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

Pull networking updates from David Miller:
 "Most notable changes in here:

   1) By far the biggest accomplishment, thanks to a large range of
      contributors, is the addition of multi-send for transmit.  This is
      the result of discussions back in Chicago, and the hard work of
      several individuals.

      Now, when the ->ndo_start_xmit() method of a driver sees
      skb->xmit_more as true, it can choose to defer the doorbell
      telling the driver to start processing the new TX queue entires.

      skb->xmit_more means that the generic networking is guaranteed to
      call the driver immediately with another SKB to send.

      There is logic added to the qdisc layer to dequeue multiple
      packets at a time, and the handling mis-predicted offloads in
      software is now done with no locks held.

      Finally, pktgen is extended to have a "burst" parameter that can
      be used to test a multi-send implementation.

      Several drivers have xmit_more support: i40e, igb, ixgbe, mlx4,
      virtio_net

      Adding support is almost trivial, so export more drivers to
      support this optimization soon.

      I want to thank, in no particular or implied order, Jesper
      Dangaard Brouer, Eric Dumazet, Alexander Duyck, Tom Herbert, Jamal
      Hadi Salim, John Fastabend, Florian Westphal, Daniel Borkmann,
      David Tat, Hannes Frederic Sowa, and Rusty Russell.

   2) PTP and timestamping support in bnx2x, from Michal Kalderon.

   3) Allow adjusting the rx_copybreak threshold for a driver via
      ethtool, and add rx_copybreak support to enic driver.  From
      Govindarajulu Varadarajan.

   4) Significant enhancements to the generic PHY layer and the bcm7xxx
      driver in particular (EEE support, auto power down, etc.) from
      Florian Fainelli.

   5) Allow raw buffers to be used for flow dissection, allowing drivers
      to determine the optimal "linear pull" size for devices that DMA
      into pools of pages.  The objective is to get exactly the
      necessary amount of headers into the linear SKB area pre-pulled,
      but no more.  The new interface drivers use is eth_get_headlen().
      From WANG Cong, with driver conversions (several had their own
      by-hand duplicated implementations) by Alexander Duyck and Eric
      Dumazet.

   6) Support checksumming more smoothly and efficiently for
      encapsulations, and add "foo over UDP" facility.  From Tom
      Herbert.

   7) Add Broadcom SF2 switch driver to DSA layer, from Florian
      Fainelli.

   8) eBPF now can load programs via a system call and has an extensive
      testsuite.  Alexei Starovoitov and Daniel Borkmann.

   9) Major overhaul of the packet scheduler to use RCU in several major
      areas such as the classifiers and rate estimators.  From John
      Fastabend.

  10) Add driver for Intel FM10000 Ethernet Switch, from Alexander
      Duyck.

  11) Rearrange TCP_SKB_CB() to reduce cache line misses, from Eric
      Dumazet.

  12) Add Datacenter TCP congestion control algorithm support, From
      Florian Westphal.

  13) Reorganize sk_buff so that __copy_skb_header() is significantly
      faster.  From Eric Dumazet"

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-next: (1558 commits)
  netlabel: directly return netlbl_unlabel_genl_init()
  net: add netdev_txq_bql_{enqueue, complete}_prefetchw() helpers
  net: description of dma_cookie cause make xmldocs warning
  cxgb4: clean up a type issue
  cxgb4: potential shift wrapping bug
  i40e: skb->xmit_more support
  net: fs_enet: Add NAPI TX
  net: fs_enet: Remove non NAPI RX
  r8169:add support for RTL8168EP
  net_sched: copy exts->type in tcf_exts_change()
  wimax: convert printk to pr_foo()
  af_unix: remove 0 assignment on static
  ipv6: Do not warn for informational ICMP messages, regardless of type.
  Update Intel Ethernet Driver maintainers list
  bridge: Save frag_max_size between PRE_ROUTING and POST_ROUTING
  tipc: fix bug in multicast congestion handling
  net: better IFF_XMIT_DST_RELEASE support
  net/mlx4_en: remove NETDEV_TX_BUSY
  3c59x: fix bad split of cpu_to_le32(pci_map_single())
  net: bcmgenet: fix Tx ring priority programming
  ...

1 files changed, 224 insertions, 37 deletions

diff --git a/net/openvswitch/actions.c b/net/openvswitch/actions.c
index 5231652a95d9..006886dbee36 100644
--- a/net/openvswitch/actions.c
+++ b/net/openvswitch/actions.c
@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2007-2013 Nicira, Inc.
+ * Copyright (c) 2007-2014 Nicira, Inc.
  *
  * This program is free software; you can redistribute it and/or
  * modify it under the terms of version 2 of the GNU General Public
@@ -35,11 +35,78 @@
 #include <net/sctp/checksum.h>
 
 #include "datapath.h"
+#include "flow.h"
 #include "vport.h"
 
 static int do_execute_actions(struct datapath *dp, struct sk_buff *skb,
+			      struct sw_flow_key *key,
 			      const struct nlattr *attr, int len);
 
+struct deferred_action {
+	struct sk_buff *skb;
+	const struct nlattr *actions;
+
+	/* Store pkt_key clone when creating deferred action. */
+	struct sw_flow_key pkt_key;
+};
+
+#define DEFERRED_ACTION_FIFO_SIZE 10
+struct action_fifo {
+	int head;
+	int tail;
+	/* Deferred action fifo queue storage. */
+	struct deferred_action fifo[DEFERRED_ACTION_FIFO_SIZE];
+};
+
+static struct action_fifo __percpu *action_fifos;
+static DEFINE_PER_CPU(int, exec_actions_level);
+
+static void action_fifo_init(struct action_fifo *fifo)
+{
+	fifo->head = 0;
+	fifo->tail = 0;
+}
+
+static bool action_fifo_is_empty(struct action_fifo *fifo)
+{
+	return (fifo->head == fifo->tail);
+}
+
+static struct deferred_action *action_fifo_get(struct action_fifo *fifo)
+{
+	if (action_fifo_is_empty(fifo))
+		return NULL;
+
+	return &fifo->fifo[fifo->tail++];
+}
+
+static struct deferred_action *action_fifo_put(struct action_fifo *fifo)
+{
+	if (fifo->head >= DEFERRED_ACTION_FIFO_SIZE - 1)
+		return NULL;
+
+	return &fifo->fifo[fifo->head++];
+}
+
+/* Return true if fifo is not full */
+static struct deferred_action *add_deferred_actions(struct sk_buff *skb,
+						    struct sw_flow_key *key,
+						    const struct nlattr *attr)
+{
+	struct action_fifo *fifo;
+	struct deferred_action *da;
+
+	fifo = this_cpu_ptr(action_fifos);
+	da = action_fifo_put(fifo);
+	if (da) {
+		da->skb = skb;
+		da->actions = attr;
+		da->pkt_key = *key;
+	}
+
+	return da;
+}
+
 static int make_writable(struct sk_buff *skb, int write_len)
 {
 	if (!pskb_may_pull(skb, write_len))
@@ -410,16 +477,14 @@ static int do_output(struct datapath *dp, struct sk_buff *skb, int out_port)
 }
 
 static int output_userspace(struct datapath *dp, struct sk_buff *skb,
-			    const struct nlattr *attr)
+			    struct sw_flow_key *key, const struct nlattr *attr)
 {
 	struct dp_upcall_info upcall;
 	const struct nlattr *a;
 	int rem;
 
-	BUG_ON(!OVS_CB(skb)->pkt_key);
-
 	upcall.cmd = OVS_PACKET_CMD_ACTION;
-	upcall.key = OVS_CB(skb)->pkt_key;
+	upcall.key = key;
 	upcall.userdata = NULL;
 	upcall.portid = 0;
 
@@ -445,11 +510,10 @@ static bool last_action(const struct nlattr *a, int rem)
 }
 
 static int sample(struct datapath *dp, struct sk_buff *skb,
-		  const struct nlattr *attr)
+		  struct sw_flow_key *key, const struct nlattr *attr)
 {
 	const struct nlattr *acts_list = NULL;
 	const struct nlattr *a;
-	struct sk_buff *sample_skb;
 	int rem;
 
 	for (a = nla_data(attr), rem = nla_len(attr); rem > 0;
@@ -469,31 +533,47 @@ static int sample(struct datapath *dp, struct sk_buff *skb,
 	rem = nla_len(acts_list);
 	a = nla_data(acts_list);
 
-	/* Actions list is either empty or only contains a single user-space
-	 * action, the latter being a special case as it is the only known
-	 * usage of the sample action.
-	 * In these special cases don't clone the skb as there are no
-	 * side-effects in the nested actions.
-	 * Otherwise, clone in case the nested actions have side effects.
+	/* Actions list is empty, do nothing */
+	if (unlikely(!rem))
+		return 0;
+
+	/* The only known usage of sample action is having a single user-space
+	 * action. Treat this usage as a special case.
+	 * The output_userspace() should clone the skb to be sent to the
+	 * user space. This skb will be consumed by its caller.
 	 */
-	if (likely(rem == 0 || (nla_type(a) == OVS_ACTION_ATTR_USERSPACE &&
-				last_action(a, rem)))) {
-		sample_skb = skb;
-		skb_get(skb);
-	} else {
-		sample_skb = skb_clone(skb, GFP_ATOMIC);
-		if (!sample_skb) /* Skip sample action when out of memory. */
-			return 0;
+	if (likely(nla_type(a) == OVS_ACTION_ATTR_USERSPACE &&
+		   last_action(a, rem)))
+		return output_userspace(dp, skb, key, a);
+
+	skb = skb_clone(skb, GFP_ATOMIC);
+	if (!skb)
+		/* Skip the sample action when out of memory. */
+		return 0;
+
+	if (!add_deferred_actions(skb, key, a)) {
+		if (net_ratelimit())
+			pr_warn("%s: deferred actions limit reached, dropping sample action\n",
+				ovs_dp_name(dp));
+
+		kfree_skb(skb);
 	}
+	return 0;
+}
 
-	/* Note that do_execute_actions() never consumes skb.
-	 * In the case where skb has been cloned above it is the clone that
-	 * is consumed.  Otherwise the skb_get(skb) call prevents
-	 * consumption by do_execute_actions(). Thus, it is safe to simply
-	 * return the error code and let the caller (also
-	 * do_execute_actions()) free skb on error.
-	 */
-	return do_execute_actions(dp, sample_skb, a, rem);
+static void execute_hash(struct sk_buff *skb, struct sw_flow_key *key,
+			 const struct nlattr *attr)
+{
+	struct ovs_action_hash *hash_act = nla_data(attr);
+	u32 hash = 0;
+
+	/* OVS_HASH_ALG_L4 is the only possible hash algorithm.  */
+	hash = skb_get_hash(skb);
+	hash = jhash_1word(hash, hash_act->hash_basis);
+	if (!hash)
+		hash = 0x1;
+
+	key->ovs_flow_hash = hash;
 }
 
 static int execute_set_action(struct sk_buff *skb,
@@ -510,8 +590,8 @@ static int execute_set_action(struct sk_buff *skb,
 		skb->mark = nla_get_u32(nested_attr);
 		break;
 
-	case OVS_KEY_ATTR_IPV4_TUNNEL:
-		OVS_CB(skb)->tun_key = nla_data(nested_attr);
+	case OVS_KEY_ATTR_TUNNEL_INFO:
+		OVS_CB(skb)->egress_tun_info = nla_data(nested_attr);
 		break;
 
 	case OVS_KEY_ATTR_ETHERNET:
@@ -542,8 +622,47 @@ static int execute_set_action(struct sk_buff *skb,
 	return err;
 }
 
+static int execute_recirc(struct datapath *dp, struct sk_buff *skb,
+			  struct sw_flow_key *key,
+			  const struct nlattr *a, int rem)
+{
+	struct deferred_action *da;
+	int err;
+
+	err = ovs_flow_key_update(skb, key);
+	if (err)
+		return err;
+
+	if (!last_action(a, rem)) {
+		/* Recirc action is the not the last action
+		 * of the action list, need to clone the skb.
+		 */
+		skb = skb_clone(skb, GFP_ATOMIC);
+
+		/* Skip the recirc action when out of memory, but
+		 * continue on with the rest of the action list.
+		 */
+		if (!skb)
+			return 0;
+	}
+
+	da = add_deferred_actions(skb, key, NULL);
+	if (da) {
+		da->pkt_key.recirc_id = nla_get_u32(a);
+	} else {
+		kfree_skb(skb);
+
+		if (net_ratelimit())
+			pr_warn("%s: deferred action limit reached, drop recirc action\n",
+				ovs_dp_name(dp));
+	}
+
+	return 0;
+}
+
 /* Execute a list of actions against 'skb'. */
 static int do_execute_actions(struct datapath *dp, struct sk_buff *skb,
+			      struct sw_flow_key *key,
 			      const struct nlattr *attr, int len)
 {
 	/* Every output action needs a separate clone of 'skb', but the common
@@ -569,7 +688,11 @@ static int do_execute_actions(struct datapath *dp, struct sk_buff *skb,
 			break;
 
 		case OVS_ACTION_ATTR_USERSPACE:
-			output_userspace(dp, skb, a);
+			output_userspace(dp, skb, key, a);
+			break;
+
+		case OVS_ACTION_ATTR_HASH:
+			execute_hash(skb, key, a);
 			break;
 
 		case OVS_ACTION_ATTR_PUSH_VLAN:
@@ -582,12 +705,23 @@ static int do_execute_actions(struct datapath *dp, struct sk_buff *skb,
 			err = pop_vlan(skb);
 			break;
 
+		case OVS_ACTION_ATTR_RECIRC:
+			err = execute_recirc(dp, skb, key, a, rem);
+			if (last_action(a, rem)) {
+				/* If this is the last action, the skb has
+				 * been consumed or freed.
+				 * Return immediately.
+				 */
+				return err;
+			}
+			break;
+
 		case OVS_ACTION_ATTR_SET:
 			err = execute_set_action(skb, nla_data(a));
 			break;
 
 		case OVS_ACTION_ATTR_SAMPLE:
-			err = sample(dp, skb, a);
+			err = sample(dp, skb, key, a);
 			if (unlikely(err)) /* skb already freed. */
 				return err;
 			break;
@@ -607,11 +741,64 @@ static int do_execute_actions(struct datapath *dp, struct sk_buff *skb,
 	return 0;
 }
 
+static void process_deferred_actions(struct datapath *dp)
+{
+	struct action_fifo *fifo = this_cpu_ptr(action_fifos);
+
+	/* Do not touch the FIFO in case there is no deferred actions. */
+	if (action_fifo_is_empty(fifo))
+		return;
+
+	/* Finishing executing all deferred actions. */
+	do {
+		struct deferred_action *da = action_fifo_get(fifo);
+		struct sk_buff *skb = da->skb;
+		struct sw_flow_key *key = &da->pkt_key;
+		const struct nlattr *actions = da->actions;
+
+		if (actions)
+			do_execute_actions(dp, skb, key, actions,
+					   nla_len(actions));
+		else
+			ovs_dp_process_packet(skb, key);
+	} while (!action_fifo_is_empty(fifo));
+
+	/* Reset FIFO for the next packet.  */
+	action_fifo_init(fifo);
+}
+
 /* Execute a list of actions against 'skb'. */
-int ovs_execute_actions(struct datapath *dp, struct sk_buff *skb)
+int ovs_execute_actions(struct datapath *dp, struct sk_buff *skb,
+			struct sw_flow_key *key)
 {
-	struct sw_flow_actions *acts = rcu_dereference(OVS_CB(skb)->flow->sf_acts);
+	int level = this_cpu_read(exec_actions_level);
+	struct sw_flow_actions *acts;
+	int err;
+
+	acts = rcu_dereference(OVS_CB(skb)->flow->sf_acts);
+
+	this_cpu_inc(exec_actions_level);
+	OVS_CB(skb)->egress_tun_info = NULL;
+	err = do_execute_actions(dp, skb, key,
+				 acts->actions, acts->actions_len);
 
-	OVS_CB(skb)->tun_key = NULL;
-	return do_execute_actions(dp, skb, acts->actions, acts->actions_len);
+	if (!level)
+		process_deferred_actions(dp);
+
+	this_cpu_dec(exec_actions_level);
+	return err;
+}
+
+int action_fifos_init(void)
+{
+	action_fifos = alloc_percpu(struct action_fifo);
+	if (!action_fifos)
+		return -ENOMEM;
+
+	return 0;
+}
+
+void action_fifos_exit(void)
+{
+	free_percpu(action_fifos);
 }