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  bridge: Update outdated comment on promiscuous mode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, 190 insertions, 55 deletions

diff --git a/drivers/net/ethernet/cisco/enic/enic_main.c b/drivers/net/ethernet/cisco/enic/enic_main.c
index f32f828b7f3d..9348febc0743 100644
--- a/drivers/net/ethernet/cisco/enic/enic_main.c
+++ b/drivers/net/ethernet/cisco/enic/enic_main.c
@@ -39,6 +39,12 @@
 #include <linux/prefetch.h>
 #include <net/ip6_checksum.h>
 #include <linux/ktime.h>
+#ifdef CONFIG_RFS_ACCEL
+#include <linux/cpu_rmap.h>
+#endif
+#ifdef CONFIG_NET_RX_BUSY_POLL
+#include <net/busy_poll.h>
+#endif
 
 #include "cq_enet_desc.h"
 #include "vnic_dev.h"
@@ -49,6 +55,7 @@
 #include "enic.h"
 #include "enic_dev.h"
 #include "enic_pp.h"
+#include "enic_clsf.h"
 
 #define ENIC_NOTIFY_TIMER_PERIOD	(2 * HZ)
 #define WQ_ENET_MAX_DESC_LEN		(1 << WQ_ENET_LEN_BITS)
@@ -309,40 +316,15 @@ static irqreturn_t enic_isr_msi(int irq, void *data)
 	return IRQ_HANDLED;
 }
 
-static irqreturn_t enic_isr_msix_rq(int irq, void *data)
+static irqreturn_t enic_isr_msix(int irq, void *data)
 {
 	struct napi_struct *napi = data;
 
-	/* schedule NAPI polling for RQ cleanup */
 	napi_schedule(napi);
 
 	return IRQ_HANDLED;
 }
 
-static irqreturn_t enic_isr_msix_wq(int irq, void *data)
-{
-	struct enic *enic = data;
-	unsigned int cq;
-	unsigned int intr;
-	unsigned int wq_work_to_do = -1; /* no limit */
-	unsigned int wq_work_done;
-	unsigned int wq_irq;
-
-	wq_irq = (u32)irq - enic->msix_entry[enic_msix_wq_intr(enic, 0)].vector;
-	cq = enic_cq_wq(enic, wq_irq);
-	intr = enic_msix_wq_intr(enic, wq_irq);
-
-	wq_work_done = vnic_cq_service(&enic->cq[cq],
-		wq_work_to_do, enic_wq_service, NULL);
-
-	vnic_intr_return_credits(&enic->intr[intr],
-		wq_work_done,
-		1 /* unmask intr */,
-		1 /* reset intr timer */);
-
-	return IRQ_HANDLED;
-}
-
 static irqreturn_t enic_isr_msix_err(int irq, void *data)
 {
 	struct enic *enic = data;
@@ -1049,10 +1031,12 @@ static void enic_rq_indicate_buf(struct vnic_rq *rq,
 		if (vlan_stripped)
 			__vlan_hwaccel_put_tag(skb, htons(ETH_P_8021Q), vlan_tci);
 
-		if (netdev->features & NETIF_F_GRO)
-			napi_gro_receive(&enic->napi[q_number], skb);
-		else
+		skb_mark_napi_id(skb, &enic->napi[rq->index]);
+		if (enic_poll_busy_polling(rq) ||
+		    !(netdev->features & NETIF_F_GRO))
 			netif_receive_skb(skb);
+		else
+			napi_gro_receive(&enic->napi[q_number], skb);
 		if (enic->rx_coalesce_setting.use_adaptive_rx_coalesce)
 			enic_intr_update_pkt_size(&cq->pkt_size_counter,
 						  bytes_written);
@@ -1089,16 +1073,22 @@ static int enic_poll(struct napi_struct *napi, int budget)
 	unsigned int  work_done, rq_work_done = 0, wq_work_done;
 	int err;
 
-	/* Service RQ (first) and WQ
-	 */
+	wq_work_done = vnic_cq_service(&enic->cq[cq_wq], wq_work_to_do,
+				       enic_wq_service, NULL);
+
+	if (!enic_poll_lock_napi(&enic->rq[cq_rq])) {
+		if (wq_work_done > 0)
+			vnic_intr_return_credits(&enic->intr[intr],
+						 wq_work_done,
+						 0 /* dont unmask intr */,
+						 0 /* dont reset intr timer */);
+		return rq_work_done;
+	}
 
 	if (budget > 0)
 		rq_work_done = vnic_cq_service(&enic->cq[cq_rq],
 			rq_work_to_do, enic_rq_service, NULL);
 
-	wq_work_done = vnic_cq_service(&enic->cq[cq_wq],
-		wq_work_to_do, enic_wq_service, NULL);
-
 	/* Accumulate intr event credits for this polling
 	 * cycle.  An intr event is the completion of a
 	 * a WQ or RQ packet.
@@ -1130,6 +1120,7 @@ static int enic_poll(struct napi_struct *napi, int budget)
 		napi_complete(napi);
 		vnic_intr_unmask(&enic->intr[intr]);
 	}
+	enic_poll_unlock_napi(&enic->rq[cq_rq]);
 
 	return rq_work_done;
 }
@@ -1192,7 +1183,102 @@ static void enic_calc_int_moderation(struct enic *enic, struct vnic_rq *rq)
 	pkt_size_counter->small_pkt_bytes_cnt = 0;
 }
 
-static int enic_poll_msix(struct napi_struct *napi, int budget)
+#ifdef CONFIG_RFS_ACCEL
+static void enic_free_rx_cpu_rmap(struct enic *enic)
+{
+	free_irq_cpu_rmap(enic->netdev->rx_cpu_rmap);
+	enic->netdev->rx_cpu_rmap = NULL;
+}
+
+static void enic_set_rx_cpu_rmap(struct enic *enic)
+{
+	int i, res;
+
+	if (vnic_dev_get_intr_mode(enic->vdev) == VNIC_DEV_INTR_MODE_MSIX) {
+		enic->netdev->rx_cpu_rmap = alloc_irq_cpu_rmap(enic->rq_count);
+		if (unlikely(!enic->netdev->rx_cpu_rmap))
+			return;
+		for (i = 0; i < enic->rq_count; i++) {
+			res = irq_cpu_rmap_add(enic->netdev->rx_cpu_rmap,
+					       enic->msix_entry[i].vector);
+			if (unlikely(res)) {
+				enic_free_rx_cpu_rmap(enic);
+				return;
+			}
+		}
+	}
+}
+
+#else
+
+static void enic_free_rx_cpu_rmap(struct enic *enic)
+{
+}
+
+static void enic_set_rx_cpu_rmap(struct enic *enic)
+{
+}
+
+#endif /* CONFIG_RFS_ACCEL */
+
+#ifdef CONFIG_NET_RX_BUSY_POLL
+int enic_busy_poll(struct napi_struct *napi)
+{
+	struct net_device *netdev = napi->dev;
+	struct enic *enic = netdev_priv(netdev);
+	unsigned int rq = (napi - &enic->napi[0]);
+	unsigned int cq = enic_cq_rq(enic, rq);
+	unsigned int intr = enic_msix_rq_intr(enic, rq);
+	unsigned int work_to_do = -1; /* clean all pkts possible */
+	unsigned int work_done;
+
+	if (!enic_poll_lock_poll(&enic->rq[rq]))
+		return LL_FLUSH_BUSY;
+	work_done = vnic_cq_service(&enic->cq[cq], work_to_do,
+				    enic_rq_service, NULL);
+
+	if (work_done > 0)
+		vnic_intr_return_credits(&enic->intr[intr],
+					 work_done, 0, 0);
+	vnic_rq_fill(&enic->rq[rq], enic_rq_alloc_buf);
+	if (enic->rx_coalesce_setting.use_adaptive_rx_coalesce)
+		enic_calc_int_moderation(enic, &enic->rq[rq]);
+	enic_poll_unlock_poll(&enic->rq[rq]);
+
+	return work_done;
+}
+#endif /* CONFIG_NET_RX_BUSY_POLL */
+
+static int enic_poll_msix_wq(struct napi_struct *napi, int budget)
+{
+	struct net_device *netdev = napi->dev;
+	struct enic *enic = netdev_priv(netdev);
+	unsigned int wq_index = (napi - &enic->napi[0]) - enic->rq_count;
+	struct vnic_wq *wq = &enic->wq[wq_index];
+	unsigned int cq;
+	unsigned int intr;
+	unsigned int wq_work_to_do = -1; /* clean all desc possible */
+	unsigned int wq_work_done;
+	unsigned int wq_irq;
+
+	wq_irq = wq->index;
+	cq = enic_cq_wq(enic, wq_irq);
+	intr = enic_msix_wq_intr(enic, wq_irq);
+	wq_work_done = vnic_cq_service(&enic->cq[cq], wq_work_to_do,
+				       enic_wq_service, NULL);
+
+	vnic_intr_return_credits(&enic->intr[intr], wq_work_done,
+				 0 /* don't unmask intr */,
+				 1 /* reset intr timer */);
+	if (!wq_work_done) {
+		napi_complete(napi);
+		vnic_intr_unmask(&enic->intr[intr]);
+	}
+
+	return 0;
+}
+
+static int enic_poll_msix_rq(struct napi_struct *napi, int budget)
 {
 	struct net_device *netdev = napi->dev;
 	struct enic *enic = netdev_priv(netdev);
@@ -1203,6 +1289,8 @@ static int enic_poll_msix(struct napi_struct *napi, int budget)
 	unsigned int work_done = 0;
 	int err;
 
+	if (!enic_poll_lock_napi(&enic->rq[rq]))
+		return work_done;
 	/* Service RQ
 	 */
 
@@ -1248,6 +1336,7 @@ static int enic_poll_msix(struct napi_struct *napi, int budget)
 			enic_set_int_moderation(enic, &enic->rq[rq]);
 		vnic_intr_unmask(&enic->intr[intr]);
 	}
+	enic_poll_unlock_napi(&enic->rq[rq]);
 
 	return work_done;
 }
@@ -1267,6 +1356,7 @@ static void enic_free_intr(struct enic *enic)
 	struct net_device *netdev = enic->netdev;
 	unsigned int i;
 
+	enic_free_rx_cpu_rmap(enic);
 	switch (vnic_dev_get_intr_mode(enic->vdev)) {
 	case VNIC_DEV_INTR_MODE_INTX:
 		free_irq(enic->pdev->irq, netdev);
@@ -1291,6 +1381,7 @@ static int enic_request_intr(struct enic *enic)
 	unsigned int i, intr;
 	int err = 0;
 
+	enic_set_rx_cpu_rmap(enic);
 	switch (vnic_dev_get_intr_mode(enic->vdev)) {
 
 	case VNIC_DEV_INTR_MODE_INTX:
@@ -1312,17 +1403,19 @@ static int enic_request_intr(struct enic *enic)
 			snprintf(enic->msix[intr].devname,
 				sizeof(enic->msix[intr].devname),
 				"%.11s-rx-%d", netdev->name, i);
-			enic->msix[intr].isr = enic_isr_msix_rq;
+			enic->msix[intr].isr = enic_isr_msix;
 			enic->msix[intr].devid = &enic->napi[i];
 		}
 
 		for (i = 0; i < enic->wq_count; i++) {
+			int wq = enic_cq_wq(enic, i);
+
 			intr = enic_msix_wq_intr(enic, i);
 			snprintf(enic->msix[intr].devname,
 				sizeof(enic->msix[intr].devname),
 				"%.11s-tx-%d", netdev->name, i);
-			enic->msix[intr].isr = enic_isr_msix_wq;
-			enic->msix[intr].devid = enic;
+			enic->msix[intr].isr = enic_isr_msix;
+			enic->msix[intr].devid = &enic->napi[wq];
 		}
 
 		intr = enic_msix_err_intr(enic);
@@ -1421,7 +1514,7 @@ static int enic_dev_notify_set(struct enic *enic)
 {
 	int err;
 
-	spin_lock(&enic->devcmd_lock);
+	spin_lock_bh(&enic->devcmd_lock);
 	switch (vnic_dev_get_intr_mode(enic->vdev)) {
 	case VNIC_DEV_INTR_MODE_INTX:
 		err = vnic_dev_notify_set(enic->vdev,
@@ -1435,7 +1528,7 @@ static int enic_dev_notify_set(struct enic *enic)
 		err = vnic_dev_notify_set(enic->vdev, -1 /* no intr */);
 		break;
 	}
-	spin_unlock(&enic->devcmd_lock);
+	spin_unlock_bh(&enic->devcmd_lock);
 
 	return err;
 }
@@ -1494,15 +1587,20 @@ static int enic_open(struct net_device *netdev)
 
 	netif_tx_wake_all_queues(netdev);
 
-	for (i = 0; i < enic->rq_count; i++)
+	for (i = 0; i < enic->rq_count; i++) {
+		enic_busy_poll_init_lock(&enic->rq[i]);
 		napi_enable(&enic->napi[i]);
-
+	}
+	if (vnic_dev_get_intr_mode(enic->vdev) == VNIC_DEV_INTR_MODE_MSIX)
+		for (i = 0; i < enic->wq_count; i++)
+			napi_enable(&enic->napi[enic_cq_wq(enic, i)]);
 	enic_dev_enable(enic);
 
 	for (i = 0; i < enic->intr_count; i++)
 		vnic_intr_unmask(&enic->intr[i]);
 
 	enic_notify_timer_start(enic);
+	enic_rfs_flw_tbl_init(enic);
 
 	return 0;
 
@@ -1529,14 +1627,23 @@ static int enic_stop(struct net_device *netdev)
 	enic_synchronize_irqs(enic);
 
 	del_timer_sync(&enic->notify_timer);
+	enic_rfs_flw_tbl_free(enic);
 
 	enic_dev_disable(enic);
 
-	for (i = 0; i < enic->rq_count; i++)
+	local_bh_disable();
+	for (i = 0; i < enic->rq_count; i++) {
 		napi_disable(&enic->napi[i]);
+		while (!enic_poll_lock_napi(&enic->rq[i]))
+			mdelay(1);
+	}
+	local_bh_enable();
 
 	netif_carrier_off(netdev);
 	netif_tx_disable(netdev);
+	if (vnic_dev_get_intr_mode(enic->vdev) == VNIC_DEV_INTR_MODE_MSIX)
+		for (i = 0; i < enic->wq_count; i++)
+			napi_disable(&enic->napi[enic_cq_wq(enic, i)]);
 
 	if (!enic_is_dynamic(enic) && !enic_is_sriov_vf(enic))
 		enic_dev_del_station_addr(enic);
@@ -1656,13 +1763,14 @@ static void enic_poll_controller(struct net_device *netdev)
 	case VNIC_DEV_INTR_MODE_MSIX:
 		for (i = 0; i < enic->rq_count; i++) {
 			intr = enic_msix_rq_intr(enic, i);
-			enic_isr_msix_rq(enic->msix_entry[intr].vector,
-				&enic->napi[i]);
+			enic_isr_msix(enic->msix_entry[intr].vector,
+				      &enic->napi[i]);
 		}
 
 		for (i = 0; i < enic->wq_count; i++) {
 			intr = enic_msix_wq_intr(enic, i);
-			enic_isr_msix_wq(enic->msix_entry[intr].vector, enic);
+			enic_isr_msix(enic->msix_entry[intr].vector,
+				      &enic->napi[enic_cq_wq(enic, i)]);
 		}
 
 		break;
@@ -1758,11 +1866,11 @@ static int enic_set_rsskey(struct enic *enic)
 
 	memcpy(rss_key_buf_va, &rss_key, sizeof(union vnic_rss_key));
 
-	spin_lock(&enic->devcmd_lock);
+	spin_lock_bh(&enic->devcmd_lock);
 	err = enic_set_rss_key(enic,
 		rss_key_buf_pa,
 		sizeof(union vnic_rss_key));
-	spin_unlock(&enic->devcmd_lock);
+	spin_unlock_bh(&enic->devcmd_lock);
 
 	pci_free_consistent(enic->pdev, sizeof(union vnic_rss_key),
 		rss_key_buf_va, rss_key_buf_pa);
@@ -1785,11 +1893,11 @@ static int enic_set_rsscpu(struct enic *enic, u8 rss_hash_bits)
 	for (i = 0; i < (1 << rss_hash_bits); i++)
 		(*rss_cpu_buf_va).cpu[i/4].b[i%4] = i % enic->rq_count;
 
-	spin_lock(&enic->devcmd_lock);
+	spin_lock_bh(&enic->devcmd_lock);
 	err = enic_set_rss_cpu(enic,
 		rss_cpu_buf_pa,
 		sizeof(union vnic_rss_cpu));
-	spin_unlock(&enic->devcmd_lock);
+	spin_unlock_bh(&enic->devcmd_lock);
 
 	pci_free_consistent(enic->pdev, sizeof(union vnic_rss_cpu),
 		rss_cpu_buf_va, rss_cpu_buf_pa);
@@ -1807,13 +1915,13 @@ static int enic_set_niccfg(struct enic *enic, u8 rss_default_cpu,
 	/* Enable VLAN tag stripping.
 	*/
 
-	spin_lock(&enic->devcmd_lock);
+	spin_lock_bh(&enic->devcmd_lock);
 	err = enic_set_nic_cfg(enic,
 		rss_default_cpu, rss_hash_type,
 		rss_hash_bits, rss_base_cpu,
 		rss_enable, tso_ipid_split_en,
 		ig_vlan_strip_en);
-	spin_unlock(&enic->devcmd_lock);
+	spin_unlock_bh(&enic->devcmd_lock);
 
 	return err;
 }
@@ -2021,6 +2129,12 @@ static const struct net_device_ops enic_netdev_dynamic_ops = {
 #ifdef CONFIG_NET_POLL_CONTROLLER
 	.ndo_poll_controller	= enic_poll_controller,
 #endif
+#ifdef CONFIG_RFS_ACCEL
+	.ndo_rx_flow_steer	= enic_rx_flow_steer,
+#endif
+#ifdef CONFIG_NET_RX_BUSY_POLL
+	.ndo_busy_poll		= enic_busy_poll,
+#endif
 };
 
 static const struct net_device_ops enic_netdev_ops = {
@@ -2041,14 +2155,25 @@ static const struct net_device_ops enic_netdev_ops = {
 #ifdef CONFIG_NET_POLL_CONTROLLER
 	.ndo_poll_controller	= enic_poll_controller,
 #endif
+#ifdef CONFIG_RFS_ACCEL
+	.ndo_rx_flow_steer	= enic_rx_flow_steer,
+#endif
+#ifdef CONFIG_NET_RX_BUSY_POLL
+	.ndo_busy_poll		= enic_busy_poll,
+#endif
 };
 
 static void enic_dev_deinit(struct enic *enic)
 {
 	unsigned int i;
 
-	for (i = 0; i < enic->rq_count; i++)
+	for (i = 0; i < enic->rq_count; i++) {
+		napi_hash_del(&enic->napi[i]);
 		netif_napi_del(&enic->napi[i]);
+	}
+	if (vnic_dev_get_intr_mode(enic->vdev) == VNIC_DEV_INTR_MODE_MSIX)
+		for (i = 0; i < enic->wq_count; i++)
+			netif_napi_del(&enic->napi[enic_cq_wq(enic, i)]);
 
 	enic_free_vnic_resources(enic);
 	enic_clear_intr_mode(enic);
@@ -2114,11 +2239,17 @@ static int enic_dev_init(struct enic *enic)
 	switch (vnic_dev_get_intr_mode(enic->vdev)) {
 	default:
 		netif_napi_add(netdev, &enic->napi[0], enic_poll, 64);
+		napi_hash_add(&enic->napi[0]);
 		break;
 	case VNIC_DEV_INTR_MODE_MSIX:
-		for (i = 0; i < enic->rq_count; i++)
+		for (i = 0; i < enic->rq_count; i++) {
 			netif_napi_add(netdev, &enic->napi[i],
-				enic_poll_msix, 64);
+				enic_poll_msix_rq, NAPI_POLL_WEIGHT);
+			napi_hash_add(&enic->napi[i]);
+		}
+		for (i = 0; i < enic->wq_count; i++)
+			netif_napi_add(netdev, &enic->napi[enic_cq_wq(enic, i)],
+				       enic_poll_msix_wq, NAPI_POLL_WEIGHT);
 		break;
 	}
 
@@ -2386,6 +2517,10 @@ static int enic_probe(struct pci_dev *pdev, const struct pci_device_id *ent)
 
 	netdev->features |= netdev->hw_features;
 
+#ifdef CONFIG_RFS_ACCEL
+	netdev->hw_features |= NETIF_F_NTUPLE;
+#endif
+
 	if (using_dac)
 		netdev->features |= NETIF_F_HIGHDMA;