socket lookup, create an array to store all sockets
+ * listening on the same port.  This allows a decision to be made after finding
+ * the first socket.  An optional BPF program can also be configured for
+ * selecting the socket index from the array of available sockets.
+ */
+
+#include <net/sock_reuseport.h>
+#include <linux/bpf.h>
+#include <linux/rcupdate.h>
+
+#define INIT_SOCKS 128
+
+static DEFINE_SPINLOCK(reuseport_lock);
+
+static struct sock_reuseport *__reuseport_alloc(u16 max_socks)
+{
+	size_t size = sizeof(struct sock_reuseport) +
+		      sizeof(struct sock *) * max_socks;
+	struct sock_reuseport *reuse = kzalloc(size, GFP_ATOMIC);
+
+	if (!reuse)
+		return NULL;
+
+	reuse->max_socks = max_socks;
+
+	RCU_INIT_POINTER(reuse->prog, NULL);
+	return reuse;
+}
+
+int reuseport_alloc(struct sock *sk)
+{
+	struct sock_reuseport *reuse;
+
+	/* bh lock used since this function call may precede hlist lock in
+	 * soft irq of receive path or setsockopt from process context
+	 */
+	spin_lock_bh(&reuseport_lock);
+	WARN_ONCE(rcu_dereference_protected(sk->sk_reuseport_cb,
+					    lockdep_is_held(&reuseport_lock)),
+		  "multiple allocations for the same socket");
+	reuse = __reuseport_alloc(INIT_SOCKS);
+	if (!reuse) {
+		spin_unlock_bh(&reuseport_lock);
+		return -ENOMEM;
+	}
+
+	reuse->socks[0] = sk;
+	reuse->num_socks = 1;
+	rcu_assign_pointer(sk->sk_reuseport_cb, reuse);
+
+	spin_unlock_bh(&reuseport_lock);
+
+	return 0;
+}
+EXPORT_SYMBOL(reuseport_alloc);
+
+static struct sock_reuseport *reuseport_grow(struct sock_reuseport *reuse)
+{
+	struct sock_reuseport *more_reuse;
+	u32 more_socks_size, i;
+
+	more_socks_size = reuse->max_socks * 2U;
+	if (more_socks_size > U16_MAX)
+		return NULL;
+
+	more_reuse = __reuseport_alloc(more_socks_size);
+	if (!more_reuse)
+		return NULL;
+
+	more_reuse->max_socks = more_socks_size;
+	more_reuse->num_socks = reuse->num_socks;
+	more_reuse->prog = reuse->prog;
+
+	memcpy(more_reuse->socks, reuse->socks,
+	       reuse->num_socks * sizeof(struct sock *));
+
+	for (i = 0; i < reuse->num_socks; ++i)
+		rcu_assign_pointer(reuse->socks[i]->sk_reuseport_cb,
+				   more_reuse);
+
+	/* Note: we use kfree_rcu here instead of reuseport_free_rcu so
+	 * that reuse and more_reuse can temporarily share a reference
+	 * to prog.
+	 */
+	kfree_rcu(reuse, rcu);
+	return more_reuse;
+}
+
+/**
+ *  reuseport_add_sock - Add a socket to the reuseport group of another.
+ *  @sk:  New socket to add to the group.
+ *  @sk2: Socket belonging to the existing reuseport group.
+ *  May return ENOMEM and not add socket to group under memory pressure.
+ */
+int reuseport_add_sock(struct sock *sk, const struct sock *sk2)
+{
+	struct sock_reuseport *reuse;
+
+	spin_lock_bh(&reuseport_lock);
+	reuse = rcu_dereference_protected(sk2->sk_reuseport_cb,
+					  lockdep_is_held(&reuseport_lock)),
+	WARN_ONCE(rcu_dereference_protected(sk->sk_reuseport_cb,
+					    lockdep_is_held(&reuseport_lock)),
+		  "socket already in reuseport group");
+
+	if (reuse->num_socks == reuse->max_socks) {
+		reuse = reuseport_grow(reuse);
+		if (!reuse) {
+			spin_unlock_bh(&reuseport_lock);
+			return -ENOMEM;
+		}
+	}
+
+	reuse->socks[reuse->num_socks] = sk;
+	/* paired with smp_rmb() in reuseport_select_sock() */
+	smp_wmb();
+	reuse->num_socks++;
+	rcu_assign_pointer(sk->sk_reuseport_cb, reuse);
+
+	spin_unlock_bh(&reuseport_lock);
+
+	return 0;
+}
+EXPORT_SYMBOL(reuseport_add_sock);
+
+static void reuseport_free_rcu(struct rcu_head *head)
+{
+	struct sock_reuseport *reuse;
+
+	reuse = container_of(head, struct sock_reuseport, rcu);
+	if (reuse->prog)
+		bpf_prog_destroy(reuse->prog);
+	kfree(reuse);
+}
+
+void reuseport_detach_sock(struct sock *sk)
+{
+	struct sock_reuseport *reuse;
+	int i;
+
+	spin_lock_bh(&reuseport_lock);
+	reuse = rcu_dereference_protected(sk->sk_reuseport_cb,
+					  lockdep_is_held(&reuseport_lock));
+	rcu_assign_pointer(sk->sk_reuseport_cb, NULL);
+
+	for (i = 0; i < reuse->num_socks; i++) {
+		if (reuse->socks[i] == sk) {
+			reuse->socks[i] = reuse->socks[reuse->num_socks - 1];
+			reuse->num_socks--;
+			if (reuse->num_socks == 0)
+				call_rcu(&reuse->rcu, reuseport_free_rcu);
+			break;
+		}
+	}
+	spin_unlock_bh(&reuseport_lock);
+}
+EXPORT_SYMBOL(reuseport_detach_sock);
+
+static struct sock *run_bpf(struct sock_reuseport *reuse, u16 socks,
+			    struct bpf_prog *prog, struct sk_buff *skb,
+			    int hdr_len)
+{
+	struct sk_buff *nskb = NULL;
+	u32 index;
+
+	if (skb_shared(skb)) {
+		nskb = skb_clone(skb, GFP_ATOMIC);
+		if (!nskb)
+			return NULL;
+		skb = nskb;
+	}
+
+	/* temporarily advance data past protocol header */
+	if (!pskb_pull(skb, hdr_len)) {
+		kfree_skb(nskb);
+		return NULL;
+	}
+	index = bpf_prog_run_save_cb(prog, skb);
+	__skb_push(skb, hdr_len);
+
+	consume_skb(nskb);
+
+	if (index >= socks)
+		return NULL;
+
+	return reuse->socks[index];
+}
+
+/**
+ *  reuseport_select_sock - Select a socket from an SO_REUSEPORT group.
+ *  @sk: First socket in the group.
+ *  @hash: When no BPF filter is available, use this hash to select.
+ *  @skb: skb to run through BPF filter.
+ *  @hdr_len: BPF filter expects skb data pointer at payload data.  If
+ *    the skb does not yet point at the payload, this parameter represents
+ *    how far the pointer needs to advance to reach the payload.
+ *  Returns a socket that should receive the packet (or NULL on error).
+ */
+struct sock *reuseport_select_sock(struct sock *sk,
+				   u32 hash,
+				   struct sk_buff *skb,
+				   int hdr_len)
+{
+	struct sock_reuseport *reuse;
+	struct bpf_prog *prog;
+	struct sock *sk2 = NULL;
+	u16 socks;
+
+	rcu_read_lock();
+	reuse = rcu_dereference(sk->sk_reuseport_cb);
+
+	/* if memory allocation failed or add call is not yet complete */
+	if (!reuse)
+		goto out;
+
+	prog = rcu_dereference(reuse->prog);
+	socks = READ_ONCE(reuse->num_socks);
+	if (likely(socks)) {
+		/* paired with smp_wmb() in reuseport_add_sock() */
+		smp_rmb();
+
+		if (prog && skb)
+			sk2 = run_bpf(reuse, socks, prog, skb, hdr_len);
+		else
+			sk2 = reuse->socks[reciprocal_scale(hash, socks)];
+	}
+
+out:
+	rcu_read_unlock();
+	return sk2;
+}
+EXPORT_SYMBOL(reuseport_select_sock);
+
+struct bpf_prog *
+reuseport_attach_prog(struct sock *sk, struct bpf_prog *prog)
+{
+	struct sock_reuseport *reuse;
+	struct bpf_prog *old_prog;
+
+	spin_lock_bh(&reuseport_lock);
+	reuse = rcu_dereference_protected(sk->sk_reuseport_cb,
+					  lockdep_is_held(&reuseport_lock));
+	old_prog = rcu_dereference_protected(reuse->prog,
+					     lockdep_is_held(&reuseport_lock));
+	rcu_assign_pointer(reuse->prog, prog);
+	spin_unlock_bh(&reuseport_lock);
+
+	return old_prog;
+}
+EXPORT_SYMBOL(reuseport_attach_prog);