1 files changed, 449 insertions, 58 deletions

diff --git a/drivers/infiniband/hw/mlx4/cq.c b/drivers/infiniband/hw/mlx4/cq.c
index 7360bbafbe8..1066eec854a 100644
--- a/drivers/infiniband/hw/mlx4/cq.c
+++ b/drivers/infiniband/hw/mlx4/cq.c
@@ -1,5 +1,6 @@
 /*
  * Copyright (c) 2007 Cisco Systems, Inc. All rights reserved.
+ * Copyright (c) 2007, 2008 Mellanox Technologies. 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
@@ -32,6 +33,8 @@
 
 #include <linux/mlx4/cq.h>
 #include <linux/mlx4/qp.h>
+#include <linux/mlx4/srq.h>
+#include <linux/slab.h>
 
 #include "mlx4_ib.h"
 #include "user.h"
@@ -48,7 +51,7 @@ static void mlx4_ib_cq_event(struct mlx4_cq *cq, enum mlx4_event type)
 	struct ib_cq *ibcq;
 
 	if (type != MLX4_EVENT_TYPE_CQ_ERROR) {
-		printk(KERN_WARNING "mlx4_ib: Unexpected event type %d "
+		pr_warn("Unexpected event type %d "
 		       "on CQ %06x\n", type, cq->cqn);
 		return;
 	}
@@ -64,7 +67,7 @@ static void mlx4_ib_cq_event(struct mlx4_cq *cq, enum mlx4_event type)
 
 static void *get_cqe_from_buf(struct mlx4_ib_cq_buf *buf, int n)
 {
-	return mlx4_buf_offset(&buf->buf, n * sizeof (struct mlx4_cqe));
+	return mlx4_buf_offset(&buf->buf, n * buf->entry_size);
 }
 
 static void *get_cqe(struct mlx4_ib_cq *cq, int n)
@@ -75,8 +78,9 @@ static void *get_cqe(struct mlx4_ib_cq *cq, int n)
 static void *get_sw_cqe(struct mlx4_ib_cq *cq, int n)
 {
 	struct mlx4_cqe *cqe = get_cqe(cq, n & cq->ibcq.cqe);
+	struct mlx4_cqe *tcqe = ((cq->buf.entry_size == 64) ? (cqe + 1) : cqe);
 
-	return (!!(cqe->owner_sr_opcode & MLX4_CQE_OWNER_MASK) ^
+	return (!!(tcqe->owner_sr_opcode & MLX4_CQE_OWNER_MASK) ^
 		!!(n & (cq->ibcq.cqe + 1))) ? NULL : cqe;
 }
 
@@ -85,6 +89,83 @@ static struct mlx4_cqe *next_cqe_sw(struct mlx4_ib_cq *cq)
 	return get_sw_cqe(cq, cq->mcq.cons_index);
 }
 
+int mlx4_ib_modify_cq(struct ib_cq *cq, u16 cq_count, u16 cq_period)
+{
+	struct mlx4_ib_cq *mcq = to_mcq(cq);
+	struct mlx4_ib_dev *dev = to_mdev(cq->device);
+
+	return mlx4_cq_modify(dev->dev, &mcq->mcq, cq_count, cq_period);
+}
+
+static int mlx4_ib_alloc_cq_buf(struct mlx4_ib_dev *dev, struct mlx4_ib_cq_buf *buf, int nent)
+{
+	int err;
+
+	err = mlx4_buf_alloc(dev->dev, nent * dev->dev->caps.cqe_size,
+			     PAGE_SIZE * 2, &buf->buf, GFP_KERNEL);
+
+	if (err)
+		goto out;
+
+	buf->entry_size = dev->dev->caps.cqe_size;
+	err = mlx4_mtt_init(dev->dev, buf->buf.npages, buf->buf.page_shift,
+				    &buf->mtt);
+	if (err)
+		goto err_buf;
+
+	err = mlx4_buf_write_mtt(dev->dev, &buf->mtt, &buf->buf, GFP_KERNEL);
+	if (err)
+		goto err_mtt;
+
+	return 0;
+
+err_mtt:
+	mlx4_mtt_cleanup(dev->dev, &buf->mtt);
+
+err_buf:
+	mlx4_buf_free(dev->dev, nent * buf->entry_size, &buf->buf);
+
+out:
+	return err;
+}
+
+static void mlx4_ib_free_cq_buf(struct mlx4_ib_dev *dev, struct mlx4_ib_cq_buf *buf, int cqe)
+{
+	mlx4_buf_free(dev->dev, (cqe + 1) * buf->entry_size, &buf->buf);
+}
+
+static int mlx4_ib_get_cq_umem(struct mlx4_ib_dev *dev, struct ib_ucontext *context,
+			       struct mlx4_ib_cq_buf *buf, struct ib_umem **umem,
+			       u64 buf_addr, int cqe)
+{
+	int err;
+	int cqe_size = dev->dev->caps.cqe_size;
+
+	*umem = ib_umem_get(context, buf_addr, cqe * cqe_size,
+			    IB_ACCESS_LOCAL_WRITE, 1);
+	if (IS_ERR(*umem))
+		return PTR_ERR(*umem);
+
+	err = mlx4_mtt_init(dev->dev, ib_umem_page_count(*umem),
+			    ilog2((*umem)->page_size), &buf->mtt);
+	if (err)
+		goto err_buf;
+
+	err = mlx4_ib_umem_write_mtt(dev, &buf->mtt, *umem);
+	if (err)
+		goto err_mtt;
+
+	return 0;
+
+err_mtt:
+	mlx4_mtt_cleanup(dev->dev, &buf->mtt);
+
+err_buf:
+	ib_umem_release(*umem);
+
+	return err;
+}
+
 struct ib_cq *mlx4_ib_create_cq(struct ib_device *ibdev, int entries, int vector,
 				struct ib_ucontext *context,
 				struct ib_udata *udata)
@@ -92,7 +173,6 @@ struct ib_cq *mlx4_ib_create_cq(struct ib_device *ibdev, int entries, int vector
 	struct mlx4_ib_dev *dev = to_mdev(ibdev);
 	struct mlx4_ib_cq *cq;
 	struct mlx4_uar *uar;
-	int buf_size;
 	int err;
 
 	if (entries < 1 || entries > dev->dev->caps.max_cqes)
@@ -104,8 +184,10 @@ struct ib_cq *mlx4_ib_create_cq(struct ib_device *ibdev, int entries, int vector
 
 	entries      = roundup_pow_of_two(entries + 1);
 	cq->ibcq.cqe = entries - 1;
-	buf_size     = entries * sizeof (struct mlx4_cqe);
+	mutex_init(&cq->resize_mutex);
 	spin_lock_init(&cq->lock);
+	cq->resize_buf = NULL;
+	cq->resize_umem = NULL;
 
 	if (context) {
 		struct mlx4_ib_create_cq ucmd;
@@ -115,21 +197,10 @@ struct ib_cq *mlx4_ib_create_cq(struct ib_device *ibdev, int entries, int vector
 			goto err_cq;
 		}
 
-		cq->umem = ib_umem_get(context, ucmd.buf_addr, buf_size,
-				       IB_ACCESS_LOCAL_WRITE);
-		if (IS_ERR(cq->umem)) {
-			err = PTR_ERR(cq->umem);
-			goto err_cq;
-		}
-
-		err = mlx4_mtt_init(dev->dev, ib_umem_page_count(cq->umem),
-				    ilog2(cq->umem->page_size), &cq->buf.mtt);
+		err = mlx4_ib_get_cq_umem(dev, context, &cq->buf, &cq->umem,
+					  ucmd.buf_addr, entries);
 		if (err)
-			goto err_buf;
-
-		err = mlx4_ib_umem_write_mtt(dev, &cq->buf.mtt, cq->umem);
-		if (err)
-			goto err_mtt;
+			goto err_cq;
 
 		err = mlx4_ib_db_map_user(to_mucontext(context), ucmd.db_addr,
 					  &cq->db);
@@ -138,7 +209,7 @@ struct ib_cq *mlx4_ib_create_cq(struct ib_device *ibdev, int entries, int vector
 
 		uar = &to_mucontext(context)->uar;
 	} else {
-		err = mlx4_ib_db_alloc(dev, &cq->db, 1);
+		err = mlx4_db_alloc(dev->dev, &cq->db, 1, GFP_KERNEL);
 		if (err)
 			goto err_cq;
 
@@ -147,25 +218,18 @@ struct ib_cq *mlx4_ib_create_cq(struct ib_device *ibdev, int entries, int vector
 		*cq->mcq.set_ci_db = 0;
 		*cq->mcq.arm_db    = 0;
 
-		if (mlx4_buf_alloc(dev->dev, buf_size, PAGE_SIZE * 2, &cq->buf.buf)) {
-			err = -ENOMEM;
-			goto err_db;
-		}
-
-		err = mlx4_mtt_init(dev->dev, cq->buf.buf.npages, cq->buf.buf.page_shift,
-				    &cq->buf.mtt);
-		if (err)
-			goto err_buf;
-
-		err = mlx4_buf_write_mtt(dev->dev, &cq->buf.mtt, &cq->buf.buf);
+		err = mlx4_ib_alloc_cq_buf(dev, &cq->buf, entries);
 		if (err)
-			goto err_mtt;
+			goto err_db;
 
 		uar = &dev->priv_uar;
 	}
 
+	if (dev->eq_table)
+		vector = dev->eq_table[vector % ibdev->num_comp_vectors];
+
 	err = mlx4_cq_alloc(dev->dev, entries, &cq->buf.mtt, uar,
-			    cq->db.dma, &cq->mcq);
+			    cq->db.dma, &cq->mcq, vector, 0, 0);
 	if (err)
 		goto err_dbmap;
 
@@ -187,16 +251,14 @@ err_dbmap:
 err_mtt:
 	mlx4_mtt_cleanup(dev->dev, &cq->buf.mtt);
 
-err_buf:
 	if (context)
 		ib_umem_release(cq->umem);
 	else
-		mlx4_buf_free(dev->dev, entries * sizeof (struct mlx4_cqe),
-			      &cq->buf.buf);
+		mlx4_ib_free_cq_buf(dev, &cq->buf, cq->ibcq.cqe);
 
 err_db:
 	if (!context)
-		mlx4_ib_db_free(dev, &cq->db);
+		mlx4_db_free(dev->dev, &cq->db);
 
 err_cq:
 	kfree(cq);
@@ -204,6 +266,197 @@ err_cq:
 	return ERR_PTR(err);
 }
 
+static int mlx4_alloc_resize_buf(struct mlx4_ib_dev *dev, struct mlx4_ib_cq *cq,
+				  int entries)
+{
+	int err;
+
+	if (cq->resize_buf)
+		return -EBUSY;
+
+	cq->resize_buf = kmalloc(sizeof *cq->resize_buf, GFP_ATOMIC);
+	if (!cq->resize_buf)
+		return -ENOMEM;
+
+	err = mlx4_ib_alloc_cq_buf(dev, &cq->resize_buf->buf, entries);
+	if (err) {
+		kfree(cq->resize_buf);
+		cq->resize_buf = NULL;
+		return err;
+	}
+
+	cq->resize_buf->cqe = entries - 1;
+
+	return 0;
+}
+
+static int mlx4_alloc_resize_umem(struct mlx4_ib_dev *dev, struct mlx4_ib_cq *cq,
+				   int entries, struct ib_udata *udata)
+{
+	struct mlx4_ib_resize_cq ucmd;
+	int err;
+
+	if (cq->resize_umem)
+		return -EBUSY;
+
+	if (ib_copy_from_udata(&ucmd, udata, sizeof ucmd))
+		return -EFAULT;
+
+	cq->resize_buf = kmalloc(sizeof *cq->resize_buf, GFP_ATOMIC);
+	if (!cq->resize_buf)
+		return -ENOMEM;
+
+	err = mlx4_ib_get_cq_umem(dev, cq->umem->context, &cq->resize_buf->buf,
+				  &cq->resize_umem, ucmd.buf_addr, entries);
+	if (err) {
+		kfree(cq->resize_buf);
+		cq->resize_buf = NULL;
+		return err;
+	}
+
+	cq->resize_buf->cqe = entries - 1;
+
+	return 0;
+}
+
+static int mlx4_ib_get_outstanding_cqes(struct mlx4_ib_cq *cq)
+{
+	u32 i;
+
+	i = cq->mcq.cons_index;
+	while (get_sw_cqe(cq, i))
+		++i;
+
+	return i - cq->mcq.cons_index;
+}
+
+static void mlx4_ib_cq_resize_copy_cqes(struct mlx4_ib_cq *cq)
+{
+	struct mlx4_cqe *cqe, *new_cqe;
+	int i;
+	int cqe_size = cq->buf.entry_size;
+	int cqe_inc = cqe_size == 64 ? 1 : 0;
+
+	i = cq->mcq.cons_index;
+	cqe = get_cqe(cq, i & cq->ibcq.cqe);
+	cqe += cqe_inc;
+
+	while ((cqe->owner_sr_opcode & MLX4_CQE_OPCODE_MASK) != MLX4_CQE_OPCODE_RESIZE) {
+		new_cqe = get_cqe_from_buf(&cq->resize_buf->buf,
+					   (i + 1) & cq->resize_buf->cqe);
+		memcpy(new_cqe, get_cqe(cq, i & cq->ibcq.cqe), cqe_size);
+		new_cqe += cqe_inc;
+
+		new_cqe->owner_sr_opcode = (cqe->owner_sr_opcode & ~MLX4_CQE_OWNER_MASK) |
+			(((i + 1) & (cq->resize_buf->cqe + 1)) ? MLX4_CQE_OWNER_MASK : 0);
+		cqe = get_cqe(cq, ++i & cq->ibcq.cqe);
+		cqe += cqe_inc;
+	}
+	++cq->mcq.cons_index;
+}
+
+int mlx4_ib_resize_cq(struct ib_cq *ibcq, int entries, struct ib_udata *udata)
+{
+	struct mlx4_ib_dev *dev = to_mdev(ibcq->device);
+	struct mlx4_ib_cq *cq = to_mcq(ibcq);
+	struct mlx4_mtt mtt;
+	int outst_cqe;
+	int err;
+
+	mutex_lock(&cq->resize_mutex);
+
+	if (entries < 1) {
+		err = -EINVAL;
+		goto out;
+	}
+
+	entries = roundup_pow_of_two(entries + 1);
+	if (entries == ibcq->cqe + 1) {
+		err = 0;
+		goto out;
+	}
+
+	if (entries > dev->dev->caps.max_cqes) {
+		err = -EINVAL;
+		goto out;
+	}
+
+	if (ibcq->uobject) {
+		err = mlx4_alloc_resize_umem(dev, cq, entries, udata);
+		if (err)
+			goto out;
+	} else {
+		/* Can't be smaller than the number of outstanding CQEs */
+		outst_cqe = mlx4_ib_get_outstanding_cqes(cq);
+		if (entries < outst_cqe + 1) {
+			err = 0;
+			goto out;
+		}
+
+		err = mlx4_alloc_resize_buf(dev, cq, entries);
+		if (err)
+			goto out;
+	}
+
+	mtt = cq->buf.mtt;
+
+	err = mlx4_cq_resize(dev->dev, &cq->mcq, entries, &cq->resize_buf->buf.mtt);
+	if (err)
+		goto err_buf;
+
+	mlx4_mtt_cleanup(dev->dev, &mtt);
+	if (ibcq->uobject) {
+		cq->buf      = cq->resize_buf->buf;
+		cq->ibcq.cqe = cq->resize_buf->cqe;
+		ib_umem_release(cq->umem);
+		cq->umem     = cq->resize_umem;
+
+		kfree(cq->resize_buf);
+		cq->resize_buf = NULL;
+		cq->resize_umem = NULL;
+	} else {
+		struct mlx4_ib_cq_buf tmp_buf;
+		int tmp_cqe = 0;
+
+		spin_lock_irq(&cq->lock);
+		if (cq->resize_buf) {
+			mlx4_ib_cq_resize_copy_cqes(cq);
+			tmp_buf = cq->buf;
+			tmp_cqe = cq->ibcq.cqe;
+			cq->buf      = cq->resize_buf->buf;
+			cq->ibcq.cqe = cq->resize_buf->cqe;
+
+			kfree(cq->resize_buf);
+			cq->resize_buf = NULL;
+		}
+		spin_unlock_irq(&cq->lock);
+
+		if (tmp_cqe)
+			mlx4_ib_free_cq_buf(dev, &tmp_buf, tmp_cqe);
+	}
+
+	goto out;
+
+err_buf:
+	mlx4_mtt_cleanup(dev->dev, &cq->resize_buf->buf.mtt);
+	if (!ibcq->uobject)
+		mlx4_ib_free_cq_buf(dev, &cq->resize_buf->buf,
+				    cq->resize_buf->cqe);
+
+	kfree(cq->resize_buf);
+	cq->resize_buf = NULL;
+
+	if (cq->resize_umem) {
+		ib_umem_release(cq->resize_umem);
+		cq->resize_umem = NULL;
+	}
+
+out:
+	mutex_unlock(&cq->resize_mutex);
+
+	return err;
+}
+
 int mlx4_ib_destroy_cq(struct ib_cq *cq)
 {
 	struct mlx4_ib_dev *dev = to_mdev(cq->device);
@@ -216,9 +469,8 @@ int mlx4_ib_destroy_cq(struct ib_cq *cq)
 		mlx4_ib_db_unmap_user(to_mucontext(cq->uobject->context), &mcq->db);
 		ib_umem_release(mcq->umem);
 	} else {
-		mlx4_buf_free(dev->dev, (cq->cqe + 1) * sizeof (struct mlx4_cqe),
-			      &mcq->buf.buf);
-		mlx4_ib_db_free(dev, &mcq->db);
+		mlx4_ib_free_cq_buf(dev, &mcq->buf, cq->cqe);
+		mlx4_db_free(dev->dev, &mcq->db);
 	}
 
 	kfree(mcq);
@@ -230,7 +482,7 @@ static void dump_cqe(void *cqe)
 {
 	__be32 *buf = cqe;
 
-	printk(KERN_DEBUG "CQE contents %08x %08x %08x %08x %08x %08x %08x %08x\n",
+	pr_debug("CQE contents %08x %08x %08x %08x %08x %08x %08x %08x\n",
 	       be32_to_cpu(buf[0]), be32_to_cpu(buf[1]), be32_to_cpu(buf[2]),
 	       be32_to_cpu(buf[3]), be32_to_cpu(buf[4]), be32_to_cpu(buf[5]),
 	       be32_to_cpu(buf[6]), be32_to_cpu(buf[7]));
@@ -240,7 +492,7 @@ static void mlx4_ib_handle_error_cqe(struct mlx4_err_cqe *cqe,
 				     struct ib_wc *wc)
 {
 	if (cqe->syndrome == MLX4_CQE_SYNDROME_LOCAL_QP_OP_ERR) {
-		printk(KERN_DEBUG "local QP operation err "
+		pr_debug("local QP operation err "
 		       "(QPN %06x, WQE index %x, vendor syndrome %02x, "
 		       "opcode = %02x)\n",
 		       be32_to_cpu(cqe->my_qpn), be16_to_cpu(cqe->wqe_index),
@@ -297,6 +549,48 @@ static void mlx4_ib_handle_error_cqe(struct mlx4_err_cqe *cqe,
 	wc->vendor_err = cqe->vendor_err_syndrome;
 }
 
+static int mlx4_ib_ipoib_csum_ok(__be16 status, __be16 checksum)
+{
+	return ((status & cpu_to_be16(MLX4_CQE_STATUS_IPV4      |
+				      MLX4_CQE_STATUS_IPV4F     |
+				      MLX4_CQE_STATUS_IPV4OPT   |
+				      MLX4_CQE_STATUS_IPV6      |
+				      MLX4_CQE_STATUS_IPOK)) ==
+		cpu_to_be16(MLX4_CQE_STATUS_IPV4        |
+			    MLX4_CQE_STATUS_IPOK))              &&
+		(status & cpu_to_be16(MLX4_CQE_STATUS_UDP       |
+				      MLX4_CQE_STATUS_TCP))     &&
+		checksum == cpu_to_be16(0xffff);
+}
+
+static int use_tunnel_data(struct mlx4_ib_qp *qp, struct mlx4_ib_cq *cq, struct ib_wc *wc,
+			   unsigned tail, struct mlx4_cqe *cqe, int is_eth)
+{
+	struct mlx4_ib_proxy_sqp_hdr *hdr;
+
+	ib_dma_sync_single_for_cpu(qp->ibqp.device,
+				   qp->sqp_proxy_rcv[tail].map,
+				   sizeof (struct mlx4_ib_proxy_sqp_hdr),
+				   DMA_FROM_DEVICE);
+	hdr = (struct mlx4_ib_proxy_sqp_hdr *) (qp->sqp_proxy_rcv[tail].addr);
+	wc->pkey_index	= be16_to_cpu(hdr->tun.pkey_index);
+	wc->src_qp	= be32_to_cpu(hdr->tun.flags_src_qp) & 0xFFFFFF;
+	wc->wc_flags   |= (hdr->tun.g_ml_path & 0x80) ? (IB_WC_GRH) : 0;
+	wc->dlid_path_bits = 0;
+
+	if (is_eth) {
+		wc->vlan_id = be16_to_cpu(hdr->tun.sl_vid);
+		memcpy(&(wc->smac[0]), (char *)&hdr->tun.mac_31_0, 4);
+		memcpy(&(wc->smac[4]), (char *)&hdr->tun.slid_mac_47_32, 2);
+		wc->wc_flags |= (IB_WC_WITH_VLAN | IB_WC_WITH_SMAC);
+	} else {
+		wc->slid        = be16_to_cpu(hdr->tun.slid_mac_47_32);
+		wc->sl          = (u8) (be16_to_cpu(hdr->tun.sl_vid) >> 12);
+	}
+
+	return 0;
+}
+
 static int mlx4_ib_poll_one(struct mlx4_ib_cq *cq,
 			    struct mlx4_ib_qp **cur_qp,
 			    struct ib_wc *wc)
@@ -305,15 +599,22 @@ static int mlx4_ib_poll_one(struct mlx4_ib_cq *cq,
 	struct mlx4_qp *mqp;
 	struct mlx4_ib_wq *wq;
 	struct mlx4_ib_srq *srq;
+	struct mlx4_srq *msrq = NULL;
 	int is_send;
 	int is_error;
+	int is_eth;
 	u32 g_mlpath_rqpn;
 	u16 wqe_ctr;
+	unsigned tail = 0;
 
+repoll:
 	cqe = next_cqe_sw(cq);
 	if (!cqe)
 		return -EAGAIN;
 
+	if (cq->buf.entry_size == 64)
+		cqe++;
+
 	++cq->mcq.cons_index;
 
 	/*
@@ -328,22 +629,38 @@ static int mlx4_ib_poll_one(struct mlx4_ib_cq *cq,
 
 	if (unlikely((cqe->owner_sr_opcode & MLX4_CQE_OPCODE_MASK) == MLX4_OPCODE_NOP &&
 		     is_send)) {
-		printk(KERN_WARNING "Completion for NOP opcode detected!\n");
+		pr_warn("Completion for NOP opcode detected!\n");
 		return -EINVAL;
 	}
 
+	/* Resize CQ in progress */
+	if (unlikely((cqe->owner_sr_opcode & MLX4_CQE_OPCODE_MASK) == MLX4_CQE_OPCODE_RESIZE)) {
+		if (cq->resize_buf) {
+			struct mlx4_ib_dev *dev = to_mdev(cq->ibcq.device);
+
+			mlx4_ib_free_cq_buf(dev, &cq->buf, cq->ibcq.cqe);
+			cq->buf      = cq->resize_buf->buf;
+			cq->ibcq.cqe = cq->resize_buf->cqe;
+
+			kfree(cq->resize_buf);
+			cq->resize_buf = NULL;
+		}
+
+		goto repoll;
+	}
+
 	if (!*cur_qp ||
-	    (be32_to_cpu(cqe->my_qpn) & 0xffffff) != (*cur_qp)->mqp.qpn) {
+	    (be32_to_cpu(cqe->vlan_my_qpn) & MLX4_CQE_QPN_MASK) != (*cur_qp)->mqp.qpn) {
 		/*
 		 * We do not have to take the QP table lock here,
 		 * because CQs will be locked while QPs are removed
 		 * from the table.
 		 */
 		mqp = __mlx4_qp_lookup(to_mdev(cq->ibcq.device)->dev,
-				       be32_to_cpu(cqe->my_qpn));
+				       be32_to_cpu(cqe->vlan_my_qpn));
 		if (unlikely(!mqp)) {
-			printk(KERN_WARNING "CQ %06x with entry for unknown QPN %06x\n",
-			       cq->mcq.cqn, be32_to_cpu(cqe->my_qpn) & 0xffffff);
+			pr_warn("CQ %06x with entry for unknown QPN %06x\n",
+			       cq->mcq.cqn, be32_to_cpu(cqe->vlan_my_qpn) & MLX4_CQE_QPN_MASK);
 			return -EINVAL;
 		}
 
@@ -352,6 +669,20 @@ static int mlx4_ib_poll_one(struct mlx4_ib_cq *cq,
 
 	wc->qp = &(*cur_qp)->ibqp;
 
+	if (wc->qp->qp_type == IB_QPT_XRC_TGT) {
+		u32 srq_num;
+		g_mlpath_rqpn = be32_to_cpu(cqe->g_mlpath_rqpn);
+		srq_num       = g_mlpath_rqpn & 0xffffff;
+		/* SRQ is also in the radix tree */
+		msrq = mlx4_srq_lookup(to_mdev(cq->ibcq.device)->dev,
+				       srq_num);
+		if (unlikely(!msrq)) {
+			pr_warn("CQ %06x with entry for unknown SRQN %06x\n",
+				cq->mcq.cqn, srq_num);
+			return -EINVAL;
+		}
+	}
+
 	if (is_send) {
 		wq = &(*cur_qp)->sq;
 		if (!(*cur_qp)->sq_signal_bits) {
@@ -365,9 +696,15 @@ static int mlx4_ib_poll_one(struct mlx4_ib_cq *cq,
 		wqe_ctr = be16_to_cpu(cqe->wqe_index);
 		wc->wr_id = srq->wrid[wqe_ctr];
 		mlx4_ib_free_srq_wqe(srq, wqe_ctr);
+	} else if (msrq) {
+		srq = to_mibsrq(msrq);
+		wqe_ctr = be16_to_cpu(cqe->wqe_index);
+		wc->wr_id = srq->wrid[wqe_ctr];
+		mlx4_ib_free_srq_wqe(srq, wqe_ctr);
 	} else {
 		wq	  = &(*cur_qp)->rq;
-		wc->wr_id = wq->wrid[wq->tail & (wq->wqe_cnt - 1)];
+		tail	  = wq->tail & (wq->wqe_cnt - 1);
+		wc->wr_id = wq->wrid[tail];
 		++wq->tail;
 	}
 
@@ -389,6 +726,7 @@ static int mlx4_ib_poll_one(struct mlx4_ib_cq *cq,
 		case MLX4_OPCODE_SEND_IMM:
 			wc->wc_flags |= IB_WC_WITH_IMM;
 		case MLX4_OPCODE_SEND:
+		case MLX4_OPCODE_SEND_INVAL:
 			wc->opcode    = IB_WC_SEND;
 			break;
 		case MLX4_OPCODE_RDMA_READ:
@@ -403,37 +741,86 @@ static int mlx4_ib_poll_one(struct mlx4_ib_cq *cq,
 			wc->opcode    = IB_WC_FETCH_ADD;
 			wc->byte_len  = 8;
 			break;
+		case MLX4_OPCODE_MASKED_ATOMIC_CS:
+			wc->opcode    = IB_WC_MASKED_COMP_SWAP;
+			wc->byte_len  = 8;
+			break;
+		case MLX4_OPCODE_MASKED_ATOMIC_FA:
+			wc->opcode    = IB_WC_MASKED_FETCH_ADD;
+			wc->byte_len  = 8;
+			break;
 		case MLX4_OPCODE_BIND_MW:
 			wc->opcode    = IB_WC_BIND_MW;
 			break;
+		case MLX4_OPCODE_LSO:
+			wc->opcode    = IB_WC_LSO;
+			break;
+		case MLX4_OPCODE_FMR:
+			wc->opcode    = IB_WC_FAST_REG_MR;
+			break;
+		case MLX4_OPCODE_LOCAL_INVAL:
+			wc->opcode    = IB_WC_LOCAL_INV;
+			break;
 		}
 	} else {
 		wc->byte_len = be32_to_cpu(cqe->byte_cnt);
 
 		switch (cqe->owner_sr_opcode & MLX4_CQE_OPCODE_MASK) {
 		case MLX4_RECV_OPCODE_RDMA_WRITE_IMM:
-			wc->opcode   = IB_WC_RECV_RDMA_WITH_IMM;
-			wc->wc_flags = IB_WC_WITH_IMM;
-			wc->imm_data = cqe->immed_rss_invalid;
+			wc->opcode	= IB_WC_RECV_RDMA_WITH_IMM;
+			wc->wc_flags	= IB_WC_WITH_IMM;
+			wc->ex.imm_data = cqe->immed_rss_invalid;
+			break;
+		case MLX4_RECV_OPCODE_SEND_INVAL:
+			wc->opcode	= IB_WC_RECV;
+			wc->wc_flags	= IB_WC_WITH_INVALIDATE;
+			wc->ex.invalidate_rkey = be32_to_cpu(cqe->immed_rss_invalid);
 			break;
 		case MLX4_RECV_OPCODE_SEND:
 			wc->opcode   = IB_WC_RECV;
 			wc->wc_flags = 0;
 			break;
 		case MLX4_RECV_OPCODE_SEND_IMM:
-			wc->opcode   = IB_WC_RECV;
-			wc->wc_flags = IB_WC_WITH_IMM;
-			wc->imm_data = cqe->immed_rss_invalid;
+			wc->opcode	= IB_WC_RECV;
+			wc->wc_flags	= IB_WC_WITH_IMM;
+			wc->ex.imm_data = cqe->immed_rss_invalid;
 			break;
 		}
 
+		is_eth = (rdma_port_get_link_layer(wc->qp->device,
+						  (*cur_qp)->port) ==
+			  IB_LINK_LAYER_ETHERNET);
+		if (mlx4_is_mfunc(to_mdev(cq->ibcq.device)->dev)) {
+			if ((*cur_qp)->mlx4_ib_qp_type &
+			    (MLX4_IB_QPT_PROXY_SMI_OWNER |
+			     MLX4_IB_QPT_PROXY_SMI | MLX4_IB_QPT_PROXY_GSI))
+				return use_tunnel_data(*cur_qp, cq, wc, tail,
+						       cqe, is_eth);
+		}
+
 		wc->slid	   = be16_to_cpu(cqe->rlid);
-		wc->sl		   = cqe->sl >> 4;
 		g_mlpath_rqpn	   = be32_to_cpu(cqe->g_mlpath_rqpn);
 		wc->src_qp	   = g_mlpath_rqpn & 0xffffff;
 		wc->dlid_path_bits = (g_mlpath_rqpn >> 24) & 0x7f;
 		wc->wc_flags	  |= g_mlpath_rqpn & 0x80000000 ? IB_WC_GRH : 0;
 		wc->pkey_index     = be32_to_cpu(cqe->immed_rss_invalid) & 0x7f;
+		wc->wc_flags	  |= mlx4_ib_ipoib_csum_ok(cqe->status,
+					cqe->checksum) ? IB_WC_IP_CSUM_OK : 0;
+		if (is_eth) {
+			wc->sl  = be16_to_cpu(cqe->sl_vid) >> 13;
+			if (be32_to_cpu(cqe->vlan_my_qpn) &
+					MLX4_CQE_VLAN_PRESENT_MASK) {
+				wc->vlan_id = be16_to_cpu(cqe->sl_vid) &
+					MLX4_CQE_VID_MASK;
+			} else {
+				wc->vlan_id = 0xffff;
+			}
+			memcpy(wc->smac, cqe->smac, ETH_ALEN);
+			wc->wc_flags |= (IB_WC_WITH_VLAN | IB_WC_WITH_SMAC);
+		} else {
+			wc->sl  = be16_to_cpu(cqe->sl_vid) >> 12;
+			wc->vlan_id = 0xffff;
+		}
 	}
 
 	return 0;
@@ -455,8 +842,7 @@ int mlx4_ib_poll_cq(struct ib_cq *ibcq, int num_entries, struct ib_wc *wc)
 			break;
 	}
 
-	if (npolled)
-		mlx4_cq_set_ci(&cq->mcq);
+	mlx4_cq_set_ci(&cq->mcq);
 
 	spin_unlock_irqrestore(&cq->lock, flags);
 
@@ -483,6 +869,7 @@ void __mlx4_ib_cq_clean(struct mlx4_ib_cq *cq, u32 qpn, struct mlx4_ib_srq *srq)
 	int nfreed = 0;
 	struct mlx4_cqe *cqe, *dest;
 	u8 owner_bit;
+	int cqe_inc = cq->buf.entry_size == 64 ? 1 : 0;
 
 	/*
 	 * First we need to find the current producer index, so we
@@ -501,12 +888,16 @@ void __mlx4_ib_cq_clean(struct mlx4_ib_cq *cq, u32 qpn, struct mlx4_ib_srq *srq)
 	 */
 	while ((int) --prod_index - (int) cq->mcq.cons_index >= 0) {
 		cqe = get_cqe(cq, prod_index & cq->ibcq.cqe);
-		if ((be32_to_cpu(cqe->my_qpn) & 0xffffff) == qpn) {
+		cqe += cqe_inc;
+
+		if ((be32_to_cpu(cqe->vlan_my_qpn) & MLX4_CQE_QPN_MASK) == qpn) {
 			if (srq && !(cqe->owner_sr_opcode & MLX4_CQE_IS_SEND_MASK))
 				mlx4_ib_free_srq_wqe(srq, be16_to_cpu(cqe->wqe_index));
 			++nfreed;
 		} else if (nfreed) {
 			dest = get_cqe(cq, (prod_index + nfreed) & cq->ibcq.cqe);
+			dest += cqe_inc;
+
 			owner_bit = dest->owner_sr_opcode & MLX4_CQE_OWNER_MASK;
 			memcpy(dest, cqe, sizeof *cqe);
 			dest->owner_sr_opcode = owner_bit |