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

Pull networking updates from David Miller:

 1) Seccomp BPF filters can now be JIT'd, from Alexei Starovoitov.

 2) Multiqueue support in xen-netback and xen-netfront, from Andrew J
    Benniston.

 3) Allow tweaking of aggregation settings in cdc_ncm driver, from Bjørn
    Mork.

 4) BPF now has a "random" opcode, from Chema Gonzalez.

 5) Add more BPF documentation and improve test framework, from Daniel
    Borkmann.

 6) Support TCP fastopen over ipv6, from Daniel Lee.

 7) Add software TSO helper functions and use them to support software
    TSO in mvneta and mv643xx_eth drivers.  From Ezequiel Garcia.

 8) Support software TSO in fec driver too, from Nimrod Andy.

 9) Add Broadcom SYSTEMPORT driver, from Florian Fainelli.

10) Handle broadcasts more gracefully over macvlan when there are large
    numbers of interfaces configured, from Herbert Xu.

11) Allow more control over fwmark used for non-socket based responses,
    from Lorenzo Colitti.

12) Do TCP congestion window limiting based upon measurements, from Neal
    Cardwell.

13) Support busy polling in SCTP, from Neal Horman.

14) Allow RSS key to be configured via ethtool, from Venkata Duvvuru.

15) Bridge promisc mode handling improvements from Vlad Yasevich.

16) Don't use inetpeer entries to implement ID generation any more, it
    performs poorly, from Eric Dumazet.

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-next: (1522 commits)
  rtnetlink: fix userspace API breakage for iproute2 < v3.9.0
  tcp: fixing TLP's FIN recovery
  net: fec: Add software TSO support
  net: fec: Add Scatter/gather support
  net: fec: Increase buffer descriptor entry number
  net: fec: Factorize feature setting
  net: fec: Enable IP header hardware checksum
  net: fec: Factorize the .xmit transmit function
  bridge: fix compile error when compiling without IPv6 support
  bridge: fix smatch warning / potential null pointer dereference
  via-rhine: fix full-duplex with autoneg disable
  bnx2x: Enlarge the dorq threshold for VFs
  bnx2x: Check for UNDI in uncommon branch
  bnx2x: Fix 1G-baseT link
  bnx2x: Fix link for KR with swapped polarity lane
  sctp: Fix sk_ack_backlog wrap-around problem
  net/core: Add VF link state control policy
  net/fsl: xgmac_mdio is dependent on OF_MDIO
  net/fsl: Make xgmac_mdio read error message useful
  net_sched: drr: warn when qdisc is not work conserving
  ...

11 files changed, 265 insertions, 64 deletions

diff --git a/drivers/net/ethernet/chelsio/cxgb/cxgb2.c b/drivers/net/ethernet/chelsio/cxgb/cxgb2.c
index 05613a85ce61..186566bfdbc8 100644
--- a/drivers/net/ethernet/chelsio/cxgb/cxgb2.c
+++ b/drivers/net/ethernet/chelsio/cxgb/cxgb2.c
@@ -580,8 +580,8 @@ static int get_settings(struct net_device *dev, struct ethtool_cmd *cmd)
 		ethtool_cmd_speed_set(cmd, p->link_config.speed);
 		cmd->duplex = p->link_config.duplex;
 	} else {
-		ethtool_cmd_speed_set(cmd, -1);
-		cmd->duplex = -1;
+		ethtool_cmd_speed_set(cmd, SPEED_UNKNOWN);
+		cmd->duplex = DUPLEX_UNKNOWN;
 	}
 
 	cmd->port = (cmd->supported & SUPPORTED_TP) ? PORT_TP : PORT_FIBRE;
@@ -1100,7 +1100,7 @@ static int init_one(struct pci_dev *pdev, const struct pci_device_id *ent)
 
 		netif_napi_add(netdev, &adapter->napi, t1_poll, 64);
 
-		SET_ETHTOOL_OPS(netdev, &t1_ethtool_ops);
+		netdev->ethtool_ops = &t1_ethtool_ops;
 	}
 
 	if (t1_init_sw_modules(adapter, bi) < 0) {
diff --git a/drivers/net/ethernet/chelsio/cxgb3/cxgb3_main.c b/drivers/net/ethernet/chelsio/cxgb3/cxgb3_main.c
index 07bbb711b7e5..5d9cce053cc9 100644
--- a/drivers/net/ethernet/chelsio/cxgb3/cxgb3_main.c
+++ b/drivers/net/ethernet/chelsio/cxgb3/cxgb3_main.c
@@ -1809,8 +1809,8 @@ static int get_settings(struct net_device *dev, struct ethtool_cmd *cmd)
 		ethtool_cmd_speed_set(cmd, p->link_config.speed);
 		cmd->duplex = p->link_config.duplex;
 	} else {
-		ethtool_cmd_speed_set(cmd, -1);
-		cmd->duplex = -1;
+		ethtool_cmd_speed_set(cmd, SPEED_UNKNOWN);
+		cmd->duplex = DUPLEX_UNKNOWN;
 	}
 
 	cmd->port = (cmd->supported & SUPPORTED_TP) ? PORT_TP : PORT_FIBRE;
@@ -3291,7 +3291,7 @@ static int init_one(struct pci_dev *pdev, const struct pci_device_id *ent)
 			netdev->features |= NETIF_F_HIGHDMA;
 
 		netdev->netdev_ops = &cxgb_netdev_ops;
-		SET_ETHTOOL_OPS(netdev, &cxgb_ethtool_ops);
+		netdev->ethtool_ops = &cxgb_ethtool_ops;
 	}
 
 	pci_set_drvdata(pdev, adapter);
diff --git a/drivers/net/ethernet/chelsio/cxgb3/cxgb3_offload.c b/drivers/net/ethernet/chelsio/cxgb3/cxgb3_offload.c
index c0a9dd55f4e5..b0cbb2b7fd48 100644
--- a/drivers/net/ethernet/chelsio/cxgb3/cxgb3_offload.c
+++ b/drivers/net/ethernet/chelsio/cxgb3/cxgb3_offload.c
@@ -185,7 +185,7 @@ static struct net_device *get_iff_from_mac(struct adapter *adapter,
 		if (ether_addr_equal(dev->dev_addr, mac)) {
 			rcu_read_lock();
 			if (vlan && vlan != VLAN_VID_MASK) {
-				dev = __vlan_find_dev_deep(dev, htons(ETH_P_8021Q), vlan);
+				dev = __vlan_find_dev_deep_rcu(dev, htons(ETH_P_8021Q), vlan);
 			} else if (netif_is_bond_slave(dev)) {
 				struct net_device *upper_dev;
 
diff --git a/drivers/net/ethernet/chelsio/cxgb4/cxgb4.h b/drivers/net/ethernet/chelsio/cxgb4/cxgb4.h
index 32db37709263..f503dce4ab17 100644
--- a/drivers/net/ethernet/chelsio/cxgb4/cxgb4.h
+++ b/drivers/net/ethernet/chelsio/cxgb4/cxgb4.h
@@ -357,11 +357,17 @@ enum {
 	MAX_OFLD_QSETS = 16,          /* # of offload Tx/Rx queue sets */
 	MAX_CTRL_QUEUES = NCHAN,      /* # of control Tx queues */
 	MAX_RDMA_QUEUES = NCHAN,      /* # of streaming RDMA Rx queues */
+	MAX_RDMA_CIQS = NCHAN,        /* # of  RDMA concentrator IQs */
+	MAX_ISCSI_QUEUES = NCHAN,     /* # of streaming iSCSI Rx queues */
 };
 
 enum {
-	MAX_EGRQ = 128,         /* max # of egress queues, including FLs */
-	MAX_INGQ = 64           /* max # of interrupt-capable ingress queues */
+	INGQ_EXTRAS = 2,        /* firmware event queue and */
+				/*   forwarded interrupts */
+	MAX_EGRQ = MAX_ETH_QSETS*2 + MAX_OFLD_QSETS*2
+		   + MAX_CTRL_QUEUES + MAX_RDMA_QUEUES + MAX_ISCSI_QUEUES,
+	MAX_INGQ = MAX_ETH_QSETS + MAX_OFLD_QSETS + MAX_RDMA_QUEUES
+		   + MAX_RDMA_CIQS + MAX_ISCSI_QUEUES + INGQ_EXTRAS,
 };
 
 struct adapter;
@@ -538,6 +544,7 @@ struct sge {
 	struct sge_eth_rxq ethrxq[MAX_ETH_QSETS];
 	struct sge_ofld_rxq ofldrxq[MAX_OFLD_QSETS];
 	struct sge_ofld_rxq rdmarxq[MAX_RDMA_QUEUES];
+	struct sge_ofld_rxq rdmaciq[MAX_RDMA_CIQS];
 	struct sge_rspq fw_evtq ____cacheline_aligned_in_smp;
 
 	struct sge_rspq intrq ____cacheline_aligned_in_smp;
@@ -548,8 +555,10 @@ struct sge {
 	u16 ethtxq_rover;           /* Tx queue to clean up next */
 	u16 ofldqsets;              /* # of active offload queue sets */
 	u16 rdmaqs;                 /* # of available RDMA Rx queues */
+	u16 rdmaciqs;               /* # of available RDMA concentrator IQs */
 	u16 ofld_rxq[MAX_OFLD_QSETS];
 	u16 rdma_rxq[NCHAN];
+	u16 rdma_ciq[NCHAN];
 	u16 timer_val[SGE_NTIMERS];
 	u8 counter_val[SGE_NCOUNTERS];
 	u32 fl_pg_order;            /* large page allocation size */
@@ -577,6 +586,7 @@ struct sge {
 #define for_each_ethrxq(sge, i) for (i = 0; i < (sge)->ethqsets; i++)
 #define for_each_ofldrxq(sge, i) for (i = 0; i < (sge)->ofldqsets; i++)
 #define for_each_rdmarxq(sge, i) for (i = 0; i < (sge)->rdmaqs; i++)
+#define for_each_rdmaciq(sge, i) for (i = 0; i < (sge)->rdmaciqs; i++)
 
 struct l2t_data;
 
diff --git a/drivers/net/ethernet/chelsio/cxgb4/cxgb4_main.c b/drivers/net/ethernet/chelsio/cxgb4/cxgb4_main.c
index 24e16e3301e0..2f8d6b910383 100644
--- a/drivers/net/ethernet/chelsio/cxgb4/cxgb4_main.c
+++ b/drivers/net/ethernet/chelsio/cxgb4/cxgb4_main.c
@@ -818,12 +818,17 @@ static void name_msix_vecs(struct adapter *adap)
 	for_each_rdmarxq(&adap->sge, i)
 		snprintf(adap->msix_info[msi_idx++].desc, n, "%s-rdma%d",
 			 adap->port[0]->name, i);
+
+	for_each_rdmaciq(&adap->sge, i)
+		snprintf(adap->msix_info[msi_idx++].desc, n, "%s-rdma-ciq%d",
+			 adap->port[0]->name, i);
 }
 
 static int request_msix_queue_irqs(struct adapter *adap)
 {
 	struct sge *s = &adap->sge;
-	int err, ethqidx, ofldqidx = 0, rdmaqidx = 0, msi_index = 2;
+	int err, ethqidx, ofldqidx = 0, rdmaqidx = 0, rdmaciqqidx = 0;
+	int msi_index = 2;
 
 	err = request_irq(adap->msix_info[1].vec, t4_sge_intr_msix, 0,
 			  adap->msix_info[1].desc, &s->fw_evtq);
@@ -857,9 +862,21 @@ static int request_msix_queue_irqs(struct adapter *adap)
 			goto unwind;
 		msi_index++;
 	}
+	for_each_rdmaciq(s, rdmaciqqidx) {
+		err = request_irq(adap->msix_info[msi_index].vec,
+				  t4_sge_intr_msix, 0,
+				  adap->msix_info[msi_index].desc,
+				  &s->rdmaciq[rdmaciqqidx].rspq);
+		if (err)
+			goto unwind;
+		msi_index++;
+	}
 	return 0;
 
 unwind:
+	while (--rdmaciqqidx >= 0)
+		free_irq(adap->msix_info[--msi_index].vec,
+			 &s->rdmaciq[rdmaciqqidx].rspq);
 	while (--rdmaqidx >= 0)
 		free_irq(adap->msix_info[--msi_index].vec,
 			 &s->rdmarxq[rdmaqidx].rspq);
@@ -885,6 +902,8 @@ static void free_msix_queue_irqs(struct adapter *adap)
 		free_irq(adap->msix_info[msi_index++].vec, &s->ofldrxq[i].rspq);
 	for_each_rdmarxq(s, i)
 		free_irq(adap->msix_info[msi_index++].vec, &s->rdmarxq[i].rspq);
+	for_each_rdmaciq(s, i)
+		free_irq(adap->msix_info[msi_index++].vec, &s->rdmaciq[i].rspq);
 }
 
 /**
@@ -1047,7 +1066,8 @@ freeout:	t4_free_sge_resources(adap);
 		if (msi_idx > 0)
 			msi_idx++;
 		err = t4_sge_alloc_rxq(adap, &q->rspq, false, dev, msi_idx,
-				       &q->fl, uldrx_handler);
+				       q->fl.size ? &q->fl : NULL,
+				       uldrx_handler);
 		if (err)
 			goto freeout;
 		memset(&q->stats, 0, sizeof(q->stats));
@@ -1064,13 +1084,28 @@ freeout:	t4_free_sge_resources(adap);
 		if (msi_idx > 0)
 			msi_idx++;
 		err = t4_sge_alloc_rxq(adap, &q->rspq, false, adap->port[i],
-				       msi_idx, &q->fl, uldrx_handler);
+				       msi_idx, q->fl.size ? &q->fl : NULL,
+				       uldrx_handler);
 		if (err)
 			goto freeout;
 		memset(&q->stats, 0, sizeof(q->stats));
 		s->rdma_rxq[i] = q->rspq.abs_id;
 	}
 
+	for_each_rdmaciq(s, i) {
+		struct sge_ofld_rxq *q = &s->rdmaciq[i];
+
+		if (msi_idx > 0)
+			msi_idx++;
+		err = t4_sge_alloc_rxq(adap, &q->rspq, false, adap->port[i],
+				       msi_idx, q->fl.size ? &q->fl : NULL,
+				       uldrx_handler);
+		if (err)
+			goto freeout;
+		memset(&q->stats, 0, sizeof(q->stats));
+		s->rdma_ciq[i] = q->rspq.abs_id;
+	}
+
 	for_each_port(adap, i) {
 		/*
 		 * Note that ->rdmarxq[i].rspq.cntxt_id below is 0 if we don't
@@ -2252,12 +2287,19 @@ static int get_settings(struct net_device *dev, struct ethtool_cmd *cmd)
 	else if (p->port_type == FW_PORT_TYPE_FIBER_XFI ||
 		 p->port_type == FW_PORT_TYPE_FIBER_XAUI)
 		cmd->port = PORT_FIBRE;
-	else if (p->port_type == FW_PORT_TYPE_SFP) {
-		if (p->mod_type == FW_PORT_MOD_TYPE_TWINAX_PASSIVE ||
-		    p->mod_type == FW_PORT_MOD_TYPE_TWINAX_ACTIVE)
+	else if (p->port_type == FW_PORT_TYPE_SFP ||
+		 p->port_type == FW_PORT_TYPE_QSFP_10G ||
+		 p->port_type == FW_PORT_TYPE_QSFP) {
+		if (p->mod_type == FW_PORT_MOD_TYPE_LR ||
+		    p->mod_type == FW_PORT_MOD_TYPE_SR ||
+		    p->mod_type == FW_PORT_MOD_TYPE_ER ||
+		    p->mod_type == FW_PORT_MOD_TYPE_LRM)
+			cmd->port = PORT_FIBRE;
+		else if (p->mod_type == FW_PORT_MOD_TYPE_TWINAX_PASSIVE ||
+			 p->mod_type == FW_PORT_MOD_TYPE_TWINAX_ACTIVE)
 			cmd->port = PORT_DA;
 		else
-			cmd->port = PORT_FIBRE;
+			cmd->port = PORT_OTHER;
 	} else
 		cmd->port = PORT_OTHER;
 
@@ -2461,8 +2503,7 @@ static unsigned int qtimer_val(const struct adapter *adap,
 }
 
 /**
- *	set_rxq_intr_params - set a queue's interrupt holdoff parameters
- *	@adap: the adapter
+ *	set_rspq_intr_params - set a queue's interrupt holdoff parameters
  *	@q: the Rx queue
  *	@us: the hold-off time in us, or 0 to disable timer
  *	@cnt: the hold-off packet count, or 0 to disable counter
@@ -2470,9 +2511,11 @@ static unsigned int qtimer_val(const struct adapter *adap,
  *	Sets an Rx queue's interrupt hold-off time and packet count.  At least
  *	one of the two needs to be enabled for the queue to generate interrupts.
  */
-static int set_rxq_intr_params(struct adapter *adap, struct sge_rspq *q,
-			       unsigned int us, unsigned int cnt)
+static int set_rspq_intr_params(struct sge_rspq *q,
+				unsigned int us, unsigned int cnt)
 {
+	struct adapter *adap = q->adap;
+
 	if ((us | cnt) == 0)
 		cnt = 1;
 
@@ -2499,24 +2542,34 @@ static int set_rxq_intr_params(struct adapter *adap, struct sge_rspq *q,
 	return 0;
 }
 
-static int set_coalesce(struct net_device *dev, struct ethtool_coalesce *c)
+/**
+ * set_rx_intr_params - set a net devices's RX interrupt holdoff paramete!
+ * @dev: the network device
+ * @us: the hold-off time in us, or 0 to disable timer
+ * @cnt: the hold-off packet count, or 0 to disable counter
+ *
+ * Set the RX interrupt hold-off parameters for a network device.
+ */
+static int set_rx_intr_params(struct net_device *dev,
+			      unsigned int us, unsigned int cnt)
 {
-	const struct port_info *pi = netdev_priv(dev);
+	int i, err;
+	struct port_info *pi = netdev_priv(dev);
 	struct adapter *adap = pi->adapter;
-	struct sge_rspq *q;
-	int i;
-	int r = 0;
-
-	for (i = pi->first_qset; i < pi->first_qset + pi->nqsets; i++) {
-		q = &adap->sge.ethrxq[i].rspq;
-		r = set_rxq_intr_params(adap, q, c->rx_coalesce_usecs,
-			c->rx_max_coalesced_frames);
-		if (r) {
-			dev_err(&dev->dev, "failed to set coalesce %d\n", r);
-			break;
-		}
+	struct sge_eth_rxq *q = &adap->sge.ethrxq[pi->first_qset];
+
+	for (i = 0; i < pi->nqsets; i++, q++) {
+		err = set_rspq_intr_params(&q->rspq, us, cnt);
+		if (err)
+			return err;
 	}
-	return r;
+	return 0;
+}
+
+static int set_coalesce(struct net_device *dev, struct ethtool_coalesce *c)
+{
+	return set_rx_intr_params(dev, c->rx_coalesce_usecs,
+				  c->rx_max_coalesced_frames);
 }
 
 static int get_coalesce(struct net_device *dev, struct ethtool_coalesce *c)
@@ -2732,7 +2785,7 @@ static u32 get_rss_table_size(struct net_device *dev)
 	return pi->rss_size;
 }
 
-static int get_rss_table(struct net_device *dev, u32 *p)
+static int get_rss_table(struct net_device *dev, u32 *p, u8 *key)
 {
 	const struct port_info *pi = netdev_priv(dev);
 	unsigned int n = pi->rss_size;
@@ -2742,7 +2795,7 @@ static int get_rss_table(struct net_device *dev, u32 *p)
 	return 0;
 }
 
-static int set_rss_table(struct net_device *dev, const u32 *p)
+static int set_rss_table(struct net_device *dev, const u32 *p, const u8 *key)
 {
 	unsigned int i;
 	struct port_info *pi = netdev_priv(dev);
@@ -2844,8 +2897,8 @@ static const struct ethtool_ops cxgb_ethtool_ops = {
 	.set_wol           = set_wol,
 	.get_rxnfc         = get_rxnfc,
 	.get_rxfh_indir_size = get_rss_table_size,
-	.get_rxfh_indir    = get_rss_table,
-	.set_rxfh_indir    = set_rss_table,
+	.get_rxfh	   = get_rss_table,
+	.set_rxfh	   = set_rss_table,
 	.flash_device      = set_flash,
 };
 
@@ -3386,6 +3439,77 @@ unsigned int cxgb4_best_mtu(const unsigned short *mtus, unsigned short mtu,
 EXPORT_SYMBOL(cxgb4_best_mtu);
 
 /**
+ *     cxgb4_best_aligned_mtu - find best MTU, [hopefully] data size aligned
+ *     @mtus: the HW MTU table
+ *     @header_size: Header Size
+ *     @data_size_max: maximum Data Segment Size
+ *     @data_size_align: desired Data Segment Size Alignment (2^N)
+ *     @mtu_idxp: HW MTU Table Index return value pointer (possibly NULL)
+ *
+ *     Similar to cxgb4_best_mtu() but instead of searching the Hardware
+ *     MTU Table based solely on a Maximum MTU parameter, we break that
+ *     parameter up into a Header Size and Maximum Data Segment Size, and
+ *     provide a desired Data Segment Size Alignment.  If we find an MTU in
+ *     the Hardware MTU Table which will result in a Data Segment Size with
+ *     the requested alignment _and_ that MTU isn't "too far" from the
+ *     closest MTU, then we'll return that rather than the closest MTU.
+ */
+unsigned int cxgb4_best_aligned_mtu(const unsigned short *mtus,
+				    unsigned short header_size,
+				    unsigned short data_size_max,
+				    unsigned short data_size_align,
+				    unsigned int *mtu_idxp)
+{
+	unsigned short max_mtu = header_size + data_size_max;
+	unsigned short data_size_align_mask = data_size_align - 1;
+	int mtu_idx, aligned_mtu_idx;
+
+	/* Scan the MTU Table till we find an MTU which is larger than our
+	 * Maximum MTU or we reach the end of the table.  Along the way,
+	 * record the last MTU found, if any, which will result in a Data
+	 * Segment Length matching the requested alignment.
+	 */
+	for (mtu_idx = 0, aligned_mtu_idx = -1; mtu_idx < NMTUS; mtu_idx++) {
+		unsigned short data_size = mtus[mtu_idx] - header_size;
+
+		/* If this MTU minus the Header Size would result in a
+		 * Data Segment Size of the desired alignment, remember it.
+		 */
+		if ((data_size & data_size_align_mask) == 0)
+			aligned_mtu_idx = mtu_idx;
+
+		/* If we're not at the end of the Hardware MTU Table and the
+		 * next element is larger than our Maximum MTU, drop out of
+		 * the loop.
+		 */
+		if (mtu_idx+1 < NMTUS && mtus[mtu_idx+1] > max_mtu)
+			break;
+	}
+
+	/* If we fell out of the loop because we ran to the end of the table,
+	 * then we just have to use the last [largest] entry.
+	 */
+	if (mtu_idx == NMTUS)
+		mtu_idx--;
+
+	/* If we found an MTU which resulted in the requested Data Segment
+	 * Length alignment and that's "not far" from the largest MTU which is
+	 * less than or equal to the maximum MTU, then use that.
+	 */
+	if (aligned_mtu_idx >= 0 &&
+	    mtu_idx - aligned_mtu_idx <= 1)
+		mtu_idx = aligned_mtu_idx;
+
+	/* If the caller has passed in an MTU Index pointer, pass the
+	 * MTU Index back.  Return the MTU value.
+	 */
+	if (mtu_idxp)
+		*mtu_idxp = mtu_idx;
+	return mtus[mtu_idx];
+}
+EXPORT_SYMBOL(cxgb4_best_aligned_mtu);
+
+/**
  *	cxgb4_port_chan - get the HW channel of a port
  *	@dev: the net device for the port
  *
@@ -3782,7 +3906,9 @@ static void uld_attach(struct adapter *adap, unsigned int uld)
 	lli.mtus = adap->params.mtus;
 	if (uld == CXGB4_ULD_RDMA) {
 		lli.rxq_ids = adap->sge.rdma_rxq;
+		lli.ciq_ids = adap->sge.rdma_ciq;
 		lli.nrxq = adap->sge.rdmaqs;
+		lli.nciq = adap->sge.rdmaciqs;
 	} else if (uld == CXGB4_ULD_ISCSI) {
 		lli.rxq_ids = adap->sge.ofld_rxq;
 		lli.nrxq = adap->sge.ofldqsets;
@@ -4061,7 +4187,7 @@ static int update_root_dev_clip(struct net_device *dev)
 
 	/* Parse all bond and vlan devices layered on top of the physical dev */
 	for (i = 0; i < VLAN_N_VID; i++) {
-		root_dev = __vlan_find_dev_deep(dev, htons(ETH_P_8021Q), i);
+		root_dev = __vlan_find_dev_deep_rcu(dev, htons(ETH_P_8021Q), i);
 		if (!root_dev)
 			continue;
 
@@ -5528,13 +5654,41 @@ static int adap_init0(struct adapter *adap)
 #undef FW_PARAM_PFVF
 #undef FW_PARAM_DEV
 
-	/*
-	 * These are finalized by FW initialization, load their values now.
+	/* The MTU/MSS Table is initialized by now, so load their values.  If
+	 * we're initializing the adapter, then we'll make any modifications
+	 * we want to the MTU/MSS Table and also initialize the congestion
+	 * parameters.
 	 */
 	t4_read_mtu_tbl(adap, adap->params.mtus, NULL);
-	t4_load_mtus(adap, adap->params.mtus, adap->params.a_wnd,
-		     adap->params.b_wnd);
+	if (state != DEV_STATE_INIT) {
+		int i;
+
+		/* The default MTU Table contains values 1492 and 1500.
+		 * However, for TCP, it's better to have two values which are
+		 * a multiple of 8 +/- 4 bytes apart near this popular MTU.
+		 * This allows us to have a TCP Data Payload which is a
+		 * multiple of 8 regardless of what combination of TCP Options
+		 * are in use (always a multiple of 4 bytes) which is
+		 * important for performance reasons.  For instance, if no
+		 * options are in use, then we have a 20-byte IP header and a
+		 * 20-byte TCP header.  In this case, a 1500-byte MSS would
+		 * result in a TCP Data Payload of 1500 - 40 == 1460 bytes
+		 * which is not a multiple of 8.  So using an MSS of 1488 in
+		 * this case results in a TCP Data Payload of 1448 bytes which
+		 * is a multiple of 8.  On the other hand, if 12-byte TCP Time
+		 * Stamps have been negotiated, then an MTU of 1500 bytes
+		 * results in a TCP Data Payload of 1448 bytes which, as
+		 * above, is a multiple of 8 bytes ...
+		 */
+		for (i = 0; i < NMTUS; i++)
+			if (adap->params.mtus[i] == 1492) {
+				adap->params.mtus[i] = 1488;
+				break;
+			}
 
+		t4_load_mtus(adap, adap->params.mtus, adap->params.a_wnd,
+			     adap->params.b_wnd);
+	}
 	t4_init_tp_params(adap);
 	adap->flags |= FW_OK;
 	return 0;
@@ -5669,12 +5823,12 @@ static inline bool is_x_10g_port(const struct link_config *lc)
 	       (lc->supported & FW_PORT_CAP_SPEED_40G) != 0;
 }
 
-static inline void init_rspq(struct sge_rspq *q, u8 timer_idx, u8 pkt_cnt_idx,
+static inline void init_rspq(struct adapter *adap, struct sge_rspq *q,
+			     unsigned int us, unsigned int cnt,
 			     unsigned int size, unsigned int iqe_size)
 {
-	q->intr_params = QINTR_TIMER_IDX(timer_idx) |
-			 (pkt_cnt_idx < SGE_NCOUNTERS ? QINTR_CNT_EN : 0);
-	q->pktcnt_idx = pkt_cnt_idx < SGE_NCOUNTERS ? pkt_cnt_idx : 0;
+	q->adap = adap;
+	set_rspq_intr_params(q, us, cnt);
 	q->iqe_len = iqe_size;
 	q->size = size;
 }
@@ -5688,6 +5842,7 @@ static void cfg_queues(struct adapter *adap)
 {
 	struct sge *s = &adap->sge;
 	int i, q10g = 0, n10g = 0, qidx = 0;
+	int ciq_size;
 
 	for_each_port(adap, i)
 		n10g += is_x_10g_port(&adap2pinfo(adap, i)->link_cfg);
@@ -5726,12 +5881,13 @@ static void cfg_queues(struct adapter *adap)
 			s->ofldqsets = adap->params.nports;
 		/* For RDMA one Rx queue per channel suffices */
 		s->rdmaqs = adap->params.nports;
+		s->rdmaciqs = adap->params.nports;
 	}
 
 	for (i = 0; i < ARRAY_SIZE(s->ethrxq); i++) {
 		struct sge_eth_rxq *r = &s->ethrxq[i];
 
-		init_rspq(&r->rspq, 0, 0, 1024, 64);
+		init_rspq(adap, &r->rspq, 5, 10, 1024, 64);
 		r->fl.size = 72;
 	}
 
@@ -5747,7 +5903,7 @@ static void cfg_queues(struct adapter *adap)
 	for (i = 0; i < ARRAY_SIZE(s->ofldrxq); i++) {
 		struct sge_ofld_rxq *r = &s->ofldrxq[i];
 
-		init_rspq(&r->rspq, 0, 0, 1024, 64);
+		init_rspq(adap, &r->rspq, 5, 1, 1024, 64);
 		r->rspq.uld = CXGB4_ULD_ISCSI;
 		r->fl.size = 72;
 	}
@@ -5755,13 +5911,26 @@ static void cfg_queues(struct adapter *adap)
 	for (i = 0; i < ARRAY_SIZE(s->rdmarxq); i++) {
 		struct sge_ofld_rxq *r = &s->rdmarxq[i];
 
-		init_rspq(&r->rspq, 0, 0, 511, 64);
+		init_rspq(adap, &r->rspq, 5, 1, 511, 64);
 		r->rspq.uld = CXGB4_ULD_RDMA;
 		r->fl.size = 72;
 	}
 
-	init_rspq(&s->fw_evtq, 6, 0, 512, 64);
-	init_rspq(&s->intrq, 6, 0, 2 * MAX_INGQ, 64);
+	ciq_size = 64 + adap->vres.cq.size + adap->tids.nftids;
+	if (ciq_size > SGE_MAX_IQ_SIZE) {
+		CH_WARN(adap, "CIQ size too small for available IQs\n");
+		ciq_size = SGE_MAX_IQ_SIZE;
+	}
+
+	for (i = 0; i < ARRAY_SIZE(s->rdmaciq); i++) {
+		struct sge_ofld_rxq *r = &s->rdmaciq[i];
+
+		init_rspq(adap, &r->rspq, 5, 1, ciq_size, 64);
+		r->rspq.uld = CXGB4_ULD_RDMA;
+	}
+
+	init_rspq(adap, &s->fw_evtq, 0, 1, 1024, 64);
+	init_rspq(adap, &s->intrq, 0, 1, 2 * MAX_INGQ, 64);
 }
 
 /*
@@ -5808,9 +5977,9 @@ static int enable_msix(struct adapter *adap)
 
 	want = s->max_ethqsets + EXTRA_VECS;
 	if (is_offload(adap)) {
-		want += s->rdmaqs + s->ofldqsets;
+		want += s->rdmaqs + s->rdmaciqs + s->ofldqsets;
 		/* need nchan for each possible ULD */
-		ofld_need = 2 * nchan;
+		ofld_need = 3 * nchan;
 	}
 	need = adap->params.nports + EXTRA_VECS + ofld_need;
 
@@ -6076,7 +6245,7 @@ static int init_one(struct pci_dev *pdev, const struct pci_device_id *ent)
 		netdev->priv_flags |= IFF_UNICAST_FLT;
 
 		netdev->netdev_ops = &cxgb4_netdev_ops;
-		SET_ETHTOOL_OPS(netdev, &cxgb_ethtool_ops);
+		netdev->ethtool_ops = &cxgb_ethtool_ops;
 	}
 
 	pci_set_drvdata(pdev, adapter);
diff --git a/drivers/net/ethernet/chelsio/cxgb4/cxgb4_uld.h b/drivers/net/ethernet/chelsio/cxgb4/cxgb4_uld.h
index e274a047528f..55e9daf7f9d4 100644
--- a/drivers/net/ethernet/chelsio/cxgb4/cxgb4_uld.h
+++ b/drivers/net/ethernet/chelsio/cxgb4/cxgb4_uld.h
@@ -232,8 +232,10 @@ struct cxgb4_lld_info {
 	const struct cxgb4_virt_res *vr;     /* assorted HW resources */
 	const unsigned short *mtus;          /* MTU table */
 	const unsigned short *rxq_ids;       /* the ULD's Rx queue ids */
+	const unsigned short *ciq_ids;       /* the ULD's concentrator IQ ids */
 	unsigned short nrxq;                 /* # of Rx queues */
 	unsigned short ntxq;                 /* # of Tx queues */
+	unsigned short nciq;		     /* # of concentrator IQ */
 	unsigned char nchan:4;               /* # of channels */
 	unsigned char nports:4;              /* # of ports */
 	unsigned char wr_cred;               /* WR 16-byte credits */
@@ -274,6 +276,11 @@ unsigned int cxgb4_port_viid(const struct net_device *dev);
 unsigned int cxgb4_port_idx(const struct net_device *dev);
 unsigned int cxgb4_best_mtu(const unsigned short *mtus, unsigned short mtu,
 			    unsigned int *idx);
+unsigned int cxgb4_best_aligned_mtu(const unsigned short *mtus,
+				    unsigned short header_size,
+				    unsigned short data_size_max,
+				    unsigned short data_size_align,
+				    unsigned int *mtu_idxp);
 void cxgb4_get_tcp_stats(struct pci_dev *pdev, struct tp_tcp_stats *v4,
 			 struct tp_tcp_stats *v6);
 void cxgb4_iscsi_init(struct net_device *dev, unsigned int tag_mask,
diff --git a/drivers/net/ethernet/chelsio/cxgb4/sge.c b/drivers/net/ethernet/chelsio/cxgb4/sge.c
index e249528c8e60..dd4355d248e4 100644
--- a/drivers/net/ethernet/chelsio/cxgb4/sge.c
+++ b/drivers/net/ethernet/chelsio/cxgb4/sge.c
@@ -1697,7 +1697,8 @@ int t4_ethrx_handler(struct sge_rspq *q, const __be64 *rsp,
 		return handle_trace_pkt(q->adap, si);
 
 	pkt = (const struct cpl_rx_pkt *)rsp;
-	csum_ok = pkt->csum_calc && !pkt->err_vec;
+	csum_ok = pkt->csum_calc && !pkt->err_vec &&
+		  (q->netdev->features & NETIF_F_RXCSUM);
 	if ((pkt->l2info & htonl(RXF_TCP)) &&
 	    (q->netdev->features & NETIF_F_GRO) && csum_ok && !pkt->ip_frag) {
 		do_gro(rxq, si, pkt);
@@ -1720,8 +1721,7 @@ int t4_ethrx_handler(struct sge_rspq *q, const __be64 *rsp,
 
 	rxq->stats.pkts++;
 
-	if (csum_ok && (q->netdev->features & NETIF_F_RXCSUM) &&
-	    (pkt->l2info & htonl(RXF_UDP | RXF_TCP))) {
+	if (csum_ok && (pkt->l2info & htonl(RXF_UDP | RXF_TCP))) {
 		if (!pkt->ip_frag) {
 			skb->ip_summed = CHECKSUM_UNNECESSARY;
 			rxq->stats.rx_cso++;
@@ -2215,7 +2215,6 @@ int t4_sge_alloc_rxq(struct adapter *adap, struct sge_rspq *iq, bool fwevtq,
 	iq->cntxt_id = ntohs(c.iqid);
 	iq->abs_id = ntohs(c.physiqid);
 	iq->size--;                           /* subtract status entry */
-	iq->adap = adap;
 	iq->netdev = dev;
 	iq->handler = hnd;
 
@@ -2515,6 +2514,10 @@ void t4_free_sge_resources(struct adapter *adap)
 		if (oq->rspq.desc)
 			free_rspq_fl(adap, &oq->rspq, &oq->fl);
 	}
+	for (i = 0, oq = adap->sge.rdmaciq; i < adap->sge.rdmaciqs; i++, oq++) {
+		if (oq->rspq.desc)
+			free_rspq_fl(adap, &oq->rspq, &oq->fl);
+	}
 
 	/* clean up offload Tx queues */
 	for (i = 0; i < ARRAY_SIZE(adap->sge.ofldtxq); i++) {
diff --git a/drivers/net/ethernet/chelsio/cxgb4/t4_hw.h b/drivers/net/ethernet/chelsio/cxgb4/t4_hw.h
index 1d1623be9f1e..71b799b5b0f4 100644
--- a/drivers/net/ethernet/chelsio/cxgb4/t4_hw.h
+++ b/drivers/net/ethernet/chelsio/cxgb4/t4_hw.h
@@ -68,6 +68,7 @@ enum {
 	SGE_MAX_WR_LEN = 512,     /* max WR size in bytes */
 	SGE_NTIMERS = 6,          /* # of interrupt holdoff timer values */
 	SGE_NCOUNTERS = 4,        /* # of interrupt packet counter values */
+	SGE_MAX_IQ_SIZE = 65520,
 
 	SGE_TIMER_RSTRT_CNTR = 6, /* restart RX packet threshold counter */
 	SGE_TIMER_UPD_CIDX = 7,   /* update cidx only */
diff --git a/drivers/net/ethernet/chelsio/cxgb4/t4_msg.h b/drivers/net/ethernet/chelsio/cxgb4/t4_msg.h
index f2738c710789..973eb11aa98a 100644
--- a/drivers/net/ethernet/chelsio/cxgb4/t4_msg.h
+++ b/drivers/net/ethernet/chelsio/cxgb4/t4_msg.h
@@ -227,6 +227,7 @@ struct cpl_pass_open_req {
 #define DELACK(x)     ((x) << 5)
 #define ULP_MODE(x)   ((x) << 8)
 #define RCV_BUFSIZ(x) ((x) << 12)
+#define RCV_BUFSIZ_MASK 0x3FFU
 #define DSCP(x)       ((x) << 22)
 #define SMAC_SEL(x)   ((u64)(x) << 28)
 #define L2T_IDX(x)    ((u64)(x) << 36)
@@ -278,6 +279,15 @@ struct cpl_pass_accept_rpl {
 	__be64 opt0;
 };
 
+struct cpl_t5_pass_accept_rpl {
+	WR_HDR;
+	union opcode_tid ot;
+	__be32 opt2;
+	__be64 opt0;
+	__be32 iss;
+	__be32 rsvd;
+};
+
 struct cpl_act_open_req {
 	WR_HDR;
 	union opcode_tid ot;
diff --git a/drivers/net/ethernet/chelsio/cxgb4vf/cxgb4vf_main.c b/drivers/net/ethernet/chelsio/cxgb4vf/cxgb4vf_main.c
index 52859288de7b..ff1cdd1788b5 100644
--- a/drivers/net/ethernet/chelsio/cxgb4vf/cxgb4vf_main.c
+++ b/drivers/net/ethernet/chelsio/cxgb4vf/cxgb4vf_main.c
@@ -2664,7 +2664,7 @@ static int cxgb4vf_pci_probe(struct pci_dev *pdev,
 		netdev->priv_flags |= IFF_UNICAST_FLT;
 
 		netdev->netdev_ops = &cxgb4vf_netdev_ops;
-		SET_ETHTOOL_OPS(netdev, &cxgb4vf_ethtool_ops);
+		netdev->ethtool_ops = &cxgb4vf_ethtool_ops;
 
 		/*
 		 * Initialize the hardware/software state for the port.
diff --git a/drivers/net/ethernet/chelsio/cxgb4vf/sge.c b/drivers/net/ethernet/chelsio/cxgb4vf/sge.c
index 9d88c1d50b49..bdfa80ca5e31 100644
--- a/drivers/net/ethernet/chelsio/cxgb4vf/sge.c
+++ b/drivers/net/ethernet/chelsio/cxgb4vf/sge.c
@@ -1510,7 +1510,8 @@ int t4vf_ethrx_handler(struct sge_rspq *rspq, const __be64 *rsp,
 {
 	struct sk_buff *skb;
 	const struct cpl_rx_pkt *pkt = (void *)rsp;
-	bool csum_ok = pkt->csum_calc && !pkt->err_vec;
+	bool csum_ok = pkt->csum_calc && !pkt->err_vec &&
+		       (rspq->netdev->features & NETIF_F_RXCSUM);
 	struct sge_eth_rxq *rxq = container_of(rspq, struct sge_eth_rxq, rspq);
 
 	/*
@@ -1538,8 +1539,8 @@ int t4vf_ethrx_handler(struct sge_rspq *rspq, const __be64 *rsp,
 	skb_record_rx_queue(skb, rspq->idx);
 	rxq->stats.pkts++;
 
-	if (csum_ok && (rspq->netdev->features & NETIF_F_RXCSUM) &&
-	    !pkt->err_vec && (be32_to_cpu(pkt->l2info) & (RXF_UDP|RXF_TCP))) {
+	if (csum_ok && !pkt->err_vec &&
+	    (be32_to_cpu(pkt->l2info) & (RXF_UDP|RXF_TCP))) {
 		if (!pkt->ip_frag)
 			skb->ip_summed = CHECKSUM_UNNECESSARY;
 		else {