Soft RoCE driver

Soft RoCE (RXE) - The software RoCE driver

ib_rxe implements the RDMA transport and registers to the RDMA core
device as a kernel verbs provider. It also implements the packet IO
layer. On the other hand ib_rxe registers to the Linux netdev stack
as a udp encapsulating protocol, in that case RDMA, for sending and
receiving packets over any Ethernet device.  This yields a RDMA
transport over the UDP/Ethernet network layer forming a RoCEv2
compatible device.

The configuration procedure of the Soft RoCE drivers requires
binding to any existing Ethernet network device. This is done with
/sys interface.

A userspace Soft RoCE library (librxe) provides user applications
the ability to run with Soft RoCE devices.  The use of rxe verbs ins
user space requires the inclusion of librxe as a device specifics
plug-in to libibverbs. librxe is packaged separately.

Architecture:

     +-----------------------------------------------------------+
     |                          Application                      |
     +-----------------------------------------------------------+
                            +-----------------------------------+
                            |             libibverbs            |
User                        +-----------------------------------+
                            +----------------+ +----------------+
                            | librxe         | | HW RoCE lib    |
                            +----------------+ +----------------+
+---------------------------------------------------------------+
     +--------------+                           +------------+
     | Sockets      |                           | RDMA ULP   |
     +--------------+                           +------------+
     +--------------+                  +---------------------+
     | TCP/IP       |                  | ib_core             |
     +--------------+                  +---------------------+
                             +------------+ +----------------+
Kernel                       | ib_rxe     | | HW RoCE driver |
                             +------------+ +----------------+
     +------------------------------------+
     | NIC driver                         |
     +------------------------------------+

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
     +-----------------------------------------------------------+
     |                          Application                      |
     +-----------------------------------------------------------+
                            +-----------------------------------+
                            |             libibverbs            |
User                        +-----------------------------------+
                            +----------------+ +----------------+
                            | librxe         | | HW RoCE lib    |
                            +----------------+ +----------------+
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
     +--------------+                           +------------+
     | Sockets      |                           | RDMA ULP   |
     +--------------+                           +------------+
     +--------------+                  +---------------------+
     | TCP/IP       |                  | ib_core             |
     +--------------+                  +---------------------+
                             +------------+ +----------------+
Kernel                       | ib_rxe     | | HW RoCE driver |
                             +------------+ +----------------+
     +------------------------------------+
     | NIC driver                         |
     +------------------------------------+
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

Soft RoCE resources:

[1[ https://github.com/SoftRoCE/librxe-dev librxe - source code in
Github
[2] https://github.com/SoftRoCE/rxe-dev/wiki/rxe-dev:-Home - Soft RoCE
Wiki page
[3] https://github.com/SoftRoCE/librxe-dev - Soft RoCE userspace library
Signed-off-by: NKamal Heib <kamalh@mellanox.com>
Signed-off-by: NAmir Vadai <amirv@mellanox.com>
Signed-off-by: NMoni Shoua <monis@mellanox.com>
Reviewed-by: NHaggai Eran <haggaie@mellanox.com>
Signed-off-by: NDoug Ledford <dledford@redhat.com>

MAINTAINERS

@@ -7444,6 +7444,15 @@ W:	http://www.mellanox.com
 Q:	http://patchwork.ozlabs.org/project/netdev/list/
 F:	drivers/net/ethernet/mellanox/mlxsw/
 
+SOFT-ROCE DRIVER (rxe)
+M:	Moni Shoua <monis@mellanox.com>
+L:	linux-rdma@vger.kernel.org
+S:	Supported
+W:	https://github.com/SoftRoCE/rxe-dev/wiki/rxe-dev:-Home
+Q:	http://patchwork.kernel.org/project/linux-rdma/list/
+F:	drivers/infiniband/hw/rxe/
+F:	include/uapi/rdma/rdma_user_rxe.h
+
 MEMBARRIER SUPPORT
 M:	Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
 M:	"Paul E. McKenney" <paulmck@linux.vnet.ibm.com>

drivers/infiniband/Kconfig

@@ -84,6 +84,7 @@ source "drivers/infiniband/ulp/iser/Kconfig"
 source "drivers/infiniband/ulp/isert/Kconfig"
 
 source "drivers/infiniband/sw/rdmavt/Kconfig"
+source "drivers/infiniband/sw/rxe/Kconfig"
 
 source "drivers/infiniband/hw/hfi1/Kconfig"
 

drivers/infiniband/sw/Makefile

 obj-$(CONFIG_INFINIBAND_RDMAVT)		+= rdmavt/
+obj-$(CONFIG_RDMA_RXE)			+= rxe/

drivers/infiniband/sw/rxe/Kconfig

+config RDMA_RXE
+	tristate "Software RDMA over Ethernet (RoCE) driver"
+	depends on INET && PCI && INFINIBAND
+	depends on NET_UDP_TUNNEL
+	---help---
+	This driver implements the InfiniBand RDMA transport over
+	the Linux network stack. It enables a system with a
+	standard Ethernet adapter to interoperate with a RoCE
+	adapter or with another system running the RXE driver.
+	Documentation on InfiniBand and RoCE can be downloaded at
+	www.infinibandta.org and www.openfabrics.org. (See also
+	siw which is a similar software driver for iWARP.)
+
+	The driver is split into two layers, one interfaces with the
+	Linux RDMA stack and implements a kernel or user space
+	verbs API. The user space verbs API requires a support
+	library named librxe which is loaded by the generic user
+	space verbs API, libibverbs. The other layer interfaces
+	with the Linux network stack at layer 3.
+
+	To configure and work with soft-RoCE driver please use the
+	following wiki page under "configure Soft-RoCE (RXE)" section:
+
+	https://github.com/SoftRoCE/rxe-dev/wiki/rxe-dev:-Home

drivers/infiniband/sw/rxe/Makefile

+obj-$(CONFIG_RDMA_RXE) += rdma_rxe.o
+
+rdma_rxe-y := \
+	rxe.o \
+	rxe_comp.o \
+	rxe_req.o \
+	rxe_resp.o \
+	rxe_recv.o \
+	rxe_pool.o \
+	rxe_queue.o \
+	rxe_verbs.o \
+	rxe_av.o \
+	rxe_srq.o \
+	rxe_qp.o \
+	rxe_cq.o \
+	rxe_mr.o \
+	rxe_dma.o \
+	rxe_opcode.o \
+	rxe_mmap.o \
+	rxe_icrc.o \
+	rxe_mcast.o \
+	rxe_task.o \
+	rxe_net.o \
+	rxe_sysfs.o

drivers/infiniband/sw/rxe/rxe.c

+/*
+ * Copyright (c) 2016 Mellanox Technologies Ltd. All rights reserved.
+ * Copyright (c) 2015 System Fabric Works, 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 "rxe.h"
+#include "rxe_loc.h"
+
+MODULE_AUTHOR("Bob Pearson, Frank Zago, John Groves, Kamal Heib");
+MODULE_DESCRIPTION("Soft RDMA transport");
+MODULE_LICENSE("Dual BSD/GPL");
+MODULE_VERSION("0.2");
+
+/* free resources for all ports on a device */
+static void rxe_cleanup_ports(struct rxe_dev *rxe)
+{
+	kfree(rxe->port.pkey_tbl);
+	rxe->port.pkey_tbl = NULL;
+
+}
+
+/* free resources for a rxe device all objects created for this device must
+ * have been destroyed
+ */
+static void rxe_cleanup(struct rxe_dev *rxe)
+{
+	rxe_pool_cleanup(&rxe->uc_pool);
+	rxe_pool_cleanup(&rxe->pd_pool);
+	rxe_pool_cleanup(&rxe->ah_pool);
+	rxe_pool_cleanup(&rxe->srq_pool);
+	rxe_pool_cleanup(&rxe->qp_pool);
+	rxe_pool_cleanup(&rxe->cq_pool);
+	rxe_pool_cleanup(&rxe->mr_pool);
+	rxe_pool_cleanup(&rxe->mw_pool);
+	rxe_pool_cleanup(&rxe->mc_grp_pool);
+	rxe_pool_cleanup(&rxe->mc_elem_pool);
+
+	rxe_cleanup_ports(rxe);
+}
+
+/* called when all references have been dropped */
+void rxe_release(struct kref *kref)
+{
+	struct rxe_dev *rxe = container_of(kref, struct rxe_dev, ref_cnt);
+
+	rxe_cleanup(rxe);
+	ib_dealloc_device(&rxe->ib_dev);
+}
+
+void rxe_dev_put(struct rxe_dev *rxe)
+{
+	kref_put(&rxe->ref_cnt, rxe_release);
+}
+EXPORT_SYMBOL_GPL(rxe_dev_put);
+
+/* initialize rxe device parameters */
+static int rxe_init_device_param(struct rxe_dev *rxe)
+{
+	rxe->max_inline_data			= RXE_MAX_INLINE_DATA;
+
+	rxe->attr.fw_ver			= RXE_FW_VER;
+	rxe->attr.max_mr_size			= RXE_MAX_MR_SIZE;
+	rxe->attr.page_size_cap			= RXE_PAGE_SIZE_CAP;
+	rxe->attr.vendor_id			= RXE_VENDOR_ID;
+	rxe->attr.vendor_part_id		= RXE_VENDOR_PART_ID;
+	rxe->attr.hw_ver			= RXE_HW_VER;
+	rxe->attr.max_qp			= RXE_MAX_QP;
+	rxe->attr.max_qp_wr			= RXE_MAX_QP_WR;
+	rxe->attr.device_cap_flags		= RXE_DEVICE_CAP_FLAGS;
+	rxe->attr.max_sge			= RXE_MAX_SGE;
+	rxe->attr.max_sge_rd			= RXE_MAX_SGE_RD;
+	rxe->attr.max_cq			= RXE_MAX_CQ;
+	rxe->attr.max_cqe			= (1 << RXE_MAX_LOG_CQE) - 1;
+	rxe->attr.max_mr			= RXE_MAX_MR;
+	rxe->attr.max_pd			= RXE_MAX_PD;
+	rxe->attr.max_qp_rd_atom		= RXE_MAX_QP_RD_ATOM;
+	rxe->attr.max_ee_rd_atom		= RXE_MAX_EE_RD_ATOM;
+	rxe->attr.max_res_rd_atom		= RXE_MAX_RES_RD_ATOM;
+	rxe->attr.max_qp_init_rd_atom		= RXE_MAX_QP_INIT_RD_ATOM;
+	rxe->attr.max_ee_init_rd_atom		= RXE_MAX_EE_INIT_RD_ATOM;
+	rxe->attr.atomic_cap			= RXE_ATOMIC_CAP;
+	rxe->attr.max_ee			= RXE_MAX_EE;
+	rxe->attr.max_rdd			= RXE_MAX_RDD;
+	rxe->attr.max_mw			= RXE_MAX_MW;
+	rxe->attr.max_raw_ipv6_qp		= RXE_MAX_RAW_IPV6_QP;
+	rxe->attr.max_raw_ethy_qp		= RXE_MAX_RAW_ETHY_QP;
+	rxe->attr.max_mcast_grp			= RXE_MAX_MCAST_GRP;
+	rxe->attr.max_mcast_qp_attach		= RXE_MAX_MCAST_QP_ATTACH;
+	rxe->attr.max_total_mcast_qp_attach	= RXE_MAX_TOT_MCAST_QP_ATTACH;
+	rxe->attr.max_ah			= RXE_MAX_AH;
+	rxe->attr.max_fmr			= RXE_MAX_FMR;
+	rxe->attr.max_map_per_fmr		= RXE_MAX_MAP_PER_FMR;
+	rxe->attr.max_srq			= RXE_MAX_SRQ;
+	rxe->attr.max_srq_wr			= RXE_MAX_SRQ_WR;
+	rxe->attr.max_srq_sge			= RXE_MAX_SRQ_SGE;
+	rxe->attr.max_fast_reg_page_list_len	= RXE_MAX_FMR_PAGE_LIST_LEN;
+	rxe->attr.max_pkeys			= RXE_MAX_PKEYS;
+	rxe->attr.local_ca_ack_delay		= RXE_LOCAL_CA_ACK_DELAY;
+
+	rxe->max_ucontext			= RXE_MAX_UCONTEXT;
+
+	return 0;
+}
+
+/* initialize port attributes */
+static int rxe_init_port_param(struct rxe_port *port)
+{
+	port->attr.state		= RXE_PORT_STATE;
+	port->attr.max_mtu		= RXE_PORT_MAX_MTU;
+	port->attr.active_mtu		= RXE_PORT_ACTIVE_MTU;
+	port->attr.gid_tbl_len		= RXE_PORT_GID_TBL_LEN;
+	port->attr.port_cap_flags	= RXE_PORT_PORT_CAP_FLAGS;
+	port->attr.max_msg_sz		= RXE_PORT_MAX_MSG_SZ;
+	port->attr.bad_pkey_cntr	= RXE_PORT_BAD_PKEY_CNTR;
+	port->attr.qkey_viol_cntr	= RXE_PORT_QKEY_VIOL_CNTR;
+	port->attr.pkey_tbl_len		= RXE_PORT_PKEY_TBL_LEN;
+	port->attr.lid			= RXE_PORT_LID;
+	port->attr.sm_lid		= RXE_PORT_SM_LID;
+	port->attr.lmc			= RXE_PORT_LMC;
+	port->attr.max_vl_num		= RXE_PORT_MAX_VL_NUM;
+	port->attr.sm_sl		= RXE_PORT_SM_SL;
+	port->attr.subnet_timeout	= RXE_PORT_SUBNET_TIMEOUT;
+	port->attr.init_type_reply	= RXE_PORT_INIT_TYPE_REPLY;
+	port->attr.active_width		= RXE_PORT_ACTIVE_WIDTH;
+	port->attr.active_speed		= RXE_PORT_ACTIVE_SPEED;
+	port->attr.phys_state		= RXE_PORT_PHYS_STATE;
+	port->mtu_cap			=
+				ib_mtu_enum_to_int(RXE_PORT_ACTIVE_MTU);
+	port->subnet_prefix		= cpu_to_be64(RXE_PORT_SUBNET_PREFIX);
+
+	return 0;
+}
+
+/* initialize port state, note IB convention that HCA ports are always
+ * numbered from 1
+ */
+static int rxe_init_ports(struct rxe_dev *rxe)
+{
+	struct rxe_port *port = &rxe->port;
+
+	rxe_init_port_param(port);
+
+	if (!port->attr.pkey_tbl_len || !port->attr.gid_tbl_len)
+		return -EINVAL;
+
+	port->pkey_tbl = kcalloc(port->attr.pkey_tbl_len,
+			sizeof(*port->pkey_tbl), GFP_KERNEL);
+
+	if (!port->pkey_tbl)
+		return -ENOMEM;
+
+	port->pkey_tbl[0] = 0xffff;
+	port->port_guid = rxe->ifc_ops->port_guid(rxe);
+
+	spin_lock_init(&port->port_lock);
+
+	return 0;
+}
+
+/* init pools of managed objects */
+static int rxe_init_pools(struct rxe_dev *rxe)
+{
+	int err;
+
+	err = rxe_pool_init(rxe, &rxe->uc_pool, RXE_TYPE_UC,
+			    rxe->max_ucontext);
+	if (err)
+		goto err1;
+
+	err = rxe_pool_init(rxe, &rxe->pd_pool, RXE_TYPE_PD,
+			    rxe->attr.max_pd);
+	if (err)
+		goto err2;
+
+	err = rxe_pool_init(rxe, &rxe->ah_pool, RXE_TYPE_AH,
+			    rxe->attr.max_ah);
+	if (err)
+		goto err3;
+
+	err = rxe_pool_init(rxe, &rxe->srq_pool, RXE_TYPE_SRQ,
+			    rxe->attr.max_srq);
+	if (err)
+		goto err4;
+
+	err = rxe_pool_init(rxe, &rxe->qp_pool, RXE_TYPE_QP,
+			    rxe->attr.max_qp);
+	if (err)
+		goto err5;
+
+	err = rxe_pool_init(rxe, &rxe->cq_pool, RXE_TYPE_CQ,
+			    rxe->attr.max_cq);
+	if (err)
+		goto err6;
+
+	err = rxe_pool_init(rxe, &rxe->mr_pool, RXE_TYPE_MR,
+			    rxe->attr.max_mr);
+	if (err)
+		goto err7;
+
+	err = rxe_pool_init(rxe, &rxe->mw_pool, RXE_TYPE_MW,
+			    rxe->attr.max_mw);
+	if (err)
+		goto err8;
+
+	err = rxe_pool_init(rxe, &rxe->mc_grp_pool, RXE_TYPE_MC_GRP,
+			    rxe->attr.max_mcast_grp);
+	if (err)
+		goto err9;
+
+	err = rxe_pool_init(rxe, &rxe->mc_elem_pool, RXE_TYPE_MC_ELEM,
+			    rxe->attr.max_total_mcast_qp_attach);
+	if (err)
+		goto err10;
+
+	return 0;
+
+err10:
+	rxe_pool_cleanup(&rxe->mc_grp_pool);
+err9:
+	rxe_pool_cleanup(&rxe->mw_pool);
+err8:
+	rxe_pool_cleanup(&rxe->mr_pool);
+err7:
+	rxe_pool_cleanup(&rxe->cq_pool);
+err6:
+	rxe_pool_cleanup(&rxe->qp_pool);
+err5:
+	rxe_pool_cleanup(&rxe->srq_pool);
+err4:
+	rxe_pool_cleanup(&rxe->ah_pool);
+err3:
+	rxe_pool_cleanup(&rxe->pd_pool);
+err2:
+	rxe_pool_cleanup(&rxe->uc_pool);
+err1:
+	return err;
+}
+
+/* initialize rxe device state */
+static int rxe_init(struct rxe_dev *rxe)
+{
+	int err;
+
+	/* init default device parameters */
+	rxe_init_device_param(rxe);
+
+	err = rxe_init_ports(rxe);
+	if (err)
+		goto err1;
+
+	err = rxe_init_pools(rxe);
+	if (err)
+		goto err2;
+
+	/* init pending mmap list */
+	spin_lock_init(&rxe->mmap_offset_lock);
+	spin_lock_init(&rxe->pending_lock);
+	INIT_LIST_HEAD(&rxe->pending_mmaps);
+	INIT_LIST_HEAD(&rxe->list);
+
+	mutex_init(&rxe->usdev_lock);
+
+	return 0;
+
+err2:
+	rxe_cleanup_ports(rxe);
+err1:
+	return err;
+}
+
+int rxe_set_mtu(struct rxe_dev *rxe, unsigned int ndev_mtu)
+{
+	struct rxe_port *port = &rxe->port;
+	enum ib_mtu mtu;
+
+	mtu = eth_mtu_int_to_enum(ndev_mtu);
+
+	/* Make sure that new MTU in range */
+	mtu = mtu ? min_t(enum ib_mtu, mtu, RXE_PORT_MAX_MTU) : IB_MTU_256;
+
+	port->attr.active_mtu = mtu;
+	port->mtu_cap = ib_mtu_enum_to_int(mtu);
+
+	return 0;
+}
+EXPORT_SYMBOL(rxe_set_mtu);
+
+/* called by ifc layer to create new rxe device.
+ * The caller should allocate memory for rxe by calling ib_alloc_device.
+ */
+int rxe_add(struct rxe_dev *rxe, unsigned int mtu)
+{
+	int err;
+
+	kref_init(&rxe->ref_cnt);
+
+	err = rxe_init(rxe);
+	if (err)
+		goto err1;
+
+	err = rxe_set_mtu(rxe, mtu);
+	if (err)
+		goto err1;
+
+	err = rxe_register_device(rxe);
+	if (err)
+		goto err1;
+
+	return 0;
+
+err1:
+	rxe_dev_put(rxe);
+	return err;
+}
+EXPORT_SYMBOL(rxe_add);
+
+/* called by the ifc layer to remove a device */
+void rxe_remove(struct rxe_dev *rxe)
+{
+	rxe_unregister_device(rxe);
+
+	rxe_dev_put(rxe);
+}
+EXPORT_SYMBOL(rxe_remove);
+
+static int __init rxe_module_init(void)
+{
+	int err;
+
+	/* initialize slab caches for managed objects */
+	err = rxe_cache_init();
+	if (err) {
+		pr_err("rxe: unable to init object pools\n");
+		return err;
+	}
+
+	err = rxe_net_init();
+	if (err) {
+		pr_err("rxe: unable to init\n");
+		rxe_cache_exit();
+		return err;
+	}
+	pr_info("rxe: loaded\n");
+
+	return 0;
+}
+
+static void __exit rxe_module_exit(void)
+{
+	rxe_remove_all();
+	rxe_net_exit();
+	rxe_cache_exit();
+
+	pr_info("rxe: unloaded\n");
+}
+
+module_init(rxe_module_init);
+module_exit(rxe_module_exit);

drivers/infiniband/sw/rxe/rxe.h

+/*
+ * Copyright (c) 2016 Mellanox Technologies Ltd. All rights reserved.
+ * Copyright (c) 2015 System Fabric Works, 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 RXE_H
+#define RXE_H
+
+#include <linux/module.h>
+#include <linux/skbuff.h>
+#include <linux/crc32.h>
+
+#include <rdma/ib_verbs.h>
+#include <rdma/ib_user_verbs.h>
+#include <rdma/ib_pack.h>
+#include <rdma/ib_smi.h>
+#include <rdma/ib_umem.h>
+#include <rdma/ib_cache.h>
+#include <rdma/ib_addr.h>
+
+#include "rxe_net.h"
+#include "rxe_opcode.h"
+#include "rxe_hdr.h"
+#include "rxe_param.h"
+#include "rxe_verbs.h"
+
+#define RXE_UVERBS_ABI_VERSION		(1)
+
+#define IB_PHYS_STATE_LINK_UP		(5)
+#define IB_PHYS_STATE_LINK_DOWN		(3)
+
+#define RXE_ROCE_V2_SPORT		(0xc000)
+
+int rxe_set_mtu(struct rxe_dev *rxe, unsigned int dev_mtu);
+
+int rxe_add(struct rxe_dev *rxe, unsigned int mtu);
+void rxe_remove(struct rxe_dev *rxe);
+void rxe_remove_all(void);
+
+int rxe_rcv(struct sk_buff *skb);
+
+void rxe_dev_put(struct rxe_dev *rxe);
+struct rxe_dev *net_to_rxe(struct net_device *ndev);
+struct rxe_dev *get_rxe_by_name(const char* name);
+
+void rxe_port_up(struct rxe_dev *rxe);
+void rxe_port_down(struct rxe_dev *rxe);
+
+#endif /* RXE_H */

drivers/infiniband/sw/rxe/rxe_av.c

+/*
+ * Copyright (c) 2016 Mellanox Technologies Ltd. All rights reserved.
+ * Copyright (c) 2015 System Fabric Works, 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 "rxe.h"
+#include "rxe_loc.h"
+
+int rxe_av_chk_attr(struct rxe_dev *rxe, struct ib_ah_attr *attr)
+{
+	struct rxe_port *port;
+
+	if (attr->port_num != 1) {
+		pr_info("rxe: invalid port_num = %d\n", attr->port_num);
+		return -EINVAL;
+	}
+
+	port = &rxe->port;
+
+	if (attr->ah_flags & IB_AH_GRH) {
+		if (attr->grh.sgid_index > port->attr.gid_tbl_len) {
+			pr_info("rxe: invalid sgid index = %d\n",
+				attr->grh.sgid_index);
+			return -EINVAL;
+		}
+	}
+
+	return 0;
+}
+
+int rxe_av_from_attr(struct rxe_dev *rxe, u8 port_num,
+		     struct rxe_av *av, struct ib_ah_attr *attr)
+{
+	memset(av, 0, sizeof(*av));
+	memcpy(&av->grh, &attr->grh, sizeof(attr->grh));
+	av->port_num = port_num;
+	return 0;
+}
+
+int rxe_av_to_attr(struct rxe_dev *rxe, struct rxe_av *av,
+		   struct ib_ah_attr *attr)
+{
+	memcpy(&attr->grh, &av->grh, sizeof(av->grh));
+	attr->port_num = av->port_num;
+	return 0;
+}
+
+int rxe_av_fill_ip_info(struct rxe_dev *rxe,
+			struct rxe_av *av,
+			struct ib_ah_attr *attr,
+			struct ib_gid_attr *sgid_attr,
+			union ib_gid *sgid)
+{
+	rdma_gid2ip(&av->sgid_addr._sockaddr, sgid);
+	rdma_gid2ip(&av->dgid_addr._sockaddr, &attr->grh.dgid);
+	av->network_type = ib_gid_to_network_type(sgid_attr->gid_type, sgid);
+
+	return 0;
+}
+
+struct rxe_av *rxe_get_av(struct rxe_pkt_info *pkt)
+{
+	if (!pkt || !pkt->qp)
+		return NULL;
+
+	if (qp_type(pkt->qp) == IB_QPT_RC || qp_type(pkt->qp) == IB_QPT_UC)
+		return &pkt->qp->pri_av;
+
+	return (pkt->wqe) ? &pkt->wqe->av : NULL;
+}

drivers/infiniband/sw/rxe/rxe_comp.c

+/*
+ * Copyright (c) 2016 Mellanox Technologies Ltd. All rights reserved.
+ * Copyright (c) 2015 System Fabric Works, 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/skbuff.h>
+
+#include "rxe.h"
+#include "rxe_loc.h"
+#include "rxe_queue.h"
+#include "rxe_task.h"
+
+enum comp_state {
+	COMPST_GET_ACK,
+	COMPST_GET_WQE,
+	COMPST_COMP_WQE,
+	COMPST_COMP_ACK,
+	COMPST_CHECK_PSN,
+	COMPST_CHECK_ACK,
+	COMPST_READ,
+	COMPST_ATOMIC,
+	COMPST_WRITE_SEND,
+	COMPST_UPDATE_COMP,
+	COMPST_ERROR_RETRY,
+	COMPST_RNR_RETRY,
+	COMPST_ERROR,
+	COMPST_EXIT, /* We have an issue, and we want to rerun the completer */
+	COMPST_DONE, /* The completer finished successflly */
+};
+
+static char *comp_state_name[] =  {
+	[COMPST_GET_ACK]		= "GET ACK",
+	[COMPST_GET_WQE]		= "GET WQE",
+	[COMPST_COMP_WQE]		= "COMP WQE",
+	[COMPST_COMP_ACK]		= "COMP ACK",
+	[COMPST_CHECK_PSN]		= "CHECK PSN",
+	[COMPST_CHECK_ACK]		= "CHECK ACK",
+	[COMPST_READ]			= "READ",
+	[COMPST_ATOMIC]			= "ATOMIC",
+	[COMPST_WRITE_SEND]		= "WRITE/SEND",
+	[COMPST_UPDATE_COMP]		= "UPDATE COMP",
+	[COMPST_ERROR_RETRY]		= "ERROR RETRY",
+	[COMPST_RNR_RETRY]		= "RNR RETRY",
+	[COMPST_ERROR]			= "ERROR",
+	[COMPST_EXIT]			= "EXIT",
+	[COMPST_DONE]			= "DONE",
+};
+
+static unsigned long rnrnak_usec[32] = {
+	[IB_RNR_TIMER_655_36] = 655360,
+	[IB_RNR_TIMER_000_01] = 10,
+	[IB_RNR_TIMER_000_02] = 20,
+	[IB_RNR_TIMER_000_03] = 30,
+	[IB_RNR_TIMER_000_04] = 40,
+	[IB_RNR_TIMER_000_06] = 60,
+	[IB_RNR_TIMER_000_08] = 80,
+	[IB_RNR_TIMER_000_12] = 120,
+	[IB_RNR_TIMER_000_16] = 160,
+	[IB_RNR_TIMER_000_24] = 240,
+	[IB_RNR_TIMER_000_32] = 320,
+	[IB_RNR_TIMER_000_48] = 480,
+	[IB_RNR_TIMER_000_64] = 640,
+	[IB_RNR_TIMER_000_96] = 960,
+	[IB_RNR_TIMER_001_28] = 1280,
+	[IB_RNR_TIMER_001_92] = 1920,
+	[IB_RNR_TIMER_002_56] = 2560,
+	[IB_RNR_TIMER_003_84] = 3840,
+	[IB_RNR_TIMER_005_12] = 5120,
+	[IB_RNR_TIMER_007_68] = 7680,
+	[IB_RNR_TIMER_010_24] = 10240,
+	[IB_RNR_TIMER_015_36] = 15360,
+	[IB_RNR_TIMER_020_48] = 20480,
+	[IB_RNR_TIMER_030_72] = 30720,
+	[IB_RNR_TIMER_040_96] = 40960,
+	[IB_RNR_TIMER_061_44] = 61410,
+	[IB_RNR_TIMER_081_92] = 81920,
+	[IB_RNR_TIMER_122_88] = 122880,
+	[IB_RNR_TIMER_163_84] = 163840,
+	[IB_RNR_TIMER_245_76] = 245760,
+	[IB_RNR_TIMER_327_68] = 327680,
+	[IB_RNR_TIMER_491_52] = 491520,
+};
+
+static inline unsigned long rnrnak_jiffies(u8 timeout)
+{
+	return max_t(unsigned long,
+		usecs_to_jiffies(rnrnak_usec[timeout]), 1);
+}
+
+static enum ib_wc_opcode wr_to_wc_opcode(enum ib_wr_opcode opcode)
+{
+	switch (opcode) {
+	case IB_WR_RDMA_WRITE:			return IB_WC_RDMA_WRITE;
+	case IB_WR_RDMA_WRITE_WITH_IMM:		return IB_WC_RDMA_WRITE;
+	case IB_WR_SEND:			return IB_WC_SEND;
+	case IB_WR_SEND_WITH_IMM:		return IB_WC_SEND;
+	case IB_WR_RDMA_READ:			return IB_WC_RDMA_READ;
+	case IB_WR_ATOMIC_CMP_AND_SWP:		return IB_WC_COMP_SWAP;
+	case IB_WR_ATOMIC_FETCH_AND_ADD:	return IB_WC_FETCH_ADD;
+	case IB_WR_LSO:				return IB_WC_LSO;
+	case IB_WR_SEND_WITH_INV:		return IB_WC_SEND;
+	case IB_WR_RDMA_READ_WITH_INV:		return IB_WC_RDMA_READ;
+	case IB_WR_LOCAL_INV:			return IB_WC_LOCAL_INV;
+	case IB_WR_REG_MR:			return IB_WC_REG_MR;
+
+	default:
+		return 0xff;
+	}
+}
+
+void retransmit_timer(unsigned long data)
+{
+	struct rxe_qp *qp = (struct rxe_qp *)data;
+
+	if (qp->valid) {
+		qp->comp.timeout = 1;
+		rxe_run_task(&qp->comp.task, 1);
+	}
+}
+
+void rxe_comp_queue_pkt(struct rxe_dev *rxe, struct rxe_qp *qp,
+			struct sk_buff *skb)
+{
+	int must_sched;
+
+	skb_queue_tail(&qp->resp_pkts, skb);
+
+	must_sched = skb_queue_len(&qp->resp_pkts) > 1;
+	rxe_run_task(&qp->comp.task, must_sched);
+}
+
+static inline enum comp_state get_wqe(struct rxe_qp *qp,
+				      struct rxe_pkt_info *pkt,
+				      struct rxe_send_wqe **wqe_p)
+{
+	struct rxe_send_wqe *wqe;
+
+	/* we come here whether or not we found a response packet to see if
+	 * there are any posted WQEs
+	 */
+	wqe = queue_head(qp->sq.queue);
+	*wqe_p = wqe;
+
+	/* no WQE or requester has not started it yet */
+	if (!wqe || wqe->state == wqe_state_posted)
+		return pkt ? COMPST_DONE : COMPST_EXIT;
+
+	/* WQE does not require an ack */
+	if (wqe->state == wqe_state_done)
+		return COMPST_COMP_WQE;
+
+	/* WQE caused an error */
+	if (wqe->state == wqe_state_error)
+		return COMPST_ERROR;
+
+	/* we have a WQE, if we also have an ack check its PSN */
+	return pkt ? COMPST_CHECK_PSN : COMPST_EXIT;
+}
+
+static inline void reset_retry_counters(struct rxe_qp *qp)
+{
+	qp->comp.retry_cnt = qp->attr.retry_cnt;
+	qp->comp.rnr_retry = qp->attr.rnr_retry;
+}
+
+static inline enum comp_state check_psn(struct rxe_qp *qp,
+					struct rxe_pkt_info *pkt,
+					struct rxe_send_wqe *wqe)
+{
+	s32 diff;
+
+	/* check to see if response is past the oldest WQE. if it is, complete
+	 * send/write or error read/atomic
+	 */
+	diff = psn_compare(pkt->psn, wqe->last_psn);
+	if (diff > 0) {
+		if (wqe->state == wqe_state_pending) {
+			if (wqe->mask & WR_ATOMIC_OR_READ_MASK)
+				return COMPST_ERROR_RETRY;
+
+			reset_retry_counters(qp);
+			return COMPST_COMP_WQE;
+		} else {
+			return COMPST_DONE;
+		}
+	}
+
+	/* compare response packet to expected response */
+	diff = psn_compare(pkt->psn, qp->comp.psn);
+	if (diff < 0) {
+		/* response is most likely a retried packet if it matches an
+		 * uncompleted WQE go complete it else ignore it
+		 */
+		if (pkt->psn == wqe->last_psn)
+			return COMPST_COMP_ACK;
+		else
+			return COMPST_DONE;
+	} else if ((diff > 0) && (wqe->mask & WR_ATOMIC_OR_READ_MASK)) {
+		return COMPST_ERROR_RETRY;
+	} else {
+		return COMPST_CHECK_ACK;
+	}
+}
+
+static inline enum comp_state check_ack(struct rxe_qp *qp,
+					struct rxe_pkt_info *pkt,
+					struct rxe_send_wqe *wqe)
+{
+	unsigned int mask = pkt->mask;
+	u8 syn;
+
+	/* Check the sequence only */
+	switch (qp->comp.opcode) {
+	case -1:
+		/* Will catch all *_ONLY cases. */
+		if (!(mask & RXE_START_MASK))
+			return COMPST_ERROR;
+
+		break;
+
+	case IB_OPCODE_RC_RDMA_READ_RESPONSE_FIRST:
+	case IB_OPCODE_RC_RDMA_READ_RESPONSE_MIDDLE:
+		if (pkt->opcode != IB_OPCODE_RC_RDMA_READ_RESPONSE_MIDDLE &&
+		    pkt->opcode != IB_OPCODE_RC_RDMA_READ_RESPONSE_LAST) {
+			return COMPST_ERROR;
+		}
+		break;
+	default:
+		WARN_ON(1);
+	}
+
+	/* Check operation validity. */
+	switch (pkt->opcode) {
+	case IB_OPCODE_RC_RDMA_READ_RESPONSE_FIRST:
+	case IB_OPCODE_RC_RDMA_READ_RESPONSE_LAST:
+	case IB_OPCODE_RC_RDMA_READ_RESPONSE_ONLY:
+		syn = aeth_syn(pkt);
+
+		if ((syn & AETH_TYPE_MASK) != AETH_ACK)
+			return COMPST_ERROR;
+
+		/* Fall through (IB_OPCODE_RC_RDMA_READ_RESPONSE_MIDDLE
+		 * doesn't have an AETH)
+		 */
+	case IB_OPCODE_RC_RDMA_READ_RESPONSE_MIDDLE:
+		if (wqe->wr.opcode != IB_WR_RDMA_READ &&
+		    wqe->wr.opcode != IB_WR_RDMA_READ_WITH_INV) {
+			return COMPST_ERROR;
+		}
+		reset_retry_counters(qp);
+		return COMPST_READ;
+
+	case IB_OPCODE_RC_ATOMIC_ACKNOWLEDGE:
+		syn = aeth_syn(pkt);
+
+		if ((syn & AETH_TYPE_MASK) != AETH_ACK)
+			return COMPST_ERROR;
+
+		if (wqe->wr.opcode != IB_WR_ATOMIC_CMP_AND_SWP &&
+		    wqe->wr.opcode != IB_WR_ATOMIC_FETCH_AND_ADD)
+			return COMPST_ERROR;
+		reset_retry_counters(qp);
+		return COMPST_ATOMIC;
+
+	case IB_OPCODE_RC_ACKNOWLEDGE:
+		syn = aeth_syn(pkt);
+		switch (syn & AETH_TYPE_MASK) {
+		case AETH_ACK:
+			reset_retry_counters(qp);
+			return COMPST_WRITE_SEND;
+
+		case AETH_RNR_NAK:
+			return COMPST_RNR_RETRY;
+
+		case AETH_NAK:
+			switch (syn) {
+			case AETH_NAK_PSN_SEQ_ERROR:
+				/* a nak implicitly acks all packets with psns
+				 * before
+				 */
+				if (psn_compare(pkt->psn, qp->comp.psn) > 0) {
+					qp->comp.psn = pkt->psn;
+					if (qp->req.wait_psn) {
+						qp->req.wait_psn = 0;
+						rxe_run_task(&qp->req.task, 1);
+					}
+				}
+				return COMPST_ERROR_RETRY;
+
+			case AETH_NAK_INVALID_REQ:
+				wqe->status = IB_WC_REM_INV_REQ_ERR;
+				return COMPST_ERROR;
+
+			case AETH_NAK_REM_ACC_ERR:
+				wqe->status = IB_WC_REM_ACCESS_ERR;
+				return COMPST_ERROR;
+
+			case AETH_NAK_REM_OP_ERR:
+				wqe->status = IB_WC_REM_OP_ERR;
+				return COMPST_ERROR;
+
+			default:
+				pr_warn("unexpected nak %x\n", syn);
+				wqe->status = IB_WC_REM_OP_ERR;
+				return COMPST_ERROR;
+			}
+
+		default:
+			return COMPST_ERROR;
+		}
+		break;
+
+	default:
+		pr_warn("unexpected opcode\n");
+	}
+
+	return COMPST_ERROR;
+}
+
+static inline enum comp_state do_read(struct rxe_qp *qp,
+				      struct rxe_pkt_info *pkt,
+				      struct rxe_send_wqe *wqe)
+{
+	struct rxe_dev *rxe = to_rdev(qp->ibqp.device);
+	int ret;
+
+	ret = copy_data(rxe, qp->pd, IB_ACCESS_LOCAL_WRITE,
+			&wqe->dma, payload_addr(pkt),
+			payload_size(pkt), to_mem_obj, NULL);
+	if (ret)
+		return COMPST_ERROR;
+
+	if (wqe->dma.resid == 0 && (pkt->mask & RXE_END_MASK))
+		return COMPST_COMP_ACK;
+	else
+		return COMPST_UPDATE_COMP;
+}
+
+static inline enum comp_state do_atomic(struct rxe_qp *qp,
+					struct rxe_pkt_info *pkt,
+					struct rxe_send_wqe *wqe)
+{
+	struct rxe_dev *rxe = to_rdev(qp->ibqp.device);
+	int ret;
+
+	u64 atomic_orig = atmack_orig(pkt);
+
+	ret = copy_data(rxe, qp->pd, IB_ACCESS_LOCAL_WRITE,
+			&wqe->dma, &atomic_orig,
+			sizeof(u64), to_mem_obj, NULL);
+	if (ret)
+		return COMPST_ERROR;
+	else
+		return COMPST_COMP_ACK;
+}
+
+static void make_send_cqe(struct rxe_qp *qp, struct rxe_send_wqe *wqe,
+			  struct rxe_cqe *cqe)
+{
+	memset(cqe, 0, sizeof(*cqe));
+
+	if (!qp->is_user) {
+		struct ib_wc		*wc	= &cqe->ibwc;
+
+		wc->wr_id		= wqe->wr.wr_id;
+		wc->status		= wqe->status;
+		wc->opcode		= wr_to_wc_opcode(wqe->wr.opcode);
+		if (wqe->wr.opcode == IB_WR_RDMA_WRITE_WITH_IMM ||
+		    wqe->wr.opcode == IB_WR_SEND_WITH_IMM)
+			wc->wc_flags = IB_WC_WITH_IMM;
+		wc->byte_len		= wqe->dma.length;
+		wc->qp			= &qp->ibqp;
+	} else {
+		struct ib_uverbs_wc	*uwc	= &cqe->uibwc;
+
+		uwc->wr_id		= wqe->wr.wr_id;
+		uwc->status		= wqe->status;
+		uwc->opcode		= wr_to_wc_opcode(wqe->wr.opcode);
+		if (wqe->wr.opcode == IB_WR_RDMA_WRITE_WITH_IMM ||
+		    wqe->wr.opcode == IB_WR_SEND_WITH_IMM)
+			uwc->wc_flags = IB_WC_WITH_IMM;
+		uwc->byte_len		= wqe->dma.length;
+		uwc->qp_num		= qp->ibqp.qp_num;
+	}
+}
+
+static void do_complete(struct rxe_qp *qp, struct rxe_send_wqe *wqe)
+{
+	struct rxe_cqe cqe;
+
+	if ((qp->sq_sig_type == IB_SIGNAL_ALL_WR) ||
+	    (wqe->wr.send_flags & IB_SEND_SIGNALED) ||
+	    (qp->req.state == QP_STATE_ERROR)) {
+		make_send_cqe(qp, wqe, &cqe);
+		rxe_cq_post(qp->scq, &cqe, 0);
+	}
+
+	advance_consumer(qp->sq.queue);
+
+	/*
+	 * we completed something so let req run again
+	 * if it is trying to fence
+	 */
+	if (qp->req.wait_fence) {
+		qp->req.wait_fence = 0;
+		rxe_run_task(&qp->req.task, 1);
+	}
+}
+
+static inline enum comp_state complete_ack(struct rxe_qp *qp,
+					   struct rxe_pkt_info *pkt,
+					   struct rxe_send_wqe *wqe)
+{
+	unsigned long flags;
+
+	if (wqe->has_rd_atomic) {
+		wqe->has_rd_atomic = 0;
+		atomic_inc(&qp->req.rd_atomic);
+		if (qp->req.need_rd_atomic) {
+			qp->comp.timeout_retry = 0;
+			qp->req.need_rd_atomic = 0;
+			rxe_run_task(&qp->req.task, 1);
+		}
+	}
+
+	if (unlikely(qp->req.state == QP_STATE_DRAIN)) {
+		/* state_lock used by requester & completer */
+		spin_lock_irqsave(&qp->state_lock, flags);
+		if ((qp->req.state == QP_STATE_DRAIN) &&
+		    (qp->comp.psn == qp->req.psn)) {
+			qp->req.state = QP_STATE_DRAINED;
+			spin_unlock_irqrestore(&qp->state_lock, flags);
+
+			if (qp->ibqp.event_handler) {
+				struct ib_event ev;
+
+				ev.device = qp->ibqp.device;
+				ev.element.qp = &qp->ibqp;
+				ev.event = IB_EVENT_SQ_DRAINED;
+				qp->ibqp.event_handler(&ev,
+					qp->ibqp.qp_context);
+			}
+		} else {
+			spin_unlock_irqrestore(&qp->state_lock, flags);
+		}
+	}
+
+	do_complete(qp, wqe);
+
+	if (psn_compare(pkt->psn, qp->comp.psn) >= 0)
+		return COMPST_UPDATE_COMP;
+	else
+		return COMPST_DONE;
+}
+
+static inline enum comp_state complete_wqe(struct rxe_qp *qp,
+					   struct rxe_pkt_info *pkt,
+					   struct rxe_send_wqe *wqe)
+{
+	qp->comp.opcode = -1;
+
+	if (pkt) {
+		if (psn_compare(pkt->psn, qp->comp.psn) >= 0)
+			qp->comp.psn = (pkt->psn + 1) & BTH_PSN_MASK;
+
+		if (qp->req.wait_psn) {
+			qp->req.wait_psn = 0;
+			rxe_run_task(&qp->req.task, 1);
+		}
+	}
+
+	do_complete(qp, wqe);
+
+	return COMPST_GET_WQE;
+}
+
+int rxe_completer(void *arg)
+{
+	struct rxe_qp *qp = (struct rxe_qp *)arg;
+	struct rxe_send_wqe *wqe = wqe;
+	struct sk_buff *skb = NULL;
+	struct rxe_pkt_info *pkt = NULL;
+	enum comp_state state;
+
+	if (!qp->valid) {
+		while ((skb = skb_dequeue(&qp->resp_pkts))) {
+			rxe_drop_ref(qp);
+			kfree_skb(skb);
+		}
+		skb = NULL;
+		pkt = NULL;
+
+		while (queue_head(qp->sq.queue))
+			advance_consumer(qp->sq.queue);
+
+		goto exit;
+	}
+
+	if (qp->req.state == QP_STATE_ERROR) {
+		while ((skb = skb_dequeue(&qp->resp_pkts))) {
+			rxe_drop_ref(qp);
+			kfree_skb(skb);
+		}
+		skb = NULL;
+		pkt = NULL;
+
+		while ((wqe = queue_head(qp->sq.queue))) {
+			wqe->status = IB_WC_WR_FLUSH_ERR;
+			do_complete(qp, wqe);
+		}
+
+		goto exit;
+	}
+
+	if (qp->req.state == QP_STATE_RESET) {
+		while ((skb = skb_dequeue(&qp->resp_pkts))) {
+			rxe_drop_ref(qp);
+			kfree_skb(skb);
+		}
+		skb = NULL;
+		pkt = NULL;
+
+		while (queue_head(qp->sq.queue))
+			advance_consumer(qp->sq.queue);
+
+		goto exit;
+	}
+
+	if (qp->comp.timeout) {
+		qp->comp.timeout_retry = 1;
+		qp->comp.timeout = 0;
+	} else {
+		qp->comp.timeout_retry = 0;
+	}
+
+	if (qp->req.need_retry)
+		goto exit;
+
+	state = COMPST_GET_ACK;
+
+	while (1) {
+		pr_debug("state = %s\n", comp_state_name[state]);
+		switch (state) {
+		case COMPST_GET_ACK:
+			skb = skb_dequeue(&qp->resp_pkts);
+			if (skb) {
+				pkt = SKB_TO_PKT(skb);
+				qp->comp.timeout_retry = 0;
+			}
+			state = COMPST_GET_WQE;
+			break;
+
+		case COMPST_GET_WQE:
+			state = get_wqe(qp, pkt, &wqe);
+			break;
+
+		case COMPST_CHECK_PSN:
+			state = check_psn(qp, pkt, wqe);
+			break;
+
+		case COMPST_CHECK_ACK:
+			state = check_ack(qp, pkt, wqe);
+			break;
+
+		case COMPST_READ:
+			state = do_read(qp, pkt, wqe);
+			break;
+
+		case COMPST_ATOMIC:
+			state = do_atomic(qp, pkt, wqe);
+			break;
+
+		case COMPST_WRITE_SEND:
+			if (wqe->state == wqe_state_pending &&
+			    wqe->last_psn == pkt->psn)
+				state = COMPST_COMP_ACK;
+			else
+				state = COMPST_UPDATE_COMP;
+			break;
+
+		case COMPST_COMP_ACK:
+			state = complete_ack(qp, pkt, wqe);
+			break;
+
+		case COMPST_COMP_WQE:
+			state = complete_wqe(qp, pkt, wqe);
+			break;
+
+		case COMPST_UPDATE_COMP:
+			if (pkt->mask & RXE_END_MASK)
+				qp->comp.opcode = -1;
+			else
+				qp->comp.opcode = pkt->opcode;
+
+			if (psn_compare(pkt->psn, qp->comp.psn) >= 0)
+				qp->comp.psn = (pkt->psn + 1) & BTH_PSN_MASK;
+
+			if (qp->req.wait_psn) {
+				qp->req.wait_psn = 0;
+				rxe_run_task(&qp->req.task, 1);
+			}
+
+			state = COMPST_DONE;
+			break;
+
+		case COMPST_DONE:
+			if (pkt) {
+				rxe_drop_ref(pkt->qp);
+				kfree_skb(skb);
+			}
+			goto done;
+
+		case COMPST_EXIT:
+			if (qp->comp.timeout_retry && wqe) {
+				state = COMPST_ERROR_RETRY;
+				break;
+			}
+
+			/* re reset the timeout counter if
+			 * (1) QP is type RC
+			 * (2) the QP is alive
+			 * (3) there is a packet sent by the requester that
+			 *     might be acked (we still might get spurious
+			 *     timeouts but try to keep them as few as possible)
+			 * (4) the timeout parameter is set
+			 */
+			if ((qp_type(qp) == IB_QPT_RC) &&
+			    (qp->req.state == QP_STATE_READY) &&
+			    (psn_compare(qp->req.psn, qp->comp.psn) > 0) &&
+			    qp->qp_timeout_jiffies)
+				mod_timer(&qp->retrans_timer,
+					  jiffies + qp->qp_timeout_jiffies);
+			goto exit;
+
+		case COMPST_ERROR_RETRY:
+			/* we come here if the retry timer fired and we did
+			 * not receive a response packet. try to retry the send
+			 * queue if that makes sense and the limits have not
+			 * been exceeded. remember that some timeouts are
+			 * spurious since we do not reset the timer but kick
+			 * it down the road or let it expire
+			 */
+
+			/* there is nothing to retry in this case */
+			if (!wqe || (wqe->state == wqe_state_posted))
+				goto exit;
+
+			if (qp->comp.retry_cnt > 0) {
+				if (qp->comp.retry_cnt != 7)
+					qp->comp.retry_cnt--;
+
+				/* no point in retrying if we have already
+				 * seen the last ack that the requester could
+				 * have caused
+				 */
+				if (psn_compare(qp->req.psn,
+						qp->comp.psn) > 0) {
+					/* tell the requester to retry the
+					 * send send queue next time around
+					 */
+					qp->req.need_retry = 1;
+					rxe_run_task(&qp->req.task, 1);
+				}
+				goto exit;
+			} else {
+				wqe->status = IB_WC_RETRY_EXC_ERR;
+				state = COMPST_ERROR;
+			}
+			break;
+
+		case COMPST_RNR_RETRY:
+			if (qp->comp.rnr_retry > 0) {
+				if (qp->comp.rnr_retry != 7)
+					qp->comp.rnr_retry--;
+
+				qp->req.need_retry = 1;
+				pr_debug("set rnr nak timer\n");
+				mod_timer(&qp->rnr_nak_timer,
+					  jiffies + rnrnak_jiffies(aeth_syn(pkt)
+						& ~AETH_TYPE_MASK));
+				goto exit;
+			} else {
+				wqe->status = IB_WC_RNR_RETRY_EXC_ERR;
+				state = COMPST_ERROR;
+			}
+			break;
+
+		case COMPST_ERROR:
+			do_complete(qp, wqe);
+			rxe_qp_error(qp);
+			goto exit;
+		}
+	}
+
+exit:
+	/* we come here if we are done with processing and want the task to
+	 * exit from the loop calling us
+	 */
+	return -EAGAIN;
+
+done:
+	/* we come here if we have processed a packet we want the task to call
+	 * us again to see if there is anything else to do
+	 */
+	return 0;
+}

drivers/infiniband/sw/rxe/rxe_cq.c

+/*
+ * Copyright (c) 2016 Mellanox Technologies Ltd. All rights reserved.
+ * Copyright (c) 2015 System Fabric Works, 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 "rxe.h"
+#include "rxe_loc.h"
+#include "rxe_queue.h"
+
+int rxe_cq_chk_attr(struct rxe_dev *rxe, struct rxe_cq *cq,
+		    int cqe, int comp_vector, struct ib_udata *udata)
+{
+	int count;
+
+	if (cqe <= 0) {
+		pr_warn("cqe(%d) <= 0\n", cqe);
+		goto err1;
+	}
+
+	if (cqe > rxe->attr.max_cqe) {
+		pr_warn("cqe(%d) > max_cqe(%d)\n",
+			cqe, rxe->attr.max_cqe);
+		goto err1;
+	}
+
+	if (cq) {
+		count = queue_count(cq->queue);
+		if (cqe < count) {
+			pr_warn("cqe(%d) < current # elements in queue (%d)",
+				cqe, count);
+			goto err1;
+		}
+	}
+
+	return 0;
+
+err1:
+	return -EINVAL;
+}
+
+static void rxe_send_complete(unsigned long data)
+{
+	struct rxe_cq *cq = (struct rxe_cq *)data;
+
+	cq->ibcq.comp_handler(&cq->ibcq, cq->ibcq.cq_context);
+}
+
+int rxe_cq_from_init(struct rxe_dev *rxe, struct rxe_cq *cq, int cqe,
+		     int comp_vector, struct ib_ucontext *context,
+		     struct ib_udata *udata)
+{
+	int err;
+
+	cq->queue = rxe_queue_init(rxe, &cqe,
+				   sizeof(struct rxe_cqe));
+	if (!cq->queue) {
+		pr_warn("unable to create cq\n");
+		return -ENOMEM;
+	}
+
+	err = do_mmap_info(rxe, udata, false, context, cq->queue->buf,
+			   cq->queue->buf_size, &cq->queue->ip);
+	if (err) {
+		kvfree(cq->queue->buf);
+		kfree(cq->queue);
+		return err;
+	}
+
+	if (udata)
+		cq->is_user = 1;
+
+	tasklet_init(&cq->comp_task, rxe_send_complete, (unsigned long)cq);
+
+	spin_lock_init(&cq->cq_lock);
+	cq->ibcq.cqe = cqe;
+	return 0;
+}
+
+int rxe_cq_resize_queue(struct rxe_cq *cq, int cqe, struct ib_udata *udata)
+{
+	int err;
+
+	err = rxe_queue_resize(cq->queue, (unsigned int *)&cqe,
+			       sizeof(struct rxe_cqe),
+			       cq->queue->ip ? cq->queue->ip->context : NULL,
+			       udata, NULL, &cq->cq_lock);
+	if (!err)
+		cq->ibcq.cqe = cqe;
+
+	return err;
+}
+
+int rxe_cq_post(struct rxe_cq *cq, struct rxe_cqe *cqe, int solicited)
+{
+	struct ib_event ev;
+	unsigned long flags;
+
+	spin_lock_irqsave(&cq->cq_lock, flags);
+
+	if (unlikely(queue_full(cq->queue))) {
+		spin_unlock_irqrestore(&cq->cq_lock, flags);
+		if (cq->ibcq.event_handler) {
+			ev.device = cq->ibcq.device;
+			ev.element.cq = &cq->ibcq;
+			ev.event = IB_EVENT_CQ_ERR;
+			cq->ibcq.event_handler(&ev, cq->ibcq.cq_context);
+		}
+
+		return -EBUSY;
+	}
+
+	memcpy(producer_addr(cq->queue), cqe, sizeof(*cqe));
+
+	/* make sure all changes to the CQ are written before we update the
+	 * producer pointer
+	 */
+	smp_wmb();
+
+	advance_producer(cq->queue);
+	spin_unlock_irqrestore(&cq->cq_lock, flags);
+
+	if ((cq->notify == IB_CQ_NEXT_COMP) ||
+	    (cq->notify == IB_CQ_SOLICITED && solicited)) {
+		cq->notify = 0;
+		tasklet_schedule(&cq->comp_task);
+	}
+
+	return 0;
+}
+
+void rxe_cq_cleanup(void *arg)
+{
+	struct rxe_cq *cq = arg;
+
+	if (cq->queue)
+		rxe_queue_cleanup(cq->queue);
+}

drivers/infiniband/sw/rxe/rxe_dma.c

+/*
+ * Copyright (c) 2016 Mellanox Technologies Ltd. All rights reserved.
+ * Copyright (c) 2015 System Fabric Works, 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 "rxe.h"
+#include "rxe_loc.h"
+
+#define DMA_BAD_ADDER ((u64)0)
+
+static int rxe_mapping_error(struct ib_device *dev, u64 dma_addr)
+{
+	return dma_addr == DMA_BAD_ADDER;
+}
+
+static u64 rxe_dma_map_single(struct ib_device *dev,
+			      void *cpu_addr, size_t size,
+			      enum dma_data_direction direction)
+{
+	WARN_ON(!valid_dma_direction(direction));
+	return (uintptr_t)cpu_addr;
+}
+
+static void rxe_dma_unmap_single(struct ib_device *dev,
+				 u64 addr, size_t size,
+				 enum dma_data_direction direction)
+{
+	WARN_ON(!valid_dma_direction(direction));
+}
+
+static u64 rxe_dma_map_page(struct ib_device *dev,
+			    struct page *page,
+			    unsigned long offset,
+			    size_t size, enum dma_data_direction direction)
+{
+	u64 addr;
+
+	WARN_ON(!valid_dma_direction(direction));
+
+	if (offset + size > PAGE_SIZE) {
+		addr = DMA_BAD_ADDER;
+		goto done;
+	}
+
+	addr = (uintptr_t)page_address(page);
+	if (addr)
+		addr += offset;
+
+done:
+	return addr;
+}
+
+static void rxe_dma_unmap_page(struct ib_device *dev,
+			       u64 addr, size_t size,
+			       enum dma_data_direction direction)
+{
+	WARN_ON(!valid_dma_direction(direction));
+}
+
+static int rxe_map_sg(struct ib_device *dev, struct scatterlist *sgl,
+		      int nents, enum dma_data_direction direction)
+{
+	struct scatterlist *sg;
+	u64 addr;
+	int i;
+	int ret = nents;
+
+	WARN_ON(!valid_dma_direction(direction));
+
+	for_each_sg(sgl, sg, nents, i) {
+		addr = (uintptr_t)page_address(sg_page(sg));
+		if (!addr) {
+			ret = 0;
+			break;
+		}
+		sg->dma_address = addr + sg->offset;
+#ifdef CONFIG_NEED_SG_DMA_LENGTH
+		sg->dma_length = sg->length;
+#endif
+	}
+
+	return ret;
+}
+
+static void rxe_unmap_sg(struct ib_device *dev,
+			 struct scatterlist *sg, int nents,
+			 enum dma_data_direction direction)
+{
+	WARN_ON(!valid_dma_direction(direction));
+}
+
+static void rxe_sync_single_for_cpu(struct ib_device *dev,
+				    u64 addr,
+				    size_t size, enum dma_data_direction dir)
+{
+}
+
+static void rxe_sync_single_for_device(struct ib_device *dev,
+				       u64 addr,
+				       size_t size, enum dma_data_direction dir)
+{
+}
+
+static void *rxe_dma_alloc_coherent(struct ib_device *dev, size_t size,
+				    u64 *dma_handle, gfp_t flag)
+{
+	struct page *p;
+	void *addr = NULL;
+
+	p = alloc_pages(flag, get_order(size));
+	if (p)
+		addr = page_address(p);
+
+	if (dma_handle)
+		*dma_handle = (uintptr_t)addr;
+
+	return addr;
+}
+
+static void rxe_dma_free_coherent(struct ib_device *dev, size_t size,
+				  void *cpu_addr, u64 dma_handle)
+{
+	free_pages((unsigned long)cpu_addr, get_order(size));
+}
+
+struct ib_dma_mapping_ops rxe_dma_mapping_ops = {
+	.mapping_error		= rxe_mapping_error,
+	.map_single		= rxe_dma_map_single,
+	.unmap_single		= rxe_dma_unmap_single,
+	.map_page		= rxe_dma_map_page,
+	.unmap_page		= rxe_dma_unmap_page,
+	.map_sg			= rxe_map_sg,
+	.unmap_sg		= rxe_unmap_sg,
+	.sync_single_for_cpu	= rxe_sync_single_for_cpu,
+	.sync_single_for_device	= rxe_sync_single_for_device,
+	.alloc_coherent		= rxe_dma_alloc_coherent,
+	.free_coherent		= rxe_dma_free_coherent
+};

drivers/infiniband/sw/rxe/rxe_icrc.c

+/*
+ * Copyright (c) 2016 Mellanox Technologies Ltd. All rights reserved.
+ * Copyright (c) 2015 System Fabric Works, 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 "rxe.h"
+#include "rxe_loc.h"
+
+/* Compute a partial ICRC for all the IB transport headers. */
+u32 rxe_icrc_hdr(struct rxe_pkt_info *pkt, struct sk_buff *skb)
+{
+	unsigned int bth_offset = 0;
+	struct iphdr *ip4h = NULL;
+	struct ipv6hdr *ip6h = NULL;
+	struct udphdr *udph;
+	struct rxe_bth *bth;
+	int crc;
+	int length;
+	int hdr_size = sizeof(struct udphdr) +
+		(skb->protocol == htons(ETH_P_IP) ?
+		sizeof(struct iphdr) : sizeof(struct ipv6hdr));
+	/* pseudo header buffer size is calculate using ipv6 header size since
+	 * it is bigger than ipv4
+	 */
+	u8 pshdr[sizeof(struct udphdr) +
+		sizeof(struct ipv6hdr) +
+		RXE_BTH_BYTES];
+
+	/* This seed is the result of computing a CRC with a seed of
+	 * 0xfffffff and 8 bytes of 0xff representing a masked LRH.
+	 */
+	crc = 0xdebb20e3;
+
+	if (skb->protocol == htons(ETH_P_IP)) { /* IPv4 */
+		memcpy(pshdr, ip_hdr(skb), hdr_size);
+		ip4h = (struct iphdr *)pshdr;
+		udph = (struct udphdr *)(ip4h + 1);
+
+		ip4h->ttl = 0xff;
+		ip4h->check = CSUM_MANGLED_0;
+		ip4h->tos = 0xff;
+	} else {				/* IPv6 */
+		memcpy(pshdr, ipv6_hdr(skb), hdr_size);
+		ip6h = (struct ipv6hdr *)pshdr;
+		udph = (struct udphdr *)(ip6h + 1);
+
+		memset(ip6h->flow_lbl, 0xff, sizeof(ip6h->flow_lbl));
+		ip6h->priority = 0xf;
+		ip6h->hop_limit = 0xff;
+	}
+	udph->check = CSUM_MANGLED_0;
+
+	bth_offset += hdr_size;
+
+	memcpy(&pshdr[bth_offset], pkt->hdr, RXE_BTH_BYTES);
+	bth = (struct rxe_bth *)&pshdr[bth_offset];
+
+	/* exclude bth.resv8a */
+	bth->qpn |= cpu_to_be32(~BTH_QPN_MASK);
+
+	length = hdr_size + RXE_BTH_BYTES;
+	crc = crc32_le(crc, pshdr, length);
+
+	/* And finish to compute the CRC on the remainder of the headers. */
+	crc = crc32_le(crc, pkt->hdr + RXE_BTH_BYTES,
+		       rxe_opcode[pkt->opcode].length - RXE_BTH_BYTES);
+	return crc;
+}

drivers/infiniband/sw/rxe/rxe_loc.h

+/*
+ * Copyright (c) 2016 Mellanox Technologies Ltd. All rights reserved.
+ * Copyright (c) 2015 System Fabric Works, 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 RXE_LOC_H
+#define RXE_LOC_H
+
+/* rxe_av.c */
+
+int rxe_av_chk_attr(struct rxe_dev *rxe, struct ib_ah_attr *attr);
+
+int rxe_av_from_attr(struct rxe_dev *rxe, u8 port_num,
+		     struct rxe_av *av, struct ib_ah_attr *attr);
+
+int rxe_av_to_attr(struct rxe_dev *rxe, struct rxe_av *av,
+		   struct ib_ah_attr *attr);
+
+int rxe_av_fill_ip_info(struct rxe_dev *rxe,
+			struct rxe_av *av,
+			struct ib_ah_attr *attr,
+			struct ib_gid_attr *sgid_attr,
+			union ib_gid *sgid);
+
+struct rxe_av *rxe_get_av(struct rxe_pkt_info *pkt);
+
+/* rxe_cq.c */
+int rxe_cq_chk_attr(struct rxe_dev *rxe, struct rxe_cq *cq,
+		    int cqe, int comp_vector, struct ib_udata *udata);
+
+int rxe_cq_from_init(struct rxe_dev *rxe, struct rxe_cq *cq, int cqe,
+		     int comp_vector, struct ib_ucontext *context,
+		     struct ib_udata *udata);
+
+int rxe_cq_resize_queue(struct rxe_cq *cq, int new_cqe, struct ib_udata *udata);
+
+int rxe_cq_post(struct rxe_cq *cq, struct rxe_cqe *cqe, int solicited);
+
+void rxe_cq_cleanup(void *arg);
+
+/* rxe_mcast.c */
+int rxe_mcast_get_grp(struct rxe_dev *rxe, union ib_gid *mgid,
+		      struct rxe_mc_grp **grp_p);
+
+int rxe_mcast_add_grp_elem(struct rxe_dev *rxe, struct rxe_qp *qp,
+			   struct rxe_mc_grp *grp);
+
+int rxe_mcast_drop_grp_elem(struct rxe_dev *rxe, struct rxe_qp *qp,
+			    union ib_gid *mgid);
+
+void rxe_drop_all_mcast_groups(struct rxe_qp *qp);
+
+void rxe_mc_cleanup(void *arg);
+
+/* rxe_mmap.c */
+struct rxe_mmap_info {
+	struct list_head	pending_mmaps;
+	struct ib_ucontext	*context;
+	struct kref		ref;
+	void			*obj;
+
+	struct mminfo info;
+};
+
+void rxe_mmap_release(struct kref *ref);
+
+struct rxe_mmap_info *rxe_create_mmap_info(struct rxe_dev *dev,
+					   u32 size,
+					   struct ib_ucontext *context,
+					   void *obj);
+
+int rxe_mmap(struct ib_ucontext *context, struct vm_area_struct *vma);
+
+/* rxe_mr.c */
+enum copy_direction {
+	to_mem_obj,
+	from_mem_obj,
+};
+
+int rxe_mem_init_dma(struct rxe_dev *rxe, struct rxe_pd *pd,
+		     int access, struct rxe_mem *mem);
+
+int rxe_mem_init_user(struct rxe_dev *rxe, struct rxe_pd *pd, u64 start,
+		      u64 length, u64 iova, int access, struct ib_udata *udata,
+		      struct rxe_mem *mr);
+
+int rxe_mem_init_fast(struct rxe_dev *rxe, struct rxe_pd *pd,
+		      int max_pages, struct rxe_mem *mem);
+
+int rxe_mem_copy(struct rxe_mem *mem, u64 iova, void *addr,
+		 int length, enum copy_direction dir, u32 *crcp);
+
+int copy_data(struct rxe_dev *rxe, struct rxe_pd *pd, int access,
+	      struct rxe_dma_info *dma, void *addr, int length,
+	      enum copy_direction dir, u32 *crcp);
+
+void *iova_to_vaddr(struct rxe_mem *mem, u64 iova, int length);
+
+enum lookup_type {
+	lookup_local,
+	lookup_remote,
+};
+
+struct rxe_mem *lookup_mem(struct rxe_pd *pd, int access, u32 key,
+			   enum lookup_type type);
+
+int mem_check_range(struct rxe_mem *mem, u64 iova, size_t length);
+
+int rxe_mem_map_pages(struct rxe_dev *rxe, struct rxe_mem *mem,
+		      u64 *page, int num_pages, u64 iova);
+
+void rxe_mem_cleanup(void *arg);
+
+int advance_dma_data(struct rxe_dma_info *dma, unsigned int length);
+
+/* rxe_qp.c */
+int rxe_qp_chk_init(struct rxe_dev *rxe, struct ib_qp_init_attr *init);
+
+int rxe_qp_from_init(struct rxe_dev *rxe, struct rxe_qp *qp, struct rxe_pd *pd,
+		     struct ib_qp_init_attr *init, struct ib_udata *udata,
+		     struct ib_pd *ibpd);
+
+int rxe_qp_to_init(struct rxe_qp *qp, struct ib_qp_init_attr *init);
+
+int rxe_qp_chk_attr(struct rxe_dev *rxe, struct rxe_qp *qp,
+		    struct ib_qp_attr *attr, int mask);
+
+int rxe_qp_from_attr(struct rxe_qp *qp, struct ib_qp_attr *attr,
+		     int mask, struct ib_udata *udata);
+
+int rxe_qp_to_attr(struct rxe_qp *qp, struct ib_qp_attr *attr, int mask);
+
+void rxe_qp_error(struct rxe_qp *qp);
+
+void rxe_qp_destroy(struct rxe_qp *qp);
+
+void rxe_qp_cleanup(void *arg);
+
+static inline int qp_num(struct rxe_qp *qp)
+{
+	return qp->ibqp.qp_num;
+}
+
+static inline enum ib_qp_type qp_type(struct rxe_qp *qp)
+{
+	return qp->ibqp.qp_type;
+}
+
+static inline enum ib_qp_state qp_state(struct rxe_qp *qp)
+{
+	return qp->attr.qp_state;
+}
+
+static inline int qp_mtu(struct rxe_qp *qp)
+{
+	if (qp->ibqp.qp_type == IB_QPT_RC || qp->ibqp.qp_type == IB_QPT_UC)
+		return qp->attr.path_mtu;
+	else
+		return RXE_PORT_MAX_MTU;
+}
+
+static inline int rcv_wqe_size(int max_sge)
+{
+	return sizeof(struct rxe_recv_wqe) +
+		max_sge * sizeof(struct ib_sge);
+}
+
+void free_rd_atomic_resource(struct rxe_qp *qp, struct resp_res *res);
+
+static inline void rxe_advance_resp_resource(struct rxe_qp *qp)
+{
+	qp->resp.res_head++;
+	if (unlikely(qp->resp.res_head == qp->attr.max_rd_atomic))
+		qp->resp.res_head = 0;
+}
+
+void retransmit_timer(unsigned long data);
+void rnr_nak_timer(unsigned long data);
+
+void dump_qp(struct rxe_qp *qp);
+
+/* rxe_srq.c */
+#define IB_SRQ_INIT_MASK (~IB_SRQ_LIMIT)
+
+int rxe_srq_chk_attr(struct rxe_dev *rxe, struct rxe_srq *srq,
+		     struct ib_srq_attr *attr, enum ib_srq_attr_mask mask);
+
+int rxe_srq_from_init(struct rxe_dev *rxe, struct rxe_srq *srq,
+		      struct ib_srq_init_attr *init,
+		      struct ib_ucontext *context, struct ib_udata *udata);
+
+int rxe_srq_from_attr(struct rxe_dev *rxe, struct rxe_srq *srq,
+		      struct ib_srq_attr *attr, enum ib_srq_attr_mask mask,
+		      struct ib_udata *udata);
+
+extern struct ib_dma_mapping_ops rxe_dma_mapping_ops;
+
+void rxe_release(struct kref *kref);
+
+int rxe_completer(void *arg);
+int rxe_requester(void *arg);
+int rxe_responder(void *arg);
+
+u32 rxe_icrc_hdr(struct rxe_pkt_info *pkt, struct sk_buff *skb);
+
+void rxe_resp_queue_pkt(struct rxe_dev *rxe,
+			struct rxe_qp *qp, struct sk_buff *skb);
+
+void rxe_comp_queue_pkt(struct rxe_dev *rxe,
+			struct rxe_qp *qp, struct sk_buff *skb);
+
+static inline unsigned wr_opcode_mask(int opcode, struct rxe_qp *qp)
+{
+	return rxe_wr_opcode_info[opcode].mask[qp->ibqp.qp_type];
+}
+
+static inline int rxe_xmit_packet(struct rxe_dev *rxe, struct rxe_qp *qp,
+				  struct rxe_pkt_info *pkt, struct sk_buff *skb)
+{
+	int err;
+	int is_request = pkt->mask & RXE_REQ_MASK;
+
+	if ((is_request && (qp->req.state != QP_STATE_READY)) ||
+	    (!is_request && (qp->resp.state != QP_STATE_READY))) {
+		pr_info("Packet dropped. QP is not in ready state\n");
+		goto drop;
+	}
+
+	if (pkt->mask & RXE_LOOPBACK_MASK) {
+		memcpy(SKB_TO_PKT(skb), pkt, sizeof(*pkt));
+		err = rxe->ifc_ops->loopback(skb);
+	} else {
+		err = rxe->ifc_ops->send(rxe, pkt, skb);
+	}
+
+	if (err) {
+		rxe->xmit_errors++;
+		return err;
+	}
+
+	atomic_inc(&qp->skb_out);
+
+	if ((qp_type(qp) != IB_QPT_RC) &&
+	    (pkt->mask & RXE_END_MASK)) {
+		pkt->wqe->state = wqe_state_done;
+		rxe_run_task(&qp->comp.task, 1);
+	}
+
+	goto done;
+
+drop:
+	kfree_skb(skb);
+	err = 0;
+done:
+	return err;
+}
+
+#endif /* RXE_LOC_H */

drivers/infiniband/sw/rxe/rxe_mcast.c

+/*
+ * Copyright (c) 2016 Mellanox Technologies Ltd. All rights reserved.
+ * Copyright (c) 2015 System Fabric Works, 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 "rxe.h"
+#include "rxe_loc.h"
+
+int rxe_mcast_get_grp(struct rxe_dev *rxe, union ib_gid *mgid,
+		      struct rxe_mc_grp **grp_p)
+{
+	int err;
+	struct rxe_mc_grp *grp;
+
+	if (rxe->attr.max_mcast_qp_attach == 0) {
+		err = -EINVAL;
+		goto err1;
+	}
+
+	grp = rxe_pool_get_key(&rxe->mc_grp_pool, mgid);
+	if (grp)
+		goto done;
+
+	grp = rxe_alloc(&rxe->mc_grp_pool);
+	if (!grp) {
+		err = -ENOMEM;
+		goto err1;
+	}
+
+	INIT_LIST_HEAD(&grp->qp_list);
+	spin_lock_init(&grp->mcg_lock);
+	grp->rxe = rxe;
+
+	rxe_add_key(grp, mgid);
+
+	err = rxe->ifc_ops->mcast_add(rxe, mgid);
+	if (err)
+		goto err2;
+
+done:
+	*grp_p = grp;
+	return 0;
+
+err2:
+	rxe_drop_ref(grp);
+err1:
+	return err;
+}
+
+int rxe_mcast_add_grp_elem(struct rxe_dev *rxe, struct rxe_qp *qp,
+			   struct rxe_mc_grp *grp)
+{
+	int err;
+	struct rxe_mc_elem *elem;
+
+	/* check to see of the qp is already a member of the group */
+	spin_lock_bh(&qp->grp_lock);
+	spin_lock_bh(&grp->mcg_lock);
+	list_for_each_entry(elem, &grp->qp_list, qp_list) {
+		if (elem->qp == qp) {
+			err = 0;
+			goto out;
+		}
+	}
+
+	if (grp->num_qp >= rxe->attr.max_mcast_qp_attach) {
+		err = -ENOMEM;
+		goto out;
+	}
+
+	elem = rxe_alloc(&rxe->mc_elem_pool);
+	if (!elem) {
+		err = -ENOMEM;
+		goto out;
+	}
+
+	/* each qp holds a ref on the grp */
+	rxe_add_ref(grp);
+
+	grp->num_qp++;
+	elem->qp = qp;
+	elem->grp = grp;
+
+	list_add(&elem->qp_list, &grp->qp_list);
+	list_add(&elem->grp_list, &qp->grp_list);
+
+	err = 0;
+out:
+	spin_unlock_bh(&grp->mcg_lock);
+	spin_unlock_bh(&qp->grp_lock);
+	return err;
+}
+
+int rxe_mcast_drop_grp_elem(struct rxe_dev *rxe, struct rxe_qp *qp,
+			    union ib_gid *mgid)
+{
+	struct rxe_mc_grp *grp;
+	struct rxe_mc_elem *elem, *tmp;
+
+	grp = rxe_pool_get_key(&rxe->mc_grp_pool, mgid);
+	if (!grp)
+		goto err1;
+
+	spin_lock_bh(&qp->grp_lock);
+	spin_lock_bh(&grp->mcg_lock);
+
+	list_for_each_entry_safe(elem, tmp, &grp->qp_list, qp_list) {
+		if (elem->qp == qp) {
+			list_del(&elem->qp_list);
+			list_del(&elem->grp_list);
+			grp->num_qp--;
+
+			spin_unlock_bh(&grp->mcg_lock);
+			spin_unlock_bh(&qp->grp_lock);
+			rxe_drop_ref(elem);
+			rxe_drop_ref(grp);	/* ref held by QP */
+			rxe_drop_ref(grp);	/* ref from get_key */
+			return 0;
+		}
+	}
+
+	spin_unlock_bh(&grp->mcg_lock);
+	spin_unlock_bh(&qp->grp_lock);
+	rxe_drop_ref(grp);			/* ref from get_key */
+err1:
+	return -EINVAL;
+}
+
+void rxe_drop_all_mcast_groups(struct rxe_qp *qp)
+{
+	struct rxe_mc_grp *grp;
+	struct rxe_mc_elem *elem;
+
+	while (1) {
+		spin_lock_bh(&qp->grp_lock);
+		if (list_empty(&qp->grp_list)) {
+			spin_unlock_bh(&qp->grp_lock);
+			break;
+		}
+		elem = list_first_entry(&qp->grp_list, struct rxe_mc_elem,
+					grp_list);
+		list_del(&elem->grp_list);
+		spin_unlock_bh(&qp->grp_lock);
+
+		grp = elem->grp;
+		spin_lock_bh(&grp->mcg_lock);
+		list_del(&elem->qp_list);
+		grp->num_qp--;
+		spin_unlock_bh(&grp->mcg_lock);
+		rxe_drop_ref(grp);
+		rxe_drop_ref(elem);
+	}
+}
+
+void rxe_mc_cleanup(void *arg)
+{
+	struct rxe_mc_grp *grp = arg;
+	struct rxe_dev *rxe = grp->rxe;
+
+	rxe_drop_key(grp);
+	rxe->ifc_ops->mcast_delete(rxe, &grp->mgid);
+}

drivers/infiniband/sw/rxe/rxe_mmap.c

+/*
+ * Copyright (c) 2016 Mellanox Technologies Ltd. All rights reserved.
+ * Copyright (c) 2015 System Fabric Works, 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/vmalloc.h>
+#include <linux/mm.h>
+#include <linux/errno.h>
+#include <asm/pgtable.h>
+
+#include "rxe.h"
+#include "rxe_loc.h"
+#include "rxe_queue.h"
+
+void rxe_mmap_release(struct kref *ref)
+{
+	struct rxe_mmap_info *ip = container_of(ref,
+					struct rxe_mmap_info, ref);
+	struct rxe_dev *rxe = to_rdev(ip->context->device);
+
+	spin_lock_bh(&rxe->pending_lock);
+
+	if (!list_empty(&ip->pending_mmaps))
+		list_del(&ip->pending_mmaps);
+
+	spin_unlock_bh(&rxe->pending_lock);
+
+	vfree(ip->obj);		/* buf */
+	kfree(ip);
+}
+
+/*
+ * open and close keep track of how many times the memory region is mapped,
+ * to avoid releasing it.
+ */
+static void rxe_vma_open(struct vm_area_struct *vma)
+{
+	struct rxe_mmap_info *ip = vma->vm_private_data;
+
+	kref_get(&ip->ref);
+}
+
+static void rxe_vma_close(struct vm_area_struct *vma)
+{
+	struct rxe_mmap_info *ip = vma->vm_private_data;
+
+	kref_put(&ip->ref, rxe_mmap_release);
+}
+
+static struct vm_operations_struct rxe_vm_ops = {
+	.open = rxe_vma_open,
+	.close = rxe_vma_close,
+};
+
+/**
+ * rxe_mmap - create a new mmap region
+ * @context: the IB user context of the process making the mmap() call
+ * @vma: the VMA to be initialized
+ * Return zero if the mmap is OK. Otherwise, return an errno.
+ */
+int rxe_mmap(struct ib_ucontext *context, struct vm_area_struct *vma)
+{
+	struct rxe_dev *rxe = to_rdev(context->device);
+	unsigned long offset = vma->vm_pgoff << PAGE_SHIFT;
+	unsigned long size = vma->vm_end - vma->vm_start;
+	struct rxe_mmap_info *ip, *pp;
+	int ret;
+
+	/*
+	 * Search the device's list of objects waiting for a mmap call.
+	 * Normally, this list is very short since a call to create a
+	 * CQ, QP, or SRQ is soon followed by a call to mmap().
+	 */
+	spin_lock_bh(&rxe->pending_lock);
+	list_for_each_entry_safe(ip, pp, &rxe->pending_mmaps, pending_mmaps) {
+		if (context != ip->context || (__u64)offset != ip->info.offset)
+			continue;
+
+		/* Don't allow a mmap larger than the object. */
+		if (size > ip->info.size) {
+			pr_err("mmap region is larger than the object!\n");
+			spin_unlock_bh(&rxe->pending_lock);
+			ret = -EINVAL;
+			goto done;
+		}
+
+		goto found_it;
+	}
+	pr_warn("unable to find pending mmap info\n");
+	spin_unlock_bh(&rxe->pending_lock);
+	ret = -EINVAL;
+	goto done;
+
+found_it:
+	list_del_init(&ip->pending_mmaps);
+	spin_unlock_bh(&rxe->pending_lock);
+
+	ret = remap_vmalloc_range(vma, ip->obj, 0);
+	if (ret) {
+		pr_err("rxe: err %d from remap_vmalloc_range\n", ret);
+		goto done;
+	}
+
+	vma->vm_ops = &rxe_vm_ops;
+	vma->vm_private_data = ip;
+	rxe_vma_open(vma);
+done:
+	return ret;
+}
+
+/*
+ * Allocate information for rxe_mmap
+ */
+struct rxe_mmap_info *rxe_create_mmap_info(struct rxe_dev *rxe,
+					   u32 size,
+					   struct ib_ucontext *context,
+					   void *obj)
+{
+	struct rxe_mmap_info *ip;
+
+	ip = kmalloc(sizeof(*ip), GFP_KERNEL);
+	if (!ip)
+		return NULL;
+
+	size = PAGE_ALIGN(size);
+
+	spin_lock_bh(&rxe->mmap_offset_lock);
+
+	if (rxe->mmap_offset == 0)
+		rxe->mmap_offset = PAGE_SIZE;
+
+	ip->info.offset = rxe->mmap_offset;
+	rxe->mmap_offset += size;
+
+	spin_unlock_bh(&rxe->mmap_offset_lock);
+
+	INIT_LIST_HEAD(&ip->pending_mmaps);
+	ip->info.size = size;
+	ip->context = context;
+	ip->obj = obj;
+	kref_init(&ip->ref);
+
+	return ip;
+}

drivers/infiniband/sw/rxe/rxe_mr.c

+/*
+ * Copyright (c) 2016 Mellanox Technologies Ltd. All rights reserved.
+ * Copyright (c) 2015 System Fabric Works, 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 "rxe.h"
+#include "rxe_loc.h"
+
+/*
+ * lfsr (linear feedback shift register) with period 255
+ */
+static u8 rxe_get_key(void)
+{
+	static unsigned key = 1;
+
+	key = key << 1;
+
+	key |= (0 != (key & 0x100)) ^ (0 != (key & 0x10))
+		^ (0 != (key & 0x80)) ^ (0 != (key & 0x40));
+
+	key &= 0xff;
+
+	return key;
+}
+
+int mem_check_range(struct rxe_mem *mem, u64 iova, size_t length)
+{
+	switch (mem->type) {
+	case RXE_MEM_TYPE_DMA:
+		return 0;
+
+	case RXE_MEM_TYPE_MR:
+	case RXE_MEM_TYPE_FMR:
+		return ((iova < mem->iova) ||
+			((iova + length) > (mem->iova + mem->length))) ?
+			-EFAULT : 0;
+
+	default:
+		return -EFAULT;
+	}
+}
+
+#define IB_ACCESS_REMOTE	(IB_ACCESS_REMOTE_READ		\
+				| IB_ACCESS_REMOTE_WRITE	\
+				| IB_ACCESS_REMOTE_ATOMIC)
+
+static void rxe_mem_init(int access, struct rxe_mem *mem)
+{
+	u32 lkey = mem->pelem.index << 8 | rxe_get_key();
+	u32 rkey = (access & IB_ACCESS_REMOTE) ? lkey : 0;
+
+	if (mem->pelem.pool->type == RXE_TYPE_MR) {
+		mem->ibmr.lkey		= lkey;
+		mem->ibmr.rkey		= rkey;
+	}
+
+	mem->lkey		= lkey;
+	mem->rkey		= rkey;
+	mem->state		= RXE_MEM_STATE_INVALID;
+	mem->type		= RXE_MEM_TYPE_NONE;
+	mem->map_shift		= ilog2(RXE_BUF_PER_MAP);
+}
+
+void rxe_mem_cleanup(void *arg)
+{
+	struct rxe_mem *mem = arg;
+	int i;
+
+	if (mem->umem)
+		ib_umem_release(mem->umem);
+
+	if (mem->map) {
+		for (i = 0; i < mem->num_map; i++)
+			kfree(mem->map[i]);
+
+		kfree(mem->map);
+	}
+}
+
+static int rxe_mem_alloc(struct rxe_dev *rxe, struct rxe_mem *mem, int num_buf)
+{
+	int i;
+	int num_map;
+	struct rxe_map **map = mem->map;
+
+	num_map = (num_buf + RXE_BUF_PER_MAP - 1) / RXE_BUF_PER_MAP;
+
+	mem->map = kmalloc_array(num_map, sizeof(*map), GFP_KERNEL);
+	if (!mem->map)
+		goto err1;
+
+	for (i = 0; i < num_map; i++) {
+		mem->map[i] = kmalloc(sizeof(**map), GFP_KERNEL);
+		if (!mem->map[i])
+			goto err2;
+	}
+
+	WARN_ON(!is_power_of_2(RXE_BUF_PER_MAP));
+
+	mem->map_shift	= ilog2(RXE_BUF_PER_MAP);
+	mem->map_mask	= RXE_BUF_PER_MAP - 1;
+
+	mem->num_buf = num_buf;
+	mem->num_map = num_map;
+	mem->max_buf = num_map * RXE_BUF_PER_MAP;
+
+	return 0;
+
+err2:
+	for (i--; i >= 0; i--)
+		kfree(mem->map[i]);
+
+	kfree(mem->map);
+err1:
+	return -ENOMEM;
+}
+
+int rxe_mem_init_dma(struct rxe_dev *rxe, struct rxe_pd *pd,
+		     int access, struct rxe_mem *mem)
+{
+	rxe_mem_init(access, mem);
+
+	mem->pd			= pd;
+	mem->access		= access;
+	mem->state		= RXE_MEM_STATE_VALID;
+	mem->type		= RXE_MEM_TYPE_DMA;
+
+	return 0;
+}
+
+int rxe_mem_init_user(struct rxe_dev *rxe, struct rxe_pd *pd, u64 start,
+		      u64 length, u64 iova, int access, struct ib_udata *udata,
+		      struct rxe_mem *mem)
+{
+	int			entry;
+	struct rxe_map		**map;
+	struct rxe_phys_buf	*buf = NULL;
+	struct ib_umem		*umem;
+	struct scatterlist	*sg;
+	int			num_buf;
+	void			*vaddr;
+	int err;
+
+	umem = ib_umem_get(pd->ibpd.uobject->context, start, length, access, 0);
+	if (IS_ERR(umem)) {
+		pr_warn("err %d from rxe_umem_get\n",
+			(int)PTR_ERR(umem));
+		err = -EINVAL;
+		goto err1;
+	}
+
+	mem->umem = umem;
+	num_buf = umem->nmap;
+
+	rxe_mem_init(access, mem);
+
+	err = rxe_mem_alloc(rxe, mem, num_buf);
+	if (err) {
+		pr_warn("err %d from rxe_mem_alloc\n", err);
+		ib_umem_release(umem);
+		goto err1;
+	}
+
+	WARN_ON(!is_power_of_2(umem->page_size));
+
+	mem->page_shift		= ilog2(umem->page_size);
+	mem->page_mask		= umem->page_size - 1;
+
+	num_buf			= 0;
+	map			= mem->map;
+	if (length > 0) {
+		buf = map[0]->buf;
+
+		for_each_sg(umem->sg_head.sgl, sg, umem->nmap, entry) {
+			vaddr = page_address(sg_page(sg));
+			if (!vaddr) {
+				pr_warn("null vaddr\n");
+				err = -ENOMEM;
+				goto err1;
+			}
+
+			buf->addr = (uintptr_t)vaddr;
+			buf->size = umem->page_size;
+			num_buf++;
+			buf++;
+
+			if (num_buf >= RXE_BUF_PER_MAP) {
+				map++;
+				buf = map[0]->buf;
+				num_buf = 0;
+			}
+		}
+	}
+
+	mem->pd			= pd;
+	mem->umem		= umem;
+	mem->access		= access;
+	mem->length		= length;
+	mem->iova		= iova;
+	mem->va			= start;
+	mem->offset		= ib_umem_offset(umem);
+	mem->state		= RXE_MEM_STATE_VALID;
+	mem->type		= RXE_MEM_TYPE_MR;
+
+	return 0;
+
+err1:
+	return err;
+}
+
+int rxe_mem_init_fast(struct rxe_dev *rxe, struct rxe_pd *pd,
+		      int max_pages, struct rxe_mem *mem)
+{
+	int err;
+
+	rxe_mem_init(0, mem);
+
+	/* In fastreg, we also set the rkey */
+	mem->ibmr.rkey = mem->ibmr.lkey;
+
+	err = rxe_mem_alloc(rxe, mem, max_pages);
+	if (err)
+		goto err1;
+
+	mem->pd			= pd;
+	mem->max_buf		= max_pages;
+	mem->state		= RXE_MEM_STATE_FREE;
+	mem->type		= RXE_MEM_TYPE_MR;
+
+	return 0;
+
+err1:
+	return err;
+}
+
+static void lookup_iova(
+	struct rxe_mem	*mem,
+	u64			iova,
+	int			*m_out,
+	int			*n_out,
+	size_t			*offset_out)
+{
+	size_t			offset = iova - mem->iova + mem->offset;
+	int			map_index;
+	int			buf_index;
+	u64			length;
+
+	if (likely(mem->page_shift)) {
+		*offset_out = offset & mem->page_mask;
+		offset >>= mem->page_shift;
+		*n_out = offset & mem->map_mask;
+		*m_out = offset >> mem->map_shift;
+	} else {
+		map_index = 0;
+		buf_index = 0;
+
+		length = mem->map[map_index]->buf[buf_index].size;
+
+		while (offset >= length) {
+			offset -= length;
+			buf_index++;
+
+			if (buf_index == RXE_BUF_PER_MAP) {
+				map_index++;
+				buf_index = 0;
+			}
+			length = mem->map[map_index]->buf[buf_index].size;
+		}
+
+		*m_out = map_index;
+		*n_out = buf_index;
+		*offset_out = offset;
+	}
+}
+
+void *iova_to_vaddr(struct rxe_mem *mem, u64 iova, int length)
+{
+	size_t offset;
+	int m, n;
+	void *addr;
+
+	if (mem->state != RXE_MEM_STATE_VALID) {
+		pr_warn("mem not in valid state\n");
+		addr = NULL;
+		goto out;
+	}
+
+	if (!mem->map) {
+		addr = (void *)(uintptr_t)iova;
+		goto out;
+	}
+
+	if (mem_check_range(mem, iova, length)) {
+		pr_warn("range violation\n");
+		addr = NULL;
+		goto out;
+	}
+
+	lookup_iova(mem, iova, &m, &n, &offset);
+
+	if (offset + length > mem->map[m]->buf[n].size) {
+		pr_warn("crosses page boundary\n");
+		addr = NULL;
+		goto out;
+	}
+
+	addr = (void *)(uintptr_t)mem->map[m]->buf[n].addr + offset;
+
+out:
+	return addr;
+}
+
+/* copy data from a range (vaddr, vaddr+length-1) to or from
+ * a mem object starting at iova. Compute incremental value of
+ * crc32 if crcp is not zero. caller must hold a reference to mem
+ */
+int rxe_mem_copy(struct rxe_mem *mem, u64 iova, void *addr, int length,
+		 enum copy_direction dir, u32 *crcp)
+{
+	int			err;
+	int			bytes;
+	u8			*va;
+	struct rxe_map		**map;
+	struct rxe_phys_buf	*buf;
+	int			m;
+	int			i;
+	size_t			offset;
+	u32			crc = crcp ? (*crcp) : 0;
+
+	if (mem->type == RXE_MEM_TYPE_DMA) {
+		u8 *src, *dest;
+
+		src  = (dir == to_mem_obj) ?
+			addr : ((void *)(uintptr_t)iova);
+
+		dest = (dir == to_mem_obj) ?
+			((void *)(uintptr_t)iova) : addr;
+
+		if (crcp)
+			*crcp = crc32_le(*crcp, src, length);
+
+		memcpy(dest, src, length);
+
+		return 0;
+	}
+
+	WARN_ON(!mem->map);
+
+	err = mem_check_range(mem, iova, length);
+	if (err) {
+		err = -EFAULT;
+		goto err1;
+	}
+
+	lookup_iova(mem, iova, &m, &i, &offset);
+
+	map	= mem->map + m;
+	buf	= map[0]->buf + i;
+
+	while (length > 0) {
+		u8 *src, *dest;
+
+		va	= (u8 *)(uintptr_t)buf->addr + offset;
+		src  = (dir == to_mem_obj) ? addr : va;
+		dest = (dir == to_mem_obj) ? va : addr;
+
+		bytes	= buf->size - offset;
+
+		if (bytes > length)
+			bytes = length;
+
+		if (crcp)
+			crc = crc32_le(crc, src, bytes);
+
+		memcpy(dest, src, bytes);
+
+		length	-= bytes;
+		addr	+= bytes;
+
+		offset	= 0;
+		buf++;
+		i++;
+
+		if (i == RXE_BUF_PER_MAP) {
+			i = 0;
+			map++;
+			buf = map[0]->buf;
+		}
+	}
+
+	if (crcp)
+		*crcp = crc;
+
+	return 0;
+
+err1:
+	return err;
+}
+
+/* copy data in or out of a wqe, i.e. sg list
+ * under the control of a dma descriptor
+ */
+int copy_data(
+	struct rxe_dev		*rxe,
+	struct rxe_pd		*pd,
+	int			access,
+	struct rxe_dma_info	*dma,
+	void			*addr,
+	int			length,
+	enum copy_direction	dir,
+	u32			*crcp)
+{
+	int			bytes;
+	struct rxe_sge		*sge	= &dma->sge[dma->cur_sge];
+	int			offset	= dma->sge_offset;
+	int			resid	= dma->resid;
+	struct rxe_mem		*mem	= NULL;
+	u64			iova;
+	int			err;
+
+	if (length == 0)
+		return 0;
+
+	if (length > resid) {
+		err = -EINVAL;
+		goto err2;
+	}
+
+	if (sge->length && (offset < sge->length)) {
+		mem = lookup_mem(pd, access, sge->lkey, lookup_local);
+		if (!mem) {
+			err = -EINVAL;
+			goto err1;
+		}
+	}
+
+	while (length > 0) {
+		bytes = length;
+
+		if (offset >= sge->length) {
+			if (mem) {
+				rxe_drop_ref(mem);
+				mem = NULL;
+			}
+			sge++;
+			dma->cur_sge++;
+			offset = 0;
+
+			if (dma->cur_sge >= dma->num_sge) {
+				err = -ENOSPC;
+				goto err2;
+			}
+
+			if (sge->length) {
+				mem = lookup_mem(pd, access, sge->lkey,
+						 lookup_local);
+				if (!mem) {
+					err = -EINVAL;
+					goto err1;
+				}
+			} else {
+				continue;
+			}
+		}
+
+		if (bytes > sge->length - offset)
+			bytes = sge->length - offset;
+
+		if (bytes > 0) {
+			iova = sge->addr + offset;
+
+			err = rxe_mem_copy(mem, iova, addr, bytes, dir, crcp);
+			if (err)
+				goto err2;
+
+			offset	+= bytes;
+			resid	-= bytes;
+			length	-= bytes;
+			addr	+= bytes;
+		}
+	}
+
+	dma->sge_offset = offset;
+	dma->resid	= resid;
+
+	if (mem)
+		rxe_drop_ref(mem);
+
+	return 0;
+
+err2:
+	if (mem)
+		rxe_drop_ref(mem);
+err1:
+	return err;
+}
+
+int advance_dma_data(struct rxe_dma_info *dma, unsigned int length)
+{
+	struct rxe_sge		*sge	= &dma->sge[dma->cur_sge];
+	int			offset	= dma->sge_offset;
+	int			resid	= dma->resid;
+
+	while (length) {
+		unsigned int bytes;
+
+		if (offset >= sge->length) {
+			sge++;
+			dma->cur_sge++;
+			offset = 0;
+			if (dma->cur_sge >= dma->num_sge)
+				return -ENOSPC;
+		}
+
+		bytes = length;
+
+		if (bytes > sge->length - offset)
+			bytes = sge->length - offset;
+
+		offset	+= bytes;
+		resid	-= bytes;
+		length	-= bytes;
+	}
+
+	dma->sge_offset = offset;
+	dma->resid	= resid;
+
+	return 0;
+}
+
+/* (1) find the mem (mr or mw) corresponding to lkey/rkey
+ *     depending on lookup_type
+ * (2) verify that the (qp) pd matches the mem pd
+ * (3) verify that the mem can support the requested access
+ * (4) verify that mem state is valid
+ */
+struct rxe_mem *lookup_mem(struct rxe_pd *pd, int access, u32 key,
+			   enum lookup_type type)
+{
+	struct rxe_mem *mem;
+	struct rxe_dev *rxe = to_rdev(pd->ibpd.device);
+	int index = key >> 8;
+
+	if (index >= RXE_MIN_MR_INDEX && index <= RXE_MAX_MR_INDEX) {
+		mem = rxe_pool_get_index(&rxe->mr_pool, index);
+		if (!mem)
+			goto err1;
+	} else {
+		goto err1;
+	}
+
+	if ((type == lookup_local && mem->lkey != key) ||
+	    (type == lookup_remote && mem->rkey != key))
+		goto err2;
+
+	if (mem->pd != pd)
+		goto err2;
+
+	if (access && !(access & mem->access))
+		goto err2;
+
+	if (mem->state != RXE_MEM_STATE_VALID)
+		goto err2;
+
+	return mem;
+
+err2:
+	rxe_drop_ref(mem);
+err1:
+	return NULL;
+}
+
+int rxe_mem_map_pages(struct rxe_dev *rxe, struct rxe_mem *mem,
+		      u64 *page, int num_pages, u64 iova)
+{
+	int i;
+	int num_buf;
+	int err;
+	struct rxe_map **map;
+	struct rxe_phys_buf *buf;
+	int page_size;
+
+	if (num_pages > mem->max_buf) {
+		err = -EINVAL;
+		goto err1;
+	}
+
+	num_buf		= 0;
+	page_size	= 1 << mem->page_shift;
+	map		= mem->map;
+	buf		= map[0]->buf;
+
+	for (i = 0; i < num_pages; i++) {
+		buf->addr = *page++;
+		buf->size = page_size;
+		buf++;
+		num_buf++;
+
+		if (num_buf == RXE_BUF_PER_MAP) {
+			map++;
+			buf = map[0]->buf;
+			num_buf = 0;
+		}
+	}
+
+	mem->iova	= iova;
+	mem->va		= iova;
+	mem->length	= num_pages << mem->page_shift;
+	mem->state	= RXE_MEM_STATE_VALID;
+
+	return 0;
+
+err1:
+	return err;
+}

drivers/infiniband/sw/rxe/rxe_net.c

+/*
+ * Copyright (c) 2016 Mellanox Technologies Ltd. All rights reserved.
+ * Copyright (c) 2015 System Fabric Works, 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/skbuff.h>
+#include <linux/if_arp.h>
+#include <linux/netdevice.h>
+#include <linux/if.h>
+#include <linux/if_vlan.h>
+#include <net/udp_tunnel.h>
+#include <net/sch_generic.h>
+#include <linux/netfilter.h>
+#include <rdma/ib_addr.h>
+
+#include "rxe.h"
+#include "rxe_net.h"
+#include "rxe_loc.h"
+
+static LIST_HEAD(rxe_dev_list);
+static spinlock_t dev_list_lock; /* spinlock for device list */
+
+struct rxe_dev *net_to_rxe(struct net_device *ndev)
+{
+	struct rxe_dev *rxe;
+	struct rxe_dev *found = NULL;
+
+	spin_lock_bh(&dev_list_lock);
+	list_for_each_entry(rxe, &rxe_dev_list, list) {
+		if (rxe->ndev == ndev) {
+			found = rxe;
+			break;
+		}
+	}
+	spin_unlock_bh(&dev_list_lock);
+
+	return found;
+}
+
+struct rxe_dev *get_rxe_by_name(const char* name)
+{
+	struct rxe_dev *rxe;
+	struct rxe_dev *found = NULL;
+
+	spin_lock_bh(&dev_list_lock);
+	list_for_each_entry(rxe, &rxe_dev_list, list) {
+		if (!strcmp(name, rxe->ib_dev.name)) {
+			found = rxe;
+			break;
+		}
+	}
+	spin_unlock_bh(&dev_list_lock);
+	return found;
+}
+
+
+struct rxe_recv_sockets recv_sockets;
+
+static __be64 rxe_mac_to_eui64(struct net_device *ndev)
+{
+	unsigned char *mac_addr = ndev->dev_addr;
+	__be64 eui64;
+	unsigned char *dst = (unsigned char *)&eui64;
+
+	dst[0] = mac_addr[0] ^ 2;
+	dst[1] = mac_addr[1];
+	dst[2] = mac_addr[2];
+	dst[3] = 0xff;
+	dst[4] = 0xfe;
+	dst[5] = mac_addr[3];
+	dst[6] = mac_addr[4];
+	dst[7] = mac_addr[5];
+
+	return eui64;
+}
+
+static __be64 node_guid(struct rxe_dev *rxe)
+{
+	return rxe_mac_to_eui64(rxe->ndev);
+}
+
+static __be64 port_guid(struct rxe_dev *rxe)
+{
+	return rxe_mac_to_eui64(rxe->ndev);
+}
+
+static struct device *dma_device(struct rxe_dev *rxe)
+{
+	struct net_device *ndev;
+
+	ndev = rxe->ndev;
+
+	if (ndev->priv_flags & IFF_802_1Q_VLAN)
+		ndev = vlan_dev_real_dev(ndev);
+
+	return ndev->dev.parent;
+}
+
+static int mcast_add(struct rxe_dev *rxe, union ib_gid *mgid)
+{
+	int err;
+	unsigned char ll_addr[ETH_ALEN];
+
+	ipv6_eth_mc_map((struct in6_addr *)mgid->raw, ll_addr);
+	err = dev_mc_add(rxe->ndev, ll_addr);
+
+	return err;
+}
+
+static int mcast_delete(struct rxe_dev *rxe, union ib_gid *mgid)
+{
+	int err;
+	unsigned char ll_addr[ETH_ALEN];
+
+	ipv6_eth_mc_map((struct in6_addr *)mgid->raw, ll_addr);
+	err = dev_mc_del(rxe->ndev, ll_addr);
+
+	return err;
+}
+
+static struct dst_entry *rxe_find_route4(struct net_device *ndev,
+				  struct in_addr *saddr,
+				  struct in_addr *daddr)
+{
+	struct rtable *rt;
+	struct flowi4 fl = { { 0 } };
+
+	memset(&fl, 0, sizeof(fl));
+	fl.flowi4_oif = ndev->ifindex;
+	memcpy(&fl.saddr, saddr, sizeof(*saddr));
+	memcpy(&fl.daddr, daddr, sizeof(*daddr));
+	fl.flowi4_proto = IPPROTO_UDP;
+
+	rt = ip_route_output_key(&init_net, &fl);
+	if (IS_ERR(rt)) {
+		pr_err_ratelimited("no route to %pI4\n", &daddr->s_addr);
+		return NULL;
+	}
+
+	return &rt->dst;
+}
+
+#if IS_ENABLED(CONFIG_IPV6)
+static struct dst_entry *rxe_find_route6(struct net_device *ndev,
+					 struct in6_addr *saddr,
+					 struct in6_addr *daddr)
+{
+	struct dst_entry *ndst;
+	struct flowi6 fl6 = { { 0 } };
+
+	memset(&fl6, 0, sizeof(fl6));
+	fl6.flowi6_oif = ndev->ifindex;
+	memcpy(&fl6.saddr, saddr, sizeof(*saddr));
+	memcpy(&fl6.daddr, daddr, sizeof(*daddr));
+	fl6.flowi6_proto = IPPROTO_UDP;
+
+	if (unlikely(ipv6_stub->ipv6_dst_lookup(sock_net(recv_sockets.sk6->sk),
+						recv_sockets.sk6->sk, &ndst, &fl6))) {
+		pr_err_ratelimited("no route to %pI6\n", daddr);
+		goto put;
+	}
+
+	if (unlikely(ndst->error)) {
+		pr_err("no route to %pI6\n", daddr);
+		goto put;
+	}
+
+	return ndst;
+put:
+	dst_release(ndst);
+	return NULL;
+}
+
+#else
+
+static struct dst_entry *rxe_find_route6(struct net_device *ndev,
+					 struct in6_addr *saddr,
+					 struct in6_addr *daddr)
+{
+	return NULL;
+}
+
+#endif
+
+static int rxe_udp_encap_recv(struct sock *sk, struct sk_buff *skb)
+{
+	struct udphdr *udph;
+	struct net_device *ndev = skb->dev;
+	struct rxe_dev *rxe = net_to_rxe(ndev);
+	struct rxe_pkt_info *pkt = SKB_TO_PKT(skb);
+
+	if (!rxe)
+		goto drop;
+
+	if (skb_linearize(skb)) {
+		pr_err("skb_linearize failed\n");
+		goto drop;
+	}
+
+	udph = udp_hdr(skb);
+	pkt->rxe = rxe;
+	pkt->port_num = 1;
+	pkt->hdr = (u8 *)(udph + 1);
+	pkt->mask = RXE_GRH_MASK;
+	pkt->paylen = be16_to_cpu(udph->len) - sizeof(*udph);
+
+	return rxe_rcv(skb);
+drop:
+	kfree_skb(skb);
+	return 0;
+}
+
+static struct socket *rxe_setup_udp_tunnel(struct net *net, __be16 port,
+					   bool ipv6)
+{
+	int err;
+	struct socket *sock;
+	struct udp_port_cfg udp_cfg;
+	struct udp_tunnel_sock_cfg tnl_cfg;
+
+	memset(&udp_cfg, 0, sizeof(udp_cfg));
+
+	if (ipv6) {
+		udp_cfg.family = AF_INET6;
+		udp_cfg.ipv6_v6only = 1;
+	} else {
+		udp_cfg.family = AF_INET;
+	}
+
+	udp_cfg.local_udp_port = port;
+
+	/* Create UDP socket */
+	err = udp_sock_create(net, &udp_cfg, &sock);
+	if (err < 0) {
+		pr_err("failed to create udp socket. err = %d\n", err);
+		return ERR_PTR(err);
+	}
+
+	tnl_cfg.sk_user_data = NULL;
+	tnl_cfg.encap_type = 1;
+	tnl_cfg.encap_rcv = rxe_udp_encap_recv;
+	tnl_cfg.encap_destroy = NULL;
+
+	/* Setup UDP tunnel */
+	setup_udp_tunnel_sock(net, sock, &tnl_cfg);
+
+	return sock;
+}
+
+static void rxe_release_udp_tunnel(struct socket *sk)
+{
+	udp_tunnel_sock_release(sk);
+}
+
+static void prepare_udp_hdr(struct sk_buff *skb, __be16 src_port,
+			    __be16 dst_port)
+{
+	struct udphdr *udph;
+
+	__skb_push(skb, sizeof(*udph));
+	skb_reset_transport_header(skb);
+	udph = udp_hdr(skb);
+
+	udph->dest = dst_port;
+	udph->source = src_port;
+	udph->len = htons(skb->len);
+	udph->check = 0;
+}
+
+static void prepare_ipv4_hdr(struct dst_entry *dst, struct sk_buff *skb,
+			     __be32 saddr, __be32 daddr, __u8 proto,
+			     __u8 tos, __u8 ttl, __be16 df, bool xnet)
+{
+	struct iphdr *iph;
+
+	skb_scrub_packet(skb, xnet);
+
+	skb_clear_hash(skb);
+	skb_dst_set(skb, dst);
+	memset(IPCB(skb), 0, sizeof(*IPCB(skb)));
+
+	skb_push(skb, sizeof(struct iphdr));
+	skb_reset_network_header(skb);
+
+	iph = ip_hdr(skb);
+
+	iph->version	=	IPVERSION;
+	iph->ihl	=	sizeof(struct iphdr) >> 2;
+	iph->frag_off	=	df;
+	iph->protocol	=	proto;
+	iph->tos	=	tos;
+	iph->daddr	=	daddr;
+	iph->saddr	=	saddr;
+	iph->ttl	=	ttl;
+	__ip_select_ident(dev_net(dst->dev), iph,
+			  skb_shinfo(skb)->gso_segs ?: 1);
+	iph->tot_len = htons(skb->len);
+	ip_send_check(iph);
+}
+
+static void prepare_ipv6_hdr(struct dst_entry *dst, struct sk_buff *skb,
+			     struct in6_addr *saddr, struct in6_addr *daddr,
+			     __u8 proto, __u8 prio, __u8 ttl)
+{
+	struct ipv6hdr *ip6h;
+
+	memset(&(IPCB(skb)->opt), 0, sizeof(IPCB(skb)->opt));
+	IPCB(skb)->flags &= ~(IPSKB_XFRM_TUNNEL_SIZE | IPSKB_XFRM_TRANSFORMED
+			    | IPSKB_REROUTED);
+	skb_dst_set(skb, dst);
+
+	__skb_push(skb, sizeof(*ip6h));
+	skb_reset_network_header(skb);
+	ip6h		  = ipv6_hdr(skb);
+	ip6_flow_hdr(ip6h, prio, htonl(0));
+	ip6h->payload_len = htons(skb->len);
+	ip6h->nexthdr     = proto;
+	ip6h->hop_limit   = ttl;
+	ip6h->daddr	  = *daddr;
+	ip6h->saddr	  = *saddr;
+	ip6h->payload_len = htons(skb->len - sizeof(*ip6h));
+}
+
+static int prepare4(struct rxe_dev *rxe, struct sk_buff *skb, struct rxe_av *av)
+{
+	struct dst_entry *dst;
+	bool xnet = false;
+	__be16 df = htons(IP_DF);
+	struct in_addr *saddr = &av->sgid_addr._sockaddr_in.sin_addr;
+	struct in_addr *daddr = &av->dgid_addr._sockaddr_in.sin_addr;
+	struct rxe_pkt_info *pkt = SKB_TO_PKT(skb);
+
+	dst = rxe_find_route4(rxe->ndev, saddr, daddr);
+	if (!dst) {
+		pr_err("Host not reachable\n");
+		return -EHOSTUNREACH;
+	}
+
+	if (!memcmp(saddr, daddr, sizeof(*daddr)))
+		pkt->mask |= RXE_LOOPBACK_MASK;
+
+	prepare_udp_hdr(skb, htons(RXE_ROCE_V2_SPORT),
+			htons(ROCE_V2_UDP_DPORT));
+
+	prepare_ipv4_hdr(dst, skb, saddr->s_addr, daddr->s_addr, IPPROTO_UDP,
+			 av->grh.traffic_class, av->grh.hop_limit, df, xnet);
+	return 0;
+}
+
+static int prepare6(struct rxe_dev *rxe, struct sk_buff *skb, struct rxe_av *av)
+{
+	struct dst_entry *dst;
+	struct in6_addr *saddr = &av->sgid_addr._sockaddr_in6.sin6_addr;
+	struct in6_addr *daddr = &av->dgid_addr._sockaddr_in6.sin6_addr;
+	struct rxe_pkt_info *pkt = SKB_TO_PKT(skb);
+
+	dst = rxe_find_route6(rxe->ndev, saddr, daddr);
+	if (!dst) {
+		pr_err("Host not reachable\n");
+		return -EHOSTUNREACH;
+	}
+
+	if (!memcmp(saddr, daddr, sizeof(*daddr)))
+		pkt->mask |= RXE_LOOPBACK_MASK;
+
+	prepare_udp_hdr(skb, htons(RXE_ROCE_V2_SPORT),
+			htons(ROCE_V2_UDP_DPORT));
+
+	prepare_ipv6_hdr(dst, skb, saddr, daddr, IPPROTO_UDP,
+			 av->grh.traffic_class,
+			 av->grh.hop_limit);
+	return 0;
+}
+
+static int prepare(struct rxe_dev *rxe, struct rxe_pkt_info *pkt,
+		   struct sk_buff *skb, u32 *crc)
+{
+	int err = 0;
+	struct rxe_av *av = rxe_get_av(pkt);
+
+	if (av->network_type == RDMA_NETWORK_IPV4)
+		err = prepare4(rxe, skb, av);
+	else if (av->network_type == RDMA_NETWORK_IPV6)
+		err = prepare6(rxe, skb, av);
+
+	*crc = rxe_icrc_hdr(pkt, skb);
+
+	return err;
+}
+
+static void rxe_skb_tx_dtor(struct sk_buff *skb)
+{
+	struct sock *sk = skb->sk;
+	struct rxe_qp *qp = sk->sk_user_data;
+	int skb_out = atomic_dec_return(&qp->skb_out);
+
+	if (unlikely(qp->need_req_skb &&
+		     skb_out < RXE_INFLIGHT_SKBS_PER_QP_LOW))
+		rxe_run_task(&qp->req.task, 1);
+}
+
+static int send(struct rxe_dev *rxe, struct rxe_pkt_info *pkt,
+		struct sk_buff *skb)
+{
+	struct sk_buff *nskb;
+	struct rxe_av *av;
+	int err;
+
+	av = rxe_get_av(pkt);
+
+	nskb = skb_clone(skb, GFP_ATOMIC);
+	if (!nskb)
+		return -ENOMEM;
+
+	nskb->destructor = rxe_skb_tx_dtor;
+	nskb->sk = pkt->qp->sk->sk;
+
+	if (av->network_type == RDMA_NETWORK_IPV4) {
+		err = ip_local_out(dev_net(skb_dst(skb)->dev), nskb->sk, nskb);
+	} else if (av->network_type == RDMA_NETWORK_IPV6) {
+		err = ip6_local_out(dev_net(skb_dst(skb)->dev), nskb->sk, nskb);
+	} else {
+		pr_err("Unknown layer 3 protocol: %d\n", av->network_type);
+		kfree_skb(nskb);
+		return -EINVAL;
+	}
+
+	if (unlikely(net_xmit_eval(err))) {
+		pr_debug("error sending packet: %d\n", err);
+		return -EAGAIN;
+	}
+
+	kfree_skb(skb);
+
+	return 0;
+}
+
+static int loopback(struct sk_buff *skb)
+{
+	return rxe_rcv(skb);
+}
+
+static inline int addr_same(struct rxe_dev *rxe, struct rxe_av *av)
+{
+	return rxe->port.port_guid == av->grh.dgid.global.interface_id;
+}
+
+static struct sk_buff *init_packet(struct rxe_dev *rxe, struct rxe_av *av,
+				   int paylen, struct rxe_pkt_info *pkt)
+{
+	unsigned int hdr_len;
+	struct sk_buff *skb;
+
+	if (av->network_type == RDMA_NETWORK_IPV4)
+		hdr_len = ETH_HLEN + sizeof(struct udphdr) +
+			sizeof(struct iphdr);
+	else
+		hdr_len = ETH_HLEN + sizeof(struct udphdr) +
+			sizeof(struct ipv6hdr);
+
+	skb = alloc_skb(paylen + hdr_len + LL_RESERVED_SPACE(rxe->ndev),
+			GFP_ATOMIC);
+	if (unlikely(!skb))
+		return NULL;
+
+	skb_reserve(skb, hdr_len + LL_RESERVED_SPACE(rxe->ndev));
+
+	skb->dev	= rxe->ndev;
+	if (av->network_type == RDMA_NETWORK_IPV4)
+		skb->protocol = htons(ETH_P_IP);
+	else
+		skb->protocol = htons(ETH_P_IPV6);
+
+	pkt->rxe	= rxe;
+	pkt->port_num	= 1;
+	pkt->hdr	= skb_put(skb, paylen);
+	pkt->mask	|= RXE_GRH_MASK;
+
+	memset(pkt->hdr, 0, paylen);
+
+	return skb;
+}
+
+/*
+ * this is required by rxe_cfg to match rxe devices in
+ * /sys/class/infiniband up with their underlying ethernet devices
+ */
+static char *parent_name(struct rxe_dev *rxe, unsigned int port_num)
+{
+	return rxe->ndev->name;
+}
+
+static enum rdma_link_layer link_layer(struct rxe_dev *rxe,
+				       unsigned int port_num)
+{
+	return IB_LINK_LAYER_ETHERNET;
+}
+
+static struct rxe_ifc_ops ifc_ops = {
+	.node_guid	= node_guid,
+	.port_guid	= port_guid,
+	.dma_device	= dma_device,
+	.mcast_add	= mcast_add,
+	.mcast_delete	= mcast_delete,
+	.prepare	= prepare,
+	.send		= send,
+	.loopback	= loopback,
+	.init_packet	= init_packet,
+	.parent_name	= parent_name,
+	.link_layer	= link_layer,
+};
+
+struct rxe_dev *rxe_net_add(struct net_device *ndev)
+{
+	int err;
+	struct rxe_dev *rxe = NULL;
+
+	rxe = (struct rxe_dev *)ib_alloc_device(sizeof(*rxe));
+	if (!rxe)
+		return NULL;
+
+	rxe->ifc_ops = &ifc_ops;
+	rxe->ndev = ndev;
+
+	err = rxe_add(rxe, ndev->mtu);
+	if (err) {
+		ib_dealloc_device(&rxe->ib_dev);
+		return NULL;
+	}
+
+	spin_lock_bh(&dev_list_lock);
+	list_add_tail(&rxe_dev_list, &rxe->list);
+	spin_unlock_bh(&dev_list_lock);
+	return rxe;
+}
+
+void rxe_remove_all(void)
+{
+	spin_lock_bh(&dev_list_lock);
+	while (!list_empty(&rxe_dev_list)) {
+		struct rxe_dev *rxe =
+			list_first_entry(&rxe_dev_list, struct rxe_dev, list);
+
+		list_del(&rxe->list);
+		spin_unlock_bh(&dev_list_lock);
+		rxe_remove(rxe);
+		spin_lock_bh(&dev_list_lock);
+	}
+	spin_unlock_bh(&dev_list_lock);
+}
+EXPORT_SYMBOL(rxe_remove_all);
+
+static void rxe_port_event(struct rxe_dev *rxe,
+			   enum ib_event_type event)
+{
+	struct ib_event ev;
+
+	ev.device = &rxe->ib_dev;
+	ev.element.port_num = 1;
+	ev.event = event;
+
+	ib_dispatch_event(&ev);
+}
+
+/* Caller must hold net_info_lock */
+void rxe_port_up(struct rxe_dev *rxe)
+{
+	struct rxe_port *port;
+
+	port = &rxe->port;
+	port->attr.state = IB_PORT_ACTIVE;
+	port->attr.phys_state = IB_PHYS_STATE_LINK_UP;
+
+	rxe_port_event(rxe, IB_EVENT_PORT_ACTIVE);
+	pr_info("rxe: set %s active\n", rxe->ib_dev.name);
+	return;
+}
+
+/* Caller must hold net_info_lock */
+void rxe_port_down(struct rxe_dev *rxe)
+{
+	struct rxe_port *port;
+
+	port = &rxe->port;
+	port->attr.state = IB_PORT_DOWN;
+	port->attr.phys_state = IB_PHYS_STATE_LINK_DOWN;
+
+	rxe_port_event(rxe, IB_EVENT_PORT_ERR);
+	pr_info("rxe: set %s down\n", rxe->ib_dev.name);
+	return;
+}
+
+static int rxe_notify(struct notifier_block *not_blk,
+		      unsigned long event,
+		      void *arg)
+{
+	struct net_device *ndev = netdev_notifier_info_to_dev(arg);
+	struct rxe_dev *rxe = net_to_rxe(ndev);
+
+	if (!rxe)
+		goto out;
+
+	switch (event) {
+	case NETDEV_UNREGISTER:
+		list_del(&rxe->list);
+		rxe_remove(rxe);
+		break;
+	case NETDEV_UP:
+		rxe_port_up(rxe);
+		break;
+	case NETDEV_DOWN:
+		rxe_port_down(rxe);
+		break;
+	case NETDEV_CHANGEMTU:
+		pr_info("rxe: %s changed mtu to %d\n", ndev->name, ndev->mtu);
+		rxe_set_mtu(rxe, ndev->mtu);
+		break;
+	case NETDEV_REBOOT:
+	case NETDEV_CHANGE:
+	case NETDEV_GOING_DOWN:
+	case NETDEV_CHANGEADDR:
+	case NETDEV_CHANGENAME:
+	case NETDEV_FEAT_CHANGE:
+	default:
+		pr_info("rxe: ignoring netdev event = %ld for %s\n",
+			event, ndev->name);
+		break;
+	}
+out:
+	return NOTIFY_OK;
+}
+
+static struct notifier_block rxe_net_notifier = {
+	.notifier_call = rxe_notify,
+};
+
+int rxe_net_init(void)
+{
+	int err;
+
+	spin_lock_init(&dev_list_lock);
+
+	recv_sockets.sk6 = rxe_setup_udp_tunnel(&init_net,
+			htons(ROCE_V2_UDP_DPORT), true);
+	if (IS_ERR(recv_sockets.sk6)) {
+		recv_sockets.sk6 = NULL;
+		pr_err("rxe: Failed to create IPv6 UDP tunnel\n");
+		return -1;
+	}
+
+	recv_sockets.sk4 = rxe_setup_udp_tunnel(&init_net,
+			htons(ROCE_V2_UDP_DPORT), false);
+	if (IS_ERR(recv_sockets.sk4)) {
+		rxe_release_udp_tunnel(recv_sockets.sk6);
+		recv_sockets.sk4 = NULL;
+		recv_sockets.sk6 = NULL;
+		pr_err("rxe: Failed to create IPv4 UDP tunnel\n");
+		return -1;
+	}
+
+	err = register_netdevice_notifier(&rxe_net_notifier);
+	if (err) {
+		rxe_release_udp_tunnel(recv_sockets.sk6);
+		rxe_release_udp_tunnel(recv_sockets.sk4);
+		pr_err("rxe: Failed to rigister netdev notifier\n");
+	}
+
+	return err;
+}
+
+void rxe_net_exit(void)
+{
+	if (recv_sockets.sk6)
+		rxe_release_udp_tunnel(recv_sockets.sk6);
+
+	if (recv_sockets.sk4)
+		rxe_release_udp_tunnel(recv_sockets.sk4);
+
+	unregister_netdevice_notifier(&rxe_net_notifier);
+}

drivers/infiniband/sw/rxe/rxe_net.h

+/*
+ * Copyright (c) 2016 Mellanox Technologies Ltd. All rights reserved.
+ * Copyright (c) 2015 System Fabric Works, 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 RXE_NET_H
+#define RXE_NET_H
+
+#include <net/sock.h>
+#include <net/if_inet6.h>
+#include <linux/module.h>
+
+struct rxe_recv_sockets {
+	struct socket *sk4;
+	struct socket *sk6;
+};
+
+extern struct rxe_recv_sockets recv_sockets;
+
+struct rxe_dev *rxe_net_add(struct net_device *ndev);
+
+int rxe_net_init(void);
+void rxe_net_exit(void);
+
+#endif /* RXE_NET_H */

drivers/infiniband/sw/rxe/rxe_opcode.h

+/*
+ * Copyright (c) 2016 Mellanox Technologies Ltd. All rights reserved.
+ * Copyright (c) 2015 System Fabric Works, 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 RXE_OPCODE_H
+#define RXE_OPCODE_H
+
+/*
+ * contains header bit mask definitions and header lengths
+ * declaration of the rxe_opcode_info struct and
+ * rxe_wr_opcode_info struct
+ */
+
+enum rxe_wr_mask {
+	WR_INLINE_MASK			= BIT(0),
+	WR_ATOMIC_MASK			= BIT(1),
+	WR_SEND_MASK			= BIT(2),
+	WR_READ_MASK			= BIT(3),
+	WR_WRITE_MASK			= BIT(4),
+	WR_LOCAL_MASK			= BIT(5),
+	WR_REG_MASK			= BIT(6),
+
+	WR_READ_OR_WRITE_MASK		= WR_READ_MASK | WR_WRITE_MASK,
+	WR_READ_WRITE_OR_SEND_MASK	= WR_READ_OR_WRITE_MASK | WR_SEND_MASK,
+	WR_WRITE_OR_SEND_MASK		= WR_WRITE_MASK | WR_SEND_MASK,
+	WR_ATOMIC_OR_READ_MASK		= WR_ATOMIC_MASK | WR_READ_MASK,
+};
+
+#define WR_MAX_QPT		(8)
+
+struct rxe_wr_opcode_info {
+	char			*name;
+	enum rxe_wr_mask	mask[WR_MAX_QPT];
+};
+
+extern struct rxe_wr_opcode_info rxe_wr_opcode_info[];
+
+enum rxe_hdr_type {
+	RXE_LRH,
+	RXE_GRH,
+	RXE_BTH,
+	RXE_RETH,
+	RXE_AETH,
+	RXE_ATMETH,
+	RXE_ATMACK,
+	RXE_IETH,
+	RXE_RDETH,
+	RXE_DETH,
+	RXE_IMMDT,
+	RXE_PAYLOAD,
+	NUM_HDR_TYPES
+};
+
+enum rxe_hdr_mask {
+	RXE_LRH_MASK		= BIT(RXE_LRH),
+	RXE_GRH_MASK		= BIT(RXE_GRH),
+	RXE_BTH_MASK		= BIT(RXE_BTH),
+	RXE_IMMDT_MASK		= BIT(RXE_IMMDT),
+	RXE_RETH_MASK		= BIT(RXE_RETH),
+	RXE_AETH_MASK		= BIT(RXE_AETH),
+	RXE_ATMETH_MASK		= BIT(RXE_ATMETH),
+	RXE_ATMACK_MASK		= BIT(RXE_ATMACK),
+	RXE_IETH_MASK		= BIT(RXE_IETH),
+	RXE_RDETH_MASK		= BIT(RXE_RDETH),
+	RXE_DETH_MASK		= BIT(RXE_DETH),
+	RXE_PAYLOAD_MASK	= BIT(RXE_PAYLOAD),
+
+	RXE_REQ_MASK		= BIT(NUM_HDR_TYPES + 0),
+	RXE_ACK_MASK		= BIT(NUM_HDR_TYPES + 1),
+	RXE_SEND_MASK		= BIT(NUM_HDR_TYPES + 2),
+	RXE_WRITE_MASK		= BIT(NUM_HDR_TYPES + 3),
+	RXE_READ_MASK		= BIT(NUM_HDR_TYPES + 4),
+	RXE_ATOMIC_MASK		= BIT(NUM_HDR_TYPES + 5),
+
+	RXE_RWR_MASK		= BIT(NUM_HDR_TYPES + 6),
+	RXE_COMP_MASK		= BIT(NUM_HDR_TYPES + 7),
+
+	RXE_START_MASK		= BIT(NUM_HDR_TYPES + 8),
+	RXE_MIDDLE_MASK		= BIT(NUM_HDR_TYPES + 9),
+	RXE_END_MASK		= BIT(NUM_HDR_TYPES + 10),
+
+	RXE_LOOPBACK_MASK	= BIT(NUM_HDR_TYPES + 12),
+
+	RXE_READ_OR_ATOMIC	= (RXE_READ_MASK | RXE_ATOMIC_MASK),
+	RXE_WRITE_OR_SEND	= (RXE_WRITE_MASK | RXE_SEND_MASK),
+};
+
+#define OPCODE_NONE		(-1)
+#define RXE_NUM_OPCODE		256
+
+struct rxe_opcode_info {
+	char			*name;
+	enum rxe_hdr_mask	mask;
+	int			length;
+	int			offset[NUM_HDR_TYPES];
+};
+
+extern struct rxe_opcode_info rxe_opcode[RXE_NUM_OPCODE];
+
+#endif /* RXE_OPCODE_H */

include/uapi/rdma/Kbuild

@@ -6,3 +6,4 @@ header-y += ib_user_verbs.h
 header-y += rdma_netlink.h
 header-y += rdma_user_cm.h
 header-y += hfi/
+header-y += rdma_user_rxe.h