Merge branch 'linus' into release

Conflicts:
	arch/x86/kernel/cpu/cpufreq/longhaul.c

Signed-off-by: Len Brown <len.brown@intel.com>

39 files changed, 897 insertions, 540 deletions

diff --git a/drivers/infiniband/hw/amso1100/c2.c b/drivers/infiniband/hw/amso1100/c2.c
index 113f3c03c5b5..7d79aa361e26 100644
--- a/drivers/infiniband/hw/amso1100/c2.c
+++ b/drivers/infiniband/hw/amso1100/c2.c
@@ -76,7 +76,6 @@ static irqreturn_t c2_interrupt(int irq, void *dev_id);
 static void c2_tx_timeout(struct net_device *netdev);
 static int c2_change_mtu(struct net_device *netdev, int new_mtu);
 static void c2_reset(struct c2_port *c2_port);
-static struct net_device_stats *c2_get_stats(struct net_device *netdev);
 
 static struct pci_device_id c2_pci_table[] = {
 	{ PCI_DEVICE(0x18b8, 0xb001) },
@@ -349,7 +348,7 @@ static void c2_tx_clean(struct c2_port *c2_port)
 					     elem->hw_desc + C2_TXP_ADDR);
 				__raw_writew((__force u16) cpu_to_be16(TXP_HTXD_DONE),
 					     elem->hw_desc + C2_TXP_FLAGS);
-				c2_port->netstats.tx_dropped++;
+				c2_port->netdev->stats.tx_dropped++;
 				break;
 			} else {
 				__raw_writew(0,
@@ -457,7 +456,7 @@ static void c2_rx_error(struct c2_port *c2_port, struct c2_element *elem)
 		     elem->hw_desc + C2_RXP_FLAGS);
 
 	pr_debug("packet dropped\n");
-	c2_port->netstats.rx_dropped++;
+	c2_port->netdev->stats.rx_dropped++;
 }
 
 static void c2_rx_interrupt(struct net_device *netdev)
@@ -532,8 +531,8 @@ static void c2_rx_interrupt(struct net_device *netdev)
 		netif_rx(skb);
 
 		netdev->last_rx = jiffies;
-		c2_port->netstats.rx_packets++;
-		c2_port->netstats.rx_bytes += buflen;
+		netdev->stats.rx_packets++;
+		netdev->stats.rx_bytes += buflen;
 	}
 
 	/* Save where we left off */
@@ -797,8 +796,8 @@ static int c2_xmit_frame(struct sk_buff *skb, struct net_device *netdev)
 	__raw_writew((__force u16) cpu_to_be16(TXP_HTXD_READY),
 		     elem->hw_desc + C2_TXP_FLAGS);
 
-	c2_port->netstats.tx_packets++;
-	c2_port->netstats.tx_bytes += maplen;
+	netdev->stats.tx_packets++;
+	netdev->stats.tx_bytes += maplen;
 
 	/* Loop thru additional data fragments and queue them */
 	if (skb_shinfo(skb)->nr_frags) {
@@ -823,8 +822,8 @@ static int c2_xmit_frame(struct sk_buff *skb, struct net_device *netdev)
 			__raw_writew((__force u16) cpu_to_be16(TXP_HTXD_READY),
 				     elem->hw_desc + C2_TXP_FLAGS);
 
-			c2_port->netstats.tx_packets++;
-			c2_port->netstats.tx_bytes += maplen;
+			netdev->stats.tx_packets++;
+			netdev->stats.tx_bytes += maplen;
 		}
 	}
 
@@ -845,13 +844,6 @@ static int c2_xmit_frame(struct sk_buff *skb, struct net_device *netdev)
 	return NETDEV_TX_OK;
 }
 
-static struct net_device_stats *c2_get_stats(struct net_device *netdev)
-{
-	struct c2_port *c2_port = netdev_priv(netdev);
-
-	return &c2_port->netstats;
-}
-
 static void c2_tx_timeout(struct net_device *netdev)
 {
 	struct c2_port *c2_port = netdev_priv(netdev);
@@ -880,6 +872,16 @@ static int c2_change_mtu(struct net_device *netdev, int new_mtu)
 	return ret;
 }
 
+static const struct net_device_ops c2_netdev = {
+	.ndo_open 		= c2_up,
+	.ndo_stop 		= c2_down,
+	.ndo_start_xmit		= c2_xmit_frame,
+	.ndo_tx_timeout		= c2_tx_timeout,
+	.ndo_change_mtu		= c2_change_mtu,
+	.ndo_set_mac_address 	= eth_mac_addr,
+	.ndo_validate_addr	= eth_validate_addr,
+};
+
 /* Initialize network device */
 static struct net_device *c2_devinit(struct c2_dev *c2dev,
 				     void __iomem * mmio_addr)
@@ -894,12 +896,7 @@ static struct net_device *c2_devinit(struct c2_dev *c2dev,
 
 	SET_NETDEV_DEV(netdev, &c2dev->pcidev->dev);
 
-	netdev->open = c2_up;
-	netdev->stop = c2_down;
-	netdev->hard_start_xmit = c2_xmit_frame;
-	netdev->get_stats = c2_get_stats;
-	netdev->tx_timeout = c2_tx_timeout;
-	netdev->change_mtu = c2_change_mtu;
+	netdev->netdev_ops = &c2_netdev;
 	netdev->watchdog_timeo = C2_TX_TIMEOUT;
 	netdev->irq = c2dev->pcidev->irq;
 
diff --git a/drivers/infiniband/hw/amso1100/c2.h b/drivers/infiniband/hw/amso1100/c2.h
index d12a24a84fd9..f7ff66f98361 100644
--- a/drivers/infiniband/hw/amso1100/c2.h
+++ b/drivers/infiniband/hw/amso1100/c2.h
@@ -369,8 +369,6 @@ struct c2_port {
 	unsigned long mem_size;
 
 	u32 rx_buf_size;
-
-	struct net_device_stats netstats;
 };
 
 /*
diff --git a/drivers/infiniband/hw/amso1100/c2_provider.c b/drivers/infiniband/hw/amso1100/c2_provider.c
index 5119d6508181..f1948fad85d7 100644
--- a/drivers/infiniband/hw/amso1100/c2_provider.c
+++ b/drivers/infiniband/hw/amso1100/c2_provider.c
@@ -708,26 +708,27 @@ static int c2_pseudo_xmit_frame(struct sk_buff *skb, struct net_device *netdev)
 
 static int c2_pseudo_change_mtu(struct net_device *netdev, int new_mtu)
 {
-	int ret = 0;
-
 	if (new_mtu < ETH_ZLEN || new_mtu > ETH_JUMBO_MTU)
 		return -EINVAL;
 
 	netdev->mtu = new_mtu;
 
 	/* TODO: Tell rnic about new rmda interface mtu */
-	return ret;
+	return 0;
 }
 
+static const struct net_device_ops c2_pseudo_netdev_ops = {
+	.ndo_open 		= c2_pseudo_up,
+	.ndo_stop 		= c2_pseudo_down,
+	.ndo_start_xmit 	= c2_pseudo_xmit_frame,
+	.ndo_change_mtu 	= c2_pseudo_change_mtu,
+	.ndo_validate_addr	= eth_validate_addr,
+};
+
 static void setup(struct net_device *netdev)
 {
-	netdev->open = c2_pseudo_up;
-	netdev->stop = c2_pseudo_down;
-	netdev->hard_start_xmit = c2_pseudo_xmit_frame;
-	netdev->get_stats = NULL;
-	netdev->tx_timeout = NULL;
-	netdev->set_mac_address = NULL;
-	netdev->change_mtu = c2_pseudo_change_mtu;
+	netdev->netdev_ops = &c2_pseudo_netdev_ops;
+
 	netdev->watchdog_timeo = 0;
 	netdev->type = ARPHRD_ETHER;
 	netdev->mtu = 1500;
@@ -735,7 +736,6 @@ static void setup(struct net_device *netdev)
 	netdev->addr_len = ETH_ALEN;
 	netdev->tx_queue_len = 0;
 	netdev->flags |= IFF_NOARP;
-	return;
 }
 
 static struct net_device *c2_pseudo_netdev_init(struct c2_dev *c2dev)
diff --git a/drivers/infiniband/hw/cxgb3/cxio_hal.c b/drivers/infiniband/hw/cxgb3/cxio_hal.c
index 4dcf08b3fd83..a4a82bff7100 100644
--- a/drivers/infiniband/hw/cxgb3/cxio_hal.c
+++ b/drivers/infiniband/hw/cxgb3/cxio_hal.c
@@ -450,7 +450,7 @@ static int cqe_completes_wr(struct t3_cqe *cqe, struct t3_wq *wq)
 	if ((CQE_OPCODE(*cqe) == T3_READ_RESP) && SQ_TYPE(*cqe))
 		return 0;
 
-	if ((CQE_OPCODE(*cqe) == T3_SEND) && RQ_TYPE(*cqe) &&
+	if (CQE_SEND_OPCODE(*cqe) && RQ_TYPE(*cqe) &&
 	    Q_EMPTY(wq->rq_rptr, wq->rq_wptr))
 		return 0;
 
@@ -701,6 +701,9 @@ static int __cxio_tpt_op(struct cxio_rdev *rdev_p, u32 reset_tpt_entry,
 	u32 stag_idx;
 	u32 wptr;
 
+	if (rdev_p->flags)
+		return -EIO;
+
 	stag_state = stag_state > 0;
 	stag_idx = (*stag) >> 8;
 
@@ -938,6 +941,23 @@ int cxio_rdev_open(struct cxio_rdev *rdev_p)
 	if (!rdev_p->t3cdev_p)
 		rdev_p->t3cdev_p = dev2t3cdev(netdev_p);
 	rdev_p->t3cdev_p->ulp = (void *) rdev_p;
+
+	err = rdev_p->t3cdev_p->ctl(rdev_p->t3cdev_p, GET_EMBEDDED_INFO,
+					 &(rdev_p->fw_info));
+	if (err) {
+		printk(KERN_ERR "%s t3cdev_p(%p)->ctl returned error %d.\n",
+		     __func__, rdev_p->t3cdev_p, err);
+		goto err1;
+	}
+	if (G_FW_VERSION_MAJOR(rdev_p->fw_info.fw_vers) != CXIO_FW_MAJ) {
+		printk(KERN_ERR MOD "fatal firmware version mismatch: "
+		       "need version %u but adapter has version %u\n",
+		       CXIO_FW_MAJ,
+		       G_FW_VERSION_MAJOR(rdev_p->fw_info.fw_vers));
+		err = -EINVAL;
+		goto err1;
+	}
+
 	err = rdev_p->t3cdev_p->ctl(rdev_p->t3cdev_p, RDMA_GET_PARAMS,
 					 &(rdev_p->rnic_info));
 	if (err) {
@@ -1204,11 +1224,12 @@ int cxio_poll_cq(struct t3_wq *wq, struct t3_cq *cq, struct t3_cqe *cqe,
 		}
 
 		/* incoming SEND with no receive posted failures */
-		if ((CQE_OPCODE(*hw_cqe) == T3_SEND) && RQ_TYPE(*hw_cqe) &&
+		if (CQE_SEND_OPCODE(*hw_cqe) && RQ_TYPE(*hw_cqe) &&
 		    Q_EMPTY(wq->rq_rptr, wq->rq_wptr)) {
 			ret = -1;
 			goto skip_cqe;
 		}
+		BUG_ON((*cqe_flushed == 0) && !SW_CQE(*hw_cqe));
 		goto proc_cqe;
 	}
 
@@ -1223,6 +1244,13 @@ int cxio_poll_cq(struct t3_wq *wq, struct t3_cq *cq, struct t3_cqe *cqe,
 		 * then we complete this with TPT_ERR_MSN and mark the wq in
 		 * error.
 		 */
+
+		if (Q_EMPTY(wq->rq_rptr, wq->rq_wptr)) {
+			wq->error = 1;
+			ret = -1;
+			goto skip_cqe;
+		}
+
 		if (unlikely((CQE_WRID_MSN(*hw_cqe) != (wq->rq_rptr + 1)))) {
 			wq->error = 1;
 			hw_cqe->header |= htonl(V_CQE_STATUS(TPT_ERR_MSN));
@@ -1277,6 +1305,7 @@ proc_cqe:
 			cxio_hal_pblpool_free(wq->rdev,
 				wq->rq[Q_PTR2IDX(wq->rq_rptr,
 				wq->rq_size_log2)].pbl_addr, T3_STAG0_PBL_SIZE);
+		BUG_ON(Q_EMPTY(wq->rq_rptr, wq->rq_wptr));
 		wq->rq_rptr++;
 	}
 
diff --git a/drivers/infiniband/hw/cxgb3/cxio_hal.h b/drivers/infiniband/hw/cxgb3/cxio_hal.h
index 656fe47bc84f..094a66d1480c 100644
--- a/drivers/infiniband/hw/cxgb3/cxio_hal.h
+++ b/drivers/infiniband/hw/cxgb3/cxio_hal.h
@@ -61,6 +61,8 @@
 
 #define T3_MAX_DEV_NAME_LEN 32
 
+#define CXIO_FW_MAJ 7
+
 struct cxio_hal_ctrl_qp {
 	u32 wptr;
 	u32 rptr;
@@ -108,6 +110,9 @@ struct cxio_rdev {
 	struct gen_pool *pbl_pool;
 	struct gen_pool *rqt_pool;
 	struct list_head entry;
+	struct ch_embedded_info fw_info;
+	u32	flags;
+#define	CXIO_ERROR_FATAL	1
 };
 
 static inline int cxio_num_stags(struct cxio_rdev *rdev_p)
diff --git a/drivers/infiniband/hw/cxgb3/cxio_wr.h b/drivers/infiniband/hw/cxgb3/cxio_wr.h
index 04618f7bfbb3..ff9be1a13106 100644
--- a/drivers/infiniband/hw/cxgb3/cxio_wr.h
+++ b/drivers/infiniband/hw/cxgb3/cxio_wr.h
@@ -604,6 +604,12 @@ struct t3_cqe {
 #define CQE_STATUS(x)     (G_CQE_STATUS(be32_to_cpu((x).header)))
 #define CQE_OPCODE(x)     (G_CQE_OPCODE(be32_to_cpu((x).header)))
 
+#define CQE_SEND_OPCODE(x)( \
+	(G_CQE_OPCODE(be32_to_cpu((x).header)) == T3_SEND) || \
+	(G_CQE_OPCODE(be32_to_cpu((x).header)) == T3_SEND_WITH_SE) || \
+	(G_CQE_OPCODE(be32_to_cpu((x).header)) == T3_SEND_WITH_INV) || \
+	(G_CQE_OPCODE(be32_to_cpu((x).header)) == T3_SEND_WITH_SE_INV))
+
 #define CQE_LEN(x)        (be32_to_cpu((x).len))
 
 /* used for RQ completion processing */
diff --git a/drivers/infiniband/hw/cxgb3/iwch.c b/drivers/infiniband/hw/cxgb3/iwch.c
index 4489c89d6710..37a4fc264a07 100644
--- a/drivers/infiniband/hw/cxgb3/iwch.c
+++ b/drivers/infiniband/hw/cxgb3/iwch.c
@@ -51,13 +51,15 @@ cxgb3_cpl_handler_func t3c_handlers[NUM_CPL_CMDS];
 
 static void open_rnic_dev(struct t3cdev *);
 static void close_rnic_dev(struct t3cdev *);
+static void iwch_err_handler(struct t3cdev *, u32, u32);
 
 struct cxgb3_client t3c_client = {
 	.name = "iw_cxgb3",
 	.add = open_rnic_dev,
 	.remove = close_rnic_dev,
 	.handlers = t3c_handlers,
-	.redirect = iwch_ep_redirect
+	.redirect = iwch_ep_redirect,
+	.err_handler = iwch_err_handler
 };
 
 static LIST_HEAD(dev_list);
@@ -160,6 +162,17 @@ static void close_rnic_dev(struct t3cdev *tdev)
 	mutex_unlock(&dev_mutex);
 }
 
+static void iwch_err_handler(struct t3cdev *tdev, u32 status, u32 error)
+{
+	struct cxio_rdev *rdev = tdev->ulp;
+
+	if (status == OFFLOAD_STATUS_DOWN)
+		rdev->flags = CXIO_ERROR_FATAL;
+
+	return;
+
+}
+
 static int __init iwch_init_module(void)
 {
 	int err;
diff --git a/drivers/infiniband/hw/cxgb3/iwch_cm.c b/drivers/infiniband/hw/cxgb3/iwch_cm.c
index 44e936e48a31..8699947aaf6c 100644
--- a/drivers/infiniband/hw/cxgb3/iwch_cm.c
+++ b/drivers/infiniband/hw/cxgb3/iwch_cm.c
@@ -1678,6 +1678,9 @@ static int terminate(struct t3cdev *tdev, struct sk_buff *skb, void *ctx)
 {
 	struct iwch_ep *ep = ctx;
 
+	if (state_read(&ep->com) != FPDU_MODE)
+		return CPL_RET_BUF_DONE;
+
 	PDBG("%s ep %p\n", __func__, ep);
 	skb_pull(skb, sizeof(struct cpl_rdma_terminate));
 	PDBG("%s saving %d bytes of term msg\n", __func__, skb->len);
diff --git a/drivers/infiniband/hw/cxgb3/iwch_ev.c b/drivers/infiniband/hw/cxgb3/iwch_ev.c
index 7b67a6771720..743c5d8b8806 100644
--- a/drivers/infiniband/hw/cxgb3/iwch_ev.c
+++ b/drivers/infiniband/hw/cxgb3/iwch_ev.c
@@ -179,11 +179,6 @@ void iwch_ev_dispatch(struct cxio_rdev *rdev_p, struct sk_buff *skb)
 	case TPT_ERR_BOUND:
 	case TPT_ERR_INVALIDATE_SHARED_MR:
 	case TPT_ERR_INVALIDATE_MR_WITH_MW_BOUND:
-		printk(KERN_ERR "%s - CQE Err qpid 0x%x opcode %d status 0x%x "
-		       "type %d wrid.hi 0x%x wrid.lo 0x%x \n", __func__,
-		       CQE_QPID(rsp_msg->cqe), CQE_OPCODE(rsp_msg->cqe),
-		       CQE_STATUS(rsp_msg->cqe), CQE_TYPE(rsp_msg->cqe),
-		       CQE_WRID_HI(rsp_msg->cqe), CQE_WRID_LOW(rsp_msg->cqe));
 		(*chp->ibcq.comp_handler)(&chp->ibcq, chp->ibcq.cq_context);
 		post_qp_event(rnicp, chp, rsp_msg, IB_EVENT_QP_ACCESS_ERR, 1);
 		break;
diff --git a/drivers/infiniband/hw/cxgb3/iwch_qp.c b/drivers/infiniband/hw/cxgb3/iwch_qp.c
index 19661b2f0406..c758fbd58478 100644
--- a/drivers/infiniband/hw/cxgb3/iwch_qp.c
+++ b/drivers/infiniband/hw/cxgb3/iwch_qp.c
@@ -99,8 +99,8 @@ static int build_rdma_write(union t3_wr *wqe, struct ib_send_wr *wr,
 	if (wr->opcode == IB_WR_RDMA_WRITE_WITH_IMM) {
 		plen = 4;
 		wqe->write.sgl[0].stag = wr->ex.imm_data;
-		wqe->write.sgl[0].len = __constant_cpu_to_be32(0);
-		wqe->write.num_sgle = __constant_cpu_to_be32(0);
+		wqe->write.sgl[0].len = cpu_to_be32(0);
+		wqe->write.num_sgle = cpu_to_be32(0);
 		*flit_cnt = 6;
 	} else {
 		plen = 0;
@@ -195,15 +195,12 @@ static int build_inv_stag(union t3_wr *wqe, struct ib_send_wr *wr,
 	return 0;
 }
 
-/*
- * TBD: this is going to be moved to firmware. Missing pdid/qpid check for now.
- */
 static int iwch_sgl2pbl_map(struct iwch_dev *rhp, struct ib_sge *sg_list,
 			    u32 num_sgle, u32 * pbl_addr, u8 * page_size)
 {
 	int i;
 	struct iwch_mr *mhp;
-	u32 offset;
+	u64 offset;
 	for (i = 0; i < num_sgle; i++) {
 
 		mhp = get_mhp(rhp, (sg_list[i].lkey) >> 8);
@@ -235,8 +232,8 @@ static int iwch_sgl2pbl_map(struct iwch_dev *rhp, struct ib_sge *sg_list,
 			return -EINVAL;
 		}
 		offset = sg_list[i].addr - mhp->attr.va_fbo;
-		offset += ((u32) mhp->attr.va_fbo) %
-		          (1UL << (12 + mhp->attr.page_size));
+		offset += mhp->attr.va_fbo &
+			  ((1UL << (12 + mhp->attr.page_size)) - 1);
 		pbl_addr[i] = ((mhp->attr.pbl_addr -
 			        rhp->rdev.rnic_info.pbl_base) >> 3) +
 			      (offset >> (12 + mhp->attr.page_size));
@@ -266,8 +263,8 @@ static int build_rdma_recv(struct iwch_qp *qhp, union t3_wr *wqe,
 		wqe->recv.sgl[i].len = cpu_to_be32(wr->sg_list[i].length);
 
 		/* to in the WQE == the offset into the page */
-		wqe->recv.sgl[i].to = cpu_to_be64(((u32) wr->sg_list[i].addr) %
-				(1UL << (12 + page_size[i])));
+		wqe->recv.sgl[i].to = cpu_to_be64(((u32)wr->sg_list[i].addr) &
+				((1UL << (12 + page_size[i])) - 1));
 
 		/* pbl_addr is the adapters address in the PBL */
 		wqe->recv.pbl_addr[i] = cpu_to_be32(pbl_addr[i]);
diff --git a/drivers/infiniband/hw/ehca/ehca_sqp.c b/drivers/infiniband/hw/ehca/ehca_sqp.c
index 44447aaa5501..c568b28f4e20 100644
--- a/drivers/infiniband/hw/ehca/ehca_sqp.c
+++ b/drivers/infiniband/hw/ehca/ehca_sqp.c
@@ -46,11 +46,11 @@
 #include "ehca_iverbs.h"
 #include "hcp_if.h"
 
-#define IB_MAD_STATUS_REDIRECT		__constant_htons(0x0002)
-#define IB_MAD_STATUS_UNSUP_VERSION	__constant_htons(0x0004)
-#define IB_MAD_STATUS_UNSUP_METHOD	__constant_htons(0x0008)
+#define IB_MAD_STATUS_REDIRECT		cpu_to_be16(0x0002)
+#define IB_MAD_STATUS_UNSUP_VERSION	cpu_to_be16(0x0004)
+#define IB_MAD_STATUS_UNSUP_METHOD	cpu_to_be16(0x0008)
 
-#define IB_PMA_CLASS_PORT_INFO		__constant_htons(0x0001)
+#define IB_PMA_CLASS_PORT_INFO		cpu_to_be16(0x0001)
 
 /**
  * ehca_define_sqp - Defines special queue pair 1 (GSI QP). When special queue
diff --git a/drivers/infiniband/hw/ipath/ipath_driver.c b/drivers/infiniband/hw/ipath/ipath_driver.c
index 69c0ce321b4e..cb9daa6ac029 100644
--- a/drivers/infiniband/hw/ipath/ipath_driver.c
+++ b/drivers/infiniband/hw/ipath/ipath_driver.c
@@ -2715,7 +2715,7 @@ static void ipath_hol_signal_up(struct ipath_devdata *dd)
  * to prevent HoL blocking, then start the HoL timer that
  * periodically continues, then stop procs, so they can detect
  * link down if they want, and do something about it.
- * Timer may already be running, so use __mod_timer, not add_timer.
+ * Timer may already be running, so use mod_timer, not add_timer.
  */
 void ipath_hol_down(struct ipath_devdata *dd)
 {
@@ -2724,7 +2724,7 @@ void ipath_hol_down(struct ipath_devdata *dd)
 	dd->ipath_hol_next = IPATH_HOL_DOWNCONT;
 	dd->ipath_hol_timer.expires = jiffies +
 		msecs_to_jiffies(ipath_hol_timeout_ms);
-	__mod_timer(&dd->ipath_hol_timer, dd->ipath_hol_timer.expires);
+	mod_timer(&dd->ipath_hol_timer, dd->ipath_hol_timer.expires);
 }
 
 /*
@@ -2763,7 +2763,7 @@ void ipath_hol_event(unsigned long opaque)
 	else {
 		dd->ipath_hol_timer.expires = jiffies +
 			msecs_to_jiffies(ipath_hol_timeout_ms);
-		__mod_timer(&dd->ipath_hol_timer,
+		mod_timer(&dd->ipath_hol_timer,
 			dd->ipath_hol_timer.expires);
 	}
 }
diff --git a/drivers/infiniband/hw/ipath/ipath_eeprom.c b/drivers/infiniband/hw/ipath/ipath_eeprom.c
index dc37277f1c80..fc7181985e8e 100644
--- a/drivers/infiniband/hw/ipath/ipath_eeprom.c
+++ b/drivers/infiniband/hw/ipath/ipath_eeprom.c
@@ -772,8 +772,8 @@ void ipath_get_eeprom_info(struct ipath_devdata *dd)
 			 "0x%x, not 0x%x\n", csum, ifp->if_csum);
 		goto done;
 	}
-	if (*(__be64 *) ifp->if_guid == 0ULL ||
-	    *(__be64 *) ifp->if_guid == __constant_cpu_to_be64(-1LL)) {
+	if (*(__be64 *) ifp->if_guid == cpu_to_be64(0) ||
+	    *(__be64 *) ifp->if_guid == ~cpu_to_be64(0)) {
 		ipath_dev_err(dd, "Invalid GUID %llx from flash; "
 			      "ignoring\n",
 			      *(unsigned long long *) ifp->if_guid);
diff --git a/drivers/infiniband/hw/ipath/ipath_init_chip.c b/drivers/infiniband/hw/ipath/ipath_init_chip.c
index 64aeefbd2a5d..077879c0bdb5 100644
--- a/drivers/infiniband/hw/ipath/ipath_init_chip.c
+++ b/drivers/infiniband/hw/ipath/ipath_init_chip.c
@@ -455,7 +455,7 @@ static void init_shadow_tids(struct ipath_devdata *dd)
 	if (!addrs) {
 		ipath_dev_err(dd, "failed to allocate shadow dma handle "
 			      "array, no expected sends!\n");
-		vfree(dd->ipath_pageshadow);
+		vfree(pages);
 		dd->ipath_pageshadow = NULL;
 		return;
 	}
diff --git a/drivers/infiniband/hw/ipath/ipath_mad.c b/drivers/infiniband/hw/ipath/ipath_mad.c
index 17a123197477..16a702d46018 100644
--- a/drivers/infiniband/hw/ipath/ipath_mad.c
+++ b/drivers/infiniband/hw/ipath/ipath_mad.c
@@ -37,10 +37,10 @@
 #include "ipath_verbs.h"
 #include "ipath_common.h"
 
-#define IB_SMP_UNSUP_VERSION	__constant_htons(0x0004)
-#define IB_SMP_UNSUP_METHOD	__constant_htons(0x0008)
-#define IB_SMP_UNSUP_METH_ATTR	__constant_htons(0x000C)
-#define IB_SMP_INVALID_FIELD	__constant_htons(0x001C)
+#define IB_SMP_UNSUP_VERSION	cpu_to_be16(0x0004)
+#define IB_SMP_UNSUP_METHOD	cpu_to_be16(0x0008)
+#define IB_SMP_UNSUP_METH_ATTR	cpu_to_be16(0x000C)
+#define IB_SMP_INVALID_FIELD	cpu_to_be16(0x001C)
 
 static int reply(struct ib_smp *smp)
 {
@@ -789,12 +789,12 @@ static int recv_subn_set_pkeytable(struct ib_smp *smp,
 	return recv_subn_get_pkeytable(smp, ibdev);
 }
 
-#define IB_PMA_CLASS_PORT_INFO		__constant_htons(0x0001)
-#define IB_PMA_PORT_SAMPLES_CONTROL	__constant_htons(0x0010)
-#define IB_PMA_PORT_SAMPLES_RESULT	__constant_htons(0x0011)
-#define IB_PMA_PORT_COUNTERS		__constant_htons(0x0012)
-#define IB_PMA_PORT_COUNTERS_EXT	__constant_htons(0x001D)
-#define IB_PMA_PORT_SAMPLES_RESULT_EXT	__constant_htons(0x001E)
+#define IB_PMA_CLASS_PORT_INFO		cpu_to_be16(0x0001)
+#define IB_PMA_PORT_SAMPLES_CONTROL	cpu_to_be16(0x0010)
+#define IB_PMA_PORT_SAMPLES_RESULT	cpu_to_be16(0x0011)
+#define IB_PMA_PORT_COUNTERS		cpu_to_be16(0x0012)
+#define IB_PMA_PORT_COUNTERS_EXT	cpu_to_be16(0x001D)
+#define IB_PMA_PORT_SAMPLES_RESULT_EXT	cpu_to_be16(0x001E)
 
 struct ib_perf {
 	u8 base_version;
@@ -884,19 +884,19 @@ struct ib_pma_portcounters {
 	__be32 port_rcv_packets;
 } __attribute__ ((packed));
 
-#define IB_PMA_SEL_SYMBOL_ERROR			__constant_htons(0x0001)
-#define IB_PMA_SEL_LINK_ERROR_RECOVERY		__constant_htons(0x0002)
-#define IB_PMA_SEL_LINK_DOWNED			__constant_htons(0x0004)
-#define IB_PMA_SEL_PORT_RCV_ERRORS		__constant_htons(0x0008)
-#define IB_PMA_SEL_PORT_RCV_REMPHYS_ERRORS	__constant_htons(0x0010)
-#define IB_PMA_SEL_PORT_XMIT_DISCARDS		__constant_htons(0x0040)
-#define IB_PMA_SEL_LOCAL_LINK_INTEGRITY_ERRORS	__constant_htons(0x0200)
-#define IB_PMA_SEL_EXCESSIVE_BUFFER_OVERRUNS	__constant_htons(0x0400)
-#define IB_PMA_SEL_PORT_VL15_DROPPED		__constant_htons(0x0800)
-#define IB_PMA_SEL_PORT_XMIT_DATA		__constant_htons(0x1000)
-#define IB_PMA_SEL_PORT_RCV_DATA		__constant_htons(0x2000)
-#define IB_PMA_SEL_PORT_XMIT_PACKETS		__constant_htons(0x4000)
-#define IB_PMA_SEL_PORT_RCV_PACKETS		__constant_htons(0x8000)
+#define IB_PMA_SEL_SYMBOL_ERROR			cpu_to_be16(0x0001)
+#define IB_PMA_SEL_LINK_ERROR_RECOVERY		cpu_to_be16(0x0002)
+#define IB_PMA_SEL_LINK_DOWNED			cpu_to_be16(0x0004)
+#define IB_PMA_SEL_PORT_RCV_ERRORS		cpu_to_be16(0x0008)
+#define IB_PMA_SEL_PORT_RCV_REMPHYS_ERRORS	cpu_to_be16(0x0010)
+#define IB_PMA_SEL_PORT_XMIT_DISCARDS		cpu_to_be16(0x0040)
+#define IB_PMA_SEL_LOCAL_LINK_INTEGRITY_ERRORS	cpu_to_be16(0x0200)
+#define IB_PMA_SEL_EXCESSIVE_BUFFER_OVERRUNS	cpu_to_be16(0x0400)
+#define IB_PMA_SEL_PORT_VL15_DROPPED		cpu_to_be16(0x0800)
+#define IB_PMA_SEL_PORT_XMIT_DATA		cpu_to_be16(0x1000)
+#define IB_PMA_SEL_PORT_RCV_DATA		cpu_to_be16(0x2000)
+#define IB_PMA_SEL_PORT_XMIT_PACKETS		cpu_to_be16(0x4000)
+#define IB_PMA_SEL_PORT_RCV_PACKETS		cpu_to_be16(0x8000)
 
 struct ib_pma_portcounters_ext {
 	u8 reserved;
@@ -913,14 +913,14 @@ struct ib_pma_portcounters_ext {
 	__be64 port_multicast_rcv_packets;
 } __attribute__ ((packed));
 
-#define IB_PMA_SELX_PORT_XMIT_DATA		__constant_htons(0x0001)
-#define IB_PMA_SELX_PORT_RCV_DATA		__constant_htons(0x0002)
-#define IB_PMA_SELX_PORT_XMIT_PACKETS		__constant_htons(0x0004)
-#define IB_PMA_SELX_PORT_RCV_PACKETS		__constant_htons(0x0008)
-#define IB_PMA_SELX_PORT_UNI_XMIT_PACKETS	__constant_htons(0x0010)
-#define IB_PMA_SELX_PORT_UNI_RCV_PACKETS	__constant_htons(0x0020)
-#define IB_PMA_SELX_PORT_MULTI_XMIT_PACKETS	__constant_htons(0x0040)
-#define IB_PMA_SELX_PORT_MULTI_RCV_PACKETS	__constant_htons(0x0080)
+#define IB_PMA_SELX_PORT_XMIT_DATA		cpu_to_be16(0x0001)
+#define IB_PMA_SELX_PORT_RCV_DATA		cpu_to_be16(0x0002)
+#define IB_PMA_SELX_PORT_XMIT_PACKETS		cpu_to_be16(0x0004)
+#define IB_PMA_SELX_PORT_RCV_PACKETS		cpu_to_be16(0x0008)
+#define IB_PMA_SELX_PORT_UNI_XMIT_PACKETS	cpu_to_be16(0x0010)
+#define IB_PMA_SELX_PORT_UNI_RCV_PACKETS	cpu_to_be16(0x0020)
+#define IB_PMA_SELX_PORT_MULTI_XMIT_PACKETS	cpu_to_be16(0x0040)
+#define IB_PMA_SELX_PORT_MULTI_RCV_PACKETS	cpu_to_be16(0x0080)
 
 static int recv_pma_get_classportinfo(struct ib_perf *pmp)
 {
@@ -933,7 +933,7 @@ static int recv_pma_get_classportinfo(struct ib_perf *pmp)
 		pmp->status |= IB_SMP_INVALID_FIELD;
 
 	/* Indicate AllPortSelect is valid (only one port anyway) */
-	p->cap_mask = __constant_cpu_to_be16(1 << 8);
+	p->cap_mask = cpu_to_be16(1 << 8);
 	p->base_version = 1;
 	p->class_version = 1;
 	/*
@@ -951,12 +951,11 @@ static int recv_pma_get_classportinfo(struct ib_perf *pmp)
  * We support 5 counters which only count the mandatory quantities.
  */
 #define COUNTER_MASK(q, n) (q << ((9 - n) * 3))
-#define COUNTER_MASK0_9 \
-	__constant_cpu_to_be32(COUNTER_MASK(1, 0) | \
-			       COUNTER_MASK(1, 1) | \
-			       COUNTER_MASK(1, 2) | \
-			       COUNTER_MASK(1, 3) | \
-			       COUNTER_MASK(1, 4))
+#define COUNTER_MASK0_9 cpu_to_be32(COUNTER_MASK(1, 0) | \
+				    COUNTER_MASK(1, 1) | \
+				    COUNTER_MASK(1, 2) | \
+				    COUNTER_MASK(1, 3) | \
+				    COUNTER_MASK(1, 4))
 
 static int recv_pma_get_portsamplescontrol(struct ib_perf *pmp,
 					   struct ib_device *ibdev, u8 port)
@@ -1137,7 +1136,7 @@ static int recv_pma_get_portsamplesresult_ext(struct ib_perf *pmp,
 		status = dev->pma_sample_status;
 	p->sample_status = cpu_to_be16(status);
 	/* 64 bits */
-	p->extended_width = __constant_cpu_to_be32(0x80000000);
+	p->extended_width = cpu_to_be32(0x80000000);
 	for (i = 0; i < ARRAY_SIZE(dev->pma_counter_select); i++)
 		p->counter[i] = (status != IB_PMA_SAMPLE_STATUS_DONE) ? 0 :
 		    cpu_to_be64(
@@ -1185,7 +1184,7 @@ static int recv_pma_get_portcounters(struct ib_perf *pmp,
 		pmp->status |= IB_SMP_INVALID_FIELD;
 
 	if (cntrs.symbol_error_counter > 0xFFFFUL)
-		p->symbol_error_counter = __constant_cpu_to_be16(0xFFFF);
+		p->symbol_error_counter = cpu_to_be16(0xFFFF);
 	else
 		p->symbol_error_counter =
 			cpu_to_be16((u16)cntrs.symbol_error_counter);
@@ -1199,17 +1198,17 @@ static int recv_pma_get_portcounters(struct ib_perf *pmp,
 	else
 		p->link_downed_counter = (u8)cntrs.link_downed_counter;
 	if (cntrs.port_rcv_errors > 0xFFFFUL)
-		p->port_rcv_errors = __constant_cpu_to_be16(0xFFFF);
+		p->port_rcv_errors = cpu_to_be16(0xFFFF);
 	else
 		p->port_rcv_errors =
 			cpu_to_be16((u16) cntrs.port_rcv_errors);
 	if (cntrs.port_rcv_remphys_errors > 0xFFFFUL)
-		p->port_rcv_remphys_errors = __constant_cpu_to_be16(0xFFFF);
+		p->port_rcv_remphys_errors = cpu_to_be16(0xFFFF);
 	else
 		p->port_rcv_remphys_errors =
 			cpu_to_be16((u16)cntrs.port_rcv_remphys_errors);
 	if (cntrs.port_xmit_discards > 0xFFFFUL)
-		p->port_xmit_discards = __constant_cpu_to_be16(0xFFFF);
+		p->port_xmit_discards = cpu_to_be16(0xFFFF);
 	else
 		p->port_xmit_discards =
 			cpu_to_be16((u16)cntrs.port_xmit_discards);
@@ -1220,24 +1219,24 @@ static int recv_pma_get_portcounters(struct ib_perf *pmp,
 	p->lli_ebor_errors = (cntrs.local_link_integrity_errors << 4) |
 		cntrs.excessive_buffer_overrun_errors;
 	if (cntrs.vl15_dropped > 0xFFFFUL)
-		p->vl15_dropped = __constant_cpu_to_be16(0xFFFF);
+		p->vl15_dropped = cpu_to_be16(0xFFFF);
 	else
 		p->vl15_dropped = cpu_to_be16((u16)cntrs.vl15_dropped);
 	if (cntrs.port_xmit_data > 0xFFFFFFFFUL)
-		p->port_xmit_data = __constant_cpu_to_be32(0xFFFFFFFF);
+		p->port_xmit_data = cpu_to_be32(0xFFFFFFFF);
 	else
 		p->port_xmit_data = cpu_to_be32((u32)cntrs.port_xmit_data);
 	if (cntrs.port_rcv_data > 0xFFFFFFFFUL)
-		p->port_rcv_data = __constant_cpu_to_be32(0xFFFFFFFF);
+		p->port_rcv_data = cpu_to_be32(0xFFFFFFFF);
 	else
 		p->port_rcv_data = cpu_to_be32((u32)cntrs.port_rcv_data);
 	if (cntrs.port_xmit_packets > 0xFFFFFFFFUL)
-		p->port_xmit_packets = __constant_cpu_to_be32(0xFFFFFFFF);
+		p->port_xmit_packets = cpu_to_be32(0xFFFFFFFF);
 	else
 		p->port_xmit_packets =
 			cpu_to_be32((u32)cntrs.port_xmit_packets);
 	if (cntrs.port_rcv_packets > 0xFFFFFFFFUL)
-		p->port_rcv_packets = __constant_cpu_to_be32(0xFFFFFFFF);
+		p->port_rcv_packets = cpu_to_be32(0xFFFFFFFF);
 	else
 		p->port_rcv_packets =
 			cpu_to_be32((u32) cntrs.port_rcv_packets);
diff --git a/drivers/infiniband/hw/ipath/ipath_rc.c b/drivers/infiniband/hw/ipath/ipath_rc.c
index 9170710b950d..79b3dbc97179 100644
--- a/drivers/infiniband/hw/ipath/ipath_rc.c
+++ b/drivers/infiniband/hw/ipath/ipath_rc.c
@@ -1744,7 +1744,7 @@ void ipath_rc_rcv(struct ipath_ibdev *dev, struct ipath_ib_header *hdr,
 		/* Signal completion event if the solicited bit is set. */
 		ipath_cq_enter(to_icq(qp->ibqp.recv_cq), &wc,
 			       (ohdr->bth[0] &
-				__constant_cpu_to_be32(1 << 23)) != 0);
+				cpu_to_be32(1 << 23)) != 0);
 		break;
 
 	case OP(RDMA_WRITE_FIRST):
diff --git a/drivers/infiniband/hw/ipath/ipath_sdma.c b/drivers/infiniband/hw/ipath/ipath_sdma.c
index 8e255adf5d9b..4b0698590850 100644
--- a/drivers/infiniband/hw/ipath/ipath_sdma.c
+++ b/drivers/infiniband/hw/ipath/ipath_sdma.c
@@ -781,10 +781,10 @@ retry:
 		descqp = &dd->ipath_sdma_descq[dd->ipath_sdma_descq_cnt].qw[0];
 	descqp -= 2;
 	/* SDmaLastDesc */
-	descqp[0] |= __constant_cpu_to_le64(1ULL << 11);
+	descqp[0] |= cpu_to_le64(1ULL << 11);
 	if (tx->txreq.flags & IPATH_SDMA_TXREQ_F_INTREQ) {
 		/* SDmaIntReq */
-		descqp[0] |= __constant_cpu_to_le64(1ULL << 15);
+		descqp[0] |= cpu_to_le64(1ULL << 15);
 	}
 
 	/* Commit writes to memory and advance the tail on the chip */
diff --git a/drivers/infiniband/hw/ipath/ipath_uc.c b/drivers/infiniband/hw/ipath/ipath_uc.c
index 82cc588b8bf2..22e60998f1a7 100644
--- a/drivers/infiniband/hw/ipath/ipath_uc.c
+++ b/drivers/infiniband/hw/ipath/ipath_uc.c
@@ -419,7 +419,7 @@ void ipath_uc_rcv(struct ipath_ibdev *dev, struct ipath_ib_header *hdr,
 		/* Signal completion event if the solicited bit is set. */
 		ipath_cq_enter(to_icq(qp->ibqp.recv_cq), &wc,
 			       (ohdr->bth[0] &
-				__constant_cpu_to_be32(1 << 23)) != 0);
+				cpu_to_be32(1 << 23)) != 0);
 		break;
 
 	case OP(RDMA_WRITE_FIRST):
diff --git a/drivers/infiniband/hw/ipath/ipath_ud.c b/drivers/infiniband/hw/ipath/ipath_ud.c
index 91c74cc797ae..6076cb61bf6a 100644
--- a/drivers/infiniband/hw/ipath/ipath_ud.c
+++ b/drivers/infiniband/hw/ipath/ipath_ud.c
@@ -370,7 +370,7 @@ int ipath_make_ud_req(struct ipath_qp *qp)
 	 */
 	ohdr->bth[1] = ah_attr->dlid >= IPATH_MULTICAST_LID_BASE &&
 		ah_attr->dlid != IPATH_PERMISSIVE_LID ?
-		__constant_cpu_to_be32(IPATH_MULTICAST_QPN) :
+		cpu_to_be32(IPATH_MULTICAST_QPN) :
 		cpu_to_be32(wqe->wr.wr.ud.remote_qpn);
 	ohdr->bth[2] = cpu_to_be32(qp->s_next_psn++ & IPATH_PSN_MASK);
 	/*
@@ -573,7 +573,7 @@ void ipath_ud_rcv(struct ipath_ibdev *dev, struct ipath_ib_header *hdr,
 	/* Signal completion event if the solicited bit is set. */
 	ipath_cq_enter(to_icq(qp->ibqp.recv_cq), &wc,
 		       (ohdr->bth[0] &
-			__constant_cpu_to_be32(1 << 23)) != 0);
+			cpu_to_be32(1 << 23)) != 0);
 
 bail:;
 }
diff --git a/drivers/infiniband/hw/ipath/ipath_user_pages.c b/drivers/infiniband/hw/ipath/ipath_user_pages.c
index 0190edc8044e..855911e7396d 100644
--- a/drivers/infiniband/hw/ipath/ipath_user_pages.c
+++ b/drivers/infiniband/hw/ipath/ipath_user_pages.c
@@ -209,20 +209,20 @@ void ipath_release_user_pages_on_close(struct page **p, size_t num_pages)
 
 	mm = get_task_mm(current);
 	if (!mm)
-		goto bail;
+		return;
 
 	work = kmalloc(sizeof(*work), GFP_KERNEL);
 	if (!work)
 		goto bail_mm;
 
-	goto bail;
-
 	INIT_WORK(&work->work, user_pages_account);
 	work->mm = mm;
 	work->num_pages = num_pages;
 
+	schedule_work(&work->work);
+	return;
+
 bail_mm:
 	mmput(mm);
-bail:
 	return;
 }
diff --git a/drivers/infiniband/hw/ipath/ipath_user_sdma.c b/drivers/infiniband/hw/ipath/ipath_user_sdma.c
index 82d9a0b5ca2f..7bff4b9baa0a 100644
--- a/drivers/infiniband/hw/ipath/ipath_user_sdma.c
+++ b/drivers/infiniband/hw/ipath/ipath_user_sdma.c
@@ -667,13 +667,13 @@ static inline __le64 ipath_sdma_make_desc0(struct ipath_devdata *dd,
 
 static inline __le64 ipath_sdma_make_first_desc0(__le64 descq)
 {
-	return descq | __constant_cpu_to_le64(1ULL << 12);
+	return descq | cpu_to_le64(1ULL << 12);
 }
 
 static inline __le64 ipath_sdma_make_last_desc0(__le64 descq)
 {
 					      /* last */  /* dma head */
-	return descq | __constant_cpu_to_le64(1ULL << 11 | 1ULL << 13);
+	return descq | cpu_to_le64(1ULL << 11 | 1ULL << 13);
 }
 
 static inline __le64 ipath_sdma_make_desc1(u64 addr)
@@ -763,7 +763,7 @@ static int ipath_user_sdma_push_pkts(struct ipath_devdata *dd,
 		if (ofs >= IPATH_SMALLBUF_DWORDS) {
 			for (i = 0; i < pkt->naddr; i++) {
 				dd->ipath_sdma_descq[dtail].qw[0] |=
-					__constant_cpu_to_le64(1ULL << 14);
+					cpu_to_le64(1ULL << 14);
 				if (++dtail == dd->ipath_sdma_descq_cnt)
 					dtail = 0;
 			}
diff --git a/drivers/infiniband/hw/ipath/ipath_verbs.c b/drivers/infiniband/hw/ipath/ipath_verbs.c
index cdf0e6abd34d..9289ab4b0ae8 100644
--- a/drivers/infiniband/hw/ipath/ipath_verbs.c
+++ b/drivers/infiniband/hw/ipath/ipath_verbs.c
@@ -1585,7 +1585,7 @@ static int ipath_query_port(struct ib_device *ibdev,
 	u64 ibcstat;
 
 	memset(props, 0, sizeof(*props));
-	props->lid = lid ? lid : __constant_be16_to_cpu(IB_LID_PERMISSIVE);
+	props->lid = lid ? lid : be16_to_cpu(IB_LID_PERMISSIVE);
 	props->lmc = dd->ipath_lmc;
 	props->sm_lid = dev->sm_lid;
 	props->sm_sl = dev->sm_sl;
diff --git a/drivers/infiniband/hw/ipath/ipath_verbs.h b/drivers/infiniband/hw/ipath/ipath_verbs.h
index 11e3f613df93..ae6cff4abffc 100644
--- a/drivers/infiniband/hw/ipath/ipath_verbs.h
+++ b/drivers/infiniband/hw/ipath/ipath_verbs.h
@@ -86,11 +86,11 @@
 #define IB_PMA_SAMPLE_STATUS_RUNNING	0x02
 
 /* Mandatory IB performance counter select values. */
-#define IB_PMA_PORT_XMIT_DATA	__constant_htons(0x0001)
-#define IB_PMA_PORT_RCV_DATA	__constant_htons(0x0002)
-#define IB_PMA_PORT_XMIT_PKTS	__constant_htons(0x0003)
-#define IB_PMA_PORT_RCV_PKTS	__constant_htons(0x0004)
-#define IB_PMA_PORT_XMIT_WAIT	__constant_htons(0x0005)
+#define IB_PMA_PORT_XMIT_DATA	cpu_to_be16(0x0001)
+#define IB_PMA_PORT_RCV_DATA	cpu_to_be16(0x0002)
+#define IB_PMA_PORT_XMIT_PKTS	cpu_to_be16(0x0003)
+#define IB_PMA_PORT_RCV_PKTS	cpu_to_be16(0x0004)
+#define IB_PMA_PORT_XMIT_WAIT	cpu_to_be16(0x0005)
 
 struct ib_reth {
 	__be64 vaddr;
diff --git a/drivers/infiniband/hw/mlx4/mad.c b/drivers/infiniband/hw/mlx4/mad.c
index 606f1e2ef284..19e68ab66168 100644
--- a/drivers/infiniband/hw/mlx4/mad.c
+++ b/drivers/infiniband/hw/mlx4/mad.c
@@ -147,7 +147,8 @@ static void update_sm_ah(struct mlx4_ib_dev *dev, u8 port_num, u16 lid, u8 sl)
  * Snoop SM MADs for port info and P_Key table sets, so we can
  * synthesize LID change and P_Key change events.
  */
-static void smp_snoop(struct ib_device *ibdev, u8 port_num, struct ib_mad *mad)
+static void smp_snoop(struct ib_device *ibdev, u8 port_num, struct ib_mad *mad,
+				u16 prev_lid)
 {
 	struct ib_event event;
 
@@ -157,6 +158,7 @@ static void smp_snoop(struct ib_device *ibdev, u8 port_num, struct ib_mad *mad)
 		if (mad->mad_hdr.attr_id == IB_SMP_ATTR_PORT_INFO) {
 			struct ib_port_info *pinfo =
 				(struct ib_port_info *) ((struct ib_smp *) mad)->data;
+			u16 lid = be16_to_cpu(pinfo->lid);
 
 			update_sm_ah(to_mdev(ibdev), port_num,
 				     be16_to_cpu(pinfo->sm_lid),
@@ -165,12 +167,15 @@ static void smp_snoop(struct ib_device *ibdev, u8 port_num, struct ib_mad *mad)
 			event.device	       = ibdev;
 			event.element.port_num = port_num;
 
-			if (pinfo->clientrereg_resv_subnetto & 0x80)
+			if (pinfo->clientrereg_resv_subnetto & 0x80) {
 				event.event    = IB_EVENT_CLIENT_REREGISTER;
-			else
-				event.event    = IB_EVENT_LID_CHANGE;
+				ib_dispatch_event(&event);
+			}
 
-			ib_dispatch_event(&event);
+			if (prev_lid != lid) {
+				event.event    = IB_EVENT_LID_CHANGE;
+				ib_dispatch_event(&event);
+			}
 		}
 
 		if (mad->mad_hdr.attr_id == IB_SMP_ATTR_PKEY_TABLE) {
@@ -228,8 +233,9 @@ int mlx4_ib_process_mad(struct ib_device *ibdev, int mad_flags,	u8 port_num,
 			struct ib_wc *in_wc, struct ib_grh *in_grh,
 			struct ib_mad *in_mad, struct ib_mad *out_mad)
 {
-	u16 slid;
+	u16 slid, prev_lid = 0;
 	int err;
+	struct ib_port_attr pattr;
 
 	slid = in_wc ? in_wc->slid : be16_to_cpu(IB_LID_PERMISSIVE);
 
@@ -263,6 +269,13 @@ int mlx4_ib_process_mad(struct ib_device *ibdev, int mad_flags,	u8 port_num,
 	} else
 		return IB_MAD_RESULT_SUCCESS;
 
+	if ((in_mad->mad_hdr.mgmt_class == IB_MGMT_CLASS_SUBN_LID_ROUTED ||
+	     in_mad->mad_hdr.mgmt_class == IB_MGMT_CLASS_SUBN_DIRECTED_ROUTE) &&
+	    in_mad->mad_hdr.method == IB_MGMT_METHOD_SET &&
+	    in_mad->mad_hdr.attr_id == IB_SMP_ATTR_PORT_INFO &&
+	    !ib_query_port(ibdev, port_num, &pattr))
+		prev_lid = pattr.lid;
+
 	err = mlx4_MAD_IFC(to_mdev(ibdev),
 			   mad_flags & IB_MAD_IGNORE_MKEY,
 			   mad_flags & IB_MAD_IGNORE_BKEY,
@@ -271,7 +284,7 @@ int mlx4_ib_process_mad(struct ib_device *ibdev, int mad_flags,	u8 port_num,
 		return IB_MAD_RESULT_FAILURE;
 
 	if (!out_mad->mad_hdr.status) {
-		smp_snoop(ibdev, port_num, in_mad);
+		smp_snoop(ibdev, port_num, in_mad, prev_lid);
 		node_desc_override(ibdev, out_mad);
 	}
 
diff --git a/drivers/infiniband/hw/mlx4/main.c b/drivers/infiniband/hw/mlx4/main.c
index 61588bd273bd..2ccb9d31771f 100644
--- a/drivers/infiniband/hw/mlx4/main.c
+++ b/drivers/infiniband/hw/mlx4/main.c
@@ -699,11 +699,12 @@ static void mlx4_ib_remove(struct mlx4_dev *dev, void *ibdev_ptr)
 	struct mlx4_ib_dev *ibdev = ibdev_ptr;
 	int p;
 
+	mlx4_ib_mad_cleanup(ibdev);
+	ib_unregister_device(&ibdev->ib_dev);
+
 	for (p = 1; p <= ibdev->num_ports; ++p)
 		mlx4_CLOSE_PORT(dev, p);
 
-	mlx4_ib_mad_cleanup(ibdev);
-	ib_unregister_device(&ibdev->ib_dev);
 	iounmap(ibdev->uar_map);
 	mlx4_uar_free(dev, &ibdev->priv_uar);
 	mlx4_pd_free(dev, ibdev->priv_pdn);
diff --git a/drivers/infiniband/hw/mlx4/qp.c b/drivers/infiniband/hw/mlx4/qp.c
index a91cb4c3fa5c..f385a24d31d2 100644
--- a/drivers/infiniband/hw/mlx4/qp.c
+++ b/drivers/infiniband/hw/mlx4/qp.c
@@ -71,17 +71,17 @@ enum {
 };
 
 static const __be32 mlx4_ib_opcode[] = {
-	[IB_WR_SEND]			= __constant_cpu_to_be32(MLX4_OPCODE_SEND),
-	[IB_WR_LSO]			= __constant_cpu_to_be32(MLX4_OPCODE_LSO),
-	[IB_WR_SEND_WITH_IMM]		= __constant_cpu_to_be32(MLX4_OPCODE_SEND_IMM),
-	[IB_WR_RDMA_WRITE]		= __constant_cpu_to_be32(MLX4_OPCODE_RDMA_WRITE),
-	[IB_WR_RDMA_WRITE_WITH_IMM]	= __constant_cpu_to_be32(MLX4_OPCODE_RDMA_WRITE_IMM),
-	[IB_WR_RDMA_READ]		= __constant_cpu_to_be32(MLX4_OPCODE_RDMA_READ),
-	[IB_WR_ATOMIC_CMP_AND_SWP]	= __constant_cpu_to_be32(MLX4_OPCODE_ATOMIC_CS),
-	[IB_WR_ATOMIC_FETCH_AND_ADD]	= __constant_cpu_to_be32(MLX4_OPCODE_ATOMIC_FA),
-	[IB_WR_SEND_WITH_INV]		= __constant_cpu_to_be32(MLX4_OPCODE_SEND_INVAL),
-	[IB_WR_LOCAL_INV]		= __constant_cpu_to_be32(MLX4_OPCODE_LOCAL_INVAL),
-	[IB_WR_FAST_REG_MR]		= __constant_cpu_to_be32(MLX4_OPCODE_FMR),
+	[IB_WR_SEND]			= cpu_to_be32(MLX4_OPCODE_SEND),
+	[IB_WR_LSO]			= cpu_to_be32(MLX4_OPCODE_LSO),
+	[IB_WR_SEND_WITH_IMM]		= cpu_to_be32(MLX4_OPCODE_SEND_IMM),
+	[IB_WR_RDMA_WRITE]		= cpu_to_be32(MLX4_OPCODE_RDMA_WRITE),
+	[IB_WR_RDMA_WRITE_WITH_IMM]	= cpu_to_be32(MLX4_OPCODE_RDMA_WRITE_IMM),
+	[IB_WR_RDMA_READ]		= cpu_to_be32(MLX4_OPCODE_RDMA_READ),
+	[IB_WR_ATOMIC_CMP_AND_SWP]	= cpu_to_be32(MLX4_OPCODE_ATOMIC_CS),
+	[IB_WR_ATOMIC_FETCH_AND_ADD]	= cpu_to_be32(MLX4_OPCODE_ATOMIC_FA),
+	[IB_WR_SEND_WITH_INV]		= cpu_to_be32(MLX4_OPCODE_SEND_INVAL),
+	[IB_WR_LOCAL_INV]		= cpu_to_be32(MLX4_OPCODE_LOCAL_INVAL),
+	[IB_WR_FAST_REG_MR]		= cpu_to_be32(MLX4_OPCODE_FMR),
 };
 
 static struct mlx4_ib_sqp *to_msqp(struct mlx4_ib_qp *mqp)
diff --git a/drivers/infiniband/hw/mthca/mthca_mad.c b/drivers/infiniband/hw/mthca/mthca_mad.c
index 640449582aba..5648659ff0b0 100644
--- a/drivers/infiniband/hw/mthca/mthca_mad.c
+++ b/drivers/infiniband/hw/mthca/mthca_mad.c
@@ -104,7 +104,8 @@ static void update_sm_ah(struct mthca_dev *dev,
  */
 static void smp_snoop(struct ib_device *ibdev,
 		      u8 port_num,
-		      struct ib_mad *mad)
+		      struct ib_mad *mad,
+		      u16 prev_lid)
 {
 	struct ib_event event;
 
@@ -114,6 +115,7 @@ static void smp_snoop(struct ib_device *ibdev,
 		if (mad->mad_hdr.attr_id == IB_SMP_ATTR_PORT_INFO) {
 			struct ib_port_info *pinfo =
 				(struct ib_port_info *) ((struct ib_smp *) mad)->data;
+			u16 lid = be16_to_cpu(pinfo->lid);
 
 			mthca_update_rate(to_mdev(ibdev), port_num);
 			update_sm_ah(to_mdev(ibdev), port_num,
@@ -123,12 +125,15 @@ static void smp_snoop(struct ib_device *ibdev,
 			event.device           = ibdev;
 			event.element.port_num = port_num;
 
-			if (pinfo->clientrereg_resv_subnetto & 0x80)
+			if (pinfo->clientrereg_resv_subnetto & 0x80) {
 				event.event    = IB_EVENT_CLIENT_REREGISTER;
-			else
-				event.event    = IB_EVENT_LID_CHANGE;
+				ib_dispatch_event(&event);
+			}
 
-			ib_dispatch_event(&event);
+			if (prev_lid != lid) {
+				event.event    = IB_EVENT_LID_CHANGE;
+				ib_dispatch_event(&event);
+			}
 		}
 
 		if (mad->mad_hdr.attr_id == IB_SMP_ATTR_PKEY_TABLE) {
@@ -196,6 +201,8 @@ int mthca_process_mad(struct ib_device *ibdev,
 	int err;
 	u8 status;
 	u16 slid = in_wc ? in_wc->slid : be16_to_cpu(IB_LID_PERMISSIVE);
+	u16 prev_lid = 0;
+	struct ib_port_attr pattr;
 
 	/* Forward locally generated traps to the SM */
 	if (in_mad->mad_hdr.method == IB_MGMT_METHOD_TRAP &&
@@ -233,6 +240,12 @@ int mthca_process_mad(struct ib_device *ibdev,
 			return IB_MAD_RESULT_SUCCESS;
 	} else
 		return IB_MAD_RESULT_SUCCESS;
+	if ((in_mad->mad_hdr.mgmt_class == IB_MGMT_CLASS_SUBN_LID_ROUTED ||
+	     in_mad->mad_hdr.mgmt_class == IB_MGMT_CLASS_SUBN_DIRECTED_ROUTE) &&
+	    in_mad->mad_hdr.method == IB_MGMT_METHOD_SET &&
+	    in_mad->mad_hdr.attr_id == IB_SMP_ATTR_PORT_INFO &&
+	    !ib_query_port(ibdev, port_num, &pattr))
+		prev_lid = pattr.lid;
 
 	err = mthca_MAD_IFC(to_mdev(ibdev),
 			    mad_flags & IB_MAD_IGNORE_MKEY,
@@ -252,7 +265,7 @@ int mthca_process_mad(struct ib_device *ibdev,
 	}
 
 	if (!out_mad->mad_hdr.status) {
-		smp_snoop(ibdev, port_num, in_mad);
+		smp_snoop(ibdev, port_num, in_mad, prev_lid);
 		node_desc_override(ibdev, out_mad);
 	}
 
diff --git a/drivers/infiniband/hw/nes/nes.c b/drivers/infiniband/hw/nes/nes.c
index b9611ade9eab..ca599767ffbd 100644
--- a/drivers/infiniband/hw/nes/nes.c
+++ b/drivers/infiniband/hw/nes/nes.c
@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2006 - 2008 NetEffect, Inc. All rights reserved.
+ * Copyright (c) 2006 - 2009 Intel-NE, Inc.  All rights reserved.
  * Copyright (c) 2005 Open Grid Computing, Inc. All rights reserved.
  *
  * This software is available to you under a choice of one of two
diff --git a/drivers/infiniband/hw/nes/nes.h b/drivers/infiniband/hw/nes/nes.h
index 13a5bb1a7bcf..04b12ad23390 100644
--- a/drivers/infiniband/hw/nes/nes.h
+++ b/drivers/infiniband/hw/nes/nes.h
@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2006 - 2008 NetEffect, Inc. All rights reserved.
+ * Copyright (c) 2006 - 2009 Intel-NE, Inc.  All rights reserved.
  * Copyright (c) 2005 Open Grid Computing, Inc. All rights reserved.
  *
  * This software is available to you under a choice of one of two
diff --git a/drivers/infiniband/hw/nes/nes_cm.c b/drivers/infiniband/hw/nes/nes_cm.c
index 4a65b96db2c8..52425154acd4 100644
--- a/drivers/infiniband/hw/nes/nes_cm.c
+++ b/drivers/infiniband/hw/nes/nes_cm.c
@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2006 - 2008 NetEffect, Inc. All rights reserved.
+ * Copyright (c) 2006 - 2009 Intel-NE, Inc.  All rights reserved.
  *
  * This software is available to you under a choice of one of two
  * licenses.  You may choose to be licensed under the terms of the GNU
@@ -103,6 +103,7 @@ static int nes_disconnect(struct nes_qp *nesqp, int abrupt);
 static void nes_disconnect_worker(struct work_struct *work);
 
 static int send_mpa_request(struct nes_cm_node *, struct sk_buff *);
+static int send_mpa_reject(struct nes_cm_node *);
 static int send_syn(struct nes_cm_node *, u32, struct sk_buff *);
 static int send_reset(struct nes_cm_node *, struct sk_buff *);
 static int send_ack(struct nes_cm_node *cm_node, struct sk_buff *skb);
@@ -113,8 +114,7 @@ static void process_packet(struct nes_cm_node *, struct sk_buff *,
 static void active_open_err(struct nes_cm_node *, struct sk_buff *, int);
 static void passive_open_err(struct nes_cm_node *, struct sk_buff *, int);
 static void cleanup_retrans_entry(struct nes_cm_node *);
-static void handle_rcv_mpa(struct nes_cm_node *, struct sk_buff *,
-	enum nes_cm_event_type);
+static void handle_rcv_mpa(struct nes_cm_node *, struct sk_buff *);
 static void free_retrans_entry(struct nes_cm_node *cm_node);
 static int handle_tcp_options(struct nes_cm_node *cm_node, struct tcphdr *tcph,
 	struct sk_buff *skb, int optionsize, int passive);
@@ -124,6 +124,8 @@ static void cm_event_connected(struct nes_cm_event *);
 static void cm_event_connect_error(struct nes_cm_event *);
 static void cm_event_reset(struct nes_cm_event *);
 static void cm_event_mpa_req(struct nes_cm_event *);
+static void cm_event_mpa_reject(struct nes_cm_event *);
+static void handle_recv_entry(struct nes_cm_node *cm_node, u32 rem_node);
 
 static void print_core(struct nes_cm_core *core);
 
@@ -196,7 +198,6 @@ static struct nes_cm_event *create_event(struct nes_cm_node *cm_node,
  */
 static int send_mpa_request(struct nes_cm_node *cm_node, struct sk_buff *skb)
 {
-	int ret;
 	if (!skb) {
 		nes_debug(NES_DBG_CM, "skb set to NULL\n");
 		return -1;
@@ -206,11 +207,27 @@ static int send_mpa_request(struct nes_cm_node *cm_node, struct sk_buff *skb)
 	form_cm_frame(skb, cm_node, NULL, 0, &cm_node->mpa_frame,
 			cm_node->mpa_frame_size, SET_ACK);
 
-	ret = schedule_nes_timer(cm_node, skb, NES_TIMER_TYPE_SEND, 1, 0);
-	if (ret < 0)
-		return ret;
+	return schedule_nes_timer(cm_node, skb, NES_TIMER_TYPE_SEND, 1, 0);
+}
 
-	return 0;
+
+
+static int send_mpa_reject(struct nes_cm_node *cm_node)
+{
+	struct sk_buff  *skb = NULL;
+
+	skb = dev_alloc_skb(MAX_CM_BUFFER);
+	if (!skb) {
+		nes_debug(NES_DBG_CM, "Failed to get a Free pkt\n");
+		return -ENOMEM;
+	}
+
+	/* send an MPA reject frame */
+	form_cm_frame(skb, cm_node, NULL, 0, &cm_node->mpa_frame,
+			cm_node->mpa_frame_size, SET_ACK | SET_FIN);
+
+	cm_node->state = NES_CM_STATE_FIN_WAIT1;
+	return schedule_nes_timer(cm_node, skb, NES_TIMER_TYPE_SEND, 1, 0);
 }
 
 
@@ -218,14 +235,17 @@ static int send_mpa_request(struct nes_cm_node *cm_node, struct sk_buff *skb)
  * recv_mpa - process a received TCP pkt, we are expecting an
  * IETF MPA frame
  */
-static int parse_mpa(struct nes_cm_node *cm_node, u8 *buffer, u32 len)
+static int parse_mpa(struct nes_cm_node *cm_node, u8 *buffer, u32 *type,
+		u32 len)
 {
 	struct ietf_mpa_frame *mpa_frame;
 
+	*type = NES_MPA_REQUEST_ACCEPT;
+
 	/* assume req frame is in tcp data payload */
 	if (len < sizeof(struct ietf_mpa_frame)) {
 		nes_debug(NES_DBG_CM, "The received ietf buffer was too small (%x)\n", len);
-		return -1;
+		return -EINVAL;
 	}
 
 	mpa_frame = (struct ietf_mpa_frame *)buffer;
@@ -234,14 +254,25 @@ static int parse_mpa(struct nes_cm_node *cm_node, u8 *buffer, u32 len)
 	if (cm_node->mpa_frame_size + sizeof(struct ietf_mpa_frame) != len) {
 		nes_debug(NES_DBG_CM, "The received ietf buffer was not right"
 				" complete (%x + %x != %x)\n",
-				cm_node->mpa_frame_size, (u32)sizeof(struct ietf_mpa_frame), len);
-		return -1;
+				cm_node->mpa_frame_size,
+				(u32)sizeof(struct ietf_mpa_frame), len);
+		return -EINVAL;
+	}
+	/* make sure it does not exceed the max size */
+	if (len > MAX_CM_BUFFER) {
+		nes_debug(NES_DBG_CM, "The received ietf buffer was too large"
+				" (%x + %x != %x)\n",
+				cm_node->mpa_frame_size,
+				(u32)sizeof(struct ietf_mpa_frame), len);
+		return -EINVAL;
 	}
 
 	/* copy entire MPA frame to our cm_node's frame */
 	memcpy(cm_node->mpa_frame_buf, buffer + sizeof(struct ietf_mpa_frame),
 			cm_node->mpa_frame_size);
 
+	if (mpa_frame->flags & IETF_MPA_FLAGS_REJECT)
+		*type = NES_MPA_REQUEST_REJECT;
 	return 0;
 }
 
@@ -380,7 +411,7 @@ int schedule_nes_timer(struct nes_cm_node *cm_node, struct sk_buff *skb,
 
 	new_send = kzalloc(sizeof(*new_send), GFP_ATOMIC);
 	if (!new_send)
-		return -1;
+		return -ENOMEM;
 
 	/* new_send->timetosend = currenttime */
 	new_send->retrycount = NES_DEFAULT_RETRYS;
@@ -394,9 +425,11 @@ int schedule_nes_timer(struct nes_cm_node *cm_node, struct sk_buff *skb,
 
 	if (type == NES_TIMER_TYPE_CLOSE) {
 		new_send->timetosend += (HZ/10);
-		spin_lock_irqsave(&cm_node->recv_list_lock, flags);
-		list_add_tail(&new_send->list, &cm_node->recv_list);
-		spin_unlock_irqrestore(&cm_node->recv_list_lock, flags);
+		if (cm_node->recv_entry) {
+			WARN_ON(1);
+			return -EINVAL;
+		}
+		cm_node->recv_entry = new_send;
 	}
 
 	if (type == NES_TIMER_TYPE_SEND) {
@@ -435,24 +468,78 @@ int schedule_nes_timer(struct nes_cm_node *cm_node, struct sk_buff *skb,
 	return ret;
 }
 
+static void nes_retrans_expired(struct nes_cm_node *cm_node)
+{
+	switch (cm_node->state) {
+	case NES_CM_STATE_SYN_RCVD:
+	case NES_CM_STATE_CLOSING:
+		rem_ref_cm_node(cm_node->cm_core, cm_node);
+		break;
+	case NES_CM_STATE_LAST_ACK:
+	case NES_CM_STATE_FIN_WAIT1:
+	case NES_CM_STATE_MPAREJ_RCVD:
+		send_reset(cm_node, NULL);
+		break;
+	default:
+		create_event(cm_node, NES_CM_EVENT_ABORTED);
+	}
+}
+
+static void handle_recv_entry(struct nes_cm_node *cm_node, u32 rem_node)
+{
+	struct nes_timer_entry *recv_entry = cm_node->recv_entry;
+	struct iw_cm_id *cm_id = cm_node->cm_id;
+	struct nes_qp *nesqp;
+	unsigned long qplockflags;
+
+	if (!recv_entry)
+		return;
+	nesqp = (struct nes_qp *)recv_entry->skb;
+	if (nesqp) {
+		spin_lock_irqsave(&nesqp->lock, qplockflags);
+		if (nesqp->cm_id) {
+			nes_debug(NES_DBG_CM, "QP%u: cm_id = %p, "
+				"refcount = %d: HIT A "
+				"NES_TIMER_TYPE_CLOSE with something "
+				"to do!!!\n", nesqp->hwqp.qp_id, cm_id,
+				atomic_read(&nesqp->refcount));
+			nesqp->hw_tcp_state = NES_AEQE_TCP_STATE_CLOSED;
+			nesqp->last_aeq = NES_AEQE_AEID_RESET_SENT;
+			nesqp->ibqp_state = IB_QPS_ERR;
+			spin_unlock_irqrestore(&nesqp->lock, qplockflags);
+			nes_cm_disconn(nesqp);
+		} else {
+			spin_unlock_irqrestore(&nesqp->lock, qplockflags);
+			nes_debug(NES_DBG_CM, "QP%u: cm_id = %p, "
+				"refcount = %d: HIT A "
+				"NES_TIMER_TYPE_CLOSE with nothing "
+				"to do!!!\n", nesqp->hwqp.qp_id, cm_id,
+				atomic_read(&nesqp->refcount));
+		}
+	} else if (rem_node) {
+		/* TIME_WAIT state */
+		rem_ref_cm_node(cm_node->cm_core, cm_node);
+	}
+	if (cm_node->cm_id)
+		cm_id->rem_ref(cm_id);
+	kfree(recv_entry);
+	cm_node->recv_entry = NULL;
+}
 
 /**
  * nes_cm_timer_tick
  */
 static void nes_cm_timer_tick(unsigned long pass)
 {
-	unsigned long flags, qplockflags;
+	unsigned long flags;
 	unsigned long nexttimeout = jiffies + NES_LONG_TIME;
-	struct iw_cm_id *cm_id;
 	struct nes_cm_node *cm_node;
 	struct nes_timer_entry *send_entry, *recv_entry;
-	struct list_head *list_core, *list_core_temp;
-	struct list_head *list_node, *list_node_temp;
+	struct list_head *list_core_temp;
+	struct list_head *list_node;
 	struct nes_cm_core *cm_core = g_cm_core;
-	struct nes_qp *nesqp;
 	u32 settimer = 0;
 	int ret = NETDEV_TX_OK;
-	enum nes_cm_node_state last_state;
 
 	struct list_head timer_list;
 	INIT_LIST_HEAD(&timer_list);
@@ -461,7 +548,7 @@ static void nes_cm_timer_tick(unsigned long pass)
 	list_for_each_safe(list_node, list_core_temp,
 				&cm_core->connected_nodes) {
 		cm_node = container_of(list_node, struct nes_cm_node, list);
-		if (!list_empty(&cm_node->recv_list) || (cm_node->send_entry)) {
+		if ((cm_node->recv_entry) || (cm_node->send_entry)) {
 			add_ref_cm_node(cm_node);
 			list_add(&cm_node->timer_entry, &timer_list);
 		}
@@ -471,54 +558,18 @@ static void nes_cm_timer_tick(unsigned long pass)
 	list_for_each_safe(list_node, list_core_temp, &timer_list) {
 		cm_node = container_of(list_node, struct nes_cm_node,
 					timer_entry);
-		spin_lock_irqsave(&cm_node->recv_list_lock, flags);
-		list_for_each_safe(list_core, list_node_temp,
-			&cm_node->recv_list) {
-			recv_entry = container_of(list_core,
-				struct nes_timer_entry, list);
-			if (!recv_entry)
-				break;
+		recv_entry = cm_node->recv_entry;
+
+		if (recv_entry) {
 			if (time_after(recv_entry->timetosend, jiffies)) {
 				if (nexttimeout > recv_entry->timetosend ||
-					!settimer) {
+						!settimer) {
 					nexttimeout = recv_entry->timetosend;
 					settimer = 1;
 				}
-				continue;
-			}
-			list_del(&recv_entry->list);
-			cm_id = cm_node->cm_id;
-			spin_unlock_irqrestore(&cm_node->recv_list_lock, flags);
-			nesqp = (struct nes_qp *)recv_entry->skb;
-			spin_lock_irqsave(&nesqp->lock, qplockflags);
-			if (nesqp->cm_id) {
-				nes_debug(NES_DBG_CM, "QP%u: cm_id = %p, "
-					"refcount = %d: HIT A "
-					"NES_TIMER_TYPE_CLOSE with something "
-					"to do!!!\n", nesqp->hwqp.qp_id, cm_id,
-					atomic_read(&nesqp->refcount));
-				nesqp->hw_tcp_state = NES_AEQE_TCP_STATE_CLOSED;
-				nesqp->last_aeq = NES_AEQE_AEID_RESET_SENT;
-				nesqp->ibqp_state = IB_QPS_ERR;
-				spin_unlock_irqrestore(&nesqp->lock,
-					qplockflags);
-				nes_cm_disconn(nesqp);
-			} else {
-				spin_unlock_irqrestore(&nesqp->lock,
-					qplockflags);
-				nes_debug(NES_DBG_CM, "QP%u: cm_id = %p, "
-					"refcount = %d: HIT A "
-					"NES_TIMER_TYPE_CLOSE with nothing "
-					"to do!!!\n", nesqp->hwqp.qp_id, cm_id,
-					atomic_read(&nesqp->refcount));
-			}
-			if (cm_id)
-				cm_id->rem_ref(cm_id);
-
-			kfree(recv_entry);
-			spin_lock_irqsave(&cm_node->recv_list_lock, flags);
+			} else
+				handle_recv_entry(cm_node, 1);
 		}
-		spin_unlock_irqrestore(&cm_node->recv_list_lock, flags);
 
 		spin_lock_irqsave(&cm_node->retrans_list_lock, flags);
 		do {
@@ -533,12 +584,11 @@ static void nes_cm_timer_tick(unsigned long pass)
 						nexttimeout =
 							send_entry->timetosend;
 						settimer = 1;
-						break;
 					}
 				} else {
 					free_retrans_entry(cm_node);
-					break;
 				}
+				break;
 			}
 
 			if ((cm_node->state == NES_CM_STATE_TSA) ||
@@ -550,16 +600,12 @@ static void nes_cm_timer_tick(unsigned long pass)
 			if (!send_entry->retranscount ||
 				!send_entry->retrycount) {
 				cm_packets_dropped++;
-				last_state = cm_node->state;
-				cm_node->state = NES_CM_STATE_CLOSED;
 				free_retrans_entry(cm_node);
+
 				spin_unlock_irqrestore(
 					&cm_node->retrans_list_lock, flags);
-				if (last_state == NES_CM_STATE_SYN_RCVD)
-					rem_ref_cm_node(cm_core, cm_node);
-				else
-					create_event(cm_node,
-						NES_CM_EVENT_ABORTED);
+				nes_retrans_expired(cm_node);
+				cm_node->state = NES_CM_STATE_CLOSED;
 				spin_lock_irqsave(&cm_node->retrans_list_lock,
 					flags);
 				break;
@@ -714,7 +760,7 @@ static int send_reset(struct nes_cm_node *cm_node, struct sk_buff *skb)
 		skb = dev_alloc_skb(MAX_CM_BUFFER);
 	if (!skb) {
 		nes_debug(NES_DBG_CM, "Failed to get a Free pkt\n");
-		return -1;
+		return -ENOMEM;
 	}
 
 	form_cm_frame(skb, cm_node, NULL, 0, NULL, 0, flags);
@@ -778,14 +824,10 @@ static struct nes_cm_node *find_node(struct nes_cm_core *cm_core,
 	unsigned long flags;
 	struct list_head *hte;
 	struct nes_cm_node *cm_node;
-	__be32 tmp_addr = cpu_to_be32(loc_addr);
 
 	/* get a handle on the hte */
 	hte = &cm_core->connected_nodes;
 
-	nes_debug(NES_DBG_CM, "Searching for an owner node: %pI4:%x from core %p->%p\n",
-		  &tmp_addr, loc_port, cm_core, hte);
-
 	/* walk list and find cm_node associated with this session ID */
 	spin_lock_irqsave(&cm_core->ht_lock, flags);
 	list_for_each_entry(cm_node, hte, list) {
@@ -875,7 +917,8 @@ static int add_hte_node(struct nes_cm_core *cm_core, struct nes_cm_node *cm_node
 static int mini_cm_dec_refcnt_listen(struct nes_cm_core *cm_core,
 	struct nes_cm_listener *listener, int free_hanging_nodes)
 {
-	int ret = 1;
+	int ret = -EINVAL;
+	int err = 0;
 	unsigned long flags;
 	struct list_head *list_pos = NULL;
 	struct list_head *list_temp = NULL;
@@ -904,10 +947,60 @@ static int mini_cm_dec_refcnt_listen(struct nes_cm_core *cm_core,
 
 	list_for_each_safe(list_pos, list_temp, &reset_list) {
 		cm_node = container_of(list_pos, struct nes_cm_node,
-					reset_entry);
-		cleanup_retrans_entry(cm_node);
-		send_reset(cm_node, NULL);
-		rem_ref_cm_node(cm_node->cm_core, cm_node);
+				reset_entry);
+		{
+			struct nes_cm_node *loopback = cm_node->loopbackpartner;
+			if (NES_CM_STATE_FIN_WAIT1 <= cm_node->state) {
+				rem_ref_cm_node(cm_node->cm_core, cm_node);
+			} else {
+				if (!loopback) {
+					cleanup_retrans_entry(cm_node);
+					err = send_reset(cm_node, NULL);
+					if (err) {
+						cm_node->state =
+							 NES_CM_STATE_CLOSED;
+						WARN_ON(1);
+					} else {
+						cm_node->state =
+							NES_CM_STATE_CLOSED;
+						rem_ref_cm_node(
+							cm_node->cm_core,
+							cm_node);
+					}
+				} else {
+					struct nes_cm_event event;
+
+					event.cm_node = loopback;
+					event.cm_info.rem_addr =
+							loopback->rem_addr;
+					event.cm_info.loc_addr =
+							loopback->loc_addr;
+					event.cm_info.rem_port =
+							loopback->rem_port;
+					event.cm_info.loc_port =
+							 loopback->loc_port;
+					event.cm_info.cm_id = loopback->cm_id;
+					cm_event_connect_error(&event);
+					loopback->state = NES_CM_STATE_CLOSED;
+
+					event.cm_node = cm_node;
+					event.cm_info.rem_addr =
+							 cm_node->rem_addr;
+					event.cm_info.loc_addr =
+							 cm_node->loc_addr;
+					event.cm_info.rem_port =
+							 cm_node->rem_port;
+					event.cm_info.loc_port =
+							 cm_node->loc_port;
+					event.cm_info.cm_id = cm_node->cm_id;
+					cm_event_reset(&event);
+
+					rem_ref_cm_node(cm_node->cm_core,
+							 cm_node);
+
+				}
+			}
+		}
 	}
 
 	spin_lock_irqsave(&cm_core->listen_list_lock, flags);
@@ -968,6 +1061,7 @@ static inline int mini_cm_accelerated(struct nes_cm_core *cm_core,
 	if (cm_node->accept_pend) {
 		BUG_ON(!cm_node->listener);
 		atomic_dec(&cm_node->listener->pend_accepts_cnt);
+		cm_node->accept_pend = 0;
 		BUG_ON(atomic_read(&cm_node->listener->pend_accepts_cnt) < 0);
 	}
 
@@ -994,7 +1088,7 @@ static int nes_addr_resolve_neigh(struct nes_vnic *nesvnic, u32 dst_ip)
 	memset(&fl, 0, sizeof fl);
 	fl.nl_u.ip4_u.daddr = htonl(dst_ip);
 	if (ip_route_output_key(&init_net, &rt, &fl)) {
-		printk("%s: ip_route_output_key failed for 0x%08X\n",
+		printk(KERN_ERR "%s: ip_route_output_key failed for 0x%08X\n",
 				__func__, dst_ip);
 		return rc;
 	}
@@ -1057,8 +1151,6 @@ static struct nes_cm_node *make_cm_node(struct nes_cm_core *cm_core,
 			cm_node->cm_id);
 
 	spin_lock_init(&cm_node->retrans_list_lock);
-	INIT_LIST_HEAD(&cm_node->recv_list);
-	spin_lock_init(&cm_node->recv_list_lock);
 
 	cm_node->loopbackpartner = NULL;
 	atomic_set(&cm_node->ref_count, 1);
@@ -1126,10 +1218,7 @@ static int add_ref_cm_node(struct nes_cm_node *cm_node)
 static int rem_ref_cm_node(struct nes_cm_core *cm_core,
 	struct nes_cm_node *cm_node)
 {
-	unsigned long flags, qplockflags;
-	struct nes_timer_entry *recv_entry;
-	struct iw_cm_id *cm_id;
-	struct list_head *list_core, *list_node_temp;
+	unsigned long flags;
 	struct nes_qp *nesqp;
 
 	if (!cm_node)
@@ -1150,38 +1239,9 @@ static int rem_ref_cm_node(struct nes_cm_core *cm_core,
 		atomic_dec(&cm_node->listener->pend_accepts_cnt);
 		BUG_ON(atomic_read(&cm_node->listener->pend_accepts_cnt) < 0);
 	}
-	BUG_ON(cm_node->send_entry);
-	spin_lock_irqsave(&cm_node->recv_list_lock, flags);
-	list_for_each_safe(list_core, list_node_temp, &cm_node->recv_list) {
-		recv_entry = container_of(list_core, struct nes_timer_entry,
-				list);
-		list_del(&recv_entry->list);
-		cm_id = cm_node->cm_id;
-		spin_unlock_irqrestore(&cm_node->recv_list_lock, flags);
-		nesqp = (struct nes_qp *)recv_entry->skb;
-		spin_lock_irqsave(&nesqp->lock, qplockflags);
-		if (nesqp->cm_id) {
-			nes_debug(NES_DBG_CM, "QP%u: cm_id = %p: HIT A "
-				"NES_TIMER_TYPE_CLOSE with something to do!\n",
-				nesqp->hwqp.qp_id, cm_id);
-			nesqp->hw_tcp_state = NES_AEQE_TCP_STATE_CLOSED;
-			nesqp->last_aeq = NES_AEQE_AEID_RESET_SENT;
-			nesqp->ibqp_state = IB_QPS_ERR;
-			spin_unlock_irqrestore(&nesqp->lock, qplockflags);
-			nes_cm_disconn(nesqp);
-		} else {
-			spin_unlock_irqrestore(&nesqp->lock, qplockflags);
-			nes_debug(NES_DBG_CM, "QP%u: cm_id = %p: HIT A "
-				"NES_TIMER_TYPE_CLOSE with nothing to do!\n",
-				nesqp->hwqp.qp_id, cm_id);
-		}
-		cm_id->rem_ref(cm_id);
-
-		kfree(recv_entry);
-		spin_lock_irqsave(&cm_node->recv_list_lock, flags);
-	}
-	spin_unlock_irqrestore(&cm_node->recv_list_lock, flags);
-
+	WARN_ON(cm_node->send_entry);
+	if (cm_node->recv_entry)
+		handle_recv_entry(cm_node, 0);
 	if (cm_node->listener) {
 		mini_cm_dec_refcnt_listen(cm_core, cm_node->listener, 0);
 	} else {
@@ -1266,8 +1326,7 @@ static void drop_packet(struct sk_buff *skb)
 	dev_kfree_skb_any(skb);
 }
 
-static void handle_fin_pkt(struct nes_cm_node *cm_node, struct sk_buff *skb,
-	struct tcphdr *tcph)
+static void handle_fin_pkt(struct nes_cm_node *cm_node)
 {
 	nes_debug(NES_DBG_CM, "Received FIN, cm_node = %p, state = %u. "
 		"refcnt=%d\n", cm_node, cm_node->state,
@@ -1279,23 +1338,30 @@ static void handle_fin_pkt(struct nes_cm_node *cm_node, struct sk_buff *skb,
 	case NES_CM_STATE_SYN_SENT:
 	case NES_CM_STATE_ESTABLISHED:
 	case NES_CM_STATE_MPAREQ_SENT:
+	case NES_CM_STATE_MPAREJ_RCVD:
 		cm_node->state = NES_CM_STATE_LAST_ACK;
-		send_fin(cm_node, skb);
+		send_fin(cm_node, NULL);
 		break;
 	case NES_CM_STATE_FIN_WAIT1:
 		cm_node->state = NES_CM_STATE_CLOSING;
-		send_ack(cm_node, skb);
+		send_ack(cm_node, NULL);
+		/* Wait for ACK as this is simultanous close..
+		* After we receive ACK, do not send anything..
+		* Just rm the node.. Done.. */
 		break;
 	case NES_CM_STATE_FIN_WAIT2:
 		cm_node->state = NES_CM_STATE_TIME_WAIT;
-		send_ack(cm_node, skb);
+		send_ack(cm_node, NULL);
+		schedule_nes_timer(cm_node, NULL,  NES_TIMER_TYPE_CLOSE, 1, 0);
+		break;
+	case NES_CM_STATE_TIME_WAIT:
 		cm_node->state = NES_CM_STATE_CLOSED;
+		rem_ref_cm_node(cm_node->cm_core, cm_node);
 		break;
 	case NES_CM_STATE_TSA:
 	default:
 		nes_debug(NES_DBG_CM, "Error Rcvd FIN for node-%p state = %d\n",
 			cm_node, cm_node->state);
-		drop_packet(skb);
 		break;
 	}
 }
@@ -1341,23 +1407,35 @@ static void handle_rst_pkt(struct nes_cm_node *cm_node, struct sk_buff *skb,
 		cleanup_retrans_entry(cm_node);
 		drop_packet(skb);
 		break;
+	case NES_CM_STATE_TIME_WAIT:
+		cleanup_retrans_entry(cm_node);
+		cm_node->state = NES_CM_STATE_CLOSED;
+		rem_ref_cm_node(cm_node->cm_core, cm_node);
+		drop_packet(skb);
+		break;
+	case NES_CM_STATE_FIN_WAIT1:
+		cleanup_retrans_entry(cm_node);
+		nes_debug(NES_DBG_CM, "Bad state %s[%u]\n", __func__, __LINE__);
 	default:
 		drop_packet(skb);
 		break;
 	}
 }
 
-static void handle_rcv_mpa(struct nes_cm_node *cm_node, struct sk_buff *skb,
-	enum nes_cm_event_type type)
+
+static void handle_rcv_mpa(struct nes_cm_node *cm_node, struct sk_buff *skb)
 {
 
-	int	ret;
+	int	ret = 0;
 	int datasize = skb->len;
 	u8 *dataloc = skb->data;
-	ret = parse_mpa(cm_node, dataloc, datasize);
-	if (ret < 0) {
+
+	enum nes_cm_event_type type = NES_CM_EVENT_UNKNOWN;
+	u32     res_type;
+	ret = parse_mpa(cm_node, dataloc, &res_type, datasize);
+	if (ret) {
 		nes_debug(NES_DBG_CM, "didn't like MPA Request\n");
-		if (type == NES_CM_EVENT_CONNECTED) {
+		if (cm_node->state == NES_CM_STATE_MPAREQ_SENT) {
 			nes_debug(NES_DBG_CM, "%s[%u] create abort for "
 				"cm_node=%p listener=%p state=%d\n", __func__,
 				__LINE__, cm_node, cm_node->listener,
@@ -1366,18 +1444,38 @@ static void handle_rcv_mpa(struct nes_cm_node *cm_node, struct sk_buff *skb,
 		} else {
 			passive_open_err(cm_node, skb, 1);
 		}
-	} else {
-		cleanup_retrans_entry(cm_node);
-		dev_kfree_skb_any(skb);
-		if (type == NES_CM_EVENT_CONNECTED)
+		return;
+	}
+
+	switch (cm_node->state) {
+	case NES_CM_STATE_ESTABLISHED:
+		if (res_type == NES_MPA_REQUEST_REJECT) {
+			/*BIG problem as we are receiving the MPA.. So should
+			* not be REJECT.. This is Passive Open.. We can
+			* only receive it Reject for Active Open...*/
+			WARN_ON(1);
+		}
+		cm_node->state = NES_CM_STATE_MPAREQ_RCVD;
+		type = NES_CM_EVENT_MPA_REQ;
+		atomic_set(&cm_node->passive_state,
+				NES_PASSIVE_STATE_INDICATED);
+		break;
+	case NES_CM_STATE_MPAREQ_SENT:
+		if (res_type == NES_MPA_REQUEST_REJECT) {
+			type = NES_CM_EVENT_MPA_REJECT;
+			cm_node->state = NES_CM_STATE_MPAREJ_RCVD;
+		} else {
+			type = NES_CM_EVENT_CONNECTED;
 			cm_node->state = NES_CM_STATE_TSA;
-		else
-			atomic_set(&cm_node->passive_state,
-					NES_PASSIVE_STATE_INDICATED);
-		create_event(cm_node, type);
+		}
 
+		break;
+	default:
+		WARN_ON(1);
+		break;
 	}
-	return ;
+	dev_kfree_skb_any(skb);
+	create_event(cm_node, type);
 }
 
 static void indicate_pkt_err(struct nes_cm_node *cm_node, struct sk_buff *skb)
@@ -1465,8 +1563,6 @@ static void handle_syn_pkt(struct nes_cm_node *cm_node, struct sk_buff *skb,
 		break;
 	case NES_CM_STATE_LISTENING:
 		/* Passive OPEN */
-		cm_node->accept_pend = 1;
-		atomic_inc(&cm_node->listener->pend_accepts_cnt);
 		if (atomic_read(&cm_node->listener->pend_accepts_cnt) >
 				cm_node->listener->backlog) {
 			nes_debug(NES_DBG_CM, "drop syn due to backlog "
@@ -1484,6 +1580,9 @@ static void handle_syn_pkt(struct nes_cm_node *cm_node, struct sk_buff *skb,
 		}
 		cm_node->tcp_cntxt.rcv_nxt = inc_sequence + 1;
 		BUG_ON(cm_node->send_entry);
+		cm_node->accept_pend = 1;
+		atomic_inc(&cm_node->listener->pend_accepts_cnt);
+
 		cm_node->state = NES_CM_STATE_SYN_RCVD;
 		send_syn(cm_node, 1, skb);
 		break;
@@ -1518,6 +1617,7 @@ static void handle_synack_pkt(struct nes_cm_node *cm_node, struct sk_buff *skb,
 	inc_sequence = ntohl(tcph->seq);
 	switch (cm_node->state) {
 	case NES_CM_STATE_SYN_SENT:
+		cleanup_retrans_entry(cm_node);
 		/* active open */
 		if (check_syn(cm_node, tcph, skb))
 			return;
@@ -1567,10 +1667,7 @@ static void handle_ack_pkt(struct nes_cm_node *cm_node, struct sk_buff *skb,
 	u32 rem_seq;
 	int ret;
 	int optionsize;
-	u32 temp_seq = cm_node->tcp_cntxt.loc_seq_num;
-
 	optionsize = (tcph->doff << 2) - sizeof(struct tcphdr);
-	cm_node->tcp_cntxt.loc_seq_num = ntohl(tcph->ack_seq);
 
 	if (check_seq(cm_node, tcph, skb))
 		return;
@@ -1580,7 +1677,7 @@ static void handle_ack_pkt(struct nes_cm_node *cm_node, struct sk_buff *skb,
 	rem_seq = ntohl(tcph->seq);
 	rem_seq_ack =  ntohl(tcph->ack_seq);
 	datasize = skb->len;
-
+	cleanup_retrans_entry(cm_node);
 	switch (cm_node->state) {
 	case NES_CM_STATE_SYN_RCVD:
 		/* Passive OPEN */
@@ -1588,7 +1685,6 @@ static void handle_ack_pkt(struct nes_cm_node *cm_node, struct sk_buff *skb,
 		if (ret)
 			break;
 		cm_node->tcp_cntxt.rem_ack_num = ntohl(tcph->ack_seq);
-		cm_node->tcp_cntxt.loc_seq_num = temp_seq;
 		if (cm_node->tcp_cntxt.rem_ack_num !=
 		    cm_node->tcp_cntxt.loc_seq_num) {
 			nes_debug(NES_DBG_CM, "rem_ack_num != loc_seq_num\n");
@@ -1597,31 +1693,30 @@ static void handle_ack_pkt(struct nes_cm_node *cm_node, struct sk_buff *skb,
 			return;
 		}
 		cm_node->state = NES_CM_STATE_ESTABLISHED;
+		cleanup_retrans_entry(cm_node);
 		if (datasize) {
 			cm_node->tcp_cntxt.rcv_nxt = inc_sequence + datasize;
-			cm_node->state = NES_CM_STATE_MPAREQ_RCVD;
-			handle_rcv_mpa(cm_node, skb, NES_CM_EVENT_MPA_REQ);
-		 } else { /* rcvd ACK only */
+			handle_rcv_mpa(cm_node, skb);
+		} else { /* rcvd ACK only */
 			dev_kfree_skb_any(skb);
 			cleanup_retrans_entry(cm_node);
 		 }
 		break;
 	case NES_CM_STATE_ESTABLISHED:
 		/* Passive OPEN */
-		/* We expect mpa frame to be received only */
+		cleanup_retrans_entry(cm_node);
 		if (datasize) {
 			cm_node->tcp_cntxt.rcv_nxt = inc_sequence + datasize;
-			cm_node->state = NES_CM_STATE_MPAREQ_RCVD;
-			handle_rcv_mpa(cm_node, skb,
-				NES_CM_EVENT_MPA_REQ);
+			handle_rcv_mpa(cm_node, skb);
 		} else
 			drop_packet(skb);
 		break;
 	case NES_CM_STATE_MPAREQ_SENT:
+		cleanup_retrans_entry(cm_node);
 		cm_node->tcp_cntxt.rem_ack_num = ntohl(tcph->ack_seq);
 		if (datasize) {
 			cm_node->tcp_cntxt.rcv_nxt = inc_sequence + datasize;
-			handle_rcv_mpa(cm_node, skb, NES_CM_EVENT_CONNECTED);
+			handle_rcv_mpa(cm_node, skb);
 		} else { /* Could be just an ack pkt.. */
 			cleanup_retrans_entry(cm_node);
 			dev_kfree_skb_any(skb);
@@ -1632,13 +1727,24 @@ static void handle_ack_pkt(struct nes_cm_node *cm_node, struct sk_buff *skb,
 		cleanup_retrans_entry(cm_node);
 		send_reset(cm_node, skb);
 		break;
+	case NES_CM_STATE_LAST_ACK:
+		cleanup_retrans_entry(cm_node);
+		cm_node->state = NES_CM_STATE_CLOSED;
+		cm_node->cm_id->rem_ref(cm_node->cm_id);
+	case NES_CM_STATE_CLOSING:
+		cleanup_retrans_entry(cm_node);
+		rem_ref_cm_node(cm_node->cm_core, cm_node);
+		drop_packet(skb);
+		break;
 	case NES_CM_STATE_FIN_WAIT1:
+		cleanup_retrans_entry(cm_node);
+		drop_packet(skb);
+		cm_node->state = NES_CM_STATE_FIN_WAIT2;
+		break;
 	case NES_CM_STATE_SYN_SENT:
 	case NES_CM_STATE_FIN_WAIT2:
 	case NES_CM_STATE_TSA:
 	case NES_CM_STATE_MPAREQ_RCVD:
-	case NES_CM_STATE_LAST_ACK:
-	case NES_CM_STATE_CLOSING:
 	case NES_CM_STATE_UNKNOWN:
 	default:
 		drop_packet(skb);
@@ -1748,6 +1854,7 @@ static void process_packet(struct nes_cm_node *cm_node, struct sk_buff *skb,
 {
 	enum nes_tcpip_pkt_type	pkt_type = NES_PKT_TYPE_UNKNOWN;
 	struct tcphdr *tcph = tcp_hdr(skb);
+	u32     fin_set = 0;
 	skb_pull(skb, ip_hdr(skb)->ihl << 2);
 
 	nes_debug(NES_DBG_CM, "process_packet: cm_node=%p state =%d syn=%d "
@@ -1760,10 +1867,10 @@ static void process_packet(struct nes_cm_node *cm_node, struct sk_buff *skb,
 		pkt_type = NES_PKT_TYPE_SYN;
 		if (tcph->ack)
 			pkt_type = NES_PKT_TYPE_SYNACK;
-	} else if (tcph->fin)
-		pkt_type = NES_PKT_TYPE_FIN;
-	else if (tcph->ack)
+	} else if (tcph->ack)
 		pkt_type = NES_PKT_TYPE_ACK;
+	if (tcph->fin)
+		fin_set = 1;
 
 	switch (pkt_type) {
 	case NES_PKT_TYPE_SYN:
@@ -1774,15 +1881,16 @@ static void process_packet(struct nes_cm_node *cm_node, struct sk_buff *skb,
 		break;
 	case NES_PKT_TYPE_ACK:
 		handle_ack_pkt(cm_node, skb, tcph);
+		if (fin_set)
+			handle_fin_pkt(cm_node);
 		break;
 	case NES_PKT_TYPE_RST:
 		handle_rst_pkt(cm_node, skb, tcph);
 		break;
-	case NES_PKT_TYPE_FIN:
-		handle_fin_pkt(cm_node, skb, tcph);
-		break;
 	default:
 		drop_packet(skb);
+		if (fin_set)
+			handle_fin_pkt(cm_node);
 		break;
 	}
 }
@@ -1925,7 +2033,7 @@ static struct nes_cm_node *mini_cm_connect(struct nes_cm_core *cm_core,
 				loopbackremotenode->tcp_cntxt.rcv_wscale;
 			loopbackremotenode->tcp_cntxt.snd_wscale =
 				cm_node->tcp_cntxt.rcv_wscale;
-
+			loopbackremotenode->state = NES_CM_STATE_MPAREQ_RCVD;
 			create_event(loopbackremotenode, NES_CM_EVENT_MPA_REQ);
 		}
 		return cm_node;
@@ -1980,7 +2088,11 @@ static int mini_cm_reject(struct nes_cm_core *cm_core,
 	struct ietf_mpa_frame *mpa_frame, struct nes_cm_node *cm_node)
 {
 	int ret = 0;
+	int err = 0;
 	int passive_state;
+	struct nes_cm_event event;
+	struct iw_cm_id *cm_id = cm_node->cm_id;
+	struct nes_cm_node *loopback = cm_node->loopbackpartner;
 
 	nes_debug(NES_DBG_CM, "%s cm_node=%p type=%d state=%d\n",
 		__func__, cm_node, cm_node->tcp_cntxt.client, cm_node->state);
@@ -1989,12 +2101,38 @@ static int mini_cm_reject(struct nes_cm_core *cm_core,
 		return ret;
 	cleanup_retrans_entry(cm_node);
 
-	passive_state = atomic_add_return(1, &cm_node->passive_state);
-	cm_node->state = NES_CM_STATE_CLOSED;
-	if (passive_state == NES_SEND_RESET_EVENT)
+	if (!loopback) {
+		passive_state = atomic_add_return(1, &cm_node->passive_state);
+		if (passive_state == NES_SEND_RESET_EVENT) {
+			cm_node->state = NES_CM_STATE_CLOSED;
+			rem_ref_cm_node(cm_core, cm_node);
+		} else {
+			ret = send_mpa_reject(cm_node);
+			if (ret) {
+				cm_node->state = NES_CM_STATE_CLOSED;
+				err = send_reset(cm_node, NULL);
+				if (err)
+					WARN_ON(1);
+			} else
+				cm_id->add_ref(cm_id);
+		}
+	} else {
+		cm_node->cm_id = NULL;
+		event.cm_node = loopback;
+		event.cm_info.rem_addr = loopback->rem_addr;
+		event.cm_info.loc_addr = loopback->loc_addr;
+		event.cm_info.rem_port = loopback->rem_port;
+		event.cm_info.loc_port = loopback->loc_port;
+		event.cm_info.cm_id = loopback->cm_id;
+		cm_event_mpa_reject(&event);
 		rem_ref_cm_node(cm_core, cm_node);
-	else
-		ret = send_reset(cm_node, NULL);
+		loopback->state = NES_CM_STATE_CLOSING;
+
+		cm_id = loopback->cm_id;
+		rem_ref_cm_node(cm_core, loopback);
+		cm_id->rem_ref(cm_id);
+	}
+
 	return ret;
 }
 
@@ -2031,6 +2169,7 @@ static int mini_cm_close(struct nes_cm_core *cm_core, struct nes_cm_node *cm_nod
 	case NES_CM_STATE_CLOSING:
 		ret = -1;
 		break;
+	case NES_CM_STATE_MPAREJ_RCVD:
 	case NES_CM_STATE_LISTENING:
 	case NES_CM_STATE_UNKNOWN:
 	case NES_CM_STATE_INITED:
@@ -2227,15 +2366,15 @@ static int mini_cm_set(struct nes_cm_core *cm_core, u32 type, u32 value)
 	int ret = 0;
 
 	switch (type) {
-		case NES_CM_SET_PKT_SIZE:
-			cm_core->mtu = value;
-			break;
-		case NES_CM_SET_FREE_PKT_Q_SIZE:
-			cm_core->free_tx_pkt_max = value;
-			break;
-		default:
-			/* unknown set option */
-			ret = -EINVAL;
+	case NES_CM_SET_PKT_SIZE:
+		cm_core->mtu = value;
+		break;
+	case NES_CM_SET_FREE_PKT_Q_SIZE:
+		cm_core->free_tx_pkt_max = value;
+		break;
+	default:
+		/* unknown set option */
+		ret = -EINVAL;
 	}
 
 	return ret;
@@ -2654,9 +2793,7 @@ int nes_accept(struct iw_cm_id *cm_id, struct iw_cm_conn_param *conn_param)
 			NES_QPCONTEXT_ORDIRD_WRPDU);
 	} else {
 		nesqp->nesqp_context->ird_ord_sizes |=
-			cpu_to_le32((NES_QPCONTEXT_ORDIRD_LSMM_PRESENT |
-			NES_QPCONTEXT_ORDIRD_WRPDU |
-			NES_QPCONTEXT_ORDIRD_ALSMM));
+			cpu_to_le32(NES_QPCONTEXT_ORDIRD_WRPDU);
 	}
 	nesqp->skip_lsmm = 1;
 
@@ -2778,23 +2915,35 @@ int nes_accept(struct iw_cm_id *cm_id, struct iw_cm_conn_param *conn_param)
 int nes_reject(struct iw_cm_id *cm_id, const void *pdata, u8 pdata_len)
 {
 	struct nes_cm_node *cm_node;
+	struct nes_cm_node *loopback;
+
 	struct nes_cm_core *cm_core;
 
 	atomic_inc(&cm_rejects);
 	cm_node = (struct nes_cm_node *) cm_id->provider_data;
+	loopback = cm_node->loopbackpartner;
 	cm_core = cm_node->cm_core;
+	cm_node->cm_id = cm_id;
 	cm_node->mpa_frame_size = sizeof(struct ietf_mpa_frame) + pdata_len;
 
+	if (cm_node->mpa_frame_size > MAX_CM_BUFFER)
+		return -EINVAL;
+
 	strcpy(&cm_node->mpa_frame.key[0], IEFT_MPA_KEY_REP);
-	memcpy(&cm_node->mpa_frame.priv_data, pdata, pdata_len);
+	if (loopback) {
+		memcpy(&loopback->mpa_frame.priv_data, pdata, pdata_len);
+		loopback->mpa_frame.priv_data_len = pdata_len;
+		loopback->mpa_frame_size = sizeof(struct ietf_mpa_frame) +
+				pdata_len;
+	} else {
+		memcpy(&cm_node->mpa_frame.priv_data, pdata, pdata_len);
+		cm_node->mpa_frame.priv_data_len = cpu_to_be16(pdata_len);
+	}
 
-	cm_node->mpa_frame.priv_data_len = cpu_to_be16(pdata_len);
 	cm_node->mpa_frame.rev = mpa_version;
 	cm_node->mpa_frame.flags = IETF_MPA_FLAGS_CRC | IETF_MPA_FLAGS_REJECT;
 
-	cm_core->api->reject(cm_core, &cm_node->mpa_frame, cm_node);
-
-	return 0;
+	return cm_core->api->reject(cm_core, &cm_node->mpa_frame, cm_node);
 }
 
 
@@ -3303,13 +3452,56 @@ static void cm_event_mpa_req(struct nes_cm_event *event)
 	cm_event.remote_addr.sin_family = AF_INET;
 	cm_event.remote_addr.sin_port = htons(event->cm_info.rem_port);
 	cm_event.remote_addr.sin_addr.s_addr = htonl(event->cm_info.rem_addr);
+	cm_event.private_data = cm_node->mpa_frame_buf;
+	cm_event.private_data_len  = (u8) cm_node->mpa_frame_size;
+
+	ret = cm_id->event_handler(cm_id, &cm_event);
+	if (ret)
+		printk(KERN_ERR "%s[%u] OFA CM event_handler returned, ret=%d\n",
+				__func__, __LINE__, ret);
+	return;
+}
+
+
+static void cm_event_mpa_reject(struct nes_cm_event *event)
+{
+	struct iw_cm_id   *cm_id;
+	struct iw_cm_event cm_event;
+	struct nes_cm_node *cm_node;
+	int ret;
+
+	cm_node = event->cm_node;
+	if (!cm_node)
+		return;
+	cm_id = cm_node->cm_id;
+
+	atomic_inc(&cm_connect_reqs);
+	nes_debug(NES_DBG_CM, "cm_node = %p - cm_id = %p, jiffies = %lu\n",
+			cm_node, cm_id, jiffies);
+
+	cm_event.event = IW_CM_EVENT_CONNECT_REPLY;
+	cm_event.status = -ECONNREFUSED;
+	cm_event.provider_data = cm_id->provider_data;
+
+	cm_event.local_addr.sin_family = AF_INET;
+	cm_event.local_addr.sin_port = htons(event->cm_info.loc_port);
+	cm_event.local_addr.sin_addr.s_addr = htonl(event->cm_info.loc_addr);
+
+	cm_event.remote_addr.sin_family = AF_INET;
+	cm_event.remote_addr.sin_port = htons(event->cm_info.rem_port);
+	cm_event.remote_addr.sin_addr.s_addr = htonl(event->cm_info.rem_addr);
 
-		cm_event.private_data                = cm_node->mpa_frame_buf;
-		cm_event.private_data_len            = (u8) cm_node->mpa_frame_size;
+	cm_event.private_data = cm_node->mpa_frame_buf;
+	cm_event.private_data_len = (u8) cm_node->mpa_frame_size;
+
+	nes_debug(NES_DBG_CM, "call CM_EVENT_MPA_REJECTED, local_addr=%08x, "
+			"remove_addr=%08x\n",
+			cm_event.local_addr.sin_addr.s_addr,
+			cm_event.remote_addr.sin_addr.s_addr);
 
 	ret = cm_id->event_handler(cm_id, &cm_event);
 	if (ret)
-		printk("%s[%u] OFA CM event_handler returned, ret=%d\n",
+		printk(KERN_ERR "%s[%u] OFA CM event_handler returned, ret=%d\n",
 				__func__, __LINE__, ret);
 
 	return;
@@ -3374,6 +3566,14 @@ static void nes_cm_event_handler(struct work_struct *work)
 		cm_event_connected(event);
 		nes_debug(NES_DBG_CM, "CM Event: CONNECTED\n");
 		break;
+	case NES_CM_EVENT_MPA_REJECT:
+		if ((!event->cm_node->cm_id) ||
+				(event->cm_node->state == NES_CM_STATE_TSA))
+			break;
+		cm_event_mpa_reject(event);
+		nes_debug(NES_DBG_CM, "CM Event: REJECT\n");
+		break;
+
 	case NES_CM_EVENT_ABORTED:
 		if ((!event->cm_node->cm_id) ||
 			(event->cm_node->state == NES_CM_STATE_TSA))
diff --git a/drivers/infiniband/hw/nes/nes_cm.h b/drivers/infiniband/hw/nes/nes_cm.h
index fafa35042ebd..d5f778202eb7 100644
--- a/drivers/infiniband/hw/nes/nes_cm.h
+++ b/drivers/infiniband/hw/nes/nes_cm.h
@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2006 - 2008 NetEffect, Inc. All rights reserved.
+ * Copyright (c) 2006 - 2009 Intel-NE, Inc.  All rights reserved.
  *
  * This software is available to you under a choice of one of two
  * licenses.  You may choose to be licensed under the terms of the GNU
@@ -39,6 +39,9 @@
 #define NES_MANAGE_APBVT_DEL 0
 #define NES_MANAGE_APBVT_ADD 1
 
+#define NES_MPA_REQUEST_ACCEPT  1
+#define NES_MPA_REQUEST_REJECT  2
+
 /* IETF MPA -- defines, enums, structs */
 #define IEFT_MPA_KEY_REQ  "MPA ID Req Frame"
 #define IEFT_MPA_KEY_REP  "MPA ID Rep Frame"
@@ -186,6 +189,7 @@ enum nes_cm_node_state {
 	NES_CM_STATE_ACCEPTING,
 	NES_CM_STATE_MPAREQ_SENT,
 	NES_CM_STATE_MPAREQ_RCVD,
+	NES_CM_STATE_MPAREJ_RCVD,
 	NES_CM_STATE_TSA,
 	NES_CM_STATE_FIN_WAIT1,
 	NES_CM_STATE_FIN_WAIT2,
@@ -278,13 +282,12 @@ struct nes_cm_node {
 	struct nes_timer_entry	*send_entry;
 
 	spinlock_t                retrans_list_lock;
-	struct list_head          recv_list;
-	spinlock_t                recv_list_lock;
+	struct nes_timer_entry  *recv_entry;
 
 	int                       send_write0;
 	union {
 		struct ietf_mpa_frame mpa_frame;
-		u8                    mpa_frame_buf[NES_CM_DEFAULT_MTU];
+		u8                    mpa_frame_buf[MAX_CM_BUFFER];
 	};
 	u16                       mpa_frame_size;
 	struct iw_cm_id           *cm_id;
@@ -326,6 +329,7 @@ enum  nes_cm_event_type {
 	NES_CM_EVENT_MPA_REQ,
 	NES_CM_EVENT_MPA_CONNECT,
 	NES_CM_EVENT_MPA_ACCEPT,
+	NES_CM_EVENT_MPA_REJECT,
 	NES_CM_EVENT_MPA_ESTABLISHED,
 	NES_CM_EVENT_CONNECTED,
 	NES_CM_EVENT_CLOSED,
diff --git a/drivers/infiniband/hw/nes/nes_context.h b/drivers/infiniband/hw/nes/nes_context.h
index da9daba8e668..0fb8d81d9a62 100644
--- a/drivers/infiniband/hw/nes/nes_context.h
+++ b/drivers/infiniband/hw/nes/nes_context.h
@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2006 - 2008 NetEffect, Inc. All rights reserved.
+ * Copyright (c) 2006 - 2009 Intel-NE, Inc.  All rights reserved.
  *
  * This software is available to you under a choice of one of two
  * licenses.  You may choose to be licensed under the terms of the GNU
diff --git a/drivers/infiniband/hw/nes/nes_hw.c b/drivers/infiniband/hw/nes/nes_hw.c
index 5d139db1b771..52e734042b8e 100644
--- a/drivers/infiniband/hw/nes/nes_hw.c
+++ b/drivers/infiniband/hw/nes/nes_hw.c
@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2006 - 2008 NetEffect, Inc. All rights reserved.
+ * Copyright (c) 2006 - 2009 Intel-NE, Inc.  All rights reserved.
  *
  * This software is available to you under a choice of one of two
  * licenses.  You may choose to be licensed under the terms of the GNU
@@ -254,6 +254,7 @@ struct nes_adapter *nes_init_adapter(struct nes_device *nesdev, u8 hw_rev) {
 	u32 adapter_size;
 	u32 arp_table_size;
 	u16 vendor_id;
+	u16 device_id;
 	u8  OneG_Mode;
 	u8  func_index;
 
@@ -356,6 +357,13 @@ struct nes_adapter *nes_init_adapter(struct nes_device *nesdev, u8 hw_rev) {
 		return NULL;
 	}
 
+	nesadapter->vendor_id = (((u32) nesadapter->mac_addr_high) << 8) |
+				(nesadapter->mac_addr_low >> 24);
+
+	pci_bus_read_config_word(nesdev->pcidev->bus, nesdev->pcidev->devfn,
+				 PCI_DEVICE_ID, &device_id);
+	nesadapter->vendor_part_id = device_id;
+
 	if (nes_init_serdes(nesdev, hw_rev, port_count, nesadapter,
 							OneG_Mode)) {
 		kfree(nesadapter);
@@ -1636,7 +1644,6 @@ int nes_init_nic_qp(struct nes_device *nesdev, struct net_device *netdev)
 	nesvnic->post_cqp_request = nes_post_cqp_request;
 	nesvnic->mcrq_mcast_filter = NULL;
 
-	spin_lock_init(&nesvnic->nic.sq_lock);
 	spin_lock_init(&nesvnic->nic.rq_lock);
 
 	/* setup the RQ */
@@ -2261,6 +2268,8 @@ static void nes_process_aeq(struct nes_device *nesdev, struct nes_hw_aeq *aeq)
 
 		if (++head >= aeq_size)
 			head = 0;
+
+		nes_write32(nesdev->regs + NES_AEQ_ALLOC, 1 << 16);
 	}
 	while (1);
 	aeq->aeq_head = head;
@@ -2541,7 +2550,7 @@ static void nes_nic_napi_ce_handler(struct nes_device *nesdev, struct nes_hw_nic
 {
 	struct nes_vnic *nesvnic = container_of(cq, struct nes_vnic, nic_cq);
 
-	netif_rx_schedule(&nesvnic->napi);
+	napi_schedule(&nesvnic->napi);
 }
 
 
@@ -2622,9 +2631,9 @@ void nes_nic_ce_handler(struct nes_device *nesdev, struct nes_hw_nic_cq *cq)
 						} else
 							break;
 					}
-					if (skb)
-						dev_kfree_skb_any(skb);
 				}
+				if (skb)
+					dev_kfree_skb_any(skb);
 				nesnic->sq_tail++;
 				nesnic->sq_tail &= nesnic->sq_size-1;
 				if (sq_cqes > 128) {
diff --git a/drivers/infiniband/hw/nes/nes_hw.h b/drivers/infiniband/hw/nes/nes_hw.h
index bc0b4de04450..f41a8710d2a8 100644
--- a/drivers/infiniband/hw/nes/nes_hw.h
+++ b/drivers/infiniband/hw/nes/nes_hw.h
@@ -1,5 +1,5 @@
 /*
-* Copyright (c) 2006 - 2008 NetEffect, Inc. All rights reserved.
+* Copyright (c) 2006 - 2009 Intel-NE, Inc.  All rights reserved.
 *
 * This software is available to you under a choice of one of two
 * licenses.  You may choose to be licensed under the terms of the GNU
@@ -61,6 +61,7 @@ enum pci_regs {
 	NES_CQ_ACK = 0x0034,
 	NES_WQE_ALLOC = 0x0040,
 	NES_CQE_ALLOC = 0x0044,
+	NES_AEQ_ALLOC = 0x0048
 };
 
 enum indexed_regs {
@@ -875,7 +876,6 @@ struct nes_hw_nic {
 	u8 replenishing_rq;
 	u8 reserved;
 
-	spinlock_t sq_lock;
 	spinlock_t rq_lock;
 };
 
@@ -1147,7 +1147,6 @@ struct nes_ib_device;
 struct nes_vnic {
 	struct nes_ib_device *nesibdev;
 	u64 sq_full;
-	u64 sq_locked;
 	u64 tso_requests;
 	u64 segmented_tso_requests;
 	u64 linearized_skbs;
diff --git a/drivers/infiniband/hw/nes/nes_nic.c b/drivers/infiniband/hw/nes/nes_nic.c
index 57a47cf7e513..ecb1f6fd6276 100644
--- a/drivers/infiniband/hw/nes/nes_nic.c
+++ b/drivers/infiniband/hw/nes/nes_nic.c
@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2006 - 2008 NetEffect, Inc. All rights reserved.
+ * Copyright (c) 2006 - 2009 Intel-NE, Inc.  All rights reserved.
  *
  * This software is available to you under a choice of one of two
  * licenses.  You may choose to be licensed under the terms of the GNU
@@ -111,7 +111,7 @@ static int nes_netdev_poll(struct napi_struct *napi, int budget)
 	nes_nic_ce_handler(nesdev, nescq);
 
 	if (nescq->cqes_pending == 0) {
-		netif_rx_complete(napi);
+		napi_complete(napi);
 		/* clear out completed cqes and arm */
 		nes_write32(nesdev->regs+NES_CQE_ALLOC, NES_CQE_ALLOC_NOTIFY_NEXT |
 				nescq->cq_number | (nescq->cqe_allocs_pending << 16));
@@ -400,8 +400,7 @@ static int nes_nic_send(struct sk_buff *skb, struct net_device *netdev)
 	if (skb_headlen(skb) == skb->len) {
 		if (skb_headlen(skb) <= NES_FIRST_FRAG_SIZE) {
 			nic_sqe->wqe_words[NES_NIC_SQ_WQE_LENGTH_2_1_IDX] = 0;
-			nesnic->tx_skb[nesnic->sq_head] = NULL;
-			dev_kfree_skb(skb);
+			nesnic->tx_skb[nesnic->sq_head] = skb;
 		}
 	} else {
 		/* Deal with Fragments */
@@ -453,7 +452,6 @@ static int nes_netdev_start_xmit(struct sk_buff *skb, struct net_device *netdev)
 	u32 wqe_count=1;
 	u32 send_rc;
 	struct iphdr *iph;
-	unsigned long flags;
 	__le16 *wqe_fragment_length;
 	u32 nr_frags;
 	u32 original_first_length;
@@ -480,13 +478,6 @@ static int nes_netdev_start_xmit(struct sk_buff *skb, struct net_device *netdev)
 	if (netif_queue_stopped(netdev))
 		return NETDEV_TX_BUSY;
 
-	local_irq_save(flags);
-	if (!spin_trylock(&nesnic->sq_lock)) {
-		local_irq_restore(flags);
-		nesvnic->sq_locked++;
-		return NETDEV_TX_LOCKED;
-	}
-
 	/* Check if SQ is full */
 	if ((((nesnic->sq_tail+(nesnic->sq_size*2))-nesnic->sq_head) & (nesnic->sq_size - 1)) == 1) {
 		if (!netif_queue_stopped(netdev)) {
@@ -498,7 +489,6 @@ static int nes_netdev_start_xmit(struct sk_buff *skb, struct net_device *netdev)
 			}
 		}
 		nesvnic->sq_full++;
-		spin_unlock_irqrestore(&nesnic->sq_lock, flags);
 		return NETDEV_TX_BUSY;
 	}
 
@@ -531,7 +521,6 @@ sq_no_longer_full:
 					}
 				}
 				nesvnic->sq_full++;
-				spin_unlock_irqrestore(&nesnic->sq_lock, flags);
 				nes_debug(NES_DBG_NIC_TX, "%s: HNIC SQ full- TSO request has too many frags!\n",
 						netdev->name);
 				return NETDEV_TX_BUSY;
@@ -656,17 +645,13 @@ tso_sq_no_longer_full:
 			skb_set_transport_header(skb, hoffset);
 			skb_set_network_header(skb, nhoffset);
 			send_rc = nes_nic_send(skb, netdev);
-			if (send_rc != NETDEV_TX_OK) {
-				spin_unlock_irqrestore(&nesnic->sq_lock, flags);
+			if (send_rc != NETDEV_TX_OK)
 				return NETDEV_TX_OK;
-			}
 		}
 	} else {
 		send_rc = nes_nic_send(skb, netdev);
-		if (send_rc != NETDEV_TX_OK) {
-			spin_unlock_irqrestore(&nesnic->sq_lock, flags);
+		if (send_rc != NETDEV_TX_OK)
 			return NETDEV_TX_OK;
-		}
 	}
 
 	barrier();
@@ -676,7 +661,6 @@ tso_sq_no_longer_full:
 				(wqe_count << 24) | (1 << 23) | nesvnic->nic.qp_id);
 
 	netdev->trans_start = jiffies;
-	spin_unlock_irqrestore(&nesnic->sq_lock, flags);
 
 	return NETDEV_TX_OK;
 }
@@ -1012,7 +996,6 @@ static const char nes_ethtool_stringset[][ETH_GSTRING_LEN] = {
 	"Pause Frames Received",
 	"Internal Routing Errors",
 	"SQ SW Dropped SKBs",
-	"SQ Locked",
 	"SQ Full",
 	"Segmented TSO Requests",
 	"Rx Symbol Errors",
@@ -1129,16 +1112,17 @@ static void nes_netdev_get_ethtool_stats(struct net_device *netdev,
 	struct nes_device *nesdev = nesvnic->nesdev;
 	u32 nic_count;
 	u32 u32temp;
+	u32 index = 0;
 
 	target_ethtool_stats->n_stats = NES_ETHTOOL_STAT_COUNT;
-	target_stat_values[0] = nesvnic->nesdev->link_status_interrupts;
-	target_stat_values[1] = nesvnic->linearized_skbs;
-	target_stat_values[2] = nesvnic->tso_requests;
+	target_stat_values[index] = nesvnic->nesdev->link_status_interrupts;
+	target_stat_values[++index] = nesvnic->linearized_skbs;
+	target_stat_values[++index] = nesvnic->tso_requests;
 
 	u32temp = nes_read_indexed(nesdev,
 			NES_IDX_MAC_TX_PAUSE_FRAMES + (nesvnic->nesdev->mac_index*0x200));
 	nesvnic->nesdev->mac_pause_frames_sent += u32temp;
-	target_stat_values[3] = nesvnic->nesdev->mac_pause_frames_sent;
+	target_stat_values[++index] = nesvnic->nesdev->mac_pause_frames_sent;
 
 	u32temp = nes_read_indexed(nesdev,
 			NES_IDX_MAC_RX_PAUSE_FRAMES + (nesvnic->nesdev->mac_index*0x200));
@@ -1209,60 +1193,59 @@ static void nes_netdev_get_ethtool_stats(struct net_device *netdev,
 		nesvnic->endnode_ipv4_tcp_retransmits += u32temp;
 	}
 
-	target_stat_values[4] = nesvnic->nesdev->mac_pause_frames_received;
-	target_stat_values[5] = nesdev->nesadapter->nic_rx_eth_route_err;
-	target_stat_values[6] = nesvnic->tx_sw_dropped;
-	target_stat_values[7] = nesvnic->sq_locked;
-	target_stat_values[8] = nesvnic->sq_full;
-	target_stat_values[9] = nesvnic->segmented_tso_requests;
-	target_stat_values[10] = nesvnic->nesdev->mac_rx_symbol_err_frames;
-	target_stat_values[11] = nesvnic->nesdev->mac_rx_jabber_frames;
-	target_stat_values[12] = nesvnic->nesdev->mac_rx_oversized_frames;
-	target_stat_values[13] = nesvnic->nesdev->mac_rx_short_frames;
-	target_stat_values[14] = nesvnic->endnode_nstat_rx_discard;
-	target_stat_values[15] = nesvnic->endnode_nstat_rx_octets;
-	target_stat_values[16] = nesvnic->endnode_nstat_rx_frames;
-	target_stat_values[17] = nesvnic->endnode_nstat_tx_octets;
-	target_stat_values[18] = nesvnic->endnode_nstat_tx_frames;
-	target_stat_values[19] = mh_detected;
-	target_stat_values[20] = mh_pauses_sent;
-	target_stat_values[21] = nesvnic->endnode_ipv4_tcp_retransmits;
-	target_stat_values[22] = atomic_read(&cm_connects);
-	target_stat_values[23] = atomic_read(&cm_accepts);
-	target_stat_values[24] = atomic_read(&cm_disconnects);
-	target_stat_values[25] = atomic_read(&cm_connecteds);
-	target_stat_values[26] = atomic_read(&cm_connect_reqs);
-	target_stat_values[27] = atomic_read(&cm_rejects);
-	target_stat_values[28] = atomic_read(&mod_qp_timouts);
-	target_stat_values[29] = atomic_read(&qps_created);
-	target_stat_values[30] = atomic_read(&sw_qps_destroyed);
-	target_stat_values[31] = atomic_read(&qps_destroyed);
-	target_stat_values[32] = atomic_read(&cm_closes);
-	target_stat_values[33] = cm_packets_sent;
-	target_stat_values[34] = cm_packets_bounced;
-	target_stat_values[35] = cm_packets_created;
-	target_stat_values[36] = cm_packets_received;
-	target_stat_values[37] = cm_packets_dropped;
-	target_stat_values[38] = cm_packets_retrans;
-	target_stat_values[39] = cm_listens_created;
-	target_stat_values[40] = cm_listens_destroyed;
-	target_stat_values[41] = cm_backlog_drops;
-	target_stat_values[42] = atomic_read(&cm_loopbacks);
-	target_stat_values[43] = atomic_read(&cm_nodes_created);
-	target_stat_values[44] = atomic_read(&cm_nodes_destroyed);
-	target_stat_values[45] = atomic_read(&cm_accel_dropped_pkts);
-	target_stat_values[46] = atomic_read(&cm_resets_recvd);
-	target_stat_values[47] = int_mod_timer_init;
-	target_stat_values[48] = int_mod_cq_depth_1;
-	target_stat_values[49] = int_mod_cq_depth_4;
-	target_stat_values[50] = int_mod_cq_depth_16;
-	target_stat_values[51] = int_mod_cq_depth_24;
-	target_stat_values[52] = int_mod_cq_depth_32;
-	target_stat_values[53] = int_mod_cq_depth_128;
-	target_stat_values[54] = int_mod_cq_depth_256;
-	target_stat_values[55] = nesvnic->lro_mgr.stats.aggregated;
-	target_stat_values[56] = nesvnic->lro_mgr.stats.flushed;
-	target_stat_values[57] = nesvnic->lro_mgr.stats.no_desc;
+	target_stat_values[++index] = nesvnic->nesdev->mac_pause_frames_received;
+	target_stat_values[++index] = nesdev->nesadapter->nic_rx_eth_route_err;
+	target_stat_values[++index] = nesvnic->tx_sw_dropped;
+	target_stat_values[++index] = nesvnic->sq_full;
+	target_stat_values[++index] = nesvnic->segmented_tso_requests;
+	target_stat_values[++index] = nesvnic->nesdev->mac_rx_symbol_err_frames;
+	target_stat_values[++index] = nesvnic->nesdev->mac_rx_jabber_frames;
+	target_stat_values[++index] = nesvnic->nesdev->mac_rx_oversized_frames;
+	target_stat_values[++index] = nesvnic->nesdev->mac_rx_short_frames;
+	target_stat_values[++index] = nesvnic->endnode_nstat_rx_discard;
+	target_stat_values[++index] = nesvnic->endnode_nstat_rx_octets;
+	target_stat_values[++index] = nesvnic->endnode_nstat_rx_frames;
+	target_stat_values[++index] = nesvnic->endnode_nstat_tx_octets;
+	target_stat_values[++index] = nesvnic->endnode_nstat_tx_frames;
+	target_stat_values[++index] = mh_detected;
+	target_stat_values[++index] = mh_pauses_sent;
+	target_stat_values[++index] = nesvnic->endnode_ipv4_tcp_retransmits;
+	target_stat_values[++index] = atomic_read(&cm_connects);
+	target_stat_values[++index] = atomic_read(&cm_accepts);
+	target_stat_values[++index] = atomic_read(&cm_disconnects);
+	target_stat_values[++index] = atomic_read(&cm_connecteds);
+	target_stat_values[++index] = atomic_read(&cm_connect_reqs);
+	target_stat_values[++index] = atomic_read(&cm_rejects);
+	target_stat_values[++index] = atomic_read(&mod_qp_timouts);
+	target_stat_values[++index] = atomic_read(&qps_created);
+	target_stat_values[++index] = atomic_read(&sw_qps_destroyed);
+	target_stat_values[++index] = atomic_read(&qps_destroyed);
+	target_stat_values[++index] = atomic_read(&cm_closes);
+	target_stat_values[++index] = cm_packets_sent;
+	target_stat_values[++index] = cm_packets_bounced;
+	target_stat_values[++index] = cm_packets_created;
+	target_stat_values[++index] = cm_packets_received;
+	target_stat_values[++index] = cm_packets_dropped;
+	target_stat_values[++index] = cm_packets_retrans;
+	target_stat_values[++index] = cm_listens_created;
+	target_stat_values[++index] = cm_listens_destroyed;
+	target_stat_values[++index] = cm_backlog_drops;
+	target_stat_values[++index] = atomic_read(&cm_loopbacks);
+	target_stat_values[++index] = atomic_read(&cm_nodes_created);
+	target_stat_values[++index] = atomic_read(&cm_nodes_destroyed);
+	target_stat_values[++index] = atomic_read(&cm_accel_dropped_pkts);
+	target_stat_values[++index] = atomic_read(&cm_resets_recvd);
+	target_stat_values[++index] = int_mod_timer_init;
+	target_stat_values[++index] = int_mod_cq_depth_1;
+	target_stat_values[++index] = int_mod_cq_depth_4;
+	target_stat_values[++index] = int_mod_cq_depth_16;
+	target_stat_values[++index] = int_mod_cq_depth_24;
+	target_stat_values[++index] = int_mod_cq_depth_32;
+	target_stat_values[++index] = int_mod_cq_depth_128;
+	target_stat_values[++index] = int_mod_cq_depth_256;
+	target_stat_values[++index] = nesvnic->lro_mgr.stats.aggregated;
+	target_stat_values[++index] = nesvnic->lro_mgr.stats.flushed;
+	target_stat_values[++index] = nesvnic->lro_mgr.stats.no_desc;
 
 }
 
@@ -1568,6 +1551,19 @@ static void nes_netdev_vlan_rx_register(struct net_device *netdev, struct vlan_g
 	spin_unlock_irqrestore(&nesadapter->phy_lock, flags);
 }
 
+static const struct net_device_ops nes_netdev_ops = {
+	.ndo_open 		= nes_netdev_open,
+	.ndo_stop		= nes_netdev_stop,
+	.ndo_start_xmit 	= nes_netdev_start_xmit,
+	.ndo_get_stats		= nes_netdev_get_stats,
+	.ndo_tx_timeout 	= nes_netdev_tx_timeout,
+	.ndo_set_mac_address	= nes_netdev_set_mac_address,
+	.ndo_set_multicast_list = nes_netdev_set_multicast_list,
+	.ndo_change_mtu		= nes_netdev_change_mtu,
+	.ndo_set_mac_address 	= eth_mac_addr,
+	.ndo_validate_addr	= eth_validate_addr,
+	.ndo_vlan_rx_register 	= nes_netdev_vlan_rx_register,
+};
 
 /**
  * nes_netdev_init - initialize network device
@@ -1576,7 +1572,7 @@ struct net_device *nes_netdev_init(struct nes_device *nesdev,
 		void __iomem *mmio_addr)
 {
 	u64 u64temp;
-	struct nes_vnic *nesvnic = NULL;
+	struct nes_vnic *nesvnic;
 	struct net_device *netdev;
 	struct nic_qp_map *curr_qp_map;
 	u32 u32temp;
@@ -1588,22 +1584,12 @@ struct net_device *nes_netdev_init(struct nes_device *nesdev,
 		printk(KERN_ERR PFX "nesvnic etherdev alloc failed");
 		return NULL;
 	}
+	nesvnic = netdev_priv(netdev);
 
 	nes_debug(NES_DBG_INIT, "netdev = %p, %s\n", netdev, netdev->name);
 
 	SET_NETDEV_DEV(netdev, &nesdev->pcidev->dev);
 
-	nesvnic = netdev_priv(netdev);
-	memset(nesvnic, 0, sizeof(*nesvnic));
-
-	netdev->open = nes_netdev_open;
-	netdev->stop = nes_netdev_stop;
-	netdev->hard_start_xmit = nes_netdev_start_xmit;
-	netdev->get_stats = nes_netdev_get_stats;
-	netdev->tx_timeout = nes_netdev_tx_timeout;
-	netdev->set_mac_address = nes_netdev_set_mac_address;
-	netdev->set_multicast_list = nes_netdev_set_multicast_list;
-	netdev->change_mtu = nes_netdev_change_mtu;
 	netdev->watchdog_timeo = NES_TX_TIMEOUT;
 	netdev->irq = nesdev->pcidev->irq;
 	netdev->mtu = ETH_DATA_LEN;
@@ -1611,12 +1597,11 @@ struct net_device *nes_netdev_init(struct nes_device *nesdev,
 	netdev->addr_len = ETH_ALEN;
 	netdev->type = ARPHRD_ETHER;
 	netdev->features = NETIF_F_HIGHDMA;
+	netdev->netdev_ops = &nes_netdev_ops;
 	netdev->ethtool_ops = &nes_ethtool_ops;
 	netif_napi_add(netdev, &nesvnic->napi, nes_netdev_poll, 128);
 	nes_debug(NES_DBG_INIT, "Enabling VLAN Insert/Delete.\n");
 	netdev->features |= NETIF_F_HW_VLAN_TX | NETIF_F_HW_VLAN_RX;
-	netdev->vlan_rx_register = nes_netdev_vlan_rx_register;
-	netdev->features |= NETIF_F_LLTX;
 
 	/* Fill in the port structure */
 	nesvnic->netdev = netdev;
diff --git a/drivers/infiniband/hw/nes/nes_user.h b/drivers/infiniband/hw/nes/nes_user.h
index e64306bce80b..cc90c14b49eb 100644
--- a/drivers/infiniband/hw/nes/nes_user.h
+++ b/drivers/infiniband/hw/nes/nes_user.h
@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2006 - 2008 NetEffect.  All rights reserved.
+ * Copyright (c) 2006 - 2009 Intel-NE, Inc.  All rights reserved.
  * Copyright (c) 2005 Topspin Communications.  All rights reserved.
  * Copyright (c) 2005 Cisco Systems.  All rights reserved.
  * Copyright (c) 2005 Open Grid Computing, Inc. All rights reserved.
diff --git a/drivers/infiniband/hw/nes/nes_utils.c b/drivers/infiniband/hw/nes/nes_utils.c
index 6f3bc1b6bf22..a282031d15c7 100644
--- a/drivers/infiniband/hw/nes/nes_utils.c
+++ b/drivers/infiniband/hw/nes/nes_utils.c
@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2006 - 2008 NetEffect, Inc. All rights reserved.
+ * Copyright (c) 2006 - 2009 Intel-NE, Inc.  All rights reserved.
  *
  * This software is available to you under a choice of one of two
  * licenses.  You may choose to be licensed under the terms of the GNU
diff --git a/drivers/infiniband/hw/nes/nes_verbs.c b/drivers/infiniband/hw/nes/nes_verbs.c
index d93a6562817c..7e5b5ba13a74 100644
--- a/drivers/infiniband/hw/nes/nes_verbs.c
+++ b/drivers/infiniband/hw/nes/nes_verbs.c
@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2006 - 2008 NetEffect, Inc. All rights reserved.
+ * Copyright (c) 2006 - 2009 Intel-NE, Inc.  All rights reserved.
  *
  * This software is available to you under a choice of one of two
  * licenses.  You may choose to be licensed under the terms of the GNU
@@ -551,6 +551,7 @@ static int nes_dealloc_fmr(struct ib_fmr *ibfmr)
 	struct nes_device *nesdev = nesvnic->nesdev;
 	struct nes_adapter *nesadapter = nesdev->nesadapter;
 	int i = 0;
+	int rc;
 
 	/* free the resources */
 	if (nesfmr->leaf_pbl_cnt == 0) {
@@ -572,7 +573,9 @@ static int nes_dealloc_fmr(struct ib_fmr *ibfmr)
 	nesmr->ibmw.rkey = ibfmr->rkey;
 	nesmr->ibmw.uobject = NULL;
 
-	if (nesfmr->nesmr.pbls_used != 0) {
+	rc = nes_dealloc_mw(&nesmr->ibmw);
+
+	if ((rc == 0) && (nesfmr->nesmr.pbls_used != 0)) {
 		spin_lock_irqsave(&nesadapter->pbl_lock, flags);
 		if (nesfmr->nesmr.pbl_4k) {
 			nesadapter->free_4kpbl += nesfmr->nesmr.pbls_used;
@@ -584,7 +587,7 @@ static int nes_dealloc_fmr(struct ib_fmr *ibfmr)
 		spin_unlock_irqrestore(&nesadapter->pbl_lock, flags);
 	}
 
-	return nes_dealloc_mw(&nesmr->ibmw);
+	return rc;
 }
 
 
@@ -1886,21 +1889,75 @@ static int nes_destroy_cq(struct ib_cq *ib_cq)
 	return ret;
 }
 
+/**
+ * root_256
+ */
+static u32 root_256(struct nes_device *nesdev,
+		    struct nes_root_vpbl *root_vpbl,
+		    struct nes_root_vpbl *new_root,
+		    u16 pbl_count_4k,
+		    u16 pbl_count_256)
+{
+	u64 leaf_pbl;
+	int i, j, k;
+
+	if (pbl_count_4k == 1) {
+		new_root->pbl_vbase = pci_alloc_consistent(nesdev->pcidev,
+						512, &new_root->pbl_pbase);
+
+		if (new_root->pbl_vbase == NULL)
+			return 0;
+
+		leaf_pbl = (u64)root_vpbl->pbl_pbase;
+		for (i = 0; i < 16; i++) {
+			new_root->pbl_vbase[i].pa_low =
+				cpu_to_le32((u32)leaf_pbl);
+			new_root->pbl_vbase[i].pa_high =
+				cpu_to_le32((u32)((((u64)leaf_pbl) >> 32)));
+			leaf_pbl += 256;
+		}
+	} else {
+		for (i = 3; i >= 0; i--) {
+			j = i * 16;
+			root_vpbl->pbl_vbase[j] = root_vpbl->pbl_vbase[i];
+			leaf_pbl = le32_to_cpu(root_vpbl->pbl_vbase[j].pa_low) +
+			    (((u64)le32_to_cpu(root_vpbl->pbl_vbase[j].pa_high))
+				<< 32);
+			for (k = 1; k < 16; k++) {
+				leaf_pbl += 256;
+				root_vpbl->pbl_vbase[j + k].pa_low =
+						cpu_to_le32((u32)leaf_pbl);
+				root_vpbl->pbl_vbase[j + k].pa_high =
+				    cpu_to_le32((u32)((((u64)leaf_pbl) >> 32)));
+			}
+		}
+	}
+
+	return 1;
+}
+
 
 /**
  * nes_reg_mr
  */
 static int nes_reg_mr(struct nes_device *nesdev, struct nes_pd *nespd,
 		u32 stag, u64 region_length, struct nes_root_vpbl *root_vpbl,
-		dma_addr_t single_buffer, u16 pbl_count, u16 residual_page_count,
-		int acc, u64 *iova_start)
+		dma_addr_t single_buffer, u16 pbl_count_4k,
+		u16 residual_page_count_4k, int acc, u64 *iova_start,
+		u16 *actual_pbl_cnt, u8 *used_4k_pbls)
 {
 	struct nes_hw_cqp_wqe *cqp_wqe;
 	struct nes_cqp_request *cqp_request;
 	unsigned long flags;
 	int ret;
 	struct nes_adapter *nesadapter = nesdev->nesadapter;
-	/* int count; */
+	uint pg_cnt = 0;
+	u16 pbl_count_256;
+	u16 pbl_count = 0;
+	u8  use_256_pbls = 0;
+	u8  use_4k_pbls = 0;
+	u16 use_two_level = (pbl_count_4k > 1) ? 1 : 0;
+	struct nes_root_vpbl new_root = {0, 0, 0};
 	u32 opcode = 0;
 	u16 major_code;
 
@@ -1913,41 +1970,70 @@ static int nes_reg_mr(struct nes_device *nesdev, struct nes_pd *nespd,
 	cqp_request->waiting = 1;
 	cqp_wqe = &cqp_request->cqp_wqe;
 
-	spin_lock_irqsave(&nesadapter->pbl_lock, flags);
-	/* track PBL resources */
-	if (pbl_count != 0) {
-		if (pbl_count > 1) {
-			/* Two level PBL */
-			if ((pbl_count+1) > nesadapter->free_4kpbl) {
-				nes_debug(NES_DBG_MR, "Out of 4KB Pbls for two level request.\n");
-				spin_unlock_irqrestore(&nesadapter->pbl_lock, flags);
-				nes_free_cqp_request(nesdev, cqp_request);
-				return -ENOMEM;
-			} else {
-				nesadapter->free_4kpbl -= pbl_count+1;
-			}
-		} else if (residual_page_count > 32) {
-			if (pbl_count > nesadapter->free_4kpbl) {
-				nes_debug(NES_DBG_MR, "Out of 4KB Pbls.\n");
-				spin_unlock_irqrestore(&nesadapter->pbl_lock, flags);
-				nes_free_cqp_request(nesdev, cqp_request);
-				return -ENOMEM;
-			} else {
-				nesadapter->free_4kpbl -= pbl_count;
+	if (pbl_count_4k) {
+		spin_lock_irqsave(&nesadapter->pbl_lock, flags);
+
+		pg_cnt = ((pbl_count_4k - 1) * 512) + residual_page_count_4k;
+		pbl_count_256 = (pg_cnt + 31) / 32;
+		if (pg_cnt <= 32) {
+			if (pbl_count_256 <= nesadapter->free_256pbl)
+				use_256_pbls = 1;
+			else if (pbl_count_4k <= nesadapter->free_4kpbl)
+				use_4k_pbls = 1;
+		} else if (pg_cnt <= 2048) {
+			if (((pbl_count_4k + use_two_level) <= nesadapter->free_4kpbl) &&
+			    (nesadapter->free_4kpbl > (nesadapter->max_4kpbl >> 1))) {
+				use_4k_pbls = 1;
+			} else if ((pbl_count_256 + 1) <= nesadapter->free_256pbl) {
+				use_256_pbls = 1;
+				use_two_level = 1;
+			} else if ((pbl_count_4k + use_two_level) <= nesadapter->free_4kpbl) {
+				use_4k_pbls = 1;
 			}
 		} else {
-			if (pbl_count > nesadapter->free_256pbl) {
-				nes_debug(NES_DBG_MR, "Out of 256B Pbls.\n");
-				spin_unlock_irqrestore(&nesadapter->pbl_lock, flags);
-				nes_free_cqp_request(nesdev, cqp_request);
-				return -ENOMEM;
-			} else {
-				nesadapter->free_256pbl -= pbl_count;
-			}
+			if ((pbl_count_4k + 1) <= nesadapter->free_4kpbl)
+				use_4k_pbls = 1;
 		}
+
+		if (use_256_pbls) {
+			pbl_count = pbl_count_256;
+			nesadapter->free_256pbl -= pbl_count + use_two_level;
+		} else if (use_4k_pbls) {
+			pbl_count =  pbl_count_4k;
+			nesadapter->free_4kpbl -= pbl_count + use_two_level;
+		} else {
+			spin_unlock_irqrestore(&nesadapter->pbl_lock, flags);
+			nes_debug(NES_DBG_MR, "Out of Pbls\n");
+			nes_free_cqp_request(nesdev, cqp_request);
+			return -ENOMEM;
+		}
+
+		spin_unlock_irqrestore(&nesadapter->pbl_lock, flags);
 	}
 
-	spin_unlock_irqrestore(&nesadapter->pbl_lock, flags);
+	if (use_256_pbls && use_two_level) {
+		if (root_256(nesdev, root_vpbl, &new_root, pbl_count_4k, pbl_count_256) == 1) {
+			if (new_root.pbl_pbase != 0)
+				root_vpbl = &new_root;
+		} else {
+			spin_lock_irqsave(&nesadapter->pbl_lock, flags);
+			nesadapter->free_256pbl += pbl_count_256 + use_two_level;
+			use_256_pbls = 0;
+
+			if (pbl_count_4k == 1)
+				use_two_level = 0;
+			pbl_count = pbl_count_4k;
+
+			if ((pbl_count_4k + use_two_level) <= nesadapter->free_4kpbl) {
+				nesadapter->free_4kpbl -= pbl_count + use_two_level;
+				use_4k_pbls = 1;
+			}
+			spin_unlock_irqrestore(&nesadapter->pbl_lock, flags);
+
+			if (use_4k_pbls == 0)
+				return -ENOMEM;
+		}
+	}
 
 	opcode = NES_CQP_REGISTER_STAG | NES_CQP_STAG_RIGHTS_LOCAL_READ |
 					NES_CQP_STAG_VA_TO | NES_CQP_STAG_MR;
@@ -1976,10 +2062,9 @@ static int nes_reg_mr(struct nes_device *nesdev, struct nes_pd *nespd,
 	} else {
 		set_wqe_64bit_value(cqp_wqe->wqe_words, NES_CQP_STAG_WQE_PA_LOW_IDX, root_vpbl->pbl_pbase);
 		set_wqe_32bit_value(cqp_wqe->wqe_words, NES_CQP_STAG_WQE_PBL_BLK_COUNT_IDX, pbl_count);
-		set_wqe_32bit_value(cqp_wqe->wqe_words, NES_CQP_STAG_WQE_PBL_LEN_IDX,
-				(((pbl_count - 1) * 4096) + (residual_page_count*8)));
+		set_wqe_32bit_value(cqp_wqe->wqe_words, NES_CQP_STAG_WQE_PBL_LEN_IDX, (pg_cnt * 8));
 
-		if ((pbl_count > 1) || (residual_page_count > 32))
+		if (use_4k_pbls)
 			cqp_wqe->wqe_words[NES_CQP_WQE_OPCODE_IDX] |= cpu_to_le32(NES_CQP_STAG_PBL_BLK_SIZE);
 	}
 	barrier();
@@ -1996,13 +2081,25 @@ static int nes_reg_mr(struct nes_device *nesdev, struct nes_pd *nespd,
 	major_code = cqp_request->major_code;
 	nes_put_cqp_request(nesdev, cqp_request);
 
+	if ((!ret || major_code) && pbl_count != 0) {
+		spin_lock_irqsave(&nesadapter->pbl_lock, flags);
+		if (use_256_pbls)
+			nesadapter->free_256pbl += pbl_count + use_two_level;
+		else if (use_4k_pbls)
+			nesadapter->free_4kpbl += pbl_count + use_two_level;
+		spin_unlock_irqrestore(&nesadapter->pbl_lock, flags);
+	}
+	if (new_root.pbl_pbase)
+		pci_free_consistent(nesdev->pcidev, 512, new_root.pbl_vbase,
+				    new_root.pbl_pbase);
+
 	if (!ret)
 		return -ETIME;
 	else if (major_code)
 		return -EIO;
-	else
-		return 0;
 
+	*actual_pbl_cnt = pbl_count + use_two_level;
+	*used_4k_pbls = use_4k_pbls;
 	return 0;
 }
 
@@ -2167,18 +2264,14 @@ static struct ib_mr *nes_reg_phys_mr(struct ib_pd *ib_pd,
 		pbl_count = root_pbl_index;
 	}
 	ret = nes_reg_mr(nesdev, nespd, stag, region_length, &root_vpbl,
-			buffer_list[0].addr, pbl_count, (u16)cur_pbl_index, acc, iova_start);
+			buffer_list[0].addr, pbl_count, (u16)cur_pbl_index, acc, iova_start,
+			&nesmr->pbls_used, &nesmr->pbl_4k);
 
 	if (ret == 0) {
 		nesmr->ibmr.rkey = stag;
 		nesmr->ibmr.lkey = stag;
 		nesmr->mode = IWNES_MEMREG_TYPE_MEM;
 		ibmr = &nesmr->ibmr;
-		nesmr->pbl_4k = ((pbl_count > 1) || (cur_pbl_index > 32)) ? 1 : 0;
-		nesmr->pbls_used = pbl_count;
-		if (pbl_count > 1) {
-			nesmr->pbls_used++;
-		}
 	} else {
 		kfree(nesmr);
 		ibmr = ERR_PTR(-ENOMEM);
@@ -2456,8 +2549,9 @@ static struct ib_mr *nes_reg_user_mr(struct ib_pd *pd, u64 start, u64 length,
 					stag, (unsigned int)iova_start,
 					(unsigned int)region_length, stag_index,
 					(unsigned long long)region->length, pbl_count);
-			ret = nes_reg_mr( nesdev, nespd, stag, region->length, &root_vpbl,
-					first_dma_addr, pbl_count, (u16)cur_pbl_index, acc, &iova_start);
+			ret = nes_reg_mr(nesdev, nespd, stag, region->length, &root_vpbl,
+					 first_dma_addr, pbl_count, (u16)cur_pbl_index, acc,
+					 &iova_start, &nesmr->pbls_used, &nesmr->pbl_4k);
 
 			nes_debug(NES_DBG_MR, "ret=%d\n", ret);
 
@@ -2466,11 +2560,6 @@ static struct ib_mr *nes_reg_user_mr(struct ib_pd *pd, u64 start, u64 length,
 				nesmr->ibmr.lkey = stag;
 				nesmr->mode = IWNES_MEMREG_TYPE_MEM;
 				ibmr = &nesmr->ibmr;
-				nesmr->pbl_4k = ((pbl_count > 1) || (cur_pbl_index > 32)) ? 1 : 0;
-				nesmr->pbls_used = pbl_count;
-				if (pbl_count > 1) {
-					nesmr->pbls_used++;
-				}
 			} else {
 				ib_umem_release(region);
 				kfree(nesmr);
@@ -2609,24 +2698,6 @@ static int nes_dereg_mr(struct ib_mr *ib_mr)
 	cqp_request->waiting = 1;
 	cqp_wqe = &cqp_request->cqp_wqe;
 
-	spin_lock_irqsave(&nesadapter->pbl_lock, flags);
-	if (nesmr->pbls_used != 0) {
-		if (nesmr->pbl_4k) {
-			nesadapter->free_4kpbl += nesmr->pbls_used;
-			if (nesadapter->free_4kpbl > nesadapter->max_4kpbl) {
-				printk(KERN_ERR PFX "free 4KB PBLs(%u) has exceeded the max(%u)\n",
-						nesadapter->free_4kpbl, nesadapter->max_4kpbl);
-			}
-		} else {
-			nesadapter->free_256pbl += nesmr->pbls_used;
-			if (nesadapter->free_256pbl > nesadapter->max_256pbl) {
-				printk(KERN_ERR PFX "free 256B PBLs(%u) has exceeded the max(%u)\n",
-						nesadapter->free_256pbl, nesadapter->max_256pbl);
-			}
-		}
-	}
-
-	spin_unlock_irqrestore(&nesadapter->pbl_lock, flags);
 	nes_fill_init_cqp_wqe(cqp_wqe, nesdev);
 	set_wqe_32bit_value(cqp_wqe->wqe_words, NES_CQP_WQE_OPCODE_IDX,
 			NES_CQP_DEALLOCATE_STAG | NES_CQP_STAG_VA_TO |
@@ -2644,11 +2715,6 @@ static int nes_dereg_mr(struct ib_mr *ib_mr)
 			" CQP Major:Minor codes = 0x%04X:0x%04X\n",
 			ib_mr->rkey, ret, cqp_request->major_code, cqp_request->minor_code);
 
-	nes_free_resource(nesadapter, nesadapter->allocated_mrs,
-			(ib_mr->rkey & 0x0fffff00) >> 8);
-
-	kfree(nesmr);
-
 	major_code = cqp_request->major_code;
 	minor_code = cqp_request->minor_code;
 
@@ -2664,8 +2730,33 @@ static int nes_dereg_mr(struct ib_mr *ib_mr)
 				" to destroy STag, ib_mr=%p, rkey = 0x%08X\n",
 				major_code, minor_code, ib_mr, ib_mr->rkey);
 		return -EIO;
-	} else
-		return 0;
+	}
+
+	if (nesmr->pbls_used != 0) {
+		spin_lock_irqsave(&nesadapter->pbl_lock, flags);
+		if (nesmr->pbl_4k) {
+			nesadapter->free_4kpbl += nesmr->pbls_used;
+			if (nesadapter->free_4kpbl > nesadapter->max_4kpbl)
+				printk(KERN_ERR PFX "free 4KB PBLs(%u) has "
+					"exceeded the max(%u)\n",
+					nesadapter->free_4kpbl,
+					nesadapter->max_4kpbl);
+		} else {
+			nesadapter->free_256pbl += nesmr->pbls_used;
+			if (nesadapter->free_256pbl > nesadapter->max_256pbl)
+				printk(KERN_ERR PFX "free 256B PBLs(%u) has "
+					"exceeded the max(%u)\n",
+					nesadapter->free_256pbl,
+					nesadapter->max_256pbl);
+		}
+		spin_unlock_irqrestore(&nesadapter->pbl_lock, flags);
+	}
+	nes_free_resource(nesadapter, nesadapter->allocated_mrs,
+			(ib_mr->rkey & 0x0fffff00) >> 8);
+
+	kfree(nesmr);
+
+	return 0;
 }
 
 
diff --git a/drivers/infiniband/hw/nes/nes_verbs.h b/drivers/infiniband/hw/nes/nes_verbs.h
index ae0ca9bc83bd..5e48f67fbe8d 100644
--- a/drivers/infiniband/hw/nes/nes_verbs.h
+++ b/drivers/infiniband/hw/nes/nes_verbs.h
@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2006 - 2008 NetEffect, Inc. All rights reserved.
+ * Copyright (c) 2006 - 2009 Intel-NE, Inc.  All rights reserved.
  * Copyright (c) 2005 Open Grid Computing, Inc. All rights reserved.
  *
  * This software is available to you under a choice of one of two