Merge tag 'net-next-6.3' of git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net-nextgrafted

Pull networking updates from Jakub Kicinski:
 "Core:

   - Add dedicated kmem_cache for typical/small skb->head, avoid having
     to access struct page at kfree time, and improve memory use.

   - Introduce sysctl to set default RPS configuration for new netdevs.

   - Define Netlink protocol specification format which can be used to
     describe messages used by each family and auto-generate parsers.
     Add tools for generating kernel data structures and uAPI headers.

   - Expose all net/core sysctls inside netns.

   - Remove 4s sleep in netpoll if carrier is instantly detected on
     boot.

   - Add configurable limit of MDB entries per port, and port-vlan.

   - Continue populating drop reasons throughout the stack.

   - Retire a handful of legacy Qdiscs and classifiers.

  Protocols:

   - Support IPv4 big TCP (TSO frames larger than 64kB).

   - Add IP_LOCAL_PORT_RANGE socket option, to control local port range
     on socket by socket basis.

   - Track and report in procfs number of MPTCP sockets used.

   - Support mixing IPv4 and IPv6 flows in the in-kernel MPTCP path
     manager.

   - IPv6: don't check net.ipv6.route.max_size and rely on garbage
     collection to free memory (similarly to IPv4).

   - Support Penultimate Segment Pop (PSP) flavor in SRv6 (RFC8986).

   - ICMP: add per-rate limit counters.

   - Add support for user scanning requests in ieee802154.

   - Remove static WEP support.

   - Support minimal Wi-Fi 7 Extremely High Throughput (EHT) rate
     reporting.

   - WiFi 7 EHT channel puncturing support (client & AP).

  BPF:

   - Add a rbtree data structure following the "next-gen data structure"
     precedent set by recently added linked list, that is, by using
     kfunc + kptr instead of adding a new BPF map type.

   - Expose XDP hints via kfuncs with initial support for RX hash and
     timestamp metadata.

   - Add BPF_F_NO_TUNNEL_KEY extension to bpf_skb_set_tunnel_key to
     better support decap on GRE tunnel devices not operating in collect
     metadata.

   - Improve x86 JIT's codegen for PROBE_MEM runtime error checks.

   - Remove the need for trace_printk_lock for bpf_trace_printk and
     bpf_trace_vprintk helpers.

   - Extend libbpf's bpf_tracing.h support for tracing arguments of
     kprobes/uprobes and syscall as a special case.

   - Significantly reduce the search time for module symbols by
     livepatch and BPF.

   - Enable cpumasks to be used as kptrs, which is useful for tracing
     programs tracking which tasks end up running on which CPUs in
     different time intervals.

   - Add support for BPF trampoline on s390x and riscv64.

   - Add capability to export the XDP features supported by the NIC.

   - Add __bpf_kfunc tag for marking kernel functions as kfuncs.

   - Add cgroup.memory=nobpf kernel parameter option to disable BPF
     memory accounting for container environments.

  Netfilter:

   - Remove the CLUSTERIP target. It has been marked as obsolete for
     years, and we still have WARN splats wrt races of the out-of-band
     /proc interface installed by this target.

   - Add 'destroy' commands to nf_tables. They are identical to the
     existing 'delete' commands, but do not return an error if the
     referenced object (set, chain, rule...) did not exist.

  Driver API:

   - Improve cpumask_local_spread() locality to help NICs set the right
     IRQ affinity on AMD platforms.

   - Separate C22 and C45 MDIO bus transactions more clearly.

   - Introduce new DCB table to control DSCP rewrite on egress.

   - Support configuration of Physical Layer Collision Avoidance (PLCA)
     Reconciliation Sublayer (RS) (802.3cg-2019). Modern version of
     shared medium Ethernet.

   - Support for MAC Merge layer (IEEE 802.3-2018 clause 99). Allowing
     preemption of low priority frames by high priority frames.

   - Add support for controlling MACSec offload using netlink SET.

   - Rework devlink instance refcounts to allow registration and
     de-registration under the instance lock. Split the code into
     multiple files, drop some of the unnecessarily granular locks and
     factor out common parts of netlink operation handling.

   - Add TX frame aggregation parameters (for USB drivers).

   - Add a new attr TCA_EXT_WARN_MSG to report TC (offload) warning
     messages with notifications for debug.

   - Allow offloading of UDP NEW connections via act_ct.

   - Add support for per action HW stats in TC.

   - Support hardware miss to TC action (continue processing in SW from
     a specific point in the action chain).

   - Warn if old Wireless Extension user space interface is used with
     modern cfg80211/mac80211 drivers. Do not support Wireless
     Extensions for Wi-Fi 7 devices at all. Everyone should switch to
     using nl80211 interface instead.

   - Improve the CAN bit timing configuration. Use extack to return
     error messages directly to user space, update the SJW handling,
     including the definition of a new default value that will benefit
     CAN-FD controllers, by increasing their oscillator tolerance.

  New hardware / drivers:

   - Ethernet:
      - nVidia BlueField-3 support (control traffic driver)
      - Ethernet support for imx93 SoCs
      - Motorcomm yt8531 gigabit Ethernet PHY
      - onsemi NCN26000 10BASE-T1S PHY (with support for PLCA)
      - Microchip LAN8841 PHY (incl. cable diagnostics and PTP)
      - Amlogic gxl MDIO mux

   - WiFi:
      - RealTek RTL8188EU (rtl8xxxu)
      - Qualcomm Wi-Fi 7 devices (ath12k)

   - CAN:
      - Renesas R-Car V4H

  Drivers:

   - Bluetooth:
      - Set Per Platform Antenna Gain (PPAG) for Intel controllers.

   - Ethernet NICs:
      - Intel (1G, igc):
         - support TSN / Qbv / packet scheduling features of i226 model
      - Intel (100G, ice):
         - use GNSS subsystem instead of TTY
         - multi-buffer XDP support
         - extend support for GPIO pins to E823 devices
      - nVidia/Mellanox:
         - update the shared buffer configuration on PFC commands
         - implement PTP adjphase function for HW offset control
         - TC support for Geneve and GRE with VF tunnel offload
         - more efficient crypto key management method
         - multi-port eswitch support
      - Netronome/Corigine:
         - add DCB IEEE support
         - support IPsec offloading for NFP3800
      - Freescale/NXP (enetc):
         - support XDP_REDIRECT for XDP non-linear buffers
         - improve reconfig, avoid link flap and waiting for idle
         - support MAC Merge layer
      - Other NICs:
         - sfc/ef100: add basic devlink support for ef100
         - ionic: rx_push mode operation (writing descriptors via MMIO)
         - bnxt: use the auxiliary bus abstraction for RDMA
         - r8169: disable ASPM and reset bus in case of tx timeout
         - cpsw: support QSGMII mode for J721e CPSW9G
         - cpts: support pulse-per-second output
         - ngbe: add an mdio bus driver
         - usbnet: optimize usbnet_bh() by avoiding unnecessary queuing
         - r8152: handle devices with FW with NCM support
         - amd-xgbe: support 10Mbps, 2.5GbE speeds and rx-adaptation
         - virtio-net: support multi buffer XDP
         - virtio/vsock: replace virtio_vsock_pkt with sk_buff
         - tsnep: XDP support

   - Ethernet high-speed switches:
      - nVidia/Mellanox (mlxsw):
         - add support for latency TLV (in FW control messages)
      - Microchip (sparx5):
         - separate explicit and implicit traffic forwarding rules, make
           the implicit rules always active
         - add support for egress DSCP rewrite
         - IS0 VCAP support (Ingress Classification)
         - IS2 VCAP filters (protos, L3 addrs, L4 ports, flags, ToS
           etc.)
         - ES2 VCAP support (Egress Access Control)
         - support for Per-Stream Filtering and Policing (802.1Q,
           8.6.5.1)

   - Ethernet embedded switches:
      - Marvell (mv88e6xxx):
         - add MAB (port auth) offload support
         - enable PTP receive for mv88e6390
      - NXP (ocelot):
         - support MAC Merge layer
         - support for the the vsc7512 internal copper phys
      - Microchip:
         - lan9303: convert to PHYLINK
         - lan966x: support TC flower filter statistics
         - lan937x: PTP support for KSZ9563/KSZ8563 and LAN937x
         - lan937x: support Credit Based Shaper configuration
         - ksz9477: support Energy Efficient Ethernet
      - other:
         - qca8k: convert to regmap read/write API, use bulk operations
         - rswitch: Improve TX timestamp accuracy

   - Intel WiFi (iwlwifi):
      - EHT (Wi-Fi 7) rate reporting
      - STEP equalizer support: transfer some STEP (connection to radio
        on platforms with integrated wifi) related parameters from the
        BIOS to the firmware.

   - Qualcomm 802.11ax WiFi (ath11k):
      - IPQ5018 support
      - Fine Timing Measurement (FTM) responder role support
      - channel 177 support

   - MediaTek WiFi (mt76):
      - per-PHY LED support
      - mt7996: EHT (Wi-Fi 7) support
      - Wireless Ethernet Dispatch (WED) reset support
      - switch to using page pool allocator

   - RealTek WiFi (rtw89):
      - support new version of Bluetooth co-existance

   - Mobile:
      - rmnet: support TX aggregation"

* tag 'net-next-6.3' of git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net-next: (1872 commits)
  page_pool: add a comment explaining the fragment counter usage
  net: ethtool: fix __ethtool_dev_mm_supported() implementation
  ethtool: pse-pd: Fix double word in comments
  xsk: add linux/vmalloc.h to xsk.c
  sefltests: netdevsim: wait for devlink instance after netns removal
  selftest: fib_tests: Always cleanup before exit
  net/mlx5e: Align IPsec ASO result memory to be as required by hardware
  net/mlx5e: TC, Set CT miss to the specific ct action instance
  net/mlx5e: Rename CHAIN_TO_REG to MAPPED_OBJ_TO_REG
  net/mlx5: Refactor tc miss handling to a single function
  net/mlx5: Kconfig: Make tc offload depend on tc skb extension
  net/sched: flower: Support hardware miss to tc action
  net/sched: flower: Move filter handle initialization earlier
  net/sched: cls_api: Support hardware miss to tc action
  net/sched: Rename user cookie and act cookie
  sfc: fix builds without CONFIG_RTC_LIB
  sfc: clean up some inconsistent indentings
  net/mlx4_en: Introduce flexible array to silence overflow warning
  net: lan966x: Fix possible deadlock inside PTP
  net/ulp: Remove redundant ->clone() test in inet_clone_ulp().
  ...

1 files changed, 1054 insertions, 0 deletions

diff --git a/drivers/infiniband/hw/vmw_pvrdma/pvrdma_qp.c b/drivers/infiniband/hw/vmw_pvrdma/pvrdma_qp.c
new file mode 100644
index 000000000..f83cd4a9d
--- /dev/null
+++ b/drivers/infiniband/hw/vmw_pvrdma/pvrdma_qp.c
@@ -0,0 +1,1054 @@
+/*
+ * Copyright (c) 2012-2016 VMware, Inc.  All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of EITHER the GNU General Public License
+ * version 2 as published by the Free Software Foundation or the BSD
+ * 2-Clause License. This program is distributed in the hope that it
+ * will be useful, but WITHOUT ANY WARRANTY; WITHOUT EVEN THE IMPLIED
+ * WARRANTY OF MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE.
+ * See the GNU General Public License version 2 for more details at
+ * http://www.gnu.org/licenses/old-licenses/gpl-2.0.en.html.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program available in the file COPYING in the main
+ * directory of this source tree.
+ *
+ * The BSD 2-Clause License
+ *
+ *     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.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS
+ * FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE
+ * COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT,
+ * INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
+ * (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
+ * SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
+ * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT,
+ * STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
+ * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED
+ * OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <asm/page.h>
+#include <linux/io.h>
+#include <linux/wait.h>
+#include <rdma/ib_addr.h>
+#include <rdma/ib_smi.h>
+#include <rdma/ib_user_verbs.h>
+
+#include "pvrdma.h"
+
+static void __pvrdma_destroy_qp(struct pvrdma_dev *dev,
+				struct pvrdma_qp *qp);
+
+static inline void get_cqs(struct pvrdma_qp *qp, struct pvrdma_cq **send_cq,
+			   struct pvrdma_cq **recv_cq)
+{
+	*send_cq = to_vcq(qp->ibqp.send_cq);
+	*recv_cq = to_vcq(qp->ibqp.recv_cq);
+}
+
+static void pvrdma_lock_cqs(struct pvrdma_cq *scq, struct pvrdma_cq *rcq,
+			    unsigned long *scq_flags,
+			    unsigned long *rcq_flags)
+	__acquires(scq->cq_lock) __acquires(rcq->cq_lock)
+{
+	if (scq == rcq) {
+		spin_lock_irqsave(&scq->cq_lock, *scq_flags);
+		__acquire(rcq->cq_lock);
+	} else if (scq->cq_handle < rcq->cq_handle) {
+		spin_lock_irqsave(&scq->cq_lock, *scq_flags);
+		spin_lock_irqsave_nested(&rcq->cq_lock, *rcq_flags,
+					 SINGLE_DEPTH_NESTING);
+	} else {
+		spin_lock_irqsave(&rcq->cq_lock, *rcq_flags);
+		spin_lock_irqsave_nested(&scq->cq_lock, *scq_flags,
+					 SINGLE_DEPTH_NESTING);
+	}
+}
+
+static void pvrdma_unlock_cqs(struct pvrdma_cq *scq, struct pvrdma_cq *rcq,
+			      unsigned long *scq_flags,
+			      unsigned long *rcq_flags)
+	__releases(scq->cq_lock) __releases(rcq->cq_lock)
+{
+	if (scq == rcq) {
+		__release(rcq->cq_lock);
+		spin_unlock_irqrestore(&scq->cq_lock, *scq_flags);
+	} else if (scq->cq_handle < rcq->cq_handle) {
+		spin_unlock_irqrestore(&rcq->cq_lock, *rcq_flags);
+		spin_unlock_irqrestore(&scq->cq_lock, *scq_flags);
+	} else {
+		spin_unlock_irqrestore(&scq->cq_lock, *scq_flags);
+		spin_unlock_irqrestore(&rcq->cq_lock, *rcq_flags);
+	}
+}
+
+static void pvrdma_reset_qp(struct pvrdma_qp *qp)
+{
+	struct pvrdma_cq *scq, *rcq;
+	unsigned long scq_flags, rcq_flags;
+
+	/* Clean up cqes */
+	get_cqs(qp, &scq, &rcq);
+	pvrdma_lock_cqs(scq, rcq, &scq_flags, &rcq_flags);
+
+	_pvrdma_flush_cqe(qp, scq);
+	if (scq != rcq)
+		_pvrdma_flush_cqe(qp, rcq);
+
+	pvrdma_unlock_cqs(scq, rcq, &scq_flags, &rcq_flags);
+
+	/*
+	 * Reset queuepair. The checks are because usermode queuepairs won't
+	 * have kernel ringstates.
+	 */
+	if (qp->rq.ring) {
+		atomic_set(&qp->rq.ring->cons_head, 0);
+		atomic_set(&qp->rq.ring->prod_tail, 0);
+	}
+	if (qp->sq.ring) {
+		atomic_set(&qp->sq.ring->cons_head, 0);
+		atomic_set(&qp->sq.ring->prod_tail, 0);
+	}
+}
+
+static int pvrdma_set_rq_size(struct pvrdma_dev *dev,
+			      struct ib_qp_cap *req_cap,
+			      struct pvrdma_qp *qp)
+{
+	if (req_cap->max_recv_wr > dev->dsr->caps.max_qp_wr ||
+	    req_cap->max_recv_sge > dev->dsr->caps.max_sge) {
+		dev_warn(&dev->pdev->dev, "recv queue size invalid\n");
+		return -EINVAL;
+	}
+
+	qp->rq.wqe_cnt = roundup_pow_of_two(max(1U, req_cap->max_recv_wr));
+	qp->rq.max_sg = roundup_pow_of_two(max(1U, req_cap->max_recv_sge));
+
+	/* Write back */
+	req_cap->max_recv_wr = qp->rq.wqe_cnt;
+	req_cap->max_recv_sge = qp->rq.max_sg;
+
+	qp->rq.wqe_size = roundup_pow_of_two(sizeof(struct pvrdma_rq_wqe_hdr) +
+					     sizeof(struct pvrdma_sge) *
+					     qp->rq.max_sg);
+	qp->npages_recv = (qp->rq.wqe_cnt * qp->rq.wqe_size + PAGE_SIZE - 1) /
+			  PAGE_SIZE;
+
+	return 0;
+}
+
+static int pvrdma_set_sq_size(struct pvrdma_dev *dev, struct ib_qp_cap *req_cap,
+			      struct pvrdma_qp *qp)
+{
+	if (req_cap->max_send_wr > dev->dsr->caps.max_qp_wr ||
+	    req_cap->max_send_sge > dev->dsr->caps.max_sge) {
+		dev_warn(&dev->pdev->dev, "send queue size invalid\n");
+		return -EINVAL;
+	}
+
+	qp->sq.wqe_cnt = roundup_pow_of_two(max(1U, req_cap->max_send_wr));
+	qp->sq.max_sg = roundup_pow_of_two(max(1U, req_cap->max_send_sge));
+
+	/* Write back */
+	req_cap->max_send_wr = qp->sq.wqe_cnt;
+	req_cap->max_send_sge = qp->sq.max_sg;
+
+	qp->sq.wqe_size = roundup_pow_of_two(sizeof(struct pvrdma_sq_wqe_hdr) +
+					     sizeof(struct pvrdma_sge) *
+					     qp->sq.max_sg);
+	/* Note: one extra page for the header. */
+	qp->npages_send = PVRDMA_QP_NUM_HEADER_PAGES +
+			  (qp->sq.wqe_cnt * qp->sq.wqe_size + PAGE_SIZE - 1) /
+								PAGE_SIZE;
+
+	return 0;
+}
+
+/**
+ * pvrdma_create_qp - create queue pair
+ * @ibqp: queue pair
+ * @init_attr: queue pair attributes
+ * @udata: user data
+ *
+ * @return: the 0 on success, otherwise returns an errno.
+ */
+int pvrdma_create_qp(struct ib_qp *ibqp, struct ib_qp_init_attr *init_attr,
+		     struct ib_udata *udata)
+{
+	struct pvrdma_qp *qp = to_vqp(ibqp);
+	struct pvrdma_dev *dev = to_vdev(ibqp->device);
+	union pvrdma_cmd_req req;
+	union pvrdma_cmd_resp rsp;
+	struct pvrdma_cmd_create_qp *cmd = &req.create_qp;
+	struct pvrdma_cmd_create_qp_resp *resp = &rsp.create_qp_resp;
+	struct pvrdma_cmd_create_qp_resp_v2 *resp_v2 = &rsp.create_qp_resp_v2;
+	struct pvrdma_create_qp ucmd;
+	struct pvrdma_create_qp_resp qp_resp = {};
+	unsigned long flags;
+	int ret;
+	bool is_srq = !!init_attr->srq;
+
+	if (init_attr->create_flags) {
+		dev_warn(&dev->pdev->dev,
+			 "invalid create queuepair flags %#x\n",
+			 init_attr->create_flags);
+		return -EOPNOTSUPP;
+	}
+
+	if (init_attr->qp_type != IB_QPT_RC &&
+	    init_attr->qp_type != IB_QPT_UD &&
+	    init_attr->qp_type != IB_QPT_GSI) {
+		dev_warn(&dev->pdev->dev, "queuepair type %d not supported\n",
+			 init_attr->qp_type);
+		return -EOPNOTSUPP;
+	}
+
+	if (is_srq && !dev->dsr->caps.max_srq) {
+		dev_warn(&dev->pdev->dev,
+			 "SRQs not supported by device\n");
+		return -EINVAL;
+	}
+
+	if (!atomic_add_unless(&dev->num_qps, 1, dev->dsr->caps.max_qp))
+		return -ENOMEM;
+
+	switch (init_attr->qp_type) {
+	case IB_QPT_GSI:
+		if (init_attr->port_num == 0 ||
+		    init_attr->port_num > ibqp->device->phys_port_cnt) {
+			dev_warn(&dev->pdev->dev, "invalid queuepair attrs\n");
+			ret = -EINVAL;
+			goto err_qp;
+		}
+		fallthrough;
+	case IB_QPT_RC:
+	case IB_QPT_UD:
+		spin_lock_init(&qp->sq.lock);
+		spin_lock_init(&qp->rq.lock);
+		mutex_init(&qp->mutex);
+		refcount_set(&qp->refcnt, 1);
+		init_completion(&qp->free);
+
+		qp->state = IB_QPS_RESET;
+		qp->is_kernel = !udata;
+
+		if (!qp->is_kernel) {
+			dev_dbg(&dev->pdev->dev,
+				"create queuepair from user space\n");
+
+			if (ib_copy_from_udata(&ucmd, udata, sizeof(ucmd))) {
+				ret = -EFAULT;
+				goto err_qp;
+			}
+
+			/* Userspace supports qpn and qp handles? */
+			if (dev->dsr_version >= PVRDMA_QPHANDLE_VERSION &&
+			    udata->outlen < sizeof(qp_resp)) {
+				dev_warn(&dev->pdev->dev,
+					 "create queuepair not supported\n");
+				ret = -EOPNOTSUPP;
+				goto err_qp;
+			}
+
+			if (!is_srq) {
+				/* set qp->sq.wqe_cnt, shift, buf_size.. */
+				qp->rumem = ib_umem_get(ibqp->device,
+							ucmd.rbuf_addr,
+							ucmd.rbuf_size, 0);
+				if (IS_ERR(qp->rumem)) {
+					ret = PTR_ERR(qp->rumem);
+					goto err_qp;
+				}
+				qp->srq = NULL;
+			} else {
+				qp->rumem = NULL;
+				qp->srq = to_vsrq(init_attr->srq);
+			}
+
+			qp->sumem = ib_umem_get(ibqp->device, ucmd.sbuf_addr,
+						ucmd.sbuf_size, 0);
+			if (IS_ERR(qp->sumem)) {
+				if (!is_srq)
+					ib_umem_release(qp->rumem);
+				ret = PTR_ERR(qp->sumem);
+				goto err_qp;
+			}
+
+			qp->npages_send =
+				ib_umem_num_dma_blocks(qp->sumem, PAGE_SIZE);
+			if (!is_srq)
+				qp->npages_recv = ib_umem_num_dma_blocks(
+					qp->rumem, PAGE_SIZE);
+			else
+				qp->npages_recv = 0;
+			qp->npages = qp->npages_send + qp->npages_recv;
+		} else {
+			ret = pvrdma_set_sq_size(to_vdev(ibqp->device),
+						 &init_attr->cap, qp);
+			if (ret)
+				goto err_qp;
+
+			ret = pvrdma_set_rq_size(to_vdev(ibqp->device),
+						 &init_attr->cap, qp);
+			if (ret)
+				goto err_qp;
+
+			qp->npages = qp->npages_send + qp->npages_recv;
+
+			/* Skip header page. */
+			qp->sq.offset = PVRDMA_QP_NUM_HEADER_PAGES * PAGE_SIZE;
+
+			/* Recv queue pages are after send pages. */
+			qp->rq.offset = qp->npages_send * PAGE_SIZE;
+		}
+
+		if (qp->npages < 0 || qp->npages > PVRDMA_PAGE_DIR_MAX_PAGES) {
+			dev_warn(&dev->pdev->dev,
+				 "overflow pages in queuepair\n");
+			ret = -EINVAL;
+			goto err_umem;
+		}
+
+		ret = pvrdma_page_dir_init(dev, &qp->pdir, qp->npages,
+					   qp->is_kernel);
+		if (ret) {
+			dev_warn(&dev->pdev->dev,
+				 "could not allocate page directory\n");
+			goto err_umem;
+		}
+
+		if (!qp->is_kernel) {
+			pvrdma_page_dir_insert_umem(&qp->pdir, qp->sumem, 0);
+			if (!is_srq)
+				pvrdma_page_dir_insert_umem(&qp->pdir,
+							    qp->rumem,
+							    qp->npages_send);
+		} else {
+			/* Ring state is always the first page. */
+			qp->sq.ring = qp->pdir.pages[0];
+			qp->rq.ring = is_srq ? NULL : &qp->sq.ring[1];
+		}
+		break;
+	default:
+		ret = -EINVAL;
+		goto err_qp;
+	}
+
+	/* Not supported */
+	init_attr->cap.max_inline_data = 0;
+
+	memset(cmd, 0, sizeof(*cmd));
+	cmd->hdr.cmd = PVRDMA_CMD_CREATE_QP;
+	cmd->pd_handle = to_vpd(ibqp->pd)->pd_handle;
+	cmd->send_cq_handle = to_vcq(init_attr->send_cq)->cq_handle;
+	cmd->recv_cq_handle = to_vcq(init_attr->recv_cq)->cq_handle;
+	if (is_srq)
+		cmd->srq_handle = to_vsrq(init_attr->srq)->srq_handle;
+	else
+		cmd->srq_handle = 0;
+	cmd->max_send_wr = init_attr->cap.max_send_wr;
+	cmd->max_recv_wr = init_attr->cap.max_recv_wr;
+	cmd->max_send_sge = init_attr->cap.max_send_sge;
+	cmd->max_recv_sge = init_attr->cap.max_recv_sge;
+	cmd->max_inline_data = init_attr->cap.max_inline_data;
+	cmd->sq_sig_all = (init_attr->sq_sig_type == IB_SIGNAL_ALL_WR) ? 1 : 0;
+	cmd->qp_type = ib_qp_type_to_pvrdma(init_attr->qp_type);
+	cmd->is_srq = is_srq;
+	cmd->lkey = 0;
+	cmd->access_flags = IB_ACCESS_LOCAL_WRITE;
+	cmd->total_chunks = qp->npages;
+	cmd->send_chunks = qp->npages_send - PVRDMA_QP_NUM_HEADER_PAGES;
+	cmd->pdir_dma = qp->pdir.dir_dma;
+
+	dev_dbg(&dev->pdev->dev, "create queuepair with %d, %d, %d, %d\n",
+		cmd->max_send_wr, cmd->max_recv_wr, cmd->max_send_sge,
+		cmd->max_recv_sge);
+
+	ret = pvrdma_cmd_post(dev, &req, &rsp, PVRDMA_CMD_CREATE_QP_RESP);
+	if (ret < 0) {
+		dev_warn(&dev->pdev->dev,
+			 "could not create queuepair, error: %d\n", ret);
+		goto err_pdir;
+	}
+
+	/* max_send_wr/_recv_wr/_send_sge/_recv_sge/_inline_data */
+	qp->port = init_attr->port_num;
+
+	if (dev->dsr_version >= PVRDMA_QPHANDLE_VERSION) {
+		qp->ibqp.qp_num = resp_v2->qpn;
+		qp->qp_handle = resp_v2->qp_handle;
+	} else {
+		qp->ibqp.qp_num = resp->qpn;
+		qp->qp_handle = resp->qpn;
+	}
+
+	spin_lock_irqsave(&dev->qp_tbl_lock, flags);
+	dev->qp_tbl[qp->qp_handle % dev->dsr->caps.max_qp] = qp;
+	spin_unlock_irqrestore(&dev->qp_tbl_lock, flags);
+
+	if (udata) {
+		qp_resp.qpn = qp->ibqp.qp_num;
+		qp_resp.qp_handle = qp->qp_handle;
+
+		if (ib_copy_to_udata(udata, &qp_resp,
+				     min(udata->outlen, sizeof(qp_resp)))) {
+			dev_warn(&dev->pdev->dev,
+				 "failed to copy back udata\n");
+			__pvrdma_destroy_qp(dev, qp);
+			return -EINVAL;
+		}
+	}
+
+	return 0;
+
+err_pdir:
+	pvrdma_page_dir_cleanup(dev, &qp->pdir);
+err_umem:
+	ib_umem_release(qp->rumem);
+	ib_umem_release(qp->sumem);
+err_qp:
+	atomic_dec(&dev->num_qps);
+	return ret;
+}
+
+static void _pvrdma_free_qp(struct pvrdma_qp *qp)
+{
+	unsigned long flags;
+	struct pvrdma_dev *dev = to_vdev(qp->ibqp.device);
+
+	spin_lock_irqsave(&dev->qp_tbl_lock, flags);
+	dev->qp_tbl[qp->qp_handle] = NULL;
+	spin_unlock_irqrestore(&dev->qp_tbl_lock, flags);
+
+	if (refcount_dec_and_test(&qp->refcnt))
+		complete(&qp->free);
+	wait_for_completion(&qp->free);
+
+	ib_umem_release(qp->rumem);
+	ib_umem_release(qp->sumem);
+
+	pvrdma_page_dir_cleanup(dev, &qp->pdir);
+
+	atomic_dec(&dev->num_qps);
+}
+
+static void pvrdma_free_qp(struct pvrdma_qp *qp)
+{
+	struct pvrdma_cq *scq;
+	struct pvrdma_cq *rcq;
+	unsigned long scq_flags, rcq_flags;
+
+	/* In case cq is polling */
+	get_cqs(qp, &scq, &rcq);
+	pvrdma_lock_cqs(scq, rcq, &scq_flags, &rcq_flags);
+
+	_pvrdma_flush_cqe(qp, scq);
+	if (scq != rcq)
+		_pvrdma_flush_cqe(qp, rcq);
+
+	/*
+	 * We're now unlocking the CQs before clearing out the qp handle this
+	 * should still be safe. We have destroyed the backend QP and flushed
+	 * the CQEs so there should be no other completions for this QP.
+	 */
+	pvrdma_unlock_cqs(scq, rcq, &scq_flags, &rcq_flags);
+
+	_pvrdma_free_qp(qp);
+}
+
+static inline void _pvrdma_destroy_qp_work(struct pvrdma_dev *dev,
+					   u32 qp_handle)
+{
+	union pvrdma_cmd_req req;
+	struct pvrdma_cmd_destroy_qp *cmd = &req.destroy_qp;
+	int ret;
+
+	memset(cmd, 0, sizeof(*cmd));
+	cmd->hdr.cmd = PVRDMA_CMD_DESTROY_QP;
+	cmd->qp_handle = qp_handle;
+
+	ret = pvrdma_cmd_post(dev, &req, NULL, 0);
+	if (ret < 0)
+		dev_warn(&dev->pdev->dev,
+			 "destroy queuepair failed, error: %d\n", ret);
+}
+
+/**
+ * pvrdma_destroy_qp - destroy a queue pair
+ * @qp: the queue pair to destroy
+ * @udata: user data or null for kernel object
+ *
+ * @return: always 0.
+ */
+int pvrdma_destroy_qp(struct ib_qp *qp, struct ib_udata *udata)
+{
+	struct pvrdma_qp *vqp = to_vqp(qp);
+
+	_pvrdma_destroy_qp_work(to_vdev(qp->device), vqp->qp_handle);
+	pvrdma_free_qp(vqp);
+
+	return 0;
+}
+
+static void __pvrdma_destroy_qp(struct pvrdma_dev *dev,
+				struct pvrdma_qp *qp)
+{
+	_pvrdma_destroy_qp_work(dev, qp->qp_handle);
+	_pvrdma_free_qp(qp);
+}
+
+/**
+ * pvrdma_modify_qp - modify queue pair attributes
+ * @ibqp: the queue pair
+ * @attr: the new queue pair's attributes
+ * @attr_mask: attributes mask
+ * @udata: user data
+ *
+ * @returns 0 on success, otherwise returns an errno.
+ */
+int pvrdma_modify_qp(struct ib_qp *ibqp, struct ib_qp_attr *attr,
+		     int attr_mask, struct ib_udata *udata)
+{
+	struct pvrdma_dev *dev = to_vdev(ibqp->device);
+	struct pvrdma_qp *qp = to_vqp(ibqp);
+	union pvrdma_cmd_req req;
+	union pvrdma_cmd_resp rsp;
+	struct pvrdma_cmd_modify_qp *cmd = &req.modify_qp;
+	enum ib_qp_state cur_state, next_state;
+	int ret;
+
+	if (attr_mask & ~IB_QP_ATTR_STANDARD_BITS)
+		return -EOPNOTSUPP;
+
+	/* Sanity checking. Should need lock here */
+	mutex_lock(&qp->mutex);
+	cur_state = (attr_mask & IB_QP_CUR_STATE) ? attr->cur_qp_state :
+		qp->state;
+	next_state = (attr_mask & IB_QP_STATE) ? attr->qp_state : cur_state;
+
+	if (!ib_modify_qp_is_ok(cur_state, next_state, ibqp->qp_type,
+				attr_mask)) {
+		ret = -EINVAL;
+		goto out;
+	}
+
+	if (attr_mask & IB_QP_PORT) {
+		if (attr->port_num == 0 ||
+		    attr->port_num > ibqp->device->phys_port_cnt) {
+			ret = -EINVAL;
+			goto out;
+		}
+	}
+
+	if (attr_mask & IB_QP_MIN_RNR_TIMER) {
+		if (attr->min_rnr_timer > 31) {
+			ret = -EINVAL;
+			goto out;
+		}
+	}
+
+	if (attr_mask & IB_QP_PKEY_INDEX) {
+		if (attr->pkey_index >= dev->dsr->caps.max_pkeys) {
+			ret = -EINVAL;
+			goto out;
+		}
+	}
+
+	if (attr_mask & IB_QP_QKEY)
+		qp->qkey = attr->qkey;
+
+	if (cur_state == next_state && cur_state == IB_QPS_RESET) {
+		ret = 0;
+		goto out;
+	}
+
+	qp->state = next_state;
+	memset(cmd, 0, sizeof(*cmd));
+	cmd->hdr.cmd = PVRDMA_CMD_MODIFY_QP;
+	cmd->qp_handle = qp->qp_handle;
+	cmd->attr_mask = ib_qp_attr_mask_to_pvrdma(attr_mask);
+	cmd->attrs.qp_state = ib_qp_state_to_pvrdma(attr->qp_state);
+	cmd->attrs.cur_qp_state =
+		ib_qp_state_to_pvrdma(attr->cur_qp_state);
+	cmd->attrs.path_mtu = ib_mtu_to_pvrdma(attr->path_mtu);
+	cmd->attrs.path_mig_state =
+		ib_mig_state_to_pvrdma(attr->path_mig_state);
+	cmd->attrs.qkey = attr->qkey;
+	cmd->attrs.rq_psn = attr->rq_psn;
+	cmd->attrs.sq_psn = attr->sq_psn;
+	cmd->attrs.dest_qp_num = attr->dest_qp_num;
+	cmd->attrs.qp_access_flags =
+		ib_access_flags_to_pvrdma(attr->qp_access_flags);
+	cmd->attrs.pkey_index = attr->pkey_index;
+	cmd->attrs.alt_pkey_index = attr->alt_pkey_index;
+	cmd->attrs.en_sqd_async_notify = attr->en_sqd_async_notify;
+	cmd->attrs.sq_draining = attr->sq_draining;
+	cmd->attrs.max_rd_atomic = attr->max_rd_atomic;
+	cmd->attrs.max_dest_rd_atomic = attr->max_dest_rd_atomic;
+	cmd->attrs.min_rnr_timer = attr->min_rnr_timer;
+	cmd->attrs.port_num = attr->port_num;
+	cmd->attrs.timeout = attr->timeout;
+	cmd->attrs.retry_cnt = attr->retry_cnt;
+	cmd->attrs.rnr_retry = attr->rnr_retry;
+	cmd->attrs.alt_port_num = attr->alt_port_num;
+	cmd->attrs.alt_timeout = attr->alt_timeout;
+	ib_qp_cap_to_pvrdma(&cmd->attrs.cap, &attr->cap);
+	rdma_ah_attr_to_pvrdma(&cmd->attrs.ah_attr, &attr->ah_attr);
+	rdma_ah_attr_to_pvrdma(&cmd->attrs.alt_ah_attr, &attr->alt_ah_attr);
+
+	ret = pvrdma_cmd_post(dev, &req, &rsp, PVRDMA_CMD_MODIFY_QP_RESP);
+	if (ret < 0) {
+		dev_warn(&dev->pdev->dev,
+			 "could not modify queuepair, error: %d\n", ret);
+	} else if (rsp.hdr.err > 0) {
+		dev_warn(&dev->pdev->dev,
+			 "cannot modify queuepair, error: %d\n", rsp.hdr.err);
+		ret = -EINVAL;
+	}
+
+	if (ret == 0 && next_state == IB_QPS_RESET)
+		pvrdma_reset_qp(qp);
+
+out:
+	mutex_unlock(&qp->mutex);
+
+	return ret;
+}
+
+static inline void *get_sq_wqe(struct pvrdma_qp *qp, unsigned int n)
+{
+	return pvrdma_page_dir_get_ptr(&qp->pdir,
+				       qp->sq.offset + n * qp->sq.wqe_size);
+}
+
+static inline void *get_rq_wqe(struct pvrdma_qp *qp, unsigned int n)
+{
+	return pvrdma_page_dir_get_ptr(&qp->pdir,
+				       qp->rq.offset + n * qp->rq.wqe_size);
+}
+
+static int set_reg_seg(struct pvrdma_sq_wqe_hdr *wqe_hdr,
+		       const struct ib_reg_wr *wr)
+{
+	struct pvrdma_user_mr *mr = to_vmr(wr->mr);
+
+	wqe_hdr->wr.fast_reg.iova_start = mr->ibmr.iova;
+	wqe_hdr->wr.fast_reg.pl_pdir_dma = mr->pdir.dir_dma;
+	wqe_hdr->wr.fast_reg.page_shift = mr->page_shift;
+	wqe_hdr->wr.fast_reg.page_list_len = mr->npages;
+	wqe_hdr->wr.fast_reg.length = mr->ibmr.length;
+	wqe_hdr->wr.fast_reg.access_flags = wr->access;
+	wqe_hdr->wr.fast_reg.rkey = wr->key;
+
+	return pvrdma_page_dir_insert_page_list(&mr->pdir, mr->pages,
+						mr->npages);
+}
+
+/**
+ * pvrdma_post_send - post send work request entries on a QP
+ * @ibqp: the QP
+ * @wr: work request list to post
+ * @bad_wr: the first bad WR returned
+ *
+ * @return: 0 on success, otherwise errno returned.
+ */
+int pvrdma_post_send(struct ib_qp *ibqp, const struct ib_send_wr *wr,
+		     const struct ib_send_wr **bad_wr)
+{
+	struct pvrdma_qp *qp = to_vqp(ibqp);
+	struct pvrdma_dev *dev = to_vdev(ibqp->device);
+	unsigned long flags;
+	struct pvrdma_sq_wqe_hdr *wqe_hdr;
+	struct pvrdma_sge *sge;
+	int i, ret;
+
+	/*
+	 * In states lower than RTS, we can fail immediately. In other states,
+	 * just post and let the device figure it out.
+	 */
+	if (qp->state < IB_QPS_RTS) {
+		*bad_wr = wr;
+		return -EINVAL;
+	}
+
+	spin_lock_irqsave(&qp->sq.lock, flags);
+
+	while (wr) {
+		unsigned int tail = 0;
+
+		if (unlikely(!pvrdma_idx_ring_has_space(
+				qp->sq.ring, qp->sq.wqe_cnt, &tail))) {
+			dev_warn_ratelimited(&dev->pdev->dev,
+					     "send queue is full\n");
+			*bad_wr = wr;
+			ret = -ENOMEM;
+			goto out;
+		}
+
+		if (unlikely(wr->num_sge > qp->sq.max_sg || wr->num_sge < 0)) {
+			dev_warn_ratelimited(&dev->pdev->dev,
+					     "send SGE overflow\n");
+			*bad_wr = wr;
+			ret = -EINVAL;
+			goto out;
+		}
+
+		if (unlikely(wr->opcode < 0)) {
+			dev_warn_ratelimited(&dev->pdev->dev,
+					     "invalid send opcode\n");
+			*bad_wr = wr;
+			ret = -EINVAL;
+			goto out;
+		}
+
+		/*
+		 * Only support UD, RC.
+		 * Need to check opcode table for thorough checking.
+		 * opcode		_UD	_UC	_RC
+		 * _SEND		x	x	x
+		 * _SEND_WITH_IMM	x	x	x
+		 * _RDMA_WRITE			x	x
+		 * _RDMA_WRITE_WITH_IMM		x	x
+		 * _LOCAL_INV			x	x
+		 * _SEND_WITH_INV		x	x
+		 * _RDMA_READ				x
+		 * _ATOMIC_CMP_AND_SWP			x
+		 * _ATOMIC_FETCH_AND_ADD		x
+		 * _MASK_ATOMIC_CMP_AND_SWP		x
+		 * _MASK_ATOMIC_FETCH_AND_ADD		x
+		 * _REG_MR				x
+		 *
+		 */
+		if (qp->ibqp.qp_type != IB_QPT_UD &&
+		    qp->ibqp.qp_type != IB_QPT_RC &&
+			wr->opcode != IB_WR_SEND) {
+			dev_warn_ratelimited(&dev->pdev->dev,
+					     "unsupported queuepair type\n");
+			*bad_wr = wr;
+			ret = -EINVAL;
+			goto out;
+		} else if (qp->ibqp.qp_type == IB_QPT_UD ||
+			   qp->ibqp.qp_type == IB_QPT_GSI) {
+			if (wr->opcode != IB_WR_SEND &&
+			    wr->opcode != IB_WR_SEND_WITH_IMM) {
+				dev_warn_ratelimited(&dev->pdev->dev,
+						     "invalid send opcode\n");
+				*bad_wr = wr;
+				ret = -EINVAL;
+				goto out;
+			}
+		}
+
+		wqe_hdr = (struct pvrdma_sq_wqe_hdr *)get_sq_wqe(qp, tail);
+		memset(wqe_hdr, 0, sizeof(*wqe_hdr));
+		wqe_hdr->wr_id = wr->wr_id;
+		wqe_hdr->num_sge = wr->num_sge;
+		wqe_hdr->opcode = ib_wr_opcode_to_pvrdma(wr->opcode);
+		wqe_hdr->send_flags = ib_send_flags_to_pvrdma(wr->send_flags);
+		if (wr->opcode == IB_WR_SEND_WITH_IMM ||
+		    wr->opcode == IB_WR_RDMA_WRITE_WITH_IMM)
+			wqe_hdr->ex.imm_data = wr->ex.imm_data;
+
+		if (unlikely(wqe_hdr->opcode == PVRDMA_WR_ERROR)) {
+			*bad_wr = wr;
+			ret = -EINVAL;
+			goto out;
+		}
+
+		switch (qp->ibqp.qp_type) {
+		case IB_QPT_GSI:
+		case IB_QPT_UD:
+			if (unlikely(!ud_wr(wr)->ah)) {
+				dev_warn_ratelimited(&dev->pdev->dev,
+						     "invalid address handle\n");
+				*bad_wr = wr;
+				ret = -EINVAL;
+				goto out;
+			}
+
+			/*
+			 * Use qkey from qp context if high order bit set,
+			 * otherwise from work request.
+			 */
+			wqe_hdr->wr.ud.remote_qpn = ud_wr(wr)->remote_qpn;
+			wqe_hdr->wr.ud.remote_qkey =
+				ud_wr(wr)->remote_qkey & 0x80000000 ?
+				qp->qkey : ud_wr(wr)->remote_qkey;
+			wqe_hdr->wr.ud.av = to_vah(ud_wr(wr)->ah)->av;
+
+			break;
+		case IB_QPT_RC:
+			switch (wr->opcode) {
+			case IB_WR_RDMA_READ:
+			case IB_WR_RDMA_WRITE:
+			case IB_WR_RDMA_WRITE_WITH_IMM:
+				wqe_hdr->wr.rdma.remote_addr =
+					rdma_wr(wr)->remote_addr;
+				wqe_hdr->wr.rdma.rkey = rdma_wr(wr)->rkey;
+				break;
+			case IB_WR_LOCAL_INV:
+			case IB_WR_SEND_WITH_INV:
+				wqe_hdr->ex.invalidate_rkey =
+					wr->ex.invalidate_rkey;
+				break;
+			case IB_WR_ATOMIC_CMP_AND_SWP:
+			case IB_WR_ATOMIC_FETCH_AND_ADD:
+				wqe_hdr->wr.atomic.remote_addr =
+					atomic_wr(wr)->remote_addr;
+				wqe_hdr->wr.atomic.rkey = atomic_wr(wr)->rkey;
+				wqe_hdr->wr.atomic.compare_add =
+					atomic_wr(wr)->compare_add;
+				if (wr->opcode == IB_WR_ATOMIC_CMP_AND_SWP)
+					wqe_hdr->wr.atomic.swap =
+						atomic_wr(wr)->swap;
+				break;
+			case IB_WR_REG_MR:
+				ret = set_reg_seg(wqe_hdr, reg_wr(wr));
+				if (ret < 0) {
+					dev_warn_ratelimited(&dev->pdev->dev,
+							     "Failed to set fast register work request\n");
+					*bad_wr = wr;
+					goto out;
+				}
+				break;
+			default:
+				break;
+			}
+
+			break;
+		default:
+			dev_warn_ratelimited(&dev->pdev->dev,
+					     "invalid queuepair type\n");
+			ret = -EINVAL;
+			*bad_wr = wr;
+			goto out;
+		}
+
+		sge = (struct pvrdma_sge *)(wqe_hdr + 1);
+		for (i = 0; i < wr->num_sge; i++) {
+			/* Need to check wqe_size 0 or max size */
+			sge->addr = wr->sg_list[i].addr;
+			sge->length = wr->sg_list[i].length;
+			sge->lkey = wr->sg_list[i].lkey;
+			sge++;
+		}
+
+		/* Make sure wqe is written before index update */
+		smp_wmb();
+
+		/* Update shared sq ring */
+		pvrdma_idx_ring_inc(&qp->sq.ring->prod_tail,
+				    qp->sq.wqe_cnt);
+
+		wr = wr->next;
+	}
+
+	ret = 0;
+
+out:
+	spin_unlock_irqrestore(&qp->sq.lock, flags);
+
+	if (!ret)
+		pvrdma_write_uar_qp(dev, PVRDMA_UAR_QP_SEND | qp->qp_handle);
+
+	return ret;
+}
+
+/**
+ * pvrdma_post_recv - post receive work request entries on a QP
+ * @ibqp: the QP
+ * @wr: the work request list to post
+ * @bad_wr: the first bad WR returned
+ *
+ * @return: 0 on success, otherwise errno returned.
+ */
+int pvrdma_post_recv(struct ib_qp *ibqp, const struct ib_recv_wr *wr,
+		     const struct ib_recv_wr **bad_wr)
+{
+	struct pvrdma_dev *dev = to_vdev(ibqp->device);
+	unsigned long flags;
+	struct pvrdma_qp *qp = to_vqp(ibqp);
+	struct pvrdma_rq_wqe_hdr *wqe_hdr;
+	struct pvrdma_sge *sge;
+	int ret = 0;
+	int i;
+
+	/*
+	 * In the RESET state, we can fail immediately. For other states,
+	 * just post and let the device figure it out.
+	 */
+	if (qp->state == IB_QPS_RESET) {
+		*bad_wr = wr;
+		return -EINVAL;
+	}
+
+	if (qp->srq) {
+		dev_warn(&dev->pdev->dev, "QP associated with SRQ\n");
+		*bad_wr = wr;
+		return -EINVAL;
+	}
+
+	spin_lock_irqsave(&qp->rq.lock, flags);
+
+	while (wr) {
+		unsigned int tail = 0;
+
+		if (unlikely(wr->num_sge > qp->rq.max_sg ||
+			     wr->num_sge < 0)) {
+			ret = -EINVAL;
+			*bad_wr = wr;
+			dev_warn_ratelimited(&dev->pdev->dev,
+					     "recv SGE overflow\n");
+			goto out;
+		}
+
+		if (unlikely(!pvrdma_idx_ring_has_space(
+				qp->rq.ring, qp->rq.wqe_cnt, &tail))) {
+			ret = -ENOMEM;
+			*bad_wr = wr;
+			dev_warn_ratelimited(&dev->pdev->dev,
+					     "recv queue full\n");
+			goto out;
+		}
+
+		wqe_hdr = (struct pvrdma_rq_wqe_hdr *)get_rq_wqe(qp, tail);
+		wqe_hdr->wr_id = wr->wr_id;
+		wqe_hdr->num_sge = wr->num_sge;
+		wqe_hdr->total_len = 0;
+
+		sge = (struct pvrdma_sge *)(wqe_hdr + 1);
+		for (i = 0; i < wr->num_sge; i++) {
+			sge->addr = wr->sg_list[i].addr;
+			sge->length = wr->sg_list[i].length;
+			sge->lkey = wr->sg_list[i].lkey;
+			sge++;
+		}
+
+		/* Make sure wqe is written before index update */
+		smp_wmb();
+
+		/* Update shared rq ring */
+		pvrdma_idx_ring_inc(&qp->rq.ring->prod_tail,
+				    qp->rq.wqe_cnt);
+
+		wr = wr->next;
+	}
+
+	spin_unlock_irqrestore(&qp->rq.lock, flags);
+
+	pvrdma_write_uar_qp(dev, PVRDMA_UAR_QP_RECV | qp->qp_handle);
+
+	return ret;
+
+out:
+	spin_unlock_irqrestore(&qp->rq.lock, flags);
+
+	return ret;
+}
+
+/**
+ * pvrdma_query_qp - query a queue pair's attributes
+ * @ibqp: the queue pair to query
+ * @attr: the queue pair's attributes
+ * @attr_mask: attributes mask
+ * @init_attr: initial queue pair attributes
+ *
+ * @returns 0 on success, otherwise returns an errno.
+ */
+int pvrdma_query_qp(struct ib_qp *ibqp, struct ib_qp_attr *attr,
+		    int attr_mask, struct ib_qp_init_attr *init_attr)
+{
+	struct pvrdma_dev *dev = to_vdev(ibqp->device);
+	struct pvrdma_qp *qp = to_vqp(ibqp);
+	union pvrdma_cmd_req req;
+	union pvrdma_cmd_resp rsp;
+	struct pvrdma_cmd_query_qp *cmd = &req.query_qp;
+	struct pvrdma_cmd_query_qp_resp *resp = &rsp.query_qp_resp;
+	int ret = 0;
+
+	mutex_lock(&qp->mutex);
+
+	if (qp->state == IB_QPS_RESET) {
+		attr->qp_state = IB_QPS_RESET;
+		goto out;
+	}
+
+	memset(cmd, 0, sizeof(*cmd));
+	cmd->hdr.cmd = PVRDMA_CMD_QUERY_QP;
+	cmd->qp_handle = qp->qp_handle;
+	cmd->attr_mask = ib_qp_attr_mask_to_pvrdma(attr_mask);
+
+	ret = pvrdma_cmd_post(dev, &req, &rsp, PVRDMA_CMD_QUERY_QP_RESP);
+	if (ret < 0) {
+		dev_warn(&dev->pdev->dev,
+			 "could not query queuepair, error: %d\n", ret);
+		goto out;
+	}
+
+	attr->qp_state = pvrdma_qp_state_to_ib(resp->attrs.qp_state);
+	attr->cur_qp_state =
+		pvrdma_qp_state_to_ib(resp->attrs.cur_qp_state);
+	attr->path_mtu = pvrdma_mtu_to_ib(resp->attrs.path_mtu);
+	attr->path_mig_state =
+		pvrdma_mig_state_to_ib(resp->attrs.path_mig_state);
+	attr->qkey = resp->attrs.qkey;
+	attr->rq_psn = resp->attrs.rq_psn;
+	attr->sq_psn = resp->attrs.sq_psn;
+	attr->dest_qp_num = resp->attrs.dest_qp_num;
+	attr->qp_access_flags =
+		pvrdma_access_flags_to_ib(resp->attrs.qp_access_flags);
+	attr->pkey_index = resp->attrs.pkey_index;
+	attr->alt_pkey_index = resp->attrs.alt_pkey_index;
+	attr->en_sqd_async_notify = resp->attrs.en_sqd_async_notify;
+	attr->sq_draining = resp->attrs.sq_draining;
+	attr->max_rd_atomic = resp->attrs.max_rd_atomic;
+	attr->max_dest_rd_atomic = resp->attrs.max_dest_rd_atomic;
+	attr->min_rnr_timer = resp->attrs.min_rnr_timer;
+	attr->port_num = resp->attrs.port_num;
+	attr->timeout = resp->attrs.timeout;
+	attr->retry_cnt = resp->attrs.retry_cnt;
+	attr->rnr_retry = resp->attrs.rnr_retry;
+	attr->alt_port_num = resp->attrs.alt_port_num;
+	attr->alt_timeout = resp->attrs.alt_timeout;
+	pvrdma_qp_cap_to_ib(&attr->cap, &resp->attrs.cap);
+	pvrdma_ah_attr_to_rdma(&attr->ah_attr, &resp->attrs.ah_attr);
+	pvrdma_ah_attr_to_rdma(&attr->alt_ah_attr, &resp->attrs.alt_ah_attr);
+
+	qp->state = attr->qp_state;
+
+	ret = 0;
+
+out:
+	attr->cur_qp_state = attr->qp_state;
+
+	init_attr->event_handler = qp->ibqp.event_handler;
+	init_attr->qp_context = qp->ibqp.qp_context;
+	init_attr->send_cq = qp->ibqp.send_cq;
+	init_attr->recv_cq = qp->ibqp.recv_cq;
+	init_attr->srq = qp->ibqp.srq;
+	init_attr->xrcd = NULL;
+	init_attr->cap = attr->cap;
+	init_attr->sq_sig_type = 0;
+	init_attr->qp_type = qp->ibqp.qp_type;
+	init_attr->create_flags = 0;
+	init_attr->port_num = qp->port;
+
+	mutex_unlock(&qp->mutex);
+	return ret;
+}