diff mbox

[for-next,7/8] mlx5: Mellanox Connect-IB driver part 7/8

Message ID 1371384149-24558-8-git-send-email-eli@dev.mellanox.co.il (mailing list archive)
State Superseded, archived
Delegated to: Roland Dreier
Headers show

Commit Message

Eli Cohen June 16, 2013, 12:02 p.m. UTC
From: Eli Cohen <eli@mellanox.com>

Signed-off-by: Eli Cohen <eli@mellanox.com>
---
 drivers/infiniband/hw/mlx5/qp.c   | 2549 +++++++++++++++++++++++++++++++++++++
 drivers/infiniband/hw/mlx5/srq.c  |  481 +++++++
 drivers/infiniband/hw/mlx5/user.h |  123 ++
 3 files changed, 3153 insertions(+)
 create mode 100644 drivers/infiniband/hw/mlx5/qp.c
 create mode 100644 drivers/infiniband/hw/mlx5/srq.c
 create mode 100644 drivers/infiniband/hw/mlx5/user.h
diff mbox

Patch

diff --git a/drivers/infiniband/hw/mlx5/qp.c b/drivers/infiniband/hw/mlx5/qp.c
new file mode 100644
index 0000000..f894774
--- /dev/null
+++ b/drivers/infiniband/hw/mlx5/qp.c
@@ -0,0 +1,2549 @@ 
+/*
+ * Copyright (c) 2013, Mellanox Technologies 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
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/module.h>
+#include <rdma/ib_umem.h>
+#include <linux/vmalloc.h>
+#include "mlx5_ib.h"
+#include "user.h"
+
+MLX5_IB_MOD_DBG_MASK(MLX5_IB_MOD_QP);
+
+/* not supported currently */
+static int wq_signature;
+
+enum {
+	MLX5_IB_ACK_REQ_FREQ	= 8,
+};
+
+enum {
+	MLX5_IB_DEFAULT_SCHED_QUEUE	= 0x83,
+	MLX5_IB_DEFAULT_QP0_SCHED_QUEUE	= 0x3f,
+	MLX5_IB_LINK_TYPE_IB		= 0,
+	MLX5_IB_LINK_TYPE_ETH		= 1
+};
+
+enum {
+	MLX5_IB_SQ_STRIDE	= 6,
+	MLX5_IB_CACHE_LINE_SIZE	= 64,
+};
+
+static const u32 mlx5_ib_opcode[] = {
+	[IB_WR_SEND]				= MLX5_OPCODE_SEND,
+	[IB_WR_SEND_WITH_IMM]			= MLX5_OPCODE_SEND_IMM,
+	[IB_WR_RDMA_WRITE]			= MLX5_OPCODE_RDMA_WRITE,
+	[IB_WR_RDMA_WRITE_WITH_IMM]		= MLX5_OPCODE_RDMA_WRITE_IMM,
+	[IB_WR_RDMA_READ]			= MLX5_OPCODE_RDMA_READ,
+	[IB_WR_ATOMIC_CMP_AND_SWP]		= MLX5_OPCODE_ATOMIC_CS,
+	[IB_WR_ATOMIC_FETCH_AND_ADD]		= MLX5_OPCODE_ATOMIC_FA,
+	[IB_WR_SEND_WITH_INV]			= MLX5_OPCODE_SEND_INVAL,
+	[IB_WR_LOCAL_INV]			= MLX5_OPCODE_UMR,
+	[IB_WR_FAST_REG_MR]			= MLX5_OPCODE_UMR,
+	[IB_WR_MASKED_ATOMIC_CMP_AND_SWP]	= MLX5_OPCODE_ATOMIC_MASKED_CS,
+	[IB_WR_MASKED_ATOMIC_FETCH_AND_ADD]	= MLX5_OPCODE_ATOMIC_MASKED_FA,
+	[IB_WR_SET_PSV]				= MLX5_OPCODE_SET_PSV,
+	[IB_WR_GET_PSV]				= MLX5_OPCODE_GET_PSV,
+	[IB_WR_CHECK_PSV]			= MLX5_OPCODE_CHECK_PSV,
+	[IB_WR_RGET_PSV]			= MLX5_OPCODE_RGET_PSV,
+	[IB_WR_RCHECK_PSV]			= MLX5_OPCODE_RCHECK_PSV,
+	[IB_WR_UMR]				= MLX5_OPCODE_UMR,
+};
+
+struct umr_wr {
+	u64				virt_addr;
+	struct ib_pd		       *pd;
+	unsigned int			page_shift;
+	unsigned int			npages;
+	u32				length;
+	int				access_flags;
+	u32				mkey;
+};
+
+static int is_qp0(enum ib_qp_type qp_type)
+{
+	return qp_type == IB_QPT_SMI;
+}
+
+static int is_qp1(enum ib_qp_type qp_type)
+{
+	return qp_type == IB_QPT_GSI;
+}
+
+static int is_sqp(enum ib_qp_type qp_type)
+{
+	return is_qp0(qp_type) || is_qp1(qp_type);
+}
+
+static void *get_wqe(struct mlx5_ib_qp *qp, int offset)
+{
+	return mlx5_buf_offset(&qp->buf, offset);
+}
+
+static void *get_recv_wqe(struct mlx5_ib_qp *qp, int n)
+{
+	return get_wqe(qp, qp->rq.offset + (n << qp->rq.wqe_shift));
+}
+
+void *mlx5_get_send_wqe(struct mlx5_ib_qp *qp, int n)
+{
+	return get_wqe(qp, qp->sq.offset + (n << MLX5_IB_SQ_STRIDE));
+}
+
+static void mlx5_ib_qp_event(struct mlx5_core_qp *qp, int type)
+{
+	struct ib_event event;
+	struct ib_qp *ibqp = &to_mibqp(qp)->ibqp;
+
+	if (type == MLX5_EVENT_TYPE_PATH_MIG)
+		to_mibqp(qp)->port = to_mibqp(qp)->alt_port;
+
+	if (ibqp->event_handler) {
+		event.device     = ibqp->device;
+		event.element.qp = ibqp;
+		switch (type) {
+		case MLX5_EVENT_TYPE_PATH_MIG:
+			event.event = IB_EVENT_PATH_MIG;
+			break;
+		case MLX5_EVENT_TYPE_COMM_EST:
+			event.event = IB_EVENT_COMM_EST;
+			break;
+		case MLX5_EVENT_TYPE_SQ_DRAINED:
+			event.event = IB_EVENT_SQ_DRAINED;
+			break;
+		case MLX5_EVENT_TYPE_SRQ_LAST_WQE:
+			event.event = IB_EVENT_QP_LAST_WQE_REACHED;
+			break;
+		case MLX5_EVENT_TYPE_WQ_CATAS_ERROR:
+			event.event = IB_EVENT_QP_FATAL;
+			break;
+		case MLX5_EVENT_TYPE_PATH_MIG_FAILED:
+			event.event = IB_EVENT_PATH_MIG_ERR;
+			break;
+		case MLX5_EVENT_TYPE_WQ_INVAL_REQ_ERROR:
+			event.event = IB_EVENT_QP_REQ_ERR;
+			break;
+		case MLX5_EVENT_TYPE_WQ_ACCESS_ERROR:
+			event.event = IB_EVENT_QP_ACCESS_ERR;
+			break;
+		default:
+			pr_warn("mlx5_ib: Unexpected event type %d on QP %06x\n", type, qp->qpn);
+			return;
+		}
+
+		ibqp->event_handler(&event, ibqp->qp_context);
+	}
+}
+
+static int set_rq_size(struct mlx5_ib_dev *dev, struct ib_qp_cap *cap,
+		       int has_rq, struct mlx5_ib_qp *qp, struct mlx5_ib_create_qp *ucmd)
+{
+	int wqe_size;
+	int wq_size;
+
+	/* Sanity check RQ size before proceeding */
+	if (cap->max_recv_wr  > dev->mdev.caps.max_wqes)
+		return -EINVAL;
+
+	if (!has_rq) {
+		qp->rq.max_gs = 0;
+		qp->rq.wqe_cnt = 0;
+		qp->rq.wqe_shift = 0;
+	} else {
+		if (ucmd) {
+			qp->rq.wqe_cnt = ucmd->rq_wqe_count;
+			qp->rq.wqe_shift = ucmd->rq_wqe_shift;
+			qp->rq.max_gs = (1 << qp->rq.wqe_shift) / sizeof(struct mlx5_wqe_data_seg) - qp->wq_sig;
+			qp->rq.max_post = qp->rq.wqe_cnt;
+		} else {
+			wqe_size = qp->wq_sig ? sizeof(struct mlx5_wqe_signature_seg) : 0;
+			wqe_size += cap->max_recv_sge * sizeof(struct mlx5_wqe_data_seg);
+			wqe_size = roundup_pow_of_two(wqe_size);
+			wq_size = roundup_pow_of_two(cap->max_recv_wr) * wqe_size;
+			wq_size = max_t(int, wq_size, MLX5_SEND_WQE_BB);
+			qp->rq.wqe_cnt = wq_size / wqe_size;
+			if (wqe_size > dev->mdev.caps.max_rq_desc_sz) {
+				mlx5_ib_dbg(dev, "wqe_size %d, max %d\n",
+					    wqe_size,
+					    dev->mdev.caps.max_rq_desc_sz);
+				return -EINVAL;
+			}
+			qp->rq.wqe_shift = ilog2(wqe_size);
+			qp->rq.max_gs = (1 << qp->rq.wqe_shift) / sizeof(struct mlx5_wqe_data_seg) - qp->wq_sig;
+			qp->rq.max_post = qp->rq.wqe_cnt;
+		}
+	}
+
+	return 0;
+}
+
+static int sq_overhead(enum ib_qp_type qp_type)
+{
+	int size;
+
+	switch (qp_type) {
+	case IB_QPT_XRC_INI:
+		size = sizeof(struct mlx5_wqe_xrc_seg);
+		/* fall through */
+	case IB_QPT_RC:
+		size += sizeof(struct mlx5_wqe_ctrl_seg) +
+			sizeof(struct mlx5_wqe_atomic_seg) +
+			sizeof(struct mlx5_wqe_raddr_seg);
+		break;
+
+	case IB_QPT_UC:
+		size = sizeof(struct mlx5_wqe_ctrl_seg) +
+			sizeof(struct mlx5_wqe_raddr_seg);
+		break;
+
+	case IB_QPT_UD:
+	case IB_QPT_SMI:
+	case IB_QPT_GSI:
+		size = sizeof(struct mlx5_wqe_ctrl_seg) +
+			sizeof(struct mlx5_wqe_datagram_seg);
+		break;
+
+	case IB_QPT_REG_UMR:
+		size = sizeof(struct mlx5_wqe_ctrl_seg) +
+			sizeof(struct mlx5_wqe_umr_ctrl_seg) +
+			sizeof(struct mlx5_mkey_seg);
+		break;
+
+	default:
+		return -EINVAL;
+	}
+
+	return size;
+}
+
+static int calc_send_wqe(struct ib_qp_init_attr *attr)
+{
+	int size;
+	int inl_size = 0;
+
+	size = sq_overhead(attr->qp_type);
+	if (size < 0)
+		return size;
+
+	if (attr->cap.max_inline_data) {
+		inl_size = size + sizeof(struct mlx5_wqe_inline_seg) +
+			attr->cap.max_inline_data;
+	}
+
+	size += attr->cap.max_send_sge * sizeof(struct mlx5_wqe_data_seg);
+
+	return ALIGN(max_t(int, inl_size, size), MLX5_SEND_WQE_BB);
+}
+
+static int calc_sq_size(struct mlx5_ib_dev *dev, struct ib_qp_init_attr *attr,
+			struct mlx5_ib_qp *qp)
+{
+	int wqe_size;
+	int wq_size;
+
+	if (!attr->cap.max_send_wr)
+		return 0;
+
+	wqe_size = calc_send_wqe(attr);
+	mlx5_ib_dbg(dev, "wqe_size %d\n", wqe_size);
+	if (wqe_size < 0)
+		return wqe_size;
+
+	if (wqe_size > dev->mdev.caps.max_sq_desc_sz) {
+		mlx5_ib_dbg(dev, "\n");
+		return -EINVAL;
+	}
+
+	qp->max_inline_data = wqe_size - sq_overhead(attr->qp_type) -
+		sizeof(struct mlx5_wqe_inline_seg);
+	attr->cap.max_inline_data = qp->max_inline_data;
+
+	wq_size = roundup_pow_of_two(attr->cap.max_send_wr * wqe_size);
+	qp->sq.wqe_cnt = wq_size / MLX5_SEND_WQE_BB;
+	qp->sq.wqe_shift = ilog2(MLX5_SEND_WQE_BB);
+	qp->sq.max_gs = attr->cap.max_send_sge;
+	qp->sq.max_post = 1 << ilog2(wq_size / wqe_size);
+
+	return wq_size;
+}
+
+static int set_user_buf_size(struct mlx5_ib_dev *dev,
+			    struct mlx5_ib_qp *qp,
+			    struct mlx5_ib_create_qp *ucmd)
+{
+	int desc_sz = 1 << qp->sq.wqe_shift;
+
+	if (desc_sz > dev->mdev.caps.max_sq_desc_sz) {
+		mlx5_ib_warn(dev, "desc_sz %d, max_sq_desc_sz %d\n",
+			     desc_sz, dev->mdev.caps.max_sq_desc_sz);
+		return -EINVAL;
+	}
+
+	if (ucmd->sq_wqe_count && ((1 << ilog2(ucmd->sq_wqe_count)) != ucmd->sq_wqe_count)) {
+		mlx5_ib_warn(dev, "sq_wqe_count %d, sq_wqe_count %d\n",
+			     ucmd->sq_wqe_count, ucmd->sq_wqe_count);
+		return -EINVAL;
+	}
+
+	qp->sq.wqe_cnt = ucmd->sq_wqe_count;
+
+	if (qp->sq.wqe_cnt > dev->mdev.caps.max_wqes) {
+		mlx5_ib_warn(dev, "wqe_cnt %d, max_wqes %d\n",
+			     qp->sq.wqe_cnt, dev->mdev.caps.max_wqes);
+		return -EINVAL;
+	}
+
+	qp->buf_size = (qp->rq.wqe_cnt << qp->rq.wqe_shift) +
+		(qp->sq.wqe_cnt << 6);
+
+	return 0;
+}
+
+static int qp_has_rq(struct ib_qp_init_attr *attr)
+{
+	if (attr->qp_type == IB_QPT_XRC_INI ||
+	    attr->qp_type == IB_QPT_XRC_TGT || attr->srq ||
+	    attr->qp_type == IB_QPT_REG_UMR ||
+	    !attr->cap.max_recv_wr)
+		return 0;
+
+	return 1;
+}
+
+static int alloc_high_class_uuar(struct mlx5_uuar_info *uuari)
+{
+	int start_uuar;
+	int i;
+	int nuuars = uuari->num_uars * MLX5_BF_REGS_PER_PAGE;
+
+	start_uuar = nuuars - uuari->num_low_latency_uuars;
+	for (i = start_uuar; i < nuuars; ++i) {
+		if (!test_bit(i, uuari->bitmap)) {
+			set_bit(i, uuari->bitmap);
+			++uuari->count[i];
+			return i;
+		}
+	}
+
+	return -ENOMEM;
+}
+
+static int alloc_med_class_uuar(struct mlx5_uuar_info *uuari)
+{
+	int end;
+	int nuuars = uuari->num_uars * MLX5_BF_REGS_PER_PAGE;
+	int i;
+	int minidx = 1;
+	int uuarn;
+
+	end = nuuars - uuari->num_low_latency_uuars;
+
+	for (i = 1; i < end; ++i) {
+		uuarn = i & 3;
+		if (uuarn == 2 || uuarn == 3)
+			continue;
+
+		if (uuari->count[i] < uuari->count[minidx])
+			minidx = i;
+	}
+
+	++uuari->count[minidx];
+	return minidx;
+}
+
+static int alloc_uuar(struct mlx5_uuar_info *uuari, enum ib_latency_class lat)
+{
+	int uuarn = -EINVAL;
+
+	mutex_lock(&uuari->lock);
+	switch (lat) {
+	case IB_LATENCY_CLASS_LOW:
+		uuarn = 0;
+		++uuari->count[uuarn];
+		break;
+
+	case IB_LATENCY_CLASS_MEDIUM:
+		uuarn = alloc_med_class_uuar(uuari);
+		break;
+
+	case IB_LATENCY_CLASS_HIGH:
+		uuarn = alloc_high_class_uuar(uuari);
+		break;
+
+	case IB_LATENCY_CLASS_FAST_PATH:
+		uuarn = 2;
+		break;
+	}
+	mutex_unlock(&uuari->lock);
+
+	return uuarn;
+}
+
+static void free_med_class_uuar(struct mlx5_uuar_info *uuari, int uuarn)
+{
+	clear_bit(uuarn, uuari->bitmap);
+	--uuari->count[uuarn];
+}
+
+static void free_high_class_uuar(struct mlx5_uuar_info *uuari, int uuarn)
+{
+	clear_bit(uuarn, uuari->bitmap);
+	--uuari->count[uuarn];
+}
+
+static void free_uuar(struct mlx5_uuar_info *uuari, int uuarn)
+{
+	int nuuars = uuari->num_uars * MLX5_BF_REGS_PER_PAGE;
+	int high_uuar = nuuars - uuari->num_low_latency_uuars;
+
+	mutex_lock(&uuari->lock);
+	if (uuarn == 0) {
+		--uuari->count[uuarn];
+		goto out;
+	}
+
+	if (uuarn < high_uuar) {
+		free_med_class_uuar(uuari, uuarn);
+		goto out;
+	}
+
+	free_high_class_uuar(uuari, uuarn);
+
+out:
+	mutex_unlock(&uuari->lock);
+}
+
+static enum mlx5_qp_state to_mlx5_state(enum ib_qp_state state)
+{
+	switch (state) {
+	case IB_QPS_RESET:	return MLX5_QP_STATE_RST;
+	case IB_QPS_INIT:	return MLX5_QP_STATE_INIT;
+	case IB_QPS_RTR:	return MLX5_QP_STATE_RTR;
+	case IB_QPS_RTS:	return MLX5_QP_STATE_RTS;
+	case IB_QPS_SQD:	return MLX5_QP_STATE_SQD;
+	case IB_QPS_SQE:	return MLX5_QP_STATE_SQER;
+	case IB_QPS_ERR:	return MLX5_QP_STATE_ERR;
+	default:		return -1;
+	}
+}
+
+static int to_mlx5_st(enum ib_qp_type type)
+{
+	switch (type) {
+	case IB_QPT_RC:			return MLX5_QP_ST_RC;
+	case IB_QPT_UC:			return MLX5_QP_ST_UC;
+	case IB_QPT_UD:			return MLX5_QP_ST_UD;
+	case IB_QPT_REG_UMR:		return MLX5_QP_ST_REG_UMR;
+	case IB_QPT_XRC_INI:
+	case IB_QPT_XRC_TGT:		return MLX5_QP_ST_XRC;
+	case IB_QPT_SMI:		return MLX5_QP_ST_QP0;
+	case IB_QPT_GSI:		return MLX5_QP_ST_QP1;
+	case IB_QPT_RAW_IPV6:		return MLX5_QP_ST_RAW_IPV6;
+	case IB_QPT_RAW_ETHERTYPE:	return MLX5_QP_ST_RAW_ETHERTYPE;
+	case IB_QPT_RAW_PACKET:
+	case IB_QPT_MAX:
+	default:		return -EINVAL;
+	}
+}
+
+static int uuarn_to_uar_index(struct mlx5_uuar_info *uuari, int uuarn)
+{
+	return uuari->uars[uuarn / MLX5_BF_REGS_PER_PAGE].index;
+}
+
+static int create_user_qp(struct mlx5_ib_dev *dev, struct ib_pd *pd,
+			  struct mlx5_ib_qp *qp, struct ib_udata *udata,
+			  struct mlx5_create_qp_mbox_in **in,
+			  struct mlx5_ib_create_qp_resp *resp, int *inlen)
+{
+	struct mlx5_ib_ucontext *context;
+	int uuarn;
+	int uar_index;
+	struct mlx5_ib_create_qp ucmd;
+	int npages;
+	int ncont;
+	int page_shift;
+	int err;
+	u32 offset;
+
+	err = ib_copy_from_udata(&ucmd, udata, sizeof(ucmd));
+	if (err) {
+		mlx5_ib_dbg(dev, "copy failed\n");
+		return err;
+	}
+
+	context = to_mucontext(pd->uobject->context);
+	/*
+	 * TBD: should come from the verbs when we have the API
+	 */
+	uuarn = alloc_uuar(&context->uuari, IB_LATENCY_CLASS_HIGH);
+	if (uuarn < 0) {
+		mlx5_ib_dbg(dev, "failed to allocate low latency UUAR\n");
+		mlx5_ib_dbg(dev, "reverting to high latency\n");
+		uuarn = alloc_uuar(&context->uuari, IB_LATENCY_CLASS_LOW);
+		if (uuarn < 0) {
+			mlx5_ib_dbg(dev, "uuar allocation failed\n");
+			return uuarn;
+		}
+	}
+
+	uar_index = uuarn_to_uar_index(&context->uuari, uuarn);
+	mlx5_ib_dbg(dev, "uuarn 0x%x, uar_index 0x%x\n", uuarn, uar_index);
+
+	err = set_user_buf_size(dev, qp, &ucmd);
+	if (err)
+		goto err_uuar;
+
+	qp->umem = ib_umem_get(pd->uobject->context, ucmd.buf_addr,
+			       qp->buf_size, 0, 0);
+	if (IS_ERR(qp->umem)) {
+		mlx5_ib_dbg(dev, "umem_get failed\n");
+		err = PTR_ERR(qp->umem);
+		goto err_uuar;
+	}
+
+	mlx5_ib_cont_pages(qp->umem, ucmd.buf_addr, &npages, &page_shift,
+			   &ncont, NULL);
+	err = mlx5_ib_get_buf_offset(ucmd.buf_addr, page_shift, &offset);
+	if (err) {
+		mlx5_ib_warn(dev, "bad offset\n");
+		goto err_umem;
+	}
+	mlx5_ib_dbg(dev, "addr 0x%llx, size %d, npages %d, page_shift %d, ncont %d, offset %d\n",
+		    ucmd.buf_addr, qp->buf_size, npages, page_shift, ncont, offset);
+
+	*inlen = sizeof **in + sizeof *(*in)->pas * ncont;
+	*in = vzalloc(*inlen);
+	if (!*in) {
+		mlx5_ib_warn(dev, "vzalloc failed\n");
+		err = -ENOMEM;
+		goto err_umem;
+	}
+	mlx5_ib_populate_pas(dev, qp->umem, page_shift, (*in)->pas, 0);
+	(*in)->ctx.log_pg_sz_remote_qpn =
+		cpu_to_be32((page_shift - PAGE_SHIFT) << 24);
+	(*in)->ctx.params2 = cpu_to_be32(offset << 6);
+
+	(*in)->ctx.qp_counter_set_usr_page = cpu_to_be32(uar_index);
+	resp->uuar_index = uuarn;
+	qp->uuarn = uuarn;
+
+	err = mlx5_ib_db_map_user(context, ucmd.db_addr, &qp->db);
+	if (err) {
+		mlx5_ib_dbg(dev, "map failed\n");
+		goto err_free;
+	}
+
+	err = ib_copy_to_udata(udata, resp, sizeof(*resp));
+	if (err) {
+		mlx5_ib_dbg(dev, "copy failed\n");
+		goto err_unmap;
+	}
+	qp->create_type = MLX5_QP_USER;
+
+	return 0;
+
+err_unmap:
+	mlx5_ib_db_unmap_user(context, &qp->db);
+
+err_free:
+	vfree(*in);
+
+err_umem:
+	ib_umem_release(qp->umem);
+
+err_uuar:
+	free_uuar(&context->uuari, uuarn);
+	return err;
+}
+
+static void destroy_qp_user(struct ib_pd *pd, struct mlx5_ib_qp *qp)
+{
+	struct mlx5_ib_ucontext *context;
+
+	context = to_mucontext(pd->uobject->context);
+	mlx5_ib_db_unmap_user(context, &qp->db);
+	ib_umem_release(qp->umem);
+	free_uuar(&context->uuari, qp->uuarn);
+}
+
+static int create_kernel_qp(struct mlx5_ib_dev *dev,
+			    struct ib_qp_init_attr *init_attr,
+			    struct mlx5_ib_qp *qp,
+			    struct mlx5_create_qp_mbox_in **in, int *inlen)
+{
+	struct mlx5_uuar_info *uuari;
+	int uuarn;
+	int uar_index;
+	int err;
+	enum ib_latency_class lc = IB_LATENCY_CLASS_LOW;
+
+	uuari = &dev->mdev.priv.uuari;
+	if (init_attr->create_flags & IB_QP_CREATE_BLOCK_MULTICAST_LOOPBACK)
+		qp->flags |= MLX5_IB_QP_BLOCK_MULTICAST_LOOPBACK;
+
+	if (init_attr->qp_type == IB_QPT_REG_UMR)
+		lc = IB_LATENCY_CLASS_FAST_PATH;
+
+	uuarn = alloc_uuar(uuari, lc);
+	if (uuarn < 0) {
+		mlx5_ib_dbg(dev, "\n");
+		return -ENOMEM;
+	}
+
+	qp->bf = &uuari->bfs[uuarn];
+	uar_index = qp->bf->uar->index;
+
+	err = calc_sq_size(dev, init_attr, qp);
+	if (err < 0) {
+		mlx5_ib_dbg(dev, "err %d\n", err);
+		goto err_uuar;
+	}
+
+	qp->rq.offset = 0;
+	qp->sq.offset = qp->rq.wqe_cnt << qp->rq.wqe_shift;
+	qp->buf_size = err + (qp->rq.wqe_cnt << qp->rq.wqe_shift);
+
+	mlx5_ib_dbg(dev, "\n");
+	err = mlx5_buf_alloc(&dev->mdev, qp->buf_size, PAGE_SIZE * 2, &qp->buf);
+	if (err) {
+		mlx5_ib_dbg(dev, "err %d\n", err);
+		goto err_uuar;
+	}
+
+	mlx5_ib_dbg(dev, "\n");
+	qp->sq.qend = mlx5_get_send_wqe(qp, qp->sq.wqe_cnt);
+	*inlen = sizeof **in + sizeof *(*in)->pas * qp->buf.npages;
+	*in = vzalloc(*inlen);
+	if (!*in) {
+		mlx5_ib_dbg(dev, "\n");
+		err = -ENOMEM;
+		goto err_buf;
+	}
+	(*in)->ctx.qp_counter_set_usr_page = cpu_to_be32(uar_index);
+	(*in)->ctx.log_pg_sz_remote_qpn = cpu_to_be32((qp->buf.page_shift - PAGE_SHIFT) << 24);
+	/* Set "fast registration enabled" for all kernel QPs */
+	(*in)->ctx.params1 |= cpu_to_be32(1 << 11);
+	(*in)->ctx.sq_crq_size |= cpu_to_be16(1 << 4);
+
+	mlx5_ib_dbg(dev, "\n");
+	mlx5_fill_page_array(&qp->buf, (*in)->pas);
+
+	mlx5_ib_dbg(dev, "\n");
+	err = mlx5_db_alloc(&dev->mdev, &qp->db);
+	if (err) {
+		mlx5_ib_dbg(dev, "err %d\n", err);
+		goto err_free;
+	}
+
+	qp->db.db[0] = 0;
+	qp->db.db[1] = 0;
+
+	qp->sq.wrid = kmalloc(qp->sq.wqe_cnt * sizeof(*qp->sq.wrid), GFP_KERNEL);
+	qp->sq.wr_data = kmalloc(qp->sq.wqe_cnt * sizeof(*qp->sq.wr_data), GFP_KERNEL);
+	qp->rq.wrid = kmalloc(qp->rq.wqe_cnt * sizeof(*qp->rq.wrid), GFP_KERNEL);
+	qp->sq.w_list = kmalloc(qp->sq.wqe_cnt * sizeof(*qp->sq.w_list), GFP_KERNEL);
+	qp->sq.wqe_head = kmalloc(qp->sq.wqe_cnt * sizeof(*qp->sq.wqe_head), GFP_KERNEL);
+
+	mlx5_ib_dbg(dev, "\n");
+	if (!qp->sq.wrid || !qp->sq.wr_data || !qp->rq.wrid ||
+	    !qp->sq.w_list || !qp->sq.wqe_head) {
+		err = -ENOMEM;
+		mlx5_ib_dbg(dev, "err %d\n", err);
+		goto err_wrid;
+	}
+	qp->create_type = MLX5_QP_KERNEL;
+
+	return 0;
+
+err_wrid:
+	mlx5_db_free(&dev->mdev, &qp->db);
+	kfree(qp->sq.wqe_head);
+	kfree(qp->sq.w_list);
+	kfree(qp->sq.wrid);
+	kfree(qp->sq.wr_data);
+	kfree(qp->rq.wrid);
+
+err_free:
+	vfree(*in);
+
+err_buf:
+	mlx5_buf_free(&dev->mdev, &qp->buf);
+
+err_uuar:
+	free_uuar(&dev->mdev.priv.uuari, uuarn);
+	return err;
+}
+
+static void destroy_qp_kernel(struct mlx5_ib_dev *dev, struct mlx5_ib_qp *qp)
+{
+	mlx5_db_free(&dev->mdev, &qp->db);
+	kfree(qp->sq.wqe_head);
+	kfree(qp->sq.w_list);
+	kfree(qp->sq.wrid);
+	kfree(qp->sq.wr_data);
+	kfree(qp->rq.wrid);
+	mlx5_buf_free(&dev->mdev, &qp->buf);
+	free_uuar(&dev->mdev.priv.uuari, qp->bf->uuarn);
+}
+
+static __be32 get_rx_type(struct mlx5_ib_qp *qp, struct ib_qp_init_attr *attr)
+{
+	if (attr->srq || (attr->qp_type == IB_QPT_XRC_TGT) ||
+	    (attr->qp_type == IB_QPT_XRC_INI))
+		return cpu_to_be32(MLX5_SRQ_RQ);
+	else if (!qp->has_rq)
+		return cpu_to_be32(MLX5_ZERO_LEN_RQ);
+	else
+		return cpu_to_be32(MLX5_NON_ZERO_RQ);
+}
+
+static int is_connected(enum ib_qp_type qp_type)
+{
+	if (qp_type == IB_QPT_RC || qp_type == IB_QPT_UC)
+		return 1;
+
+	return 0;
+}
+
+static int create_qp_common(struct mlx5_ib_dev *dev, struct ib_pd *pd,
+			    struct ib_qp_init_attr *init_attr,
+			    struct ib_udata *udata, struct mlx5_ib_qp *qp)
+{
+	int err;
+	struct mlx5_create_qp_mbox_in *in;
+	struct mlx5_ib_create_qp_resp resp;
+	int inlen = sizeof(*in);
+	struct mlx5_ib_create_qp ucmd;
+	struct mlx5_ib_resources *devr = &dev->devr;
+
+	mutex_init(&qp->mutex);
+	spin_lock_init(&qp->sq.lock);
+	spin_lock_init(&qp->rq.lock);
+
+	if (init_attr->sq_sig_type == IB_SIGNAL_ALL_WR)
+		qp->sq_signal_bits = MLX5_WQE_CTRL_CQ_UPDATE;
+
+	if (pd && pd->uobject) {
+		if (ib_copy_from_udata(&ucmd, udata, sizeof(ucmd))) {
+			mlx5_ib_dbg(dev, "copy failed\n");
+			return -EFAULT;
+		}
+
+		qp->wq_sig = !!(ucmd.flags & MLX5_QP_FLAG_SIGNATURE);
+		qp->scat_cqe = !!(ucmd.flags & MLX5_QP_FLAG_SCATTER_CQE);
+	} else {
+		qp->wq_sig = !!wq_signature;
+	}
+
+	qp->has_rq = qp_has_rq(init_attr);
+	err = set_rq_size(dev, &init_attr->cap, qp->has_rq,
+			  qp, (pd && pd->uobject) ? &ucmd : NULL);
+	if (err) {
+		mlx5_ib_dbg(dev, "err %d\n", err);
+		return err;
+	}
+
+	if (pd) {
+		if (pd->uobject) {
+			mlx5_ib_dbg(dev, "requested sq_wqe_count (%d)\n", ucmd.sq_wqe_count);
+			if (ucmd.rq_wqe_shift != qp->rq.wqe_shift ||
+			    ucmd.rq_wqe_count != qp->rq.wqe_cnt) {
+				mlx5_ib_dbg(dev, "invalid rq params\n");
+				return -EINVAL;
+			}
+			if (ucmd.sq_wqe_count > dev->mdev.caps.max_wqes) {
+				mlx5_ib_dbg(dev, "requested sq_wqe_count (%d) > max allowed (%d)\n",
+					    ucmd.sq_wqe_count, dev->mdev.caps.max_wqes);
+				return -EINVAL;
+			}
+			err = create_user_qp(dev, pd, qp, udata, &in, &resp, &inlen);
+			if (err)
+				mlx5_ib_dbg(dev, "err %d\n", err);
+		} else {
+			err = create_kernel_qp(dev, init_attr, qp, &in, &inlen);
+			if (err)
+				mlx5_ib_dbg(dev, "err %d\n", err);
+			else
+				qp->pa_lkey = to_mpd(pd)->pa_lkey;
+		}
+
+		if (err)
+			return err;
+	} else {
+		in = vzalloc(sizeof(*in));
+		if (!in)
+			return -ENOMEM;
+
+		qp->create_type = MLX5_QP_EMPTY;
+	}
+
+	if (is_sqp(init_attr->qp_type))
+		qp->port = init_attr->port_num;
+
+	in->ctx.flags = cpu_to_be32(to_mlx5_st(init_attr->qp_type) << 16 |
+				    MLX5_QP_PM_MIGRATED << 11);
+
+	if (init_attr->qp_type != IB_QPT_REG_UMR)
+		in->ctx.flags_pd = cpu_to_be32(to_mpd(pd ? pd : devr->p0)->pdn);
+	else
+		in->ctx.flags_pd = cpu_to_be32(MLX5_QP_LAT_SENSITIVE);
+
+	if (qp->wq_sig)
+		in->ctx.flags_pd |= cpu_to_be32(MLX5_QP_ENABLE_SIG);
+
+	if (qp->scat_cqe && is_connected(init_attr->qp_type)) {
+		int rcqe_sz;
+		int scqe_sz;
+
+		rcqe_sz = mlx5_ib_get_cqe_size(dev, init_attr->recv_cq);
+		scqe_sz = mlx5_ib_get_cqe_size(dev, init_attr->send_cq);
+
+		if (rcqe_sz == 128)
+			in->ctx.cs_res = MLX5_RES_SCAT_DATA64_CQE;
+		else
+			in->ctx.cs_res = MLX5_RES_SCAT_DATA32_CQE;
+
+		if (init_attr->sq_sig_type == IB_SIGNAL_ALL_WR) {
+			if (scqe_sz == 128)
+				in->ctx.cs_req = MLX5_REQ_SCAT_DATA64_CQE;
+			else
+				in->ctx.cs_req = MLX5_REQ_SCAT_DATA32_CQE;
+		}
+	}
+
+	if (qp->rq.wqe_cnt) {
+		in->ctx.rq_size_stride = (qp->rq.wqe_shift - 4);
+		in->ctx.rq_size_stride |= ilog2(qp->rq.wqe_cnt) << 3;
+	}
+
+	in->ctx.rq_type_srqn = get_rx_type(qp, init_attr);
+
+	if (qp->sq.wqe_cnt)
+		in->ctx.sq_crq_size |= cpu_to_be16(ilog2(qp->sq.wqe_cnt) << 11);
+	else
+		in->ctx.sq_crq_size |= cpu_to_be16(0x8000);
+
+	/* Set default resources */
+	switch (init_attr->qp_type) {
+	case IB_QPT_XRC_TGT:
+		in->ctx.cqn_recv = cpu_to_be32(to_mcq(devr->c0)->mcq.cqn);
+		in->ctx.cqn_send = cpu_to_be32(to_mcq(devr->c0)->mcq.cqn);
+		in->ctx.rq_type_srqn |= cpu_to_be32(to_msrq(devr->s0)->msrq.srqn);
+		in->ctx.xrcd = cpu_to_be32(to_mxrcd(init_attr->xrcd)->xrcdn);
+		break;
+	case IB_QPT_XRC_INI:
+		in->ctx.cqn_recv = cpu_to_be32(to_mcq(devr->c0)->mcq.cqn);
+		in->ctx.xrcd = cpu_to_be32(to_mxrcd(devr->x1)->xrcdn);
+		in->ctx.rq_type_srqn |= cpu_to_be32(to_msrq(devr->s0)->msrq.srqn);
+		break;
+	default:
+		if (init_attr->srq) {
+			in->ctx.xrcd = cpu_to_be32(to_mxrcd(devr->x0)->xrcdn);
+			in->ctx.rq_type_srqn |= cpu_to_be32(to_msrq(init_attr->srq)->msrq.srqn);
+		} else {
+			in->ctx.xrcd = cpu_to_be32(to_mxrcd(devr->x1)->xrcdn);
+			in->ctx.rq_type_srqn |= cpu_to_be32(to_msrq(devr->s0)->msrq.srqn);
+		}
+	}
+
+	if (init_attr->send_cq)
+		in->ctx.cqn_send = cpu_to_be32(to_mcq(init_attr->send_cq)->mcq.cqn);
+
+	if (init_attr->recv_cq)
+		in->ctx.cqn_recv = cpu_to_be32(to_mcq(init_attr->recv_cq)->mcq.cqn);
+
+	in->ctx.db_rec_addr = cpu_to_be64(qp->db.dma);
+
+	err = mlx5_core_create_qp(&dev->mdev, &qp->mqp, in, inlen);
+	if (err) {
+		mlx5_ib_dbg(dev, "create qp failed\n");
+		goto err_create;
+	}
+
+	vfree(in);
+	/*
+	 * Hardware wants QPN written in big-endian order (after
+	 * shifting) for send doorbell.  Precompute this value to save
+	 * a little bit when posting sends.
+	 */
+	qp->doorbell_qpn = swab32(qp->mqp.qpn << 8);
+
+	qp->mqp.event = mlx5_ib_qp_event;
+
+	return 0;
+
+err_create:
+	if (qp->create_type == MLX5_QP_USER)
+		destroy_qp_user(pd, qp);
+	else if (qp->create_type == MLX5_QP_KERNEL)
+		destroy_qp_kernel(dev, qp);
+
+	vfree(in);
+	return err;
+}
+
+static void mlx5_ib_lock_cqs(struct mlx5_ib_cq *send_cq, struct mlx5_ib_cq *recv_cq)
+	__acquires(&send_cq->lock) __acquires(&recv_cq->lock)
+{
+	if (send_cq) {
+		if (recv_cq) {
+			if (send_cq->mcq.cqn < recv_cq->mcq.cqn)  {
+				spin_lock_irq(&send_cq->lock);
+				spin_lock_nested(&recv_cq->lock,
+						 SINGLE_DEPTH_NESTING);
+			} else if (send_cq->mcq.cqn == recv_cq->mcq.cqn) {
+				spin_lock_irq(&send_cq->lock);
+				__acquire(&recv_cq->lock);
+			} else {
+				spin_lock_irq(&recv_cq->lock);
+				spin_lock_nested(&send_cq->lock,
+						 SINGLE_DEPTH_NESTING);
+			}
+		} else {
+			spin_lock_irq(&send_cq->lock);
+		}
+	} else if (recv_cq) {
+		spin_lock_irq(&recv_cq->lock);
+	}
+}
+
+static void mlx5_ib_unlock_cqs(struct mlx5_ib_cq *send_cq, struct mlx5_ib_cq *recv_cq)
+	__releases(&send_cq->lock) __releases(&recv_cq->lock)
+{
+	if (send_cq) {
+		if (recv_cq) {
+			if (send_cq->mcq.cqn < recv_cq->mcq.cqn)  {
+				spin_unlock(&recv_cq->lock);
+				spin_unlock_irq(&send_cq->lock);
+			} else if (send_cq->mcq.cqn == recv_cq->mcq.cqn) {
+				__release(&recv_cq->lock);
+				spin_unlock_irq(&send_cq->lock);
+			} else {
+				spin_unlock(&send_cq->lock);
+				spin_unlock_irq(&recv_cq->lock);
+			}
+		} else {
+			spin_unlock_irq(&send_cq->lock);
+		}
+	} else if (recv_cq) {
+		spin_unlock_irq(&recv_cq->lock);
+	}
+}
+
+static struct mlx5_ib_pd *get_pd(struct mlx5_ib_qp *qp)
+{
+	return to_mpd(qp->ibqp.pd);
+}
+
+static void get_cqs(struct mlx5_ib_qp *qp,
+		    struct mlx5_ib_cq **send_cq, struct mlx5_ib_cq **recv_cq)
+{
+	switch (qp->ibqp.qp_type) {
+	case IB_QPT_XRC_TGT:
+		*send_cq = NULL;
+		*recv_cq = NULL;
+		break;
+	case IB_QPT_REG_UMR:
+	case IB_QPT_XRC_INI:
+		*send_cq = to_mcq(qp->ibqp.send_cq);
+		*recv_cq = NULL;
+		break;
+
+	case IB_QPT_SMI:
+	case IB_QPT_GSI:
+	case IB_QPT_RC:
+	case IB_QPT_UC:
+	case IB_QPT_UD:
+	case IB_QPT_RAW_IPV6:
+	case IB_QPT_RAW_ETHERTYPE:
+		*send_cq = to_mcq(qp->ibqp.send_cq);
+		*recv_cq = to_mcq(qp->ibqp.recv_cq);
+		break;
+
+	case IB_QPT_RAW_PACKET:
+	case IB_QPT_MAX:
+	default:
+		*send_cq = NULL;
+		*recv_cq = NULL;
+		break;
+	}
+}
+
+static void destroy_qp_common(struct mlx5_ib_dev *dev, struct mlx5_ib_qp *qp)
+{
+	struct mlx5_ib_cq *send_cq, *recv_cq;
+	struct mlx5_modify_qp_mbox_in *in;
+	int err;
+
+	in = kzalloc(sizeof(*in), GFP_KERNEL);
+	if (!in) {
+		mlx5_ib_warn(dev, "allocation failed\n");
+		return;
+	}
+	if (qp->state != IB_QPS_RESET)
+		if (mlx5_core_qp_modify(&dev->mdev, to_mlx5_state(qp->state),
+					MLX5_QP_STATE_RST, in, sizeof(*in), &qp->mqp))
+			mlx5_ib_warn(dev, "mlx5_ib: modify QP %06x to RESET failed\n",
+				     qp->mqp.qpn);
+
+	get_cqs(qp, &send_cq, &recv_cq);
+
+	if (qp->create_type == MLX5_QP_KERNEL) {
+		mlx5_ib_lock_cqs(send_cq, recv_cq);
+		__mlx5_ib_cq_clean(recv_cq, qp->mqp.qpn,
+				   qp->ibqp.srq ? to_msrq(qp->ibqp.srq) : NULL);
+		if (send_cq != recv_cq)
+			__mlx5_ib_cq_clean(send_cq, qp->mqp.qpn, NULL);
+		mlx5_ib_unlock_cqs(send_cq, recv_cq);
+	}
+
+	err = mlx5_core_destroy_qp(&dev->mdev, &qp->mqp);
+	if (err)
+		mlx5_ib_warn(dev, "failed to destroy QP 0x%x\n", qp->mqp.qpn);
+	kfree(in);
+
+
+	if (qp->create_type == MLX5_QP_KERNEL)
+		destroy_qp_kernel(dev, qp);
+	else if (qp->create_type == MLX5_QP_USER)
+		destroy_qp_user(&get_pd(qp)->ibpd, qp);
+}
+
+static const char *ib_qp_type_str(enum ib_qp_type type)
+{
+	switch (type) {
+	case IB_QPT_SMI:
+		return "IB_QPT_SMI";
+	case IB_QPT_GSI:
+		return "IB_QPT_GSI";
+	case IB_QPT_RC:
+		return "IB_QPT_RC";
+	case IB_QPT_UC:
+		return "IB_QPT_UC";
+	case IB_QPT_UD:
+		return "IB_QPT_UD";
+	case IB_QPT_RAW_IPV6:
+		return "IB_QPT_RAW_IPV6";
+	case IB_QPT_RAW_ETHERTYPE:
+		return "IB_QPT_RAW_ETHERTYPE";
+	case IB_QPT_XRC_INI:
+		return "IB_QPT_XRC_INI";
+	case IB_QPT_XRC_TGT:
+		return "IB_QPT_XRC_TGT";
+	case IB_QPT_RAW_PACKET:
+		return "IB_QPT_RAW_PACKET";
+	case IB_QPT_REG_UMR:
+		return "IB_QPT_REG_UMR";
+	case IB_QPT_MAX:
+	default:
+		return "Invalid QP type";
+	}
+}
+
+struct ib_qp *mlx5_ib_create_qp(struct ib_pd *pd,
+				struct ib_qp_init_attr *init_attr,
+				struct ib_udata *udata)
+{
+	struct mlx5_ib_qp *qp;
+	int err;
+	u16 xrcdn = 0;
+	struct mlx5_ib_dev *dev;
+
+	if (pd) {
+		dev = to_mdev(pd->device);
+	} else {
+		/* being cautious here */
+		if (init_attr->qp_type != IB_QPT_XRC_TGT &&
+		    init_attr->qp_type != IB_QPT_REG_UMR) {
+			pr_warn("%s: no PD for transport %s\n", __func__,
+				ib_qp_type_str(init_attr->qp_type));
+			return ERR_PTR(-EINVAL);
+		}
+		dev = to_mdev(to_mxrcd(init_attr->xrcd)->ibxrcd.device);
+	}
+
+	switch (init_attr->qp_type) {
+	case IB_QPT_XRC_TGT:
+	case IB_QPT_XRC_INI:
+		if (!(dev->mdev.caps.flags & MLX5_DEV_CAP_FLAG_XRC)) {
+			mlx5_ib_dbg(dev, "XRC not supported\n");
+			return ERR_PTR(-ENOSYS);
+		}
+		init_attr->recv_cq = NULL;
+		if (init_attr->qp_type == IB_QPT_XRC_TGT) {
+			xrcdn = to_mxrcd(init_attr->xrcd)->xrcdn;
+			init_attr->send_cq = NULL;
+		}
+
+		/* fall through */
+	case IB_QPT_RC:
+	case IB_QPT_UC:
+	case IB_QPT_UD:
+	case IB_QPT_SMI:
+	case IB_QPT_GSI:
+	case IB_QPT_REG_UMR:
+		qp = kzalloc(sizeof(*qp), GFP_KERNEL);
+		if (!qp) {
+			mlx5_ib_dbg(dev, "XRC not supported\n");
+			return ERR_PTR(-ENOMEM);
+		}
+
+		err = create_qp_common(dev, pd, init_attr, udata, qp);
+		if (err) {
+			mlx5_ib_dbg(dev, "create_qp_common failed\n");
+			kfree(qp);
+			return ERR_PTR(err);
+		}
+
+		if (is_qp0(init_attr->qp_type))
+			qp->ibqp.qp_num = 0;
+		else if (is_qp1(init_attr->qp_type))
+			qp->ibqp.qp_num = 1;
+		else
+			qp->ibqp.qp_num = qp->mqp.qpn;
+
+		mlx5_ib_dbg(dev, "ib qpnum 0x%x, mlx qpn 0x%x, rcqn 0x%x, scqn 0x%x\n",
+			    qp->ibqp.qp_num, qp->mqp.qpn, to_mcq(init_attr->recv_cq)->mcq.cqn,
+			    to_mcq(init_attr->send_cq)->mcq.cqn);
+
+		qp->xrcdn = xrcdn;
+
+		break;
+
+	case IB_QPT_RAW_IPV6:
+	case IB_QPT_RAW_ETHERTYPE:
+	case IB_QPT_RAW_PACKET:
+	case IB_QPT_MAX:
+	default:
+		mlx5_ib_dbg(dev, "unsupported qp type %d\n",
+			    init_attr->qp_type);
+		/* Don't support raw QPs */
+		return ERR_PTR(-EINVAL);
+	}
+
+	return &qp->ibqp;
+}
+
+int mlx5_ib_destroy_qp(struct ib_qp *qp)
+{
+	struct mlx5_ib_dev *dev = to_mdev(qp->device);
+	struct mlx5_ib_qp *mqp = to_mqp(qp);
+
+	destroy_qp_common(dev, mqp);
+
+	kfree(mqp);
+
+	return 0;
+}
+
+static __be32 to_mlx5_access_flags(struct mlx5_ib_qp *qp, const struct ib_qp_attr *attr,
+				   int attr_mask)
+{
+	u8 dest_rd_atomic;
+	u32 access_flags;
+	u32 hw_access_flags = 0;
+
+	if (attr_mask & IB_QP_MAX_DEST_RD_ATOMIC)
+		dest_rd_atomic = attr->max_dest_rd_atomic;
+	else
+		dest_rd_atomic = qp->resp_depth;
+
+	if (attr_mask & IB_QP_ACCESS_FLAGS)
+		access_flags = attr->qp_access_flags;
+	else
+		access_flags = qp->atomic_rd_en;
+
+	if (!dest_rd_atomic)
+		access_flags &= IB_ACCESS_REMOTE_WRITE;
+
+	if (access_flags & IB_ACCESS_REMOTE_READ)
+		hw_access_flags |= MLX5_QP_BIT_RRE;
+	if (access_flags & IB_ACCESS_REMOTE_ATOMIC)
+		hw_access_flags |= (MLX5_QP_BIT_RAE | MLX5_ATOMIC_MODE_CX);
+	if (access_flags & IB_ACCESS_REMOTE_WRITE)
+		hw_access_flags |= MLX5_QP_BIT_RWE;
+
+	return cpu_to_be32(hw_access_flags);
+}
+
+enum {
+	MLX5_PATH_FLAG_FL	= 1 << 0,
+	MLX5_PATH_FLAG_FREE_AR	= 1 << 1,
+	MLX5_PATH_FLAG_COUNTER	= 1 << 2,
+};
+
+static int ib_rate_to_mlx5(struct mlx5_ib_dev *dev, u8 rate)
+{
+	if (rate == IB_RATE_PORT_CURRENT) {
+		return 0;
+	} else if (rate < IB_RATE_2_5_GBPS || rate > IB_RATE_300_GBPS) {
+		return -EINVAL;
+	} else {
+		while (rate != IB_RATE_2_5_GBPS &&
+		       !(1 << (rate + MLX5_STAT_RATE_OFFSET) &
+			 dev->mdev.caps.stat_rate_support))
+			--rate;
+	}
+
+	return rate + MLX5_STAT_RATE_OFFSET;
+}
+
+static int mlx5_set_path(struct mlx5_ib_dev *dev, const struct ib_ah_attr *ah,
+			 struct mlx5_qp_path *path, u8 port, int attr_mask,
+			 u32 path_flags, const struct ib_qp_attr *attr)
+{
+	int err;
+
+	path->fl = (path_flags & MLX5_PATH_FLAG_FL) ? 0x80 : 0;
+	path->free_ar = (path_flags & MLX5_PATH_FLAG_FREE_AR) ? 0x80 : 0;
+
+	if (attr_mask & IB_QP_PKEY_INDEX)
+		path->pkey_index = attr->pkey_index;
+
+	path->grh_mlid	= ah->src_path_bits & 0x7f;
+	path->rlid	= cpu_to_be16(ah->dlid);
+
+	if (ah->ah_flags & IB_AH_GRH) {
+		path->grh_mlid |= 1 << 7;
+		path->mgid_index = ah->grh.sgid_index;
+		path->hop_limit  = ah->grh.hop_limit;
+		path->tclass_flowlabel =
+			cpu_to_be32((ah->grh.traffic_class << 20) |
+				    (ah->grh.flow_label));
+		memcpy(path->rgid, ah->grh.dgid.raw, 16);
+	}
+
+	err = ib_rate_to_mlx5(dev, ah->static_rate);
+	if (err < 0)
+		return err;
+	path->static_rate = err;
+	path->port = port;
+
+	if (ah->ah_flags & IB_AH_GRH) {
+		if (ah->grh.sgid_index >= dev->mdev.caps.port[port - 1].gid_table_len) {
+			pr_err(KERN_ERR "sgid_index (%u) too large. max is %d\n",
+			       ah->grh.sgid_index, dev->mdev.caps.port[port - 1].gid_table_len);
+			return -EINVAL;
+		}
+
+		path->grh_mlid |= 1 << 7;
+		path->mgid_index = ah->grh.sgid_index;
+		path->hop_limit  = ah->grh.hop_limit;
+		path->tclass_flowlabel =
+			cpu_to_be32((ah->grh.traffic_class << 20) |
+				    (ah->grh.flow_label));
+		memcpy(path->rgid, ah->grh.dgid.raw, 16);
+	}
+
+	if (attr_mask & IB_QP_TIMEOUT)
+		path->ackto_lt = attr->timeout << 3;
+
+	path->sl = ah->sl & 0xf;
+
+	return 0;
+}
+
+static enum mlx5_qp_optpar opt_mask[MLX5_QP_NUM_STATE][MLX5_QP_NUM_STATE][MLX5_QP_ST_MAX] = {
+	[MLX5_QP_STATE_INIT] = {
+		[MLX5_QP_STATE_INIT] = {
+			[MLX5_QP_ST_RC] = MLX5_QP_OPTPAR_RRE		|
+					  MLX5_QP_OPTPAR_RAE		|
+					  MLX5_QP_OPTPAR_RWE		|
+					  MLX5_QP_OPTPAR_PKEY_INDEX	|
+					  MLX5_QP_OPTPAR_PRI_PORT,
+			[MLX5_QP_ST_UC] = MLX5_QP_OPTPAR_RWE		|
+					  MLX5_QP_OPTPAR_PKEY_INDEX	|
+					  MLX5_QP_OPTPAR_PRI_PORT,
+			[MLX5_QP_ST_UD] = MLX5_QP_OPTPAR_PKEY_INDEX	|
+					  MLX5_QP_OPTPAR_Q_KEY		|
+					  MLX5_QP_OPTPAR_PRI_PORT,
+		},
+		[MLX5_QP_STATE_RTR] = {
+			[MLX5_QP_ST_RC] = MLX5_QP_OPTPAR_ALT_ADDR_PATH  |
+					  MLX5_QP_OPTPAR_RRE            |
+					  MLX5_QP_OPTPAR_RAE            |
+					  MLX5_QP_OPTPAR_RWE            |
+					  MLX5_QP_OPTPAR_PKEY_INDEX,
+			[MLX5_QP_ST_UC] = MLX5_QP_OPTPAR_ALT_ADDR_PATH  |
+					  MLX5_QP_OPTPAR_RWE            |
+					  MLX5_QP_OPTPAR_PKEY_INDEX,
+			[MLX5_QP_ST_UD] = MLX5_QP_OPTPAR_PKEY_INDEX     |
+					  MLX5_QP_OPTPAR_Q_KEY,
+			[MLX5_QP_ST_MLX] = MLX5_QP_OPTPAR_PKEY_INDEX	|
+					   MLX5_QP_OPTPAR_Q_KEY,
+		},
+	},
+	[MLX5_QP_STATE_RTR] = {
+		[MLX5_QP_STATE_RTS] = {
+			[MLX5_QP_ST_RC] = MLX5_QP_OPTPAR_ALT_ADDR_PATH	|
+					  MLX5_QP_OPTPAR_RRE		|
+					  MLX5_QP_OPTPAR_RAE		|
+					  MLX5_QP_OPTPAR_RWE		|
+					  MLX5_QP_OPTPAR_PM_STATE	|
+					  MLX5_QP_OPTPAR_RNR_TIMEOUT,
+			[MLX5_QP_ST_UC] = MLX5_QP_OPTPAR_ALT_ADDR_PATH	|
+					  MLX5_QP_OPTPAR_RWE		|
+					  MLX5_QP_OPTPAR_PM_STATE,
+			[MLX5_QP_ST_UD] = MLX5_QP_OPTPAR_Q_KEY,
+		},
+	},
+	[MLX5_QP_STATE_RTS] = {
+		[MLX5_QP_STATE_RTS] = {
+			[MLX5_QP_ST_RC] = MLX5_QP_OPTPAR_RRE		|
+					  MLX5_QP_OPTPAR_RAE		|
+					  MLX5_QP_OPTPAR_RWE		|
+					  MLX5_QP_OPTPAR_RNR_TIMEOUT	|
+					  MLX5_QP_OPTPAR_PM_STATE,
+			[MLX5_QP_ST_UC] = MLX5_QP_OPTPAR_RWE		|
+					  MLX5_QP_OPTPAR_PM_STATE,
+			[MLX5_QP_ST_UD] = MLX5_QP_OPTPAR_Q_KEY		|
+					  MLX5_QP_OPTPAR_SRQN		|
+					  MLX5_QP_OPTPAR_CQN_RCV,
+		},
+	},
+	[MLX5_QP_STATE_SQER] = {
+		[MLX5_QP_STATE_RTS] = {
+			[MLX5_QP_ST_UD]	 = MLX5_QP_OPTPAR_Q_KEY,
+			[MLX5_QP_ST_MLX] = MLX5_QP_OPTPAR_Q_KEY,
+		},
+	},
+};
+
+static int ib_nr_to_mlx5_nr(int ib_mask)
+{
+	switch (ib_mask) {
+	case IB_QP_STATE:
+		return 0;
+	case IB_QP_CUR_STATE:
+		return 0;
+	case IB_QP_EN_SQD_ASYNC_NOTIFY:
+		return 0;
+	case IB_QP_ACCESS_FLAGS:
+		return MLX5_QP_OPTPAR_RWE | MLX5_QP_OPTPAR_RRE |
+			MLX5_QP_OPTPAR_RAE;
+	case IB_QP_PKEY_INDEX:
+		return MLX5_QP_OPTPAR_PKEY_INDEX;
+	case IB_QP_PORT:
+		return MLX5_QP_OPTPAR_PRI_PORT;
+	case IB_QP_QKEY:
+		return MLX5_QP_OPTPAR_Q_KEY;
+	case IB_QP_AV:
+		return MLX5_QP_OPTPAR_PRIMARY_ADDR_PATH |
+			MLX5_QP_OPTPAR_PRI_PORT;
+	case IB_QP_PATH_MTU:
+		return 0;
+	case IB_QP_TIMEOUT:
+		return MLX5_QP_OPTPAR_ACK_TIMEOUT;
+	case IB_QP_RETRY_CNT:
+		return MLX5_QP_OPTPAR_RETRY_COUNT;
+	case IB_QP_RNR_RETRY:
+		return MLX5_QP_OPTPAR_RNR_RETRY;
+	case IB_QP_RQ_PSN:
+		return 0;
+	case IB_QP_MAX_QP_RD_ATOMIC:
+		return MLX5_QP_OPTPAR_SRA_MAX;
+	case IB_QP_ALT_PATH:
+		return MLX5_QP_OPTPAR_ALT_ADDR_PATH;
+	case IB_QP_MIN_RNR_TIMER:
+		return MLX5_QP_OPTPAR_RNR_TIMEOUT;
+	case IB_QP_SQ_PSN:
+		return 0;
+	case IB_QP_MAX_DEST_RD_ATOMIC:
+		return MLX5_QP_OPTPAR_RRA_MAX | MLX5_QP_OPTPAR_RWE |
+			MLX5_QP_OPTPAR_RRE | MLX5_QP_OPTPAR_RAE;
+	case IB_QP_PATH_MIG_STATE:
+		return MLX5_QP_OPTPAR_PM_STATE;
+	case IB_QP_CAP:
+		return 0;
+	case IB_QP_DEST_QPN:
+		return 0;
+	}
+	return 0;
+}
+
+static int ib_mask_to_mlx5_opt(int ib_mask)
+{
+	int i;
+	int result = 0;
+
+	for (i = 0; i < 8 * sizeof(int); ++i) {
+		if ((1 << i) & ib_mask)
+			result |= ib_nr_to_mlx5_nr(1 << i);
+	}
+
+	return result;
+}
+
+static int __mlx5_ib_modify_qp(struct ib_qp *ibqp,
+			       const struct ib_qp_attr *attr, int attr_mask,
+			       enum ib_qp_state cur_state, enum ib_qp_state new_state)
+{
+	struct mlx5_ib_dev *dev = to_mdev(ibqp->device);
+	struct mlx5_ib_qp *qp = to_mqp(ibqp);
+	struct mlx5_ib_pd *pd;
+	struct mlx5_ib_cq *send_cq, *recv_cq;
+	struct mlx5_qp_context *context;
+	struct mlx5_modify_qp_mbox_in *in;
+	enum mlx5_qp_optpar optpar;
+	int sqd_event;
+	int err;
+	enum mlx5_qp_state mlx5_cur, mlx5_new;
+	int mlx5_st;
+
+	in = kzalloc(sizeof(*in), GFP_KERNEL);
+	if (!in)
+		return -ENOMEM;
+
+	context = &in->ctx;
+	err = to_mlx5_st(ibqp->qp_type);
+	if (err < 0)
+		goto out;
+
+	context->flags = cpu_to_be32(err << 16);
+
+	if (!(attr_mask & IB_QP_PATH_MIG_STATE)) {
+		context->flags |= cpu_to_be32(MLX5_QP_PM_MIGRATED << 11);
+	} else {
+		switch (attr->path_mig_state) {
+		case IB_MIG_MIGRATED:
+			context->flags |= cpu_to_be32(MLX5_QP_PM_MIGRATED << 11);
+			break;
+		case IB_MIG_REARM:
+			context->flags |= cpu_to_be32(MLX5_QP_PM_REARM << 11);
+			break;
+		case IB_MIG_ARMED:
+			context->flags |= cpu_to_be32(MLX5_QP_PM_ARMED << 11);
+			break;
+		}
+	}
+
+	if (ibqp->qp_type == IB_QPT_GSI || ibqp->qp_type == IB_QPT_SMI) {
+		context->mtu_msgmax = (IB_MTU_256 << 5) | 8;
+	} else if (ibqp->qp_type == IB_QPT_UD ||
+		   ibqp->qp_type == IB_QPT_REG_UMR) {
+		context->mtu_msgmax = (IB_MTU_4096 << 5) | 12;
+	} else if (attr_mask & IB_QP_PATH_MTU) {
+		if (attr->path_mtu < IB_MTU_256 ||
+		    attr->path_mtu > IB_MTU_4096) {
+			mlx5_ib_warn(dev, "invalid mtu %d\n", attr->path_mtu);
+			err = -EINVAL;
+			goto out;
+		}
+		context->mtu_msgmax = (attr->path_mtu << 5) | dev->mdev.caps.log_max_msg;
+	}
+
+	if (attr_mask & IB_QP_DEST_QPN)
+		context->log_pg_sz_remote_qpn = cpu_to_be32(attr->dest_qp_num);
+
+	if (attr_mask & IB_QP_PKEY_INDEX)
+		context->pri_path.pkey_index = attr->pkey_index;
+
+	/* todo implement counter_index functionality */
+
+	if (is_sqp(ibqp->qp_type))
+		context->pri_path.port = qp->port;
+
+	if (attr_mask & IB_QP_PORT)
+		context->pri_path.port = attr->port_num;
+
+	if (attr_mask & IB_QP_AV) {
+		err = mlx5_set_path(dev, &attr->ah_attr, &context->pri_path,
+				    attr_mask & IB_QP_PORT ? attr->port_num : qp->port,
+				    attr_mask, 0, attr);
+		if (err)
+			goto out;
+	}
+
+	if (attr_mask & IB_QP_TIMEOUT)
+		context->pri_path.ackto_lt |= attr->timeout << 3;
+
+	if (attr_mask & IB_QP_ALT_PATH) {
+		err = mlx5_set_path(dev, &attr->alt_ah_attr, &context->alt_path,
+				    attr->alt_port_num, attr_mask, 0, attr);
+		if (err)
+			goto out;
+	}
+
+	pd = get_pd(qp);
+	get_cqs(qp, &send_cq, &recv_cq);
+
+	context->flags_pd = cpu_to_be32(pd ? pd->pdn : to_mpd(dev->devr.p0)->pdn);
+	context->cqn_send = send_cq ? cpu_to_be32(send_cq->mcq.cqn) : 0;
+	context->cqn_recv = recv_cq ? cpu_to_be32(recv_cq->mcq.cqn) : 0;
+	context->params1  = cpu_to_be32(MLX5_IB_ACK_REQ_FREQ << 28);
+
+	if (attr_mask & IB_QP_RNR_RETRY)
+		context->params1 |= cpu_to_be32(attr->rnr_retry << 13);
+
+	if (attr_mask & IB_QP_RETRY_CNT)
+		context->params1 |= cpu_to_be32(attr->retry_cnt << 16);
+
+	if (attr_mask & IB_QP_MAX_QP_RD_ATOMIC) {
+		if (attr->max_rd_atomic)
+			context->params1 |=
+				cpu_to_be32(fls(attr->max_rd_atomic - 1) << 21);
+	}
+
+	if (attr_mask & IB_QP_SQ_PSN)
+		context->next_send_psn = cpu_to_be32(attr->sq_psn);
+
+	if (attr_mask & IB_QP_MAX_DEST_RD_ATOMIC) {
+		if (attr->max_dest_rd_atomic)
+			context->params2 |=
+				cpu_to_be32(fls(attr->max_dest_rd_atomic - 1) << 21);
+	}
+
+	if (attr_mask & (IB_QP_ACCESS_FLAGS | IB_QP_MAX_DEST_RD_ATOMIC))
+		context->params2 |= to_mlx5_access_flags(qp, attr, attr_mask);
+
+	if (attr_mask & IB_QP_MIN_RNR_TIMER)
+		context->rnr_nextrecvpsn |= cpu_to_be32(attr->min_rnr_timer << 24);
+
+	if (attr_mask & IB_QP_RQ_PSN)
+		context->rnr_nextrecvpsn |= cpu_to_be32(attr->rq_psn);
+
+	if (attr_mask & IB_QP_QKEY)
+		context->qkey = cpu_to_be32(attr->qkey);
+
+	if (qp->rq.wqe_cnt && cur_state == IB_QPS_RESET && new_state == IB_QPS_INIT)
+		context->db_rec_addr = cpu_to_be64(qp->db.dma);
+
+	if (cur_state == IB_QPS_RTS && new_state == IB_QPS_SQD	&&
+	    attr_mask & IB_QP_EN_SQD_ASYNC_NOTIFY && attr->en_sqd_async_notify)
+		sqd_event = 1;
+	else
+		sqd_event = 0;
+
+	if (!ibqp->uobject && cur_state == IB_QPS_RESET && new_state == IB_QPS_INIT)
+		context->sq_crq_size |= cpu_to_be16(1 << 4);
+
+
+	mlx5_cur = to_mlx5_state(cur_state);
+	mlx5_new = to_mlx5_state(new_state);
+	mlx5_st = to_mlx5_st(ibqp->qp_type);
+	if (mlx5_cur < 0 || mlx5_new < 0 || mlx5_st < 0)
+		goto out;
+
+	optpar = ib_mask_to_mlx5_opt(attr_mask);
+	optpar &= opt_mask[mlx5_cur][mlx5_new][mlx5_st];
+	in->optparam = cpu_to_be32(optpar);
+	err = mlx5_core_qp_modify(&dev->mdev, to_mlx5_state(cur_state),
+				  to_mlx5_state(new_state), in, sqd_event,
+				  &qp->mqp);
+	if (err)
+		goto out;
+
+	qp->state = new_state;
+
+	if (attr_mask & IB_QP_ACCESS_FLAGS)
+		qp->atomic_rd_en = attr->qp_access_flags;
+	if (attr_mask & IB_QP_MAX_DEST_RD_ATOMIC)
+		qp->resp_depth = attr->max_dest_rd_atomic;
+	if (attr_mask & IB_QP_PORT)
+		qp->port = attr->port_num;
+	if (attr_mask & IB_QP_ALT_PATH)
+		qp->alt_port = attr->alt_port_num;
+
+	/*
+	 * If we moved a kernel QP to RESET, clean up all old CQ
+	 * entries and reinitialize the QP.
+	 */
+	if (new_state == IB_QPS_RESET && !ibqp->uobject) {
+		mlx5_ib_cq_clean(recv_cq, qp->mqp.qpn,
+				 ibqp->srq ? to_msrq(ibqp->srq) : NULL);
+		if (send_cq != recv_cq)
+			mlx5_ib_cq_clean(send_cq, qp->mqp.qpn, NULL);
+
+		qp->rq.head = 0;
+		qp->rq.tail = 0;
+		qp->sq.head = 0;
+		qp->sq.tail = 0;
+		qp->sq.cur_post = 0;
+		qp->sq.last_poll = 0;
+		qp->db.db[MLX5_RCV_DBR] = 0;
+		qp->db.db[MLX5_SND_DBR] = 0;
+	}
+
+out:
+	kfree(in);
+	return err;
+}
+
+int mlx5_ib_modify_qp(struct ib_qp *ibqp, struct ib_qp_attr *attr,
+		      int attr_mask, struct ib_udata *udata)
+{
+	struct mlx5_ib_dev *dev = to_mdev(ibqp->device);
+	struct mlx5_ib_qp *qp = to_mqp(ibqp);
+	enum ib_qp_state cur_state, new_state;
+	int port;
+	int err = -EINVAL;
+
+	mutex_lock(&qp->mutex);
+
+	cur_state = attr_mask & IB_QP_CUR_STATE ? attr->cur_qp_state : qp->state;
+	new_state = attr_mask & IB_QP_STATE ? attr->qp_state : cur_state;
+
+	if (ibqp->qp_type != IB_QPT_REG_UMR &&
+	    !ib_modify_qp_is_ok(cur_state, new_state, ibqp->qp_type, attr_mask))
+		goto out;
+
+	if ((attr_mask & IB_QP_PORT) &&
+	    (attr->port_num == 0 || attr->port_num > dev->mdev.caps.num_ports))
+		goto out;
+
+	if (attr_mask & IB_QP_PKEY_INDEX) {
+		port = attr_mask & IB_QP_PORT ? attr->port_num : qp->port;
+		if (attr->pkey_index >= dev->mdev.caps.port[port - 1].pkey_table_len)
+			goto out;
+	}
+
+	if (attr_mask & IB_QP_MAX_QP_RD_ATOMIC &&
+	    attr->max_rd_atomic > dev->mdev.caps.max_ra_res_qp)
+		goto out;
+
+	if (attr_mask & IB_QP_MAX_DEST_RD_ATOMIC &&
+	    attr->max_dest_rd_atomic > dev->mdev.caps.max_ra_req_qp)
+		goto out;
+
+	if (cur_state == new_state && cur_state == IB_QPS_RESET) {
+		err = 0;
+		goto out;
+	}
+
+	err = __mlx5_ib_modify_qp(ibqp, attr, attr_mask, cur_state, new_state);
+
+out:
+	mutex_unlock(&qp->mutex);
+	return err;
+}
+
+static int mlx5_wq_overflow(struct mlx5_ib_wq *wq, int nreq, struct ib_cq *ib_cq)
+{
+	unsigned cur;
+	struct mlx5_ib_cq *cq;
+
+	cur = wq->head - wq->tail;
+	if (likely(cur + nreq < wq->max_post))
+		return 0;
+
+	cq = to_mcq(ib_cq);
+	spin_lock(&cq->lock);
+	cur = wq->head - wq->tail;
+	spin_unlock(&cq->lock);
+
+	return cur + nreq >= wq->max_post;
+}
+
+static __always_inline void set_raddr_seg(struct mlx5_wqe_raddr_seg *rseg,
+					  u64 remote_addr, u32 rkey)
+{
+	rseg->raddr    = cpu_to_be64(remote_addr);
+	rseg->rkey     = cpu_to_be32(rkey);
+	rseg->reserved = 0;
+}
+
+static void set_atomic_seg(struct mlx5_wqe_atomic_seg *aseg, struct ib_send_wr *wr)
+{
+	if (wr->opcode == IB_WR_ATOMIC_CMP_AND_SWP) {
+		aseg->swap_add = cpu_to_be64(wr->wr.atomic.swap);
+		aseg->compare  = cpu_to_be64(wr->wr.atomic.compare_add);
+	} else if (wr->opcode == IB_WR_MASKED_ATOMIC_FETCH_AND_ADD) {
+		aseg->swap_add = cpu_to_be64(wr->wr.atomic.compare_add);
+		aseg->compare  = cpu_to_be64(wr->wr.atomic.compare_add_mask);
+	} else {
+		aseg->swap_add = cpu_to_be64(wr->wr.atomic.compare_add);
+		aseg->compare  = 0;
+	}
+}
+
+static void set_masked_atomic_seg(struct mlx5_wqe_masked_atomic_seg *aseg,
+				  struct ib_send_wr *wr)
+{
+	aseg->swap_add		= cpu_to_be64(wr->wr.atomic.swap);
+	aseg->swap_add_mask	= cpu_to_be64(wr->wr.atomic.swap_mask);
+	aseg->compare		= cpu_to_be64(wr->wr.atomic.compare_add);
+	aseg->compare_mask	= cpu_to_be64(wr->wr.atomic.compare_add_mask);
+}
+
+static void set_datagram_seg(struct mlx5_wqe_datagram_seg *dseg,
+			     struct ib_send_wr *wr)
+{
+	memcpy(&dseg->av, &to_mah(wr->wr.ud.ah)->av, sizeof(struct mlx5_av));
+	dseg->av.dqp_dct = cpu_to_be32(wr->wr.ud.remote_qpn | MLX5_EXTENDED_UD_AV);
+	dseg->av.key.qkey.qkey = cpu_to_be32(wr->wr.ud.remote_qkey);
+}
+
+static void set_data_ptr_seg(struct mlx5_wqe_data_seg *dseg, struct ib_sge *sg)
+{
+	dseg->byte_count = cpu_to_be32(sg->length);
+	dseg->lkey       = cpu_to_be32(sg->lkey);
+	dseg->addr       = cpu_to_be64(sg->addr);
+}
+
+static __be16 get_klm_octo(int npages)
+{
+	return cpu_to_be16(ALIGN(npages, 8) / 2);
+}
+
+static __be64 frwr_mkey_mask(void)
+{
+	u64 result;
+
+	result = MLX5_MKEY_MASK_LEN		|
+		MLX5_MKEY_MASK_PAGE_SIZE	|
+		MLX5_MKEY_MASK_START_ADDR	|
+		MLX5_MKEY_MASK_EN_RINVAL	|
+		MLX5_MKEY_MASK_KEY		|
+		MLX5_MKEY_MASK_LR		|
+		MLX5_MKEY_MASK_LW		|
+		MLX5_MKEY_MASK_RR		|
+		MLX5_MKEY_MASK_RW		|
+		MLX5_MKEY_MASK_A		|
+		MLX5_MKEY_MASK_SMALL_FENCE	|
+		MLX5_MKEY_MASK_FREE;
+
+	return cpu_to_be64(result);
+}
+
+static void set_frwr_umr_segment(struct mlx5_wqe_umr_ctrl_seg *umr,
+				 struct ib_send_wr *wr, int li)
+{
+	memset(umr, 0, sizeof(*umr));
+
+	if (li) {
+		umr->mkey_mask = cpu_to_be64(MLX5_MKEY_MASK_FREE);
+		umr->flags = 1 << 7;
+		return;
+	}
+
+	umr->flags = (1 << 5); /* fail if not free */
+	umr->klm_octowords = get_klm_octo(wr->wr.fast_reg.page_list_len);
+	umr->mkey_mask = frwr_mkey_mask();
+}
+
+static void set_reg_umr_segment(struct mlx5_wqe_umr_ctrl_seg *umr,
+				struct ib_send_wr *wr)
+{
+	u64 mask;
+	struct umr_wr *umrwr = (struct umr_wr *)&wr->wr.fast_reg;
+
+	memset(umr, 0, sizeof(*umr));
+
+	if (!(wr->send_flags & IB_SEND_UMR_UNREG)) {
+		umr->flags = 1 << 5; /* fail if not free */
+		umr->klm_octowords = get_klm_octo(umrwr->npages);
+		mask =  MLX5_MKEY_MASK_LEN		|
+			MLX5_MKEY_MASK_PAGE_SIZE	|
+			MLX5_MKEY_MASK_START_ADDR	|
+			MLX5_MKEY_MASK_PD		|
+			MLX5_MKEY_MASK_LR		|
+			MLX5_MKEY_MASK_LW		|
+			MLX5_MKEY_MASK_RR		|
+			MLX5_MKEY_MASK_RW		|
+			MLX5_MKEY_MASK_A		|
+			MLX5_MKEY_MASK_FREE;
+		umr->mkey_mask = cpu_to_be64(mask);
+	} else {
+		umr->flags = 2 << 5; /* fail if free */
+		mask = MLX5_MKEY_MASK_FREE;
+		umr->mkey_mask = cpu_to_be64(mask);
+	}
+
+	if (!wr->num_sge)
+		umr->flags |= (1 << 7); /* inline */
+}
+
+static u8 get_umr_flags(int acc)
+{
+	return (acc & IB_ACCESS_REMOTE_ATOMIC ? MLX5_PERM_ATOMIC       : 0) |
+	       (acc & IB_ACCESS_REMOTE_WRITE  ? MLX5_PERM_REMOTE_WRITE : 0) |
+	       (acc & IB_ACCESS_REMOTE_READ   ? MLX5_PERM_REMOTE_READ  : 0) |
+	       (acc & IB_ACCESS_LOCAL_WRITE   ? MLX5_PERM_LOCAL_WRITE  : 0) |
+		MLX5_PERM_LOCAL_READ | MLX5_PERM_UMR_EN | MLX5_ACCESS_MODE_MTT;
+}
+
+static void set_mkey_segment(struct mlx5_mkey_seg *seg, struct ib_send_wr *wr,
+			     int li, int *writ)
+{
+	memset(seg, 0, sizeof(*seg));
+	if (li) {
+		seg->status = 1 << 6;
+		return;
+	}
+
+	seg->flags = get_umr_flags(wr->wr.fast_reg.access_flags);
+	*writ = seg->flags & (MLX5_PERM_LOCAL_WRITE | IB_ACCESS_REMOTE_WRITE);
+	seg->qpn_mkey7_0 = cpu_to_be32((wr->wr.fast_reg.rkey & 0xff) | 0xffffff00);
+	seg->flags_pd = cpu_to_be32(MLX5_MKEY_REMOTE_INVAL);
+	seg->start_addr = cpu_to_be64(wr->wr.fast_reg.iova_start);
+	seg->len = cpu_to_be64(wr->wr.fast_reg.length);
+	seg->xlt_oct_size = cpu_to_be32((wr->wr.fast_reg.page_list_len + 1) / 2);
+	seg->log2_page_size = wr->wr.fast_reg.page_shift;
+}
+
+static void set_reg_mkey_segment(struct mlx5_mkey_seg *seg, struct ib_send_wr *wr)
+{
+	memset(seg, 0, sizeof(*seg));
+	if (wr->send_flags & IB_SEND_UMR_UNREG) {
+		seg->status = 1 << 6;
+		return;
+	}
+
+	seg->flags = convert_access(wr->wr.fast_reg.access_flags);
+	seg->flags_pd = cpu_to_be32(to_mpd((struct ib_pd *)wr->wr.fast_reg.page_list)->pdn);
+	seg->start_addr = cpu_to_be64(wr->wr.fast_reg.iova_start);
+	seg->len = cpu_to_be64(wr->wr.fast_reg.length);
+	seg->log2_page_size = wr->wr.fast_reg.page_shift;
+	seg->qpn_mkey7_0 = cpu_to_be32(0xffffff << 8);
+}
+
+static void set_frwr_pages(struct mlx5_wqe_data_seg *dseg,
+			   struct ib_send_wr *wr,
+			   struct mlx5_core_dev *mdev,
+			   struct mlx5_ib_pd *pd,
+			   int writ)
+{
+	struct mlx5_ib_fast_reg_page_list *mfrpl = to_mfrpl(wr->wr.fast_reg.page_list);
+	int i;
+	u64 *page_list = wr->wr.fast_reg.page_list->page_list;
+	u64 perm = MLX5_EN_RD | (writ ? MLX5_EN_WR : 0);
+
+	for (i = 0; i < wr->wr.fast_reg.page_list_len; ++i)
+		mfrpl->mapped_page_list[i] = cpu_to_be64(page_list[i] | perm);
+	dseg->addr = cpu_to_be64(mfrpl->map);
+	dseg->byte_count = cpu_to_be32(ALIGN(sizeof(u64) * wr->wr.fast_reg.page_list_len, 64));
+	dseg->lkey = cpu_to_be32(pd->pa_lkey);
+}
+
+static __be32 send_ieth(struct ib_send_wr *wr)
+{
+	switch (wr->opcode) {
+	case IB_WR_SEND_WITH_IMM:
+	case IB_WR_RDMA_WRITE_WITH_IMM:
+		return wr->ex.imm_data;
+
+	case IB_WR_SEND_WITH_INV:
+		return cpu_to_be32(wr->ex.invalidate_rkey);
+
+	default:
+		return 0;
+	}
+}
+
+static u8 calc_sig(void *wqe, int size)
+{
+	int i;
+	u8 *p = wqe;
+	u8 res = 0;
+
+	for (i = 0; i < size; ++i)
+		res ^= p[i];
+
+	return ~res;
+}
+
+static u8 wq_sig(void *wqe)
+{
+	return calc_sig(wqe, (*((u8 *)wqe + 8) & 0x3f) << 4);
+}
+
+static int set_data_inl_seg(struct mlx5_ib_qp *qp, struct ib_send_wr *wr,
+			    void *wqe, int *sz)
+{
+	struct mlx5_wqe_inline_seg *seg;
+	void *addr;
+	int len;
+	int i;
+	int inl = 0;
+	void *qend = qp->sq.qend;
+	int copy;
+
+	seg = wqe;
+	wqe += sizeof(*seg);
+	for (i = 0; i < wr->num_sge; ++i) {
+		addr = (void *)(unsigned long)(wr->sg_list[i].addr);
+		len  = wr->sg_list[i].length;
+		inl += len;
+
+		if (unlikely(inl > qp->max_inline_data))
+			return -ENOMEM;
+
+		if (unlikely(wqe + len > qend)) {
+			copy = qend - wqe;
+			memcpy(wqe, addr, copy);
+			addr += copy;
+			len -= copy;
+			wqe = mlx5_get_send_wqe(qp, 0);
+		}
+		memcpy(wqe, addr, len);
+		wqe += len;
+	}
+
+	seg->byte_count = cpu_to_be32(inl | MLX5_INLINE_SEG);
+
+	*sz = ALIGN(inl + sizeof(seg->byte_count), 16) / 16;
+
+	return 0;
+}
+
+static int set_frwr_li_wr(void **seg, struct ib_send_wr *wr, int *size,
+			  struct mlx5_core_dev *mdev, struct mlx5_ib_pd *pd, struct mlx5_ib_qp *qp)
+{
+	int li;
+	int writ = 0;
+
+	li = wr->opcode == IB_WR_LOCAL_INV ? 1 : 0;
+	if (unlikely(wr->send_flags & IB_SEND_INLINE))
+		return -EINVAL;
+
+	set_frwr_umr_segment(*seg, wr, li);
+	*seg += sizeof(struct mlx5_wqe_umr_ctrl_seg);
+	*size += sizeof(struct mlx5_wqe_umr_ctrl_seg) / 16;
+	if (unlikely((*seg == qp->sq.qend)))
+		*seg = mlx5_get_send_wqe(qp, 0);
+	set_mkey_segment(*seg, wr, li, &writ);
+	*seg += sizeof(struct mlx5_mkey_seg);
+	*size += sizeof(struct mlx5_mkey_seg) / 16;
+	if (unlikely((*seg == qp->sq.qend)))
+		*seg = mlx5_get_send_wqe(qp, 0);
+	if (!li) {
+		set_frwr_pages(*seg, wr, mdev, pd, writ);
+		*seg += sizeof(struct mlx5_wqe_data_seg);
+		*size += (sizeof(struct mlx5_wqe_data_seg) / 16);
+	}
+	return 0;
+}
+
+static void dump_wqe(struct mlx5_ib_qp *qp, int idx, int size_16)
+{
+	__be32 *p = NULL;
+	int i, j;
+	int tidx = idx;
+
+	pr_debug("dump wqe at %p\n", mlx5_get_send_wqe(qp, tidx));
+	for (i = 0, j = 0; i < size_16 * 4; i += 4, j += 4) {
+		if ((i & 0xf) == 0) {
+			void *buf = mlx5_get_send_wqe(qp, tidx);
+			tidx = (tidx + 1) & (qp->sq.wqe_cnt - 1);
+			p = buf;
+			j = 0;
+		}
+		pr_debug("%08x %08x %08x %08x\n", be32_to_cpu(p[j]),
+			 be32_to_cpu(p[j + 1]), be32_to_cpu(p[j + 2]),
+			 be32_to_cpu(p[j + 3]));
+	}
+}
+
+static void mlx5_bf_copy(u64 __iomem *dst, u64 *src,
+			 unsigned bytecnt, struct mlx5_ib_qp *qp)
+{
+	while (bytecnt > 0) {
+		__iowrite64_copy(dst++, src++, 8);
+		__iowrite64_copy(dst++, src++, 8);
+		__iowrite64_copy(dst++, src++, 8);
+		__iowrite64_copy(dst++, src++, 8);
+		__iowrite64_copy(dst++, src++, 8);
+		__iowrite64_copy(dst++, src++, 8);
+		__iowrite64_copy(dst++, src++, 8);
+		__iowrite64_copy(dst++, src++, 8);
+		bytecnt -= 64;
+		if (unlikely(src == qp->sq.qend))
+			src = mlx5_get_send_wqe(qp, 0);
+	}
+}
+
+static u8 get_fence(u8 fence, struct ib_send_wr *wr)
+{
+	if (unlikely(wr->opcode == IB_WR_LOCAL_INV &&
+		     wr->send_flags & IB_SEND_FENCE))
+		return MLX5_FENCE_MODE_STRONG_ORDERING;
+
+	if (unlikely(fence)) {
+		if (wr->send_flags & IB_SEND_FENCE)
+			return MLX5_FENCE_MODE_SMALL_AND_FENCE;
+		else
+			return fence;
+
+	} else {
+		return 0;
+	}
+}
+
+int mlx5_ib_post_send(struct ib_qp *ibqp, struct ib_send_wr *wr,
+		      struct ib_send_wr **bad_wr)
+{
+	struct mlx5_ib_qp *qp = to_mqp(ibqp);
+	void *seg;
+	struct mlx5_wqe_ctrl_seg *ctrl = NULL;  /* compiler warning */
+	struct mlx5_wqe_data_seg *dpseg;
+	struct mlx5_ib_dev *dev = to_mdev(ibqp->device);
+	struct mlx5_core_dev *mdev = &dev->mdev;
+	struct mlx5_bf *bf = qp->bf;
+	unsigned long flags;
+	int nreq;
+	int err = 0;
+	int uninitialized_var(size);
+	int i;
+	u8 opmod = 0;
+	int inl = 0;
+	unsigned idx;
+	void *qend = qp->sq.qend;
+	u32 mlx5_opcode;
+	struct mlx5_wqe_xrc_seg *xrc;
+	u8 fence;
+	u8 next_fence = 0;
+	int num_sge;
+
+	spin_lock_irqsave(&qp->sq.lock, flags);
+
+	for (nreq = 0; wr; ++nreq, wr = wr->next) {
+		if (unlikely(wr->opcode >= sizeof(mlx5_ib_opcode) / sizeof(mlx5_ib_opcode[0]))) {
+			mlx5_ib_warn(dev, "\n");
+			err = -EINVAL;
+			*bad_wr = wr;
+			goto out;
+		}
+
+		if (unlikely(mlx5_wq_overflow(&qp->sq, nreq, qp->ibqp.send_cq))) {
+			mlx5_ib_warn(dev, "\n");
+			err = -ENOMEM;
+			*bad_wr = wr;
+			goto out;
+		}
+
+		fence = qp->fm_cache;
+		num_sge = wr->num_sge;
+		if (unlikely(num_sge > qp->sq.max_gs)) {
+			mlx5_ib_warn(dev, "\n");
+			err = -ENOMEM;
+			*bad_wr = wr;
+			goto out;
+		}
+
+		idx = qp->sq.cur_post & (qp->sq.wqe_cnt - 1);
+		seg = mlx5_get_send_wqe(qp, idx);
+		ctrl = seg;
+		*(uint32_t *)(seg + 8) = 0;
+		ctrl->imm = send_ieth(wr);
+		ctrl->fm_ce_se = qp->sq_signal_bits |
+			(wr->send_flags & IB_SEND_SIGNALED ?
+			 MLX5_WQE_CTRL_CQ_UPDATE : 0) |
+			(wr->send_flags & IB_SEND_SOLICITED ?
+			 MLX5_WQE_CTRL_SOLICITED : 0);
+
+		seg += sizeof(*ctrl);
+		size = sizeof(*ctrl) / 16;
+
+		switch (ibqp->qp_type) {
+		case IB_QPT_XRC_INI:
+			xrc = seg;
+			xrc->xrc_srqn = htonl(wr->xrc_remote_srq_num);
+			seg += sizeof(*xrc);
+			size += sizeof(*xrc) / 16;
+			/* fall through */
+		case IB_QPT_RC:
+			switch (wr->opcode) {
+			case IB_WR_RDMA_READ:
+			case IB_WR_RDMA_WRITE:
+			case IB_WR_RDMA_WRITE_WITH_IMM:
+				set_raddr_seg(seg, wr->wr.rdma.remote_addr,
+					      wr->wr.rdma.rkey);
+				seg  += sizeof(struct mlx5_wqe_raddr_seg);
+				size += sizeof(struct mlx5_wqe_raddr_seg) / 16;
+				break;
+
+			case IB_WR_ATOMIC_CMP_AND_SWP:
+			case IB_WR_ATOMIC_FETCH_AND_ADD:
+				set_raddr_seg(seg, wr->wr.atomic.remote_addr,
+					      wr->wr.atomic.rkey);
+				seg  += sizeof(struct mlx5_wqe_raddr_seg);
+
+				set_atomic_seg(seg, wr);
+				seg  += sizeof(struct mlx5_wqe_atomic_seg);
+
+				size += (sizeof(struct mlx5_wqe_raddr_seg) +
+					 sizeof(struct mlx5_wqe_atomic_seg)) / 16;
+				break;
+
+			case IB_WR_MASKED_ATOMIC_CMP_AND_SWP:
+				set_raddr_seg(seg, wr->wr.atomic.remote_addr,
+					      wr->wr.atomic.rkey);
+				seg  += sizeof(struct mlx5_wqe_raddr_seg);
+
+				set_masked_atomic_seg(seg, wr);
+				seg  += sizeof(struct mlx5_wqe_masked_atomic_seg);
+
+				size += (sizeof(struct mlx5_wqe_raddr_seg) +
+					 sizeof(struct mlx5_wqe_masked_atomic_seg)) / 16;
+				break;
+
+			case IB_WR_LOCAL_INV:
+				next_fence = MLX5_FENCE_MODE_INITIATOR_SMALL;
+				qp->sq.wr_data[idx] = IB_WR_LOCAL_INV;
+				ctrl->imm = cpu_to_be32(wr->ex.invalidate_rkey);
+				err = set_frwr_li_wr(&seg, wr, &size, mdev, to_mpd(ibqp->pd), qp);
+				if (err) {
+					mlx5_ib_warn(dev, "\n");
+					*bad_wr = wr;
+					goto out;
+				}
+				num_sge = 0;
+				break;
+
+			case IB_WR_FAST_REG_MR:
+				next_fence = MLX5_FENCE_MODE_INITIATOR_SMALL;
+				qp->sq.wr_data[idx] = IB_WR_FAST_REG_MR;
+				ctrl->imm = cpu_to_be32(wr->wr.fast_reg.rkey);
+				err = set_frwr_li_wr(&seg, wr, &size, mdev, to_mpd(ibqp->pd), qp);
+				if (err) {
+					mlx5_ib_warn(dev, "\n");
+					*bad_wr = wr;
+					goto out;
+				}
+				num_sge = 0;
+				break;
+
+			default:
+				break;
+			}
+			break;
+
+		case IB_QPT_UC:
+			switch (wr->opcode) {
+			case IB_WR_RDMA_WRITE:
+			case IB_WR_RDMA_WRITE_WITH_IMM:
+				set_raddr_seg(seg, wr->wr.rdma.remote_addr,
+					      wr->wr.rdma.rkey);
+				seg  += sizeof(struct mlx5_wqe_raddr_seg);
+				size += sizeof(struct mlx5_wqe_raddr_seg) / 16;
+				break;
+
+			default:
+				break;
+			}
+			break;
+
+		case IB_QPT_UD:
+		case IB_QPT_SMI:
+		case IB_QPT_GSI:
+			set_datagram_seg(seg, wr);
+			seg  += sizeof(struct mlx5_wqe_datagram_seg);
+			size += sizeof(struct mlx5_wqe_datagram_seg) / 16;
+			if (unlikely((seg == qend)))
+				seg = mlx5_get_send_wqe(qp, 0);
+			break;
+
+		case IB_QPT_REG_UMR:
+			if (wr->opcode != IB_WR_UMR) {
+				err = -EINVAL;
+				mlx5_ib_warn(dev, "bad opcode\n");
+				goto out;
+			}
+			qp->sq.wr_data[idx] = IB_WR_UMR;
+			ctrl->imm = cpu_to_be32(wr->wr.fast_reg.rkey);
+			set_reg_umr_segment(seg, wr);
+			seg += sizeof(struct mlx5_wqe_umr_ctrl_seg);
+			size += sizeof(struct mlx5_wqe_umr_ctrl_seg) / 16;
+			if (unlikely((seg == qend)))
+				seg = mlx5_get_send_wqe(qp, 0);
+			set_reg_mkey_segment(seg, wr);
+			seg += sizeof(struct mlx5_mkey_seg);
+			size += sizeof(struct mlx5_mkey_seg) / 16;
+			if (unlikely((seg == qend)))
+				seg = mlx5_get_send_wqe(qp, 0);
+			break;
+
+		default:
+			break;
+		}
+
+		if (wr->send_flags & IB_SEND_INLINE && num_sge) {
+			int uninitialized_var(sz);
+
+			err = set_data_inl_seg(qp, wr, seg, &sz);
+			if (unlikely(err)) {
+				mlx5_ib_warn(dev, "\n");
+				*bad_wr = wr;
+				goto out;
+			}
+			inl = 1;
+			size += sz;
+		} else {
+			dpseg = seg;
+			for (i = 0; i < num_sge; ++i) {
+				if (unlikely(dpseg == qend)) {
+					seg = mlx5_get_send_wqe(qp, 0);
+					dpseg = seg;
+				}
+				if (likely(wr->sg_list[i].length)) {
+					set_data_ptr_seg(dpseg, wr->sg_list + i);
+					size += sizeof(struct mlx5_wqe_data_seg) / 16;
+					++dpseg;
+				}
+			}
+		}
+
+		mlx5_opcode = mlx5_ib_opcode[wr->opcode];
+		ctrl->opmod_idx_opcode = cpu_to_be32(((u32)(qp->sq.cur_post) << 8)	|
+						     mlx5_opcode			|
+						     ((u32)opmod << 24));
+		ctrl->qpn_ds = cpu_to_be32(size | (qp->mqp.qpn << 8));
+		ctrl->fm_ce_se |= get_fence(fence, wr);
+		qp->fm_cache = next_fence;
+		if (unlikely(qp->wq_sig))
+			ctrl->signature = wq_sig(ctrl);
+
+		qp->sq.wrid[idx] = wr->wr_id;
+		qp->sq.w_list[idx].opcode = mlx5_opcode;
+		qp->sq.wqe_head[idx] = qp->sq.head + nreq;
+		qp->sq.cur_post += DIV_ROUND_UP(size * 16, MLX5_SEND_WQE_BB);
+		qp->sq.w_list[idx].next = qp->sq.cur_post;
+
+		if (0)
+			dump_wqe(qp, idx, size);
+	}
+
+out:
+	if (likely(nreq)) {
+		qp->sq.head += nreq;
+
+		/*
+		 * Make sure that descriptors are written before
+		 * updating doorbell record and ringing the doorbell
+		 */
+		wmb();
+
+		qp->db.db[MLX5_SND_DBR] = cpu_to_be32(qp->sq.cur_post);
+
+		if (bf->need_lock)
+			spin_lock(&bf->lock);
+
+		/* TBD enable WC */
+		if (0 && nreq == 1 && bf->uuarn && inl && size > 1 && size <= bf->buf_size / 16) {
+			mlx5_bf_copy(bf->reg + bf->offset, (u64 *)ctrl, ALIGN(size * 16, 64), qp);
+			/* wc_wmb(); */
+		} else {
+			mlx5_write64((__be32 *)ctrl, bf->regreg + bf->offset,
+				     MLX5_GET_DOORBELL_LOCK(&bf->lock32));
+			/*
+			 * Make sure doorbells don't leak out of SQ spinlock
+			 * and reach the HCA out of order.
+			 */
+			mmiowb();
+		}
+		bf->offset ^= bf->buf_size;
+		if (bf->need_lock)
+			spin_unlock(&bf->lock);
+	}
+
+	spin_unlock_irqrestore(&qp->sq.lock, flags);
+
+	return err;
+}
+
+static void set_sig_seg(struct mlx5_rwqe_sig *sig, int size)
+{
+	sig->signature = calc_sig(sig, size);
+}
+
+int mlx5_ib_post_recv(struct ib_qp *ibqp, struct ib_recv_wr *wr,
+		      struct ib_recv_wr **bad_wr)
+{
+	struct mlx5_ib_qp *qp = to_mqp(ibqp);
+	struct mlx5_wqe_data_seg *scat;
+	unsigned long flags;
+	int err = 0;
+	int nreq;
+	int ind;
+	int i;
+	struct mlx5_rwqe_sig *sig;
+
+
+	spin_lock_irqsave(&qp->rq.lock, flags);
+
+	ind = qp->rq.head & (qp->rq.wqe_cnt - 1);
+
+	for (nreq = 0; wr; ++nreq, wr = wr->next) {
+		if (mlx5_wq_overflow(&qp->rq, nreq, qp->ibqp.recv_cq)) {
+			err = -ENOMEM;
+			*bad_wr = wr;
+			goto out;
+		}
+
+		if (unlikely(wr->num_sge > qp->rq.max_gs)) {
+			err = -EINVAL;
+			*bad_wr = wr;
+			goto out;
+		}
+
+		scat = get_recv_wqe(qp, ind);
+		if (qp->wq_sig)
+			++scat;
+
+		for (i = 0; i < wr->num_sge; ++i)
+			set_data_ptr_seg(scat + i, wr->sg_list + i);
+
+		if (i < qp->rq.max_gs) {
+			scat[i].byte_count = 0;
+			scat[i].lkey       = cpu_to_be32(MLX5_INVALID_LKEY);
+			scat[i].addr       = 0;
+		}
+
+		if (qp->wq_sig) {
+			sig = (struct mlx5_rwqe_sig *)scat;
+			set_sig_seg(sig, (qp->rq.max_gs + 1) << 2);
+		}
+
+		qp->rq.wrid[ind] = wr->wr_id;
+
+		ind = (ind + 1) & (qp->rq.wqe_cnt - 1);
+	}
+
+out:
+	if (likely(nreq)) {
+		qp->rq.head += nreq;
+
+		/*
+		 * Make sure that descriptors are written before
+		 * doorbell record.
+		 */
+		wmb();
+
+		*qp->db.db = cpu_to_be32(qp->rq.head & 0xffff);
+	}
+
+	spin_unlock_irqrestore(&qp->rq.lock, flags);
+
+	return err;
+}
+
+static inline enum ib_qp_state to_ib_qp_state(enum mlx5_qp_state mlx5_state)
+{
+	switch (mlx5_state) {
+	case MLX5_QP_STATE_RST:      return IB_QPS_RESET;
+	case MLX5_QP_STATE_INIT:     return IB_QPS_INIT;
+	case MLX5_QP_STATE_RTR:      return IB_QPS_RTR;
+	case MLX5_QP_STATE_RTS:      return IB_QPS_RTS;
+	case MLX5_QP_STATE_SQ_DRAINING:
+	case MLX5_QP_STATE_SQD:      return IB_QPS_SQD;
+	case MLX5_QP_STATE_SQER:     return IB_QPS_SQE;
+	case MLX5_QP_STATE_ERR:      return IB_QPS_ERR;
+	default:		     return -1;
+	}
+}
+
+static inline enum ib_mig_state to_ib_mig_state(int mlx5_mig_state)
+{
+	switch (mlx5_mig_state) {
+	case MLX5_QP_PM_ARMED:		return IB_MIG_ARMED;
+	case MLX5_QP_PM_REARM:		return IB_MIG_REARM;
+	case MLX5_QP_PM_MIGRATED:	return IB_MIG_MIGRATED;
+	default: return -1;
+	}
+}
+
+static int to_ib_qp_access_flags(int mlx5_flags)
+{
+	int ib_flags = 0;
+
+	if (mlx5_flags & MLX5_QP_BIT_RRE)
+		ib_flags |= IB_ACCESS_REMOTE_READ;
+	if (mlx5_flags & MLX5_QP_BIT_RWE)
+		ib_flags |= IB_ACCESS_REMOTE_WRITE;
+	if (mlx5_flags & MLX5_QP_BIT_RAE)
+		ib_flags |= IB_ACCESS_REMOTE_ATOMIC;
+
+	return ib_flags;
+}
+
+static void to_ib_ah_attr(struct mlx5_ib_dev *ibdev, struct ib_ah_attr *ib_ah_attr,
+				struct mlx5_qp_path *path)
+{
+	struct mlx5_core_dev *dev = &ibdev->mdev;
+
+	memset(ib_ah_attr, 0, sizeof(*ib_ah_attr));
+	ib_ah_attr->port_num	  = path->port;
+
+	if (ib_ah_attr->port_num == 0 || ib_ah_attr->port_num > dev->caps.num_ports)
+		return;
+
+	ib_ah_attr->sl = path->sl & 0xf;
+
+	ib_ah_attr->dlid	  = be16_to_cpu(path->rlid);
+	ib_ah_attr->src_path_bits = path->grh_mlid & 0x7f;
+	ib_ah_attr->static_rate   = path->static_rate ? path->static_rate - 5 : 0;
+	ib_ah_attr->ah_flags      = (path->grh_mlid & (1 << 7)) ? IB_AH_GRH : 0;
+	if (ib_ah_attr->ah_flags) {
+		ib_ah_attr->grh.sgid_index = path->mgid_index;
+		ib_ah_attr->grh.hop_limit  = path->hop_limit;
+		ib_ah_attr->grh.traffic_class =
+			(be32_to_cpu(path->tclass_flowlabel) >> 20) & 0xff;
+		ib_ah_attr->grh.flow_label =
+			be32_to_cpu(path->tclass_flowlabel) & 0xfffff;
+		memcpy(ib_ah_attr->grh.dgid.raw,
+		       path->rgid, sizeof(ib_ah_attr->grh.dgid.raw));
+	}
+}
+
+int mlx5_ib_query_qp(struct ib_qp *ibqp, struct ib_qp_attr *qp_attr, int qp_attr_mask,
+		     struct ib_qp_init_attr *qp_init_attr)
+{
+	struct mlx5_ib_dev *dev = to_mdev(ibqp->device);
+	struct mlx5_ib_qp *qp = to_mqp(ibqp);
+	struct mlx5_query_qp_mbox_out *outb;
+	int mlx5_state;
+	int err = 0;
+	struct mlx5_qp_context *context;
+
+	mutex_lock(&qp->mutex);
+	outb = kzalloc(sizeof(*outb), GFP_KERNEL);
+	if (!outb) {
+		err = -ENOMEM;
+		goto out;
+	}
+	context = &outb->ctx;
+	err = mlx5_core_qp_query(&dev->mdev, &qp->mqp, outb, sizeof(*outb));
+	if (err)
+		goto out_free;
+
+	mlx5_state = be32_to_cpu(context->flags) >> 28;
+
+	qp->state		     = to_ib_qp_state(mlx5_state);
+	qp_attr->qp_state	     = qp->state;
+	qp_attr->path_mtu	     = context->mtu_msgmax >> 5;
+	qp_attr->path_mig_state	     =
+		to_ib_mig_state((be32_to_cpu(context->flags) >> 11) & 0x3);
+	qp_attr->qkey		     = be32_to_cpu(context->qkey);
+	qp_attr->rq_psn		     = be32_to_cpu(context->rnr_nextrecvpsn) & 0xffffff;
+	qp_attr->sq_psn		     = be32_to_cpu(context->next_send_psn) & 0xffffff;
+	qp_attr->dest_qp_num	     = be32_to_cpu(context->log_pg_sz_remote_qpn) & 0xffffff;
+	qp_attr->qp_access_flags     =
+		to_ib_qp_access_flags(be32_to_cpu(context->params2));
+
+	if (qp->ibqp.qp_type == IB_QPT_RC || qp->ibqp.qp_type == IB_QPT_UC) {
+		to_ib_ah_attr(dev, &qp_attr->ah_attr, &context->pri_path);
+		to_ib_ah_attr(dev, &qp_attr->alt_ah_attr, &context->alt_path);
+		qp_attr->alt_pkey_index = context->alt_path.pkey_index & 0x7f;
+		qp_attr->alt_port_num	= qp_attr->alt_ah_attr.port_num;
+	}
+
+	qp_attr->pkey_index = context->pri_path.pkey_index & 0x7f;
+	qp_attr->port_num = context->pri_path.port;
+
+	/* qp_attr->en_sqd_async_notify is only applicable in modify qp */
+	qp_attr->sq_draining = mlx5_state == MLX5_QP_STATE_SQ_DRAINING;
+
+	qp_attr->max_rd_atomic = 1 << ((be32_to_cpu(context->params1) >> 21) & 0x7);
+
+	qp_attr->max_dest_rd_atomic =
+		1 << ((be32_to_cpu(context->params2) >> 21) & 0x7);
+	qp_attr->min_rnr_timer	    =
+		(be32_to_cpu(context->rnr_nextrecvpsn) >> 24) & 0x1f;
+	qp_attr->timeout	    = context->pri_path.ackto_lt >> 3;
+	qp_attr->retry_cnt	    = (be32_to_cpu(context->params1) >> 16) & 0x7;
+	qp_attr->rnr_retry	    = (be32_to_cpu(context->params1) >> 13) & 0x7;
+	qp_attr->alt_timeout	    = context->alt_path.ackto_lt >> 3;
+	qp_attr->cur_qp_state	     = qp_attr->qp_state;
+	qp_attr->cap.max_recv_wr     = qp->rq.wqe_cnt;
+	qp_attr->cap.max_recv_sge    = qp->rq.max_gs;
+
+	if (!ibqp->uobject) {
+		qp_attr->cap.max_send_wr  = qp->sq.wqe_cnt;
+		qp_attr->cap.max_send_sge = qp->sq.max_gs;
+	} else {
+		qp_attr->cap.max_send_wr  = 0;
+		qp_attr->cap.max_send_sge = 0;
+	}
+
+	/*
+	 * We don't support inline sends for kernel QPs (yet), and we
+	 * don't know what userspace's value should be.
+	 */
+	qp_attr->cap.max_inline_data = 0;
+
+	qp_init_attr->cap	     = qp_attr->cap;
+
+	qp_init_attr->create_flags = 0;
+	if (qp->flags & MLX5_IB_QP_BLOCK_MULTICAST_LOOPBACK)
+		qp_init_attr->create_flags |= IB_QP_CREATE_BLOCK_MULTICAST_LOOPBACK;
+
+	qp_init_attr->sq_sig_type = qp->sq_signal_bits & MLX5_WQE_CTRL_CQ_UPDATE ?
+		IB_SIGNAL_ALL_WR : IB_SIGNAL_REQ_WR;
+
+out_free:
+	kfree(outb);
+
+out:
+	mutex_unlock(&qp->mutex);
+	return err;
+}
+
+struct ib_xrcd *mlx5_ib_alloc_xrcd(struct ib_device *ibdev,
+					  struct ib_ucontext *context,
+					  struct ib_udata *udata)
+{
+	struct mlx5_ib_xrcd *xrcd;
+	struct mlx5_ib_dev *dev = to_mdev(ibdev);
+	int err;
+
+	if (!(dev->mdev.caps.flags & MLX5_DEV_CAP_FLAG_XRC))
+		return ERR_PTR(-ENOSYS);
+
+	xrcd = kmalloc(sizeof(*xrcd), GFP_KERNEL);
+	if (!xrcd)
+		return ERR_PTR(-ENOMEM);
+
+	err = mlx5_core_xrcd_alloc(&dev->mdev, &xrcd->xrcdn);
+	if (err) {
+		kfree(xrcd);
+		return ERR_PTR(-ENOMEM);
+	}
+
+	return &xrcd->ibxrcd;
+}
+
+int mlx5_ib_dealloc_xrcd(struct ib_xrcd *xrcd)
+{
+	int err;
+	struct mlx5_ib_dev *dev = to_mdev(xrcd->device);
+	u32 xrcdn = to_mxrcd(xrcd)->xrcdn;
+
+	err = mlx5_core_xrcd_dealloc(&dev->mdev, xrcdn);
+	if (err) {
+		mlx5_ib_warn(dev, "failed to dealloc xrcdn 0x%x\n", xrcdn);
+		return err;
+	}
+
+	kfree(xrcd);
+
+	return 0;
+}
diff --git a/drivers/infiniband/hw/mlx5/srq.c b/drivers/infiniband/hw/mlx5/srq.c
new file mode 100644
index 0000000..63e3fdc
--- /dev/null
+++ b/drivers/infiniband/hw/mlx5/srq.c
@@ -0,0 +1,481 @@ 
+/*
+ * Copyright (c) 2013, Mellanox Technologies 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
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/module.h>
+#include <linux/mlx5/qp.h>
+#include <linux/mlx5/srq.h>
+#include <linux/slab.h>
+#include <linux/vmalloc.h>
+#include <rdma/ib_umem.h>
+
+#include "mlx5_ib.h"
+#include "user.h"
+
+MLX5_IB_MOD_DBG_MASK(MLX5_IB_MOD_SRQ);
+
+/* not supported currently */
+static int srq_signature;
+
+static void *get_wqe(struct mlx5_ib_srq *srq, int n)
+{
+	return mlx5_buf_offset(&srq->buf, n << srq->msrq.wqe_shift);
+}
+
+static void mlx5_ib_srq_event(struct mlx5_core_srq *srq, enum mlx5_event type)
+{
+	struct ib_event event;
+	struct ib_srq *ibsrq = &to_mibsrq(srq)->ibsrq;
+
+	if (ibsrq->event_handler) {
+		event.device      = ibsrq->device;
+		event.element.srq = ibsrq;
+		switch (type) {
+		case MLX5_EVENT_TYPE_SRQ_RQ_LIMIT:
+			event.event = IB_EVENT_SRQ_LIMIT_REACHED;
+			break;
+		case MLX5_EVENT_TYPE_SRQ_CATAS_ERROR:
+			event.event = IB_EVENT_SRQ_ERR;
+			break;
+		default:
+			pr_warn("mlx5_ib: Unexpected event type %d on SRQ %06x\n",
+				type, srq->srqn);
+			return;
+		}
+
+		ibsrq->event_handler(&event, ibsrq->srq_context);
+	}
+}
+
+static int create_srq_user(struct ib_pd *pd, struct mlx5_ib_srq *srq,
+			   struct mlx5_create_srq_mbox_in **in,
+			   struct ib_udata *udata, int buf_size, int *inlen)
+{
+	struct mlx5_ib_dev *dev = to_mdev(pd->device);
+	struct mlx5_ib_create_srq ucmd;
+	int err;
+	int npages;
+	int page_shift;
+	int ncont;
+	u32 offset;
+
+	if (ib_copy_from_udata(&ucmd, udata, sizeof(ucmd))) {
+		mlx5_ib_dbg(dev, "failed copy udata\n");
+		return -EFAULT;
+	}
+	srq->wq_sig = !!(ucmd.flags & MLX5_SRQ_FLAG_SIGNATURE);
+
+	srq->umem = ib_umem_get(pd->uobject->context, ucmd.buf_addr, buf_size,
+				0, 0);
+	if (IS_ERR(srq->umem)) {
+		mlx5_ib_dbg(dev, "failed umem get, size %d\n", buf_size);
+		err = PTR_ERR(srq->umem);
+		return err;
+	}
+
+	mlx5_ib_cont_pages(srq->umem, ucmd.buf_addr, &npages,
+			   &page_shift, &ncont, NULL);
+	err = mlx5_ib_get_buf_offset(ucmd.buf_addr, page_shift,
+				     &offset);
+	if (err) {
+		mlx5_ib_warn(dev, "bad offset\n");
+		goto err_umem;
+	}
+
+	*inlen = sizeof(**in) + sizeof(*(*in)->pas) * ncont;
+	*in = vzalloc(*inlen);
+	if (!(*in)) {
+		mlx5_ib_dbg(dev, "failed allocation, size %d\n", *inlen);
+		err = -ENOMEM;
+		goto err_umem;
+	}
+
+	mlx5_ib_populate_pas(dev, srq->umem, page_shift, (*in)->pas, 0);
+
+	err = mlx5_ib_db_map_user(to_mucontext(pd->uobject->context),
+				  ucmd.db_addr, &srq->db);
+	if (err) {
+		mlx5_ib_dbg(dev, "map doorbell failed\n");
+		goto err_in;
+	}
+
+	(*in)->ctx.log_pg_sz = page_shift - PAGE_SHIFT;
+	(*in)->ctx.pgoff_cqn = cpu_to_be32(offset << 26);
+
+	return 0;
+
+err_in:
+	vfree(*in);
+
+err_umem:
+	ib_umem_release(srq->umem);
+
+	return err;
+}
+
+static int create_srq_kernel(struct mlx5_ib_dev *dev, struct mlx5_ib_srq *srq,
+			     struct mlx5_create_srq_mbox_in **in, int buf_size,
+			     int *inlen)
+{
+	int err;
+	int i;
+	struct mlx5_wqe_srq_next_seg *next;
+	int page_shift;
+	int npages;
+
+	err = mlx5_db_alloc(&dev->mdev, &srq->db);
+	if (err) {
+		mlx5_ib_warn(dev, "alloc dbell rec failed\n");
+		return err;
+	}
+
+	*srq->db.db = 0;
+
+	if (mlx5_buf_alloc(&dev->mdev, buf_size, PAGE_SIZE * 2, &srq->buf)) {
+		mlx5_ib_dbg(dev, "buf alloc failed\n");
+		err = -ENOMEM;
+		goto err_db;
+	}
+	page_shift = srq->buf.page_shift;
+
+	srq->head    = 0;
+	srq->tail    = srq->msrq.max - 1;
+	srq->wqe_ctr = 0;
+
+	for (i = 0; i < srq->msrq.max; ++i) {
+		next = get_wqe(srq, i);
+		next->next_wqe_index =
+			cpu_to_be16((i + 1) & (srq->msrq.max - 1));
+	}
+
+	npages = DIV_ROUND_UP(srq->buf.npages, 1 << (page_shift - PAGE_SHIFT));
+	mlx5_ib_dbg(dev, "buf_size %d, page_shift %d, npages %d, calc npages %d\n",
+		    buf_size, page_shift, srq->buf.npages, npages);
+	*inlen = sizeof(**in) + sizeof(*(*in)->pas) * npages;
+	*in = vzalloc(*inlen);
+	if (!*in) {
+		mlx5_ib_dbg(dev, "vzalloc failed %d\n", *inlen);
+		err = -ENOMEM;
+		goto err_buf;
+	}
+	mlx5_fill_page_array(&srq->buf, (*in)->pas);
+
+	srq->wrid = kmalloc(srq->msrq.max * sizeof(u64), GFP_KERNEL);
+	if (!srq->wrid) {
+		mlx5_ib_dbg(dev, "kmalloc failed %lu\n",
+			    (unsigned long)(srq->msrq.max * sizeof(u64)));
+		err = -ENOMEM;
+		goto err_in;
+	}
+	srq->wq_sig = !!srq_signature;
+
+	(*in)->ctx.log_pg_sz = page_shift - PAGE_SHIFT;
+
+	return 0;
+
+err_in:
+	vfree(*in);
+
+err_buf:
+	mlx5_buf_free(&dev->mdev, &srq->buf);
+
+err_db:
+	mlx5_db_free(&dev->mdev, &srq->db);
+	return err;
+}
+
+static void destroy_srq_user(struct ib_pd *pd, struct mlx5_ib_srq *srq)
+{
+	mlx5_ib_db_unmap_user(to_mucontext(pd->uobject->context), &srq->db);
+	ib_umem_release(srq->umem);
+}
+
+
+static void destroy_srq_kernel(struct mlx5_ib_dev *dev, struct mlx5_ib_srq *srq)
+{
+	kfree(srq->wrid);
+	mlx5_buf_free(&dev->mdev, &srq->buf);
+	mlx5_db_free(&dev->mdev, &srq->db);
+}
+
+struct ib_srq *mlx5_ib_create_srq(struct ib_pd *pd,
+				  struct ib_srq_init_attr *init_attr,
+				  struct ib_udata *udata)
+{
+	struct mlx5_ib_dev *dev = to_mdev(pd->device);
+	struct mlx5_ib_srq *srq;
+	int desc_size;
+	int buf_size;
+	int err;
+	struct mlx5_create_srq_mbox_in *uninitialized_var(in);
+	int uninitialized_var(inlen);
+	int is_xrc;
+	u32 flgs, xrcdn;
+
+	/* Sanity check SRQ size before proceeding */
+	if (init_attr->attr.max_wr >= dev->mdev.caps.max_srq_wqes) {
+		mlx5_ib_dbg(dev, "max_wr %d, cap %d\n",
+			    init_attr->attr.max_wr,
+			    dev->mdev.caps.max_srq_wqes);
+		return ERR_PTR(-EINVAL);
+	}
+
+	srq = kmalloc(sizeof(*srq), GFP_KERNEL);
+	if (!srq) {
+		mlx5_ib_warn(dev, "failed allocation\n");
+		return ERR_PTR(-ENOMEM);
+	}
+
+	mutex_init(&srq->mutex);
+	spin_lock_init(&srq->lock);
+	srq->msrq.max    = roundup_pow_of_two(init_attr->attr.max_wr + 1);
+	srq->msrq.max_gs = init_attr->attr.max_sge;
+
+	desc_size = sizeof(struct mlx5_wqe_srq_next_seg) +
+		    srq->msrq.max_gs * sizeof(struct mlx5_wqe_data_seg);
+	desc_size = roundup_pow_of_two(desc_size);
+	desc_size = max_t(int, 32, desc_size);
+	srq->msrq.max_avail_gather = (desc_size - sizeof(struct mlx5_wqe_srq_next_seg)) /
+		sizeof(struct mlx5_wqe_data_seg);
+	srq->msrq.wqe_shift = ilog2(desc_size);
+	buf_size = srq->msrq.max * desc_size;
+	mlx5_ib_dbg(dev, "desc_size 0x%x, req wr 0x%x, srq size 0x%x, max_gs 0x%x, max_avail_gather 0x%x\n",
+		    desc_size, init_attr->attr.max_wr, srq->msrq.max, srq->msrq.max_gs,
+		    srq->msrq.max_avail_gather);
+
+	if (pd->uobject)
+		err = create_srq_user(pd, srq, &in, udata, buf_size, &inlen);
+	else
+		err = create_srq_kernel(dev, srq, &in, buf_size, &inlen);
+
+	if (err) {
+		mlx5_ib_warn(dev, "create srq %s failed, err %d\n",
+			     pd->uobject ? "user" : "kernel", err);
+		goto err_srq;
+	}
+
+	is_xrc = (init_attr->srq_type == IB_SRQT_XRC);
+	in->ctx.state_log_sz = ilog2(srq->msrq.max);
+	flgs = ((srq->msrq.wqe_shift - 4) | (is_xrc << 5) | (srq->wq_sig << 7)) << 24;
+	xrcdn = 0;
+	if (is_xrc) {
+		xrcdn = to_mxrcd(init_attr->ext.xrc.xrcd)->xrcdn;
+		in->ctx.pgoff_cqn |= cpu_to_be32(to_mcq(init_attr->ext.xrc.cq)->mcq.cqn);
+	} else if (init_attr->srq_type == IB_SRQT_BASIC) {
+		xrcdn = to_mxrcd(dev->devr.x0)->xrcdn;
+		in->ctx.pgoff_cqn |= cpu_to_be32(to_mcq(dev->devr.c0)->mcq.cqn);
+	}
+
+	in->ctx.flags_xrcd = cpu_to_be32((flgs & 0xFF000000) | (xrcdn & 0xFFFFFF));
+
+	in->ctx.pd = cpu_to_be32(to_mpd(pd)->pdn);
+	in->ctx.db_record = cpu_to_be64(srq->db.dma);
+	err = mlx5_core_create_srq(&dev->mdev, &srq->msrq, in, inlen);
+	vfree(in);
+	if (err) {
+		mlx5_ib_dbg(dev, "create SRQ failed, err %d\n", err);
+		goto err_srq;
+	}
+
+	mlx5_ib_dbg(dev, "create SRQ with srqn 0x%x\n", srq->msrq.srqn);
+
+	srq->msrq.event = mlx5_ib_srq_event;
+	srq->ibsrq.ext.xrc.srq_num = srq->msrq.srqn;
+
+	if (pd->uobject)
+		if (ib_copy_to_udata(udata, &srq->msrq.srqn, sizeof(__u32))) {
+			mlx5_ib_dbg(dev, "copy to user failed\n");
+			err = -EFAULT;
+			goto err_core;
+		}
+
+	init_attr->attr.max_wr = srq->msrq.max - 1;
+
+	return &srq->ibsrq;
+
+err_core:
+	mlx5_core_destroy_srq(&dev->mdev, &srq->msrq);
+	if (pd->uobject)
+		destroy_srq_user(pd, srq);
+	else
+		destroy_srq_kernel(dev, srq);
+
+err_srq:
+	kfree(srq);
+
+	return ERR_PTR(err);
+}
+
+int mlx5_ib_modify_srq(struct ib_srq *ibsrq, struct ib_srq_attr *attr,
+		       enum ib_srq_attr_mask attr_mask, struct ib_udata *udata)
+{
+	struct mlx5_ib_dev *dev = to_mdev(ibsrq->device);
+	struct mlx5_ib_srq *srq = to_msrq(ibsrq);
+	int ret;
+
+	/* We don't support resizing SRQs yet */
+	if (attr_mask & IB_SRQ_MAX_WR)
+		return -EINVAL;
+
+	if (attr_mask & IB_SRQ_LIMIT) {
+		if (attr->srq_limit >= srq->msrq.max)
+			return -EINVAL;
+
+		mutex_lock(&srq->mutex);
+		ret = mlx5_core_arm_srq(&dev->mdev, &srq->msrq, attr->srq_limit, 1);
+		mutex_unlock(&srq->mutex);
+
+		if (ret)
+			return ret;
+	}
+
+	return 0;
+}
+
+int mlx5_ib_query_srq(struct ib_srq *ibsrq, struct ib_srq_attr *srq_attr)
+{
+	struct mlx5_ib_dev *dev = to_mdev(ibsrq->device);
+	struct mlx5_ib_srq *srq = to_msrq(ibsrq);
+	int ret;
+	struct mlx5_query_srq_mbox_out *out;
+
+	out = kzalloc(sizeof(*out), GFP_KERNEL);
+	if (!out)
+		return -ENOMEM;
+
+	ret = mlx5_core_query_srq(&dev->mdev, &srq->msrq, out);
+	if (ret)
+		goto out_box;
+
+	srq_attr->srq_limit = be16_to_cpu(out->ctx.lwm);
+	srq_attr->max_wr    = srq->msrq.max - 1;
+	srq_attr->max_sge   = srq->msrq.max_gs;
+
+out_box:
+	kfree(out);
+	return ret;
+}
+
+int mlx5_ib_destroy_srq(struct ib_srq *srq)
+{
+	struct mlx5_ib_dev *dev = to_mdev(srq->device);
+	struct mlx5_ib_srq *msrq = to_msrq(srq);
+
+	mlx5_core_destroy_srq(&dev->mdev, &msrq->msrq);
+
+	if (srq->uobject) {
+		mlx5_ib_db_unmap_user(to_mucontext(srq->uobject->context), &msrq->db);
+		ib_umem_release(msrq->umem);
+	} else {
+		kfree(msrq->wrid);
+		mlx5_buf_free(&dev->mdev, &msrq->buf);
+		mlx5_db_free(&dev->mdev, &msrq->db);
+	}
+
+	kfree(srq);
+	return 0;
+}
+
+void mlx5_ib_free_srq_wqe(struct mlx5_ib_srq *srq, int wqe_index)
+{
+	struct mlx5_wqe_srq_next_seg *next;
+
+	/* always called with interrupts disabled. */
+	spin_lock(&srq->lock);
+
+	next = get_wqe(srq, srq->tail);
+	next->next_wqe_index = cpu_to_be16(wqe_index);
+	srq->tail = wqe_index;
+
+	spin_unlock(&srq->lock);
+}
+
+int mlx5_ib_post_srq_recv(struct ib_srq *ibsrq, struct ib_recv_wr *wr,
+			  struct ib_recv_wr **bad_wr)
+{
+	struct mlx5_ib_srq *srq = to_msrq(ibsrq);
+	struct mlx5_wqe_srq_next_seg *next;
+	struct mlx5_wqe_data_seg *scat;
+	unsigned long flags;
+	int err = 0;
+	int nreq;
+	int i;
+
+	spin_lock_irqsave(&srq->lock, flags);
+
+	for (nreq = 0; wr; ++nreq, wr = wr->next) {
+		if (unlikely(wr->num_sge > srq->msrq.max_gs)) {
+			err = -EINVAL;
+			*bad_wr = wr;
+			break;
+		}
+
+		if (unlikely(srq->head == srq->tail)) {
+			err = -ENOMEM;
+			*bad_wr = wr;
+			break;
+		}
+
+		srq->wrid[srq->head] = wr->wr_id;
+
+		next      = get_wqe(srq, srq->head);
+		srq->head = be16_to_cpu(next->next_wqe_index);
+		scat      = (struct mlx5_wqe_data_seg *)(next + 1);
+
+		for (i = 0; i < wr->num_sge; ++i) {
+			scat[i].byte_count = cpu_to_be32(wr->sg_list[i].length);
+			scat[i].lkey       = cpu_to_be32(wr->sg_list[i].lkey);
+			scat[i].addr       = cpu_to_be64(wr->sg_list[i].addr);
+		}
+
+		if (i < srq->msrq.max_avail_gather) {
+			scat[i].byte_count = 0;
+			scat[i].lkey       = cpu_to_be32(MLX5_INVALID_LKEY);
+			scat[i].addr       = 0;
+		}
+	}
+
+	if (likely(nreq)) {
+		srq->wqe_ctr += nreq;
+
+		/*
+		 * Make sure that descriptors are written before
+		 * doorbell record.
+		 */
+		wmb();
+
+		*srq->db.db = cpu_to_be32(srq->wqe_ctr);
+	}
+
+	spin_unlock_irqrestore(&srq->lock, flags);
+
+	return err;
+}
diff --git a/drivers/infiniband/hw/mlx5/user.h b/drivers/infiniband/hw/mlx5/user.h
new file mode 100644
index 0000000..fa58c39
--- /dev/null
+++ b/drivers/infiniband/hw/mlx5/user.h
@@ -0,0 +1,123 @@ 
+/*
+ * Copyright (c) 2013, Mellanox Technologies 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
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#ifndef MLX5_IB_USER_H
+#define MLX5_IB_USER_H
+
+#include <linux/types.h>
+
+enum {
+	MLX5_QP_FLAG_SIGNATURE		= 1 << 0,
+	MLX5_QP_FLAG_SCATTER_CQE	= 1 << 1,
+};
+
+enum {
+	MLX5_SRQ_FLAG_SIGNATURE		= 1 << 0,
+};
+
+
+/*
+ * Increment this value if any changes that break userspace ABI
+ * compatibility are made.
+ */
+#define MLX5_IB_UVERBS_ABI_VERSION	1
+
+/*
+ * Make sure that all structs defined in this file remain laid out so
+ * that they pack the same way on 32-bit and 64-bit architectures (to
+ * avoid incompatibility between 32-bit userspace and 64-bit kernels).
+ * In particular do not use pointer types -- pass pointers in __u64
+ * instead.
+ */
+
+struct mlx5_ib_alloc_ucontext_req {
+	__u32	total_num_uuars;
+	__u32	num_low_latency_uuars;
+};
+
+struct mlx5_ib_alloc_ucontext_resp {
+	__u32	qp_tab_size;
+	__u32	bf_reg_size;
+	__u32	tot_uuars;
+	__u32	cache_line_size;
+	__u16	max_sq_desc_sz;
+	__u16	max_rq_desc_sz;
+	__u32	max_send_wqebb;
+	__u32	max_recv_wr;
+	__u32	max_srq_recv_wr;
+	__u16	num_ports;
+	__u16	reserved;
+};
+
+struct mlx5_ib_alloc_pd_resp {
+	__u32	pdn;
+};
+
+struct mlx5_ib_create_cq {
+	__u64	buf_addr;
+	__u64	db_addr;
+	__u32	cqe_size;
+};
+
+struct mlx5_ib_create_cq_resp {
+	__u32	cqn;
+	__u32	reserved;
+};
+
+struct mlx5_ib_resize_cq {
+	__u64	buf_addr;
+};
+
+struct mlx5_ib_create_srq {
+	__u64	buf_addr;
+	__u64	db_addr;
+	__u32	flags;
+};
+
+struct mlx5_ib_create_srq_resp {
+	__u32	srqn;
+	__u32	reserved;
+};
+
+struct mlx5_ib_create_qp {
+	__u64	buf_addr;
+	__u64	db_addr;
+	__u32	sq_wqe_count;
+	__u32	rq_wqe_count;
+	__u32	rq_wqe_shift;
+	__u32	flags;
+};
+
+struct mlx5_ib_create_qp_resp {
+	__u32	uuar_index;
+};
+#endif /* MLX5_IB_USER_H */