main.c

/*
 * Copyright (c) 2004, 2005 Topspin Communications.  All rights reserved.
 * Copyright (c) 2005 Sun Microsystems, Inc. All rights reserved.
 * Copyright (c) 2005, 2006, 2007, 2008 Mellanox Technologies. All rights reserved.
 * Copyright (c) 2006, 2007 Cisco Systems, 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/kernel.h>
#include <linux/init.h>
#include <linux/errno.h>
#include <linux/pci.h>
#include <linux/dma-mapping.h>
#include <linux/slab.h>
#include <linux/io-mapping.h>
#include <linux/delay.h>
#include <linux/kmod.h>
#include <linux/etherdevice.h>
#include <net/devlink.h>

#include <uapi/rdma/mlx4-abi.h>
#include <linux/mlx4/device.h>
#include <linux/mlx4/doorbell.h>

#include "mlx4.h"
#include "fw.h"
#include "icm.h"

MODULE_AUTHOR("Roland Dreier");
MODULE_DESCRIPTION("Mellanox ConnectX HCA low-level driver");
MODULE_LICENSE("Dual BSD/GPL");
MODULE_VERSION(DRV_VERSION);

struct workqueue_struct *mlx4_wq;

#ifdef CONFIG_MLX4_DEBUG

int mlx4_debug_level; /* 0 by default */
module_param_named(debug_level, mlx4_debug_level, int, 0644);
MODULE_PARM_DESC(debug_level, "Enable debug tracing if > 0");

#endif /* CONFIG_MLX4_DEBUG */

#ifdef CONFIG_PCI_MSI

static int msi_x = 1;
module_param(msi_x, int, 0444);
MODULE_PARM_DESC(msi_x, "0 - don't use MSI-X, 1 - use MSI-X, >1 - limit number of MSI-X irqs to msi_x");

#else /* CONFIG_PCI_MSI */

#define msi_x (0)

#endif /* CONFIG_PCI_MSI */

static uint8_t num_vfs[3] = {0, 0, 0};
static int num_vfs_argc;
module_param_array(num_vfs, byte, &num_vfs_argc, 0444);
MODULE_PARM_DESC(num_vfs, "enable #num_vfs functions if num_vfs > 0\n"
			  "num_vfs=port1,port2,port1+2");

static uint8_t probe_vf[3] = {0, 0, 0};
static int probe_vfs_argc;
module_param_array(probe_vf, byte, &probe_vfs_argc, 0444);
MODULE_PARM_DESC(probe_vf, "number of vfs to probe by pf driver (num_vfs > 0)\n"
			   "probe_vf=port1,port2,port1+2");

static int mlx4_log_num_mgm_entry_size = MLX4_DEFAULT_MGM_LOG_ENTRY_SIZE;
module_param_named(log_num_mgm_entry_size,
			mlx4_log_num_mgm_entry_size, int, 0444);
MODULE_PARM_DESC(log_num_mgm_entry_size, "log mgm size, that defines the num"
					 " of qp per mcg, for example:"
					 " 10 gives 248.range: 7 <="
					 " log_num_mgm_entry_size <= 12."
					 " To activate device managed"
					 " flow steering when available, set to -1");

static bool enable_64b_cqe_eqe = true;
module_param(enable_64b_cqe_eqe, bool, 0444);
MODULE_PARM_DESC(enable_64b_cqe_eqe,
		 "Enable 64 byte CQEs/EQEs when the FW supports this (default: True)");

static bool enable_4k_uar;
module_param(enable_4k_uar, bool, 0444);
MODULE_PARM_DESC(enable_4k_uar,
		 "Enable using 4K UAR. Should not be enabled if have VFs which do not support 4K UARs (default: false)");

#define PF_CONTEXT_BEHAVIOUR_MASK	(MLX4_FUNC_CAP_64B_EQE_CQE | \
					 MLX4_FUNC_CAP_EQE_CQE_STRIDE | \
					 MLX4_FUNC_CAP_DMFS_A0_STATIC)

#define RESET_PERSIST_MASK_FLAGS	(MLX4_FLAG_SRIOV)

static char mlx4_version[] =
	DRV_NAME ": Mellanox ConnectX core driver v"
	DRV_VERSION "\n";

static const struct mlx4_profile default_profile = {
	.num_qp		= 1 << 18,
	.num_srq	= 1 << 16,
	.rdmarc_per_qp	= 1 << 4,
	.num_cq		= 1 << 16,
	.num_mcg	= 1 << 13,
	.num_mpt	= 1 << 19,
	.num_mtt	= 1 << 20, /* It is really num mtt segements */
};

static const struct mlx4_profile low_mem_profile = {
	.num_qp		= 1 << 17,
	.num_srq	= 1 << 6,
	.rdmarc_per_qp	= 1 << 4,
	.num_cq		= 1 << 8,
	.num_mcg	= 1 << 8,
	.num_mpt	= 1 << 9,
	.num_mtt	= 1 << 7,
};

static int log_num_mac = 7;
module_param_named(log_num_mac, log_num_mac, int, 0444);
MODULE_PARM_DESC(log_num_mac, "Log2 max number of MACs per ETH port (1-7)");

static int log_num_vlan;
module_param_named(log_num_vlan, log_num_vlan, int, 0444);
MODULE_PARM_DESC(log_num_vlan, "Log2 max number of VLANs per ETH port (0-7)");
/* Log2 max number of VLANs per ETH port (0-7) */
#define MLX4_LOG_NUM_VLANS 7
#define MLX4_MIN_LOG_NUM_VLANS 0
#define MLX4_MIN_LOG_NUM_MAC 1

static bool use_prio;
module_param_named(use_prio, use_prio, bool, 0444);
MODULE_PARM_DESC(use_prio, "Enable steering by VLAN priority on ETH ports (deprecated)");

int log_mtts_per_seg = ilog2(1);
module_param_named(log_mtts_per_seg, log_mtts_per_seg, int, 0444);
MODULE_PARM_DESC(log_mtts_per_seg, "Log2 number of MTT entries per segment "
		 "(0-7) (default: 0)");

static int port_type_array[2] = {MLX4_PORT_TYPE_NONE, MLX4_PORT_TYPE_NONE};
static int arr_argc = 2;
module_param_array(port_type_array, int, &arr_argc, 0444);
MODULE_PARM_DESC(port_type_array, "Array of port types: HW_DEFAULT (0) is default "
				"1 for IB, 2 for Ethernet");

struct mlx4_port_config {
	struct list_head list;
	enum mlx4_port_type port_type[MLX4_MAX_PORTS + 1];
	struct pci_dev *pdev;
};

static atomic_t pf_loading = ATOMIC_INIT(0);

static int mlx4_devlink_ierr_reset_get(struct devlink *devlink, u32 id,
				       struct devlink_param_gset_ctx *ctx)
{
	ctx->val.vbool = !!mlx4_internal_err_reset;
	return 0;
}

static int mlx4_devlink_ierr_reset_set(struct devlink *devlink, u32 id,
				       struct devlink_param_gset_ctx *ctx)
{
	mlx4_internal_err_reset = ctx->val.vbool;
	return 0;
}

static int mlx4_devlink_crdump_snapshot_get(struct devlink *devlink, u32 id,
					    struct devlink_param_gset_ctx *ctx)
{
	struct mlx4_priv *priv = devlink_priv(devlink);
	struct mlx4_dev *dev = &priv->dev;

	ctx->val.vbool = dev->persist->crdump.snapshot_enable;
	return 0;
}

static int mlx4_devlink_crdump_snapshot_set(struct devlink *devlink, u32 id,
					    struct devlink_param_gset_ctx *ctx)
{
	struct mlx4_priv *priv = devlink_priv(devlink);
	struct mlx4_dev *dev = &priv->dev;

	dev->persist->crdump.snapshot_enable = ctx->val.vbool;
	return 0;
}

static int
mlx4_devlink_max_macs_validate(struct devlink *devlink, u32 id,
			       union devlink_param_value val,
			       struct netlink_ext_ack *extack)
{
	u32 value = val.vu32;

	if (value < 1 || value > 128)
		return -ERANGE;

	if (!is_power_of_2(value)) {
		NL_SET_ERR_MSG_MOD(extack, "max_macs supported must be power of 2");
		return -EINVAL;
	}

	return 0;
}

enum mlx4_devlink_param_id {
	MLX4_DEVLINK_PARAM_ID_BASE = DEVLINK_PARAM_GENERIC_ID_MAX,
	MLX4_DEVLINK_PARAM_ID_ENABLE_64B_CQE_EQE,
	MLX4_DEVLINK_PARAM_ID_ENABLE_4K_UAR,
};

static const struct devlink_param mlx4_devlink_params[] = {
	DEVLINK_PARAM_GENERIC(INT_ERR_RESET,
			      BIT(DEVLINK_PARAM_CMODE_RUNTIME) |
			      BIT(DEVLINK_PARAM_CMODE_DRIVERINIT),
			      mlx4_devlink_ierr_reset_get,
			      mlx4_devlink_ierr_reset_set, NULL),
	DEVLINK_PARAM_GENERIC(MAX_MACS,
			      BIT(DEVLINK_PARAM_CMODE_DRIVERINIT),
			      NULL, NULL, mlx4_devlink_max_macs_validate),
	DEVLINK_PARAM_GENERIC(REGION_SNAPSHOT,
			      BIT(DEVLINK_PARAM_CMODE_RUNTIME) |
			      BIT(DEVLINK_PARAM_CMODE_DRIVERINIT),
			      mlx4_devlink_crdump_snapshot_get,
			      mlx4_devlink_crdump_snapshot_set, NULL),
	DEVLINK_PARAM_DRIVER(MLX4_DEVLINK_PARAM_ID_ENABLE_64B_CQE_EQE,
			     "enable_64b_cqe_eqe", DEVLINK_PARAM_TYPE_BOOL,
			     BIT(DEVLINK_PARAM_CMODE_DRIVERINIT),
			     NULL, NULL, NULL),
	DEVLINK_PARAM_DRIVER(MLX4_DEVLINK_PARAM_ID_ENABLE_4K_UAR,
			     "enable_4k_uar", DEVLINK_PARAM_TYPE_BOOL,
			     BIT(DEVLINK_PARAM_CMODE_DRIVERINIT),
			     NULL, NULL, NULL),
};

static void mlx4_devlink_set_params_init_values(struct devlink *devlink)
{
	union devlink_param_value value;

	value.vbool = !!mlx4_internal_err_reset;
	devlink_param_driverinit_value_set(devlink,
					   DEVLINK_PARAM_GENERIC_ID_INT_ERR_RESET,
					   value);

	value.vu32 = 1UL << log_num_mac;
	devlink_param_driverinit_value_set(devlink,
					   DEVLINK_PARAM_GENERIC_ID_MAX_MACS,
					   value);

	value.vbool = enable_64b_cqe_eqe;
	devlink_param_driverinit_value_set(devlink,
					   MLX4_DEVLINK_PARAM_ID_ENABLE_64B_CQE_EQE,
					   value);

	value.vbool = enable_4k_uar;
	devlink_param_driverinit_value_set(devlink,
					   MLX4_DEVLINK_PARAM_ID_ENABLE_4K_UAR,
					   value);

	value.vbool = false;
	devlink_param_driverinit_value_set(devlink,
					   DEVLINK_PARAM_GENERIC_ID_REGION_SNAPSHOT,
					   value);
}

static inline void mlx4_set_num_reserved_uars(struct mlx4_dev *dev,
					      struct mlx4_dev_cap *dev_cap)
{
	/* The reserved_uars is calculated by system page size unit.
	 * Therefore, adjustment is added when the uar page size is less
	 * than the system page size
	 */
	dev->caps.reserved_uars	=
		max_t(int,
		      mlx4_get_num_reserved_uar(dev),
		      dev_cap->reserved_uars /
			(1 << (PAGE_SHIFT - dev->uar_page_shift)));
}

int mlx4_check_port_params(struct mlx4_dev *dev,
			   enum mlx4_port_type *port_type)
{
	int i;

	if (!(dev->caps.flags & MLX4_DEV_CAP_FLAG_DPDP)) {
		for (i = 0; i < dev->caps.num_ports - 1; i++) {
			if (port_type[i] != port_type[i + 1]) {
				mlx4_err(dev, "Only same port types supported on this HCA, aborting\n");
				return -EOPNOTSUPP;
			}
		}
	}

	for (i = 0; i < dev->caps.num_ports; i++) {
		if (!(port_type[i] & dev->caps.supported_type[i+1])) {
			mlx4_err(dev, "Requested port type for port %d is not supported on this HCA\n",
				 i + 1);
			return -EOPNOTSUPP;
		}
	}
	return 0;
}

static void mlx4_set_port_mask(struct mlx4_dev *dev)
{
	int i;

	for (i = 1; i <= dev->caps.num_ports; ++i)
		dev->caps.port_mask[i] = dev->caps.port_type[i];
}

enum {
	MLX4_QUERY_FUNC_NUM_SYS_EQS = 1 << 0,
};

static int mlx4_query_func(struct mlx4_dev *dev, struct mlx4_dev_cap *dev_cap)
{
	int err = 0;
	struct mlx4_func func;

	if (dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_SYS_EQS) {
		err = mlx4_QUERY_FUNC(dev, &func, 0);
		if (err) {
			mlx4_err(dev, "QUERY_DEV_CAP command failed, aborting.\n");
			return err;
		}
		dev_cap->max_eqs = func.max_eq;
		dev_cap->reserved_eqs = func.rsvd_eqs;
		dev_cap->reserved_uars = func.rsvd_uars;
		err |= MLX4_QUERY_FUNC_NUM_SYS_EQS;
	}
	return err;
}

static void mlx4_enable_cqe_eqe_stride(struct mlx4_dev *dev)
{
	struct mlx4_caps *dev_cap = &dev->caps;

	/* FW not supporting or cancelled by user */
	if (!(dev_cap->flags2 & MLX4_DEV_CAP_FLAG2_EQE_STRIDE) ||
	    !(dev_cap->flags2 & MLX4_DEV_CAP_FLAG2_CQE_STRIDE))
		return;

	/* Must have 64B CQE_EQE enabled by FW to use bigger stride
	 * When FW has NCSI it may decide not to report 64B CQE/EQEs
	 */
	if (!(dev_cap->flags & MLX4_DEV_CAP_FLAG_64B_EQE) ||
	    !(dev_cap->flags & MLX4_DEV_CAP_FLAG_64B_CQE)) {
		dev_cap->flags2 &= ~MLX4_DEV_CAP_FLAG2_CQE_STRIDE;
		dev_cap->flags2 &= ~MLX4_DEV_CAP_FLAG2_EQE_STRIDE;
		return;
	}

	if (cache_line_size() == 128 || cache_line_size() == 256) {
		mlx4_dbg(dev, "Enabling CQE stride cacheLine supported\n");
		/* Changing the real data inside CQE size to 32B */
		dev_cap->flags &= ~MLX4_DEV_CAP_FLAG_64B_CQE;
		dev_cap->flags &= ~MLX4_DEV_CAP_FLAG_64B_EQE;

		if (mlx4_is_master(dev))
			dev_cap->function_caps |= MLX4_FUNC_CAP_EQE_CQE_STRIDE;
	} else {
		if (cache_line_size() != 32  && cache_line_size() != 64)
			mlx4_dbg(dev, "Disabling CQE stride, cacheLine size unsupported\n");
		dev_cap->flags2 &= ~MLX4_DEV_CAP_FLAG2_CQE_STRIDE;
		dev_cap->flags2 &= ~MLX4_DEV_CAP_FLAG2_EQE_STRIDE;
	}
}

static int _mlx4_dev_port(struct mlx4_dev *dev, int port,
			  struct mlx4_port_cap *port_cap)
{
	dev->caps.vl_cap[port]	    = port_cap->max_vl;
	dev->caps.ib_mtu_cap[port]	    = port_cap->ib_mtu;
	dev->phys_caps.gid_phys_table_len[port]  = port_cap->max_gids;
	dev->phys_caps.pkey_phys_table_len[port] = port_cap->max_pkeys;
	/* set gid and pkey table operating lengths by default
	 * to non-sriov values
	 */
	dev->caps.gid_table_len[port]  = port_cap->max_gids;
	dev->caps.pkey_table_len[port] = port_cap->max_pkeys;
	dev->caps.port_width_cap[port] = port_cap->max_port_width;
	dev->caps.eth_mtu_cap[port]    = port_cap->eth_mtu;
	dev->caps.max_tc_eth	       = port_cap->max_tc_eth;
	dev->caps.def_mac[port]        = port_cap->def_mac;
	dev->caps.supported_type[port] = port_cap->supported_port_types;
	dev->caps.suggested_type[port] = port_cap->suggested_type;
	dev->caps.default_sense[port] = port_cap->default_sense;
	dev->caps.trans_type[port]	    = port_cap->trans_type;
	dev->caps.vendor_oui[port]     = port_cap->vendor_oui;
	dev->caps.wavelength[port]     = port_cap->wavelength;
	dev->caps.trans_code[port]     = port_cap->trans_code;

	return 0;
}

static int mlx4_dev_port(struct mlx4_dev *dev, int port,
			 struct mlx4_port_cap *port_cap)
{
	int err = 0;

	err = mlx4_QUERY_PORT(dev, port, port_cap);

	if (err)
		mlx4_err(dev, "QUERY_PORT command failed.\n");

	return err;
}

static inline void mlx4_enable_ignore_fcs(struct mlx4_dev *dev)
{
	if (!(dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_IGNORE_FCS))
		return;

	if (mlx4_is_mfunc(dev)) {
		mlx4_dbg(dev, "SRIOV mode - Disabling Ignore FCS");
		dev->caps.flags2 &= ~MLX4_DEV_CAP_FLAG2_IGNORE_FCS;
		return;
	}

	if (!(dev->caps.flags & MLX4_DEV_CAP_FLAG_FCS_KEEP)) {
		mlx4_dbg(dev,
			 "Keep FCS is not supported - Disabling Ignore FCS");
		dev->caps.flags2 &= ~MLX4_DEV_CAP_FLAG2_IGNORE_FCS;
		return;
	}
}

#define MLX4_A0_STEERING_TABLE_SIZE	256
static int mlx4_dev_cap(struct mlx4_dev *dev, struct mlx4_dev_cap *dev_cap)
{
	int err;
	int i;

	err = mlx4_QUERY_DEV_CAP(dev, dev_cap);
	if (err) {
		mlx4_err(dev, "QUERY_DEV_CAP command failed, aborting\n");
		return err;
	}
	mlx4_dev_cap_dump(dev, dev_cap);

	if (dev_cap->min_page_sz > PAGE_SIZE) {
		mlx4_err(dev, "HCA minimum page size of %d bigger than kernel PAGE_SIZE of %ld, aborting\n",
			 dev_cap->min_page_sz, PAGE_SIZE);
		return -ENODEV;
	}
	if (dev_cap->num_ports > MLX4_MAX_PORTS) {
		mlx4_err(dev, "HCA has %d ports, but we only support %d, aborting\n",
			 dev_cap->num_ports, MLX4_MAX_PORTS);
		return -ENODEV;
	}

	if (dev_cap->uar_size > pci_resource_len(dev->persist->pdev, 2)) {
		mlx4_err(dev, "HCA reported UAR size of 0x%x bigger than PCI resource 2 size of 0x%llx, aborting\n",
			 dev_cap->uar_size,
			 (unsigned long long)
			 pci_resource_len(dev->persist->pdev, 2));
		return -ENODEV;
	}

	dev->caps.num_ports	     = dev_cap->num_ports;
	dev->caps.num_sys_eqs = dev_cap->num_sys_eqs;
	dev->phys_caps.num_phys_eqs = dev_cap->flags2 & MLX4_DEV_CAP_FLAG2_SYS_EQS ?
				      dev->caps.num_sys_eqs :
				      MLX4_MAX_EQ_NUM;
	for (i = 1; i <= dev->caps.num_ports; ++i) {
		err = _mlx4_dev_port(dev, i, dev_cap->port_cap + i);
		if (err) {
			mlx4_err(dev, "QUERY_PORT command failed, aborting\n");
			return err;
		}
	}

	dev->caps.uar_page_size	     = PAGE_SIZE;
	dev->caps.num_uars	     = dev_cap->uar_size / PAGE_SIZE;
	dev->caps.local_ca_ack_delay = dev_cap->local_ca_ack_delay;
	dev->caps.bf_reg_size	     = dev_cap->bf_reg_size;
	dev->caps.bf_regs_per_page   = dev_cap->bf_regs_per_page;
	dev->caps.max_sq_sg	     = dev_cap->max_sq_sg;
	dev->caps.max_rq_sg	     = dev_cap->max_rq_sg;
	dev->caps.max_wqes	     = dev_cap->max_qp_sz;
	dev->caps.max_qp_init_rdma   = dev_cap->max_requester_per_qp;
	dev->caps.max_srq_wqes	     = dev_cap->max_srq_sz;
	dev->caps.max_srq_sge	     = dev_cap->max_rq_sg - 1;
	dev->caps.reserved_srqs	     = dev_cap->reserved_srqs;
	dev->caps.max_sq_desc_sz     = dev_cap->max_sq_desc_sz;
	dev->caps.max_rq_desc_sz     = dev_cap->max_rq_desc_sz;
	/*
	 * Subtract 1 from the limit because we need to allocate a
	 * spare CQE to enable resizing the CQ.
	 */
	dev->caps.max_cqes	     = dev_cap->max_cq_sz - 1;
	dev->caps.reserved_cqs	     = dev_cap->reserved_cqs;
	dev->caps.reserved_eqs	     = dev_cap->reserved_eqs;
	dev->caps.reserved_mtts      = dev_cap->reserved_mtts;
	dev->caps.reserved_mrws	     = dev_cap->reserved_mrws;

	dev->caps.reserved_pds	     = dev_cap->reserved_pds;
	dev->caps.reserved_xrcds     = (dev->caps.flags & MLX4_DEV_CAP_FLAG_XRC) ?
					dev_cap->reserved_xrcds : 0;
	dev->caps.max_xrcds          = (dev->caps.flags & MLX4_DEV_CAP_FLAG_XRC) ?
					dev_cap->max_xrcds : 0;
	dev->caps.mtt_entry_sz       = dev_cap->mtt_entry_sz;

	dev->caps.max_msg_sz         = dev_cap->max_msg_sz;
	dev->caps.page_size_cap	     = ~(u32) (dev_cap->min_page_sz - 1);
	dev->caps.flags		     = dev_cap->flags;
	dev->caps.flags2	     = dev_cap->flags2;
	dev->caps.bmme_flags	     = dev_cap->bmme_flags;
	dev->caps.reserved_lkey	     = dev_cap->reserved_lkey;
	dev->caps.stat_rate_support  = dev_cap->stat_rate_support;
	dev->caps.max_gso_sz	     = dev_cap->max_gso_sz;
	dev->caps.max_rss_tbl_sz     = dev_cap->max_rss_tbl_sz;
	dev->caps.wol_port[1]          = dev_cap->wol_port[1];
	dev->caps.wol_port[2]          = dev_cap->wol_port[2];
	dev->caps.health_buffer_addrs  = dev_cap->health_buffer_addrs;

	/* Save uar page shift */
	if (!mlx4_is_slave(dev)) {
		/* Virtual PCI function needs to determine UAR page size from
		 * firmware. Only master PCI function can set the uar page size
		 */
		if (enable_4k_uar || !dev->persist->num_vfs)
			dev->uar_page_shift = DEFAULT_UAR_PAGE_SHIFT;
		else
			dev->uar_page_shift = PAGE_SHIFT;

		mlx4_set_num_reserved_uars(dev, dev_cap);
	}

	if (dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_PHV_EN) {
		struct mlx4_init_hca_param hca_param;

		memset(&hca_param, 0, sizeof(hca_param));
		err = mlx4_QUERY_HCA(dev, &hca_param);
		/* Turn off PHV_EN flag in case phv_check_en is set.
		 * phv_check_en is a HW check that parse the packet and verify
		 * phv bit was reported correctly in the wqe. To allow QinQ
		 * PHV_EN flag should be set and phv_check_en must be cleared
		 * otherwise QinQ packets will be drop by the HW.
		 */
		if (err || hca_param.phv_check_en)
			dev->caps.flags2 &= ~MLX4_DEV_CAP_FLAG2_PHV_EN;
	}

	/* Sense port always allowed on supported devices for ConnectX-1 and -2 */
	if (mlx4_priv(dev)->pci_dev_data & MLX4_PCI_DEV_FORCE_SENSE_PORT)
		dev->caps.flags |= MLX4_DEV_CAP_FLAG_SENSE_SUPPORT;
	/* Don't do sense port on multifunction devices (for now at least) */
	if (mlx4_is_mfunc(dev))
		dev->caps.flags &= ~MLX4_DEV_CAP_FLAG_SENSE_SUPPORT;

	if (mlx4_low_memory_profile()) {
		dev->caps.log_num_macs  = MLX4_MIN_LOG_NUM_MAC;
		dev->caps.log_num_vlans = MLX4_MIN_LOG_NUM_VLANS;
	} else {
		dev->caps.log_num_macs  = log_num_mac;
		dev->caps.log_num_vlans = MLX4_LOG_NUM_VLANS;
	}

	for (i = 1; i <= dev->caps.num_ports; ++i) {
		dev->caps.port_type[i] = MLX4_PORT_TYPE_NONE;
		if (dev->caps.supported_type[i]) {
			/* if only ETH is supported - assign ETH */
			if (dev->caps.supported_type[i] == MLX4_PORT_TYPE_ETH)
				dev->caps.port_type[i] = MLX4_PORT_TYPE_ETH;
			/* if only IB is supported, assign IB */
			else if (dev->caps.supported_type[i] ==
				 MLX4_PORT_TYPE_IB)
				dev->caps.port_type[i] = MLX4_PORT_TYPE_IB;
			else {
				/* if IB and ETH are supported, we set the port
				 * type according to user selection of port type;
				 * if user selected none, take the FW hint */
				if (port_type_array[i - 1] == MLX4_PORT_TYPE_NONE)
					dev->caps.port_type[i] = dev->caps.suggested_type[i] ?
						MLX4_PORT_TYPE_ETH : MLX4_PORT_TYPE_IB;
				else
					dev->caps.port_type[i] = port_type_array[i - 1];
			}
		}
		/*
		 * Link sensing is allowed on the port if 3 conditions are true:
		 * 1. Both protocols are supported on the port.
		 * 2. Different types are supported on the port
		 * 3. FW declared that it supports link sensing
		 */
		mlx4_priv(dev)->sense.sense_allowed[i] =
			((dev->caps.supported_type[i] == MLX4_PORT_TYPE_AUTO) &&
			 (dev->caps.flags & MLX4_DEV_CAP_FLAG_DPDP) &&
			 (dev->caps.flags & MLX4_DEV_CAP_FLAG_SENSE_SUPPORT));

		/*
		 * If "default_sense" bit is set, we move the port to "AUTO" mode
		 * and perform sense_port FW command to try and set the correct
		 * port type from beginning
		 */
		if (mlx4_priv(dev)->sense.sense_allowed[i] && dev->caps.default_sense[i]) {
			enum mlx4_port_type sensed_port = MLX4_PORT_TYPE_NONE;
			dev->caps.possible_type[i] = MLX4_PORT_TYPE_AUTO;
			mlx4_SENSE_PORT(dev, i, &sensed_port);
			if (sensed_port != MLX4_PORT_TYPE_NONE)
				dev->caps.port_type[i] = sensed_port;
		} else {
			dev->caps.possible_type[i] = dev->caps.port_type[i];
		}

		if (dev->caps.log_num_macs > dev_cap->port_cap[i].log_max_macs) {
			dev->caps.log_num_macs = dev_cap->port_cap[i].log_max_macs;
			mlx4_warn(dev, "Requested number of MACs is too much for port %d, reducing to %d\n",
				  i, 1 << dev->caps.log_num_macs);
		}
		if (dev->caps.log_num_vlans > dev_cap->port_cap[i].log_max_vlans) {
			dev->caps.log_num_vlans = dev_cap->port_cap[i].log_max_vlans;
			mlx4_warn(dev, "Requested number of VLANs is too much for port %d, reducing to %d\n",
				  i, 1 << dev->caps.log_num_vlans);
		}
	}

	if (mlx4_is_master(dev) && (dev->caps.num_ports == 2) &&
	    (port_type_array[0] == MLX4_PORT_TYPE_IB) &&
	    (port_type_array[1] == MLX4_PORT_TYPE_ETH)) {
		mlx4_warn(dev,
			  "Granular QoS per VF not supported with IB/Eth configuration\n");
		dev->caps.flags2 &= ~MLX4_DEV_CAP_FLAG2_QOS_VPP;
	}

	dev->caps.max_counters = dev_cap->max_counters;

	dev->caps.reserved_qps_cnt[MLX4_QP_REGION_FW] = dev_cap->reserved_qps;
	dev->caps.reserved_qps_cnt[MLX4_QP_REGION_ETH_ADDR] =
		dev->caps.reserved_qps_cnt[MLX4_QP_REGION_FC_ADDR] =
		(1 << dev->caps.log_num_macs) *
		(1 << dev->caps.log_num_vlans) *
		dev->caps.num_ports;
	dev->caps.reserved_qps_cnt[MLX4_QP_REGION_FC_EXCH] = MLX4_NUM_FEXCH;

	if (dev_cap->dmfs_high_rate_qpn_base > 0 &&
	    dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_FS_EN)
		dev->caps.dmfs_high_rate_qpn_base = dev_cap->dmfs_high_rate_qpn_base;
	else
		dev->caps.dmfs_high_rate_qpn_base =
			dev->caps.reserved_qps_cnt[MLX4_QP_REGION_FW];

	if (dev_cap->dmfs_high_rate_qpn_range > 0 &&
	    dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_FS_EN) {
		dev->caps.dmfs_high_rate_qpn_range = dev_cap->dmfs_high_rate_qpn_range;
		dev->caps.dmfs_high_steer_mode = MLX4_STEERING_DMFS_A0_DEFAULT;
		dev->caps.flags2 |= MLX4_DEV_CAP_FLAG2_FS_A0;
	} else {
		dev->caps.dmfs_high_steer_mode = MLX4_STEERING_DMFS_A0_NOT_SUPPORTED;
		dev->caps.dmfs_high_rate_qpn_base =
			dev->caps.reserved_qps_cnt[MLX4_QP_REGION_FW];
		dev->caps.dmfs_high_rate_qpn_range = MLX4_A0_STEERING_TABLE_SIZE;
	}

	dev->caps.rl_caps = dev_cap->rl_caps;

	dev->caps.reserved_qps_cnt[MLX4_QP_REGION_RSS_RAW_ETH] =
		dev->caps.dmfs_high_rate_qpn_range;

	dev->caps.reserved_qps = dev->caps.reserved_qps_cnt[MLX4_QP_REGION_FW] +
		dev->caps.reserved_qps_cnt[MLX4_QP_REGION_ETH_ADDR] +
		dev->caps.reserved_qps_cnt[MLX4_QP_REGION_FC_ADDR] +
		dev->caps.reserved_qps_cnt[MLX4_QP_REGION_FC_EXCH];

	dev->caps.sqp_demux = (mlx4_is_master(dev)) ? MLX4_MAX_NUM_SLAVES : 0;

	if (!enable_64b_cqe_eqe && !mlx4_is_slave(dev)) {
		if (dev_cap->flags &
		    (MLX4_DEV_CAP_FLAG_64B_CQE | MLX4_DEV_CAP_FLAG_64B_EQE)) {
			mlx4_warn(dev, "64B EQEs/CQEs supported by the device but not enabled\n");
			dev->caps.flags &= ~MLX4_DEV_CAP_FLAG_64B_CQE;
			dev->caps.flags &= ~MLX4_DEV_CAP_FLAG_64B_EQE;
		}

		if (dev_cap->flags2 &
		    (MLX4_DEV_CAP_FLAG2_CQE_STRIDE |
		     MLX4_DEV_CAP_FLAG2_EQE_STRIDE)) {
			mlx4_warn(dev, "Disabling EQE/CQE stride per user request\n");
			dev_cap->flags2 &= ~MLX4_DEV_CAP_FLAG2_CQE_STRIDE;
			dev_cap->flags2 &= ~MLX4_DEV_CAP_FLAG2_EQE_STRIDE;
		}
	}

	if ((dev->caps.flags &
	    (MLX4_DEV_CAP_FLAG_64B_CQE | MLX4_DEV_CAP_FLAG_64B_EQE)) &&
	    mlx4_is_master(dev))
		dev->caps.function_caps |= MLX4_FUNC_CAP_64B_EQE_CQE;

	if (!mlx4_is_slave(dev)) {
		mlx4_enable_cqe_eqe_stride(dev);
		dev->caps.alloc_res_qp_mask =
			(dev->caps.bf_reg_size ? MLX4_RESERVE_ETH_BF_QP : 0) |
			MLX4_RESERVE_A0_QP;

		if (!(dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_ETS_CFG) &&
		    dev->caps.flags & MLX4_DEV_CAP_FLAG_SET_ETH_SCHED) {
			mlx4_warn(dev, "Old device ETS support detected\n");
			mlx4_warn(dev, "Consider upgrading device FW.\n");
			dev->caps.flags2 |= MLX4_DEV_CAP_FLAG2_ETS_CFG;
		}

	} else {
		dev->caps.alloc_res_qp_mask = 0;
	}

	mlx4_enable_ignore_fcs(dev);

	return 0;
}

/*The function checks if there are live vf, return the num of them*/
static int mlx4_how_many_lives_vf(struct mlx4_dev *dev)
{
	struct mlx4_priv *priv = mlx4_priv(dev);
	struct mlx4_slave_state *s_state;
	int i;
	int ret = 0;

	for (i = 1/*the ppf is 0*/; i < dev->num_slaves; ++i) {
		s_state = &priv->mfunc.master.slave_state[i];
		if (s_state->active && s_state->last_cmd !=
		    MLX4_COMM_CMD_RESET) {
			mlx4_warn(dev, "%s: slave: %d is still active\n",
				  __func__, i);
			ret++;
		}
	}
	return ret;
}

int mlx4_get_parav_qkey(struct mlx4_dev *dev, u32 qpn, u32 *qkey)
{
	u32 qk = MLX4_RESERVED_QKEY_BASE;

	if (qpn >= dev->phys_caps.base_tunnel_sqpn + 8 * MLX4_MFUNC_MAX ||
	    qpn < dev->phys_caps.base_proxy_sqpn)
		return -EINVAL;

	if (qpn >= dev->phys_caps.base_tunnel_sqpn)
		/* tunnel qp */
		qk += qpn - dev->phys_caps.base_tunnel_sqpn;
	else
		qk += qpn - dev->phys_caps.base_proxy_sqpn;
	*qkey = qk;
	return 0;
}
EXPORT_SYMBOL(mlx4_get_parav_qkey);

void mlx4_sync_pkey_table(struct mlx4_dev *dev, int slave, int port, int i, int val)
{
	struct mlx4_priv *priv = container_of(dev, struct mlx4_priv, dev);

	if (!mlx4_is_master(dev))
		return;

	priv->virt2phys_pkey[slave][port - 1][i] = val;
}
EXPORT_SYMBOL(mlx4_sync_pkey_table);

void mlx4_put_slave_node_guid(struct mlx4_dev *dev, int slave, __be64 guid)
{
	struct mlx4_priv *priv = container_of(dev, struct mlx4_priv, dev);

	if (!mlx4_is_master(dev))
		return;

	priv->slave_node_guids[slave] = guid;
}
EXPORT_SYMBOL(mlx4_put_slave_node_guid);

__be64 mlx4_get_slave_node_guid(struct mlx4_dev *dev, int slave)
{
	struct mlx4_priv *priv = container_of(dev, struct mlx4_priv, dev);

	if (!mlx4_is_master(dev))
		return 0;

	return priv->slave_node_guids[slave];
}
EXPORT_SYMBOL(mlx4_get_slave_node_guid);

int mlx4_is_slave_active(struct mlx4_dev *dev, int slave)
{
	struct mlx4_priv *priv = mlx4_priv(dev);
	struct mlx4_slave_state *s_slave;

	if (!mlx4_is_master(dev))
		return 0;

	s_slave = &priv->mfunc.master.slave_state[slave];
	return !!s_slave->active;
}
EXPORT_SYMBOL(mlx4_is_slave_active);

void mlx4_handle_eth_header_mcast_prio(struct mlx4_net_trans_rule_hw_ctrl *ctrl,
				       struct _rule_hw *eth_header)
{
	if (is_multicast_ether_addr(eth_header->eth.dst_mac) ||
	    is_broadcast_ether_addr(eth_header->eth.dst_mac)) {
		struct mlx4_net_trans_rule_hw_eth *eth =
			(struct mlx4_net_trans_rule_hw_eth *)eth_header;
		struct _rule_hw *next_rule = (struct _rule_hw *)(eth + 1);
		bool last_rule = next_rule->size == 0 && next_rule->id == 0 &&
			next_rule->rsvd == 0;

		if (last_rule)
			ctrl->prio = cpu_to_be16(MLX4_DOMAIN_NIC);
	}
}
EXPORT_SYMBOL(mlx4_handle_eth_header_mcast_prio);

static void slave_adjust_steering_mode(struct mlx4_dev *dev,
				       struct mlx4_dev_cap *dev_cap,
				       struct mlx4_init_hca_param *hca_param)
{
	dev->caps.steering_mode = hca_param->steering_mode;
	if (dev->caps.steering_mode == MLX4_STEERING_MODE_DEVICE_MANAGED) {
		dev->caps.num_qp_per_mgm = dev_cap->fs_max_num_qp_per_entry;
		dev->caps.fs_log_max_ucast_qp_range_size =
			dev_cap->fs_log_max_ucast_qp_range_size;
	} else
		dev->caps.num_qp_per_mgm =
			4 * ((1 << hca_param->log_mc_entry_sz)/16 - 2);

	mlx4_dbg(dev, "Steering mode is: %s\n",
		 mlx4_steering_mode_str(dev->caps.steering_mode));
}

static void mlx4_slave_destroy_special_qp_cap(struct mlx4_dev *dev)
{
	kfree(dev->caps.spec_qps);
	dev->caps.spec_qps = NULL;
}

static int mlx4_slave_special_qp_cap(struct mlx4_dev *dev)
{
	struct mlx4_func_cap *func_cap = NULL;
	struct mlx4_caps *caps = &dev->caps;
	int i, err = 0;

	func_cap = kzalloc(sizeof(*func_cap), GFP_KERNEL);
	caps->spec_qps = kcalloc(caps->num_ports, sizeof(*caps->spec_qps), GFP_KERNEL);

	if (!func_cap || !caps->spec_qps) {
		mlx4_err(dev, "Failed to allocate memory for special qps cap\n");
		err = -ENOMEM;
		goto err_mem;
	}

	for (i = 1; i <= caps->num_ports; ++i) {
		err = mlx4_QUERY_FUNC_CAP(dev, i, func_cap);
		if (err) {
			mlx4_err(dev, "QUERY_FUNC_CAP port command failed for port %d, aborting (%d)\n",
				 i, err);
			goto err_mem;
		}
		caps->spec_qps[i - 1] = func_cap->spec_qps;
		caps->port_mask[i] = caps->port_type[i];
		caps->phys_port_id[i] = func_cap->phys_port_id;
		err = mlx4_get_slave_pkey_gid_tbl_len(dev, i,
						      &caps->gid_table_len[i],
						      &caps->pkey_table_len[i]);
		if (err) {
			mlx4_err(dev, "QUERY_PORT command failed for port %d, aborting (%d)\n",
				 i, err);
			goto err_mem;
		}
	}

err_mem:
	if (err)
		mlx4_slave_destroy_special_qp_cap(dev);
	kfree(func_cap);
	return err;
}

static int mlx4_slave_cap(struct mlx4_dev *dev)
{
	int			   err;
	u32			   page_size;
	struct mlx4_dev_cap	   *dev_cap = NULL;
	struct mlx4_func_cap	   *func_cap = NULL;
	struct mlx4_init_hca_param *hca_param = NULL;

	hca_param = kzalloc(sizeof(*hca_param), GFP_KERNEL);
	func_cap = kzalloc(sizeof(*func_cap), GFP_KERNEL);
	dev_cap = kzalloc(sizeof(*dev_cap), GFP_KERNEL);
	if (!hca_param || !func_cap || !dev_cap) {
		mlx4_err(dev, "Failed to allocate memory for slave_cap\n");
		err = -ENOMEM;
		goto free_mem;
	}

	err = mlx4_QUERY_HCA(dev, hca_param);
	if (err) {
		mlx4_err(dev, "QUERY_HCA command failed, aborting\n");
		goto free_mem;
	}

	/* fail if the hca has an unknown global capability
	 * at this time global_caps should be always zeroed
	 */
	if (hca_param->global_caps) {
		mlx4_err(dev, "Unknown hca global capabilities\n");
		err = -EINVAL;
		goto free_mem;
	}

	dev->caps.hca_core_clock = hca_param->hca_core_clock;

	dev->caps.max_qp_dest_rdma = 1 << hca_param->log_rd_per_qp;
	err = mlx4_dev_cap(dev, dev_cap);
	if (err) {
		mlx4_err(dev, "QUERY_DEV_CAP command failed, aborting\n");
		goto free_mem;
	}

	err = mlx4_QUERY_FW(dev);
	if (err)
		mlx4_err(dev, "QUERY_FW command failed: could not get FW version\n");

	page_size = ~dev->caps.page_size_cap + 1;
	mlx4_warn(dev, "HCA minimum page size:%d\n", page_size);
	if (page_size > PAGE_SIZE) {
		mlx4_err(dev, "HCA minimum page size of %d bigger than kernel PAGE_SIZE of %ld, aborting\n",
			 page_size, PAGE_SIZE);
		err = -ENODEV;
		goto free_mem;
	}

	/* Set uar_page_shift for VF */
	dev->uar_page_shift = hca_param->uar_page_sz + 12;

	/* Make sure the master uar page size is valid */
	if (dev->uar_page_shift > PAGE_SHIFT) {
		mlx4_err(dev,
			 "Invalid configuration: uar page size is larger than system page size\n");
		err = -ENODEV;
		goto free_mem;
	}

	/* Set reserved_uars based on the uar_page_shift */
	mlx4_set_num_reserved_uars(dev, dev_cap);

	/* Although uar page size in FW differs from system page size,
	 * upper software layers (mlx4_ib, mlx4_en and part of mlx4_core)
	 * still works with assumption that uar page size == system page size
	 */
	dev->caps.uar_page_size = PAGE_SIZE;

	err = mlx4_QUERY_FUNC_CAP(dev, 0, func_cap);
	if (err) {
		mlx4_err(dev, "QUERY_FUNC_CAP general command failed, aborting (%d)\n",
			 err);
		goto free_mem;
	}

	if ((func_cap->pf_context_behaviour | PF_CONTEXT_BEHAVIOUR_MASK) !=
	    PF_CONTEXT_BEHAVIOUR_MASK) {
		mlx4_err(dev, "Unknown pf context behaviour %x known flags %x\n",
			 func_cap->pf_context_behaviour,
			 PF_CONTEXT_BEHAVIOUR_MASK);
		err = -EINVAL;
		goto free_mem;
	}

	dev->caps.num_ports		= func_cap->num_ports;
	dev->quotas.qp			= func_cap->qp_quota;
	dev->quotas.srq			= func_cap->srq_quota;
	dev->quotas.cq			= func_cap->cq_quota;
	dev->quotas.mpt			= func_cap->mpt_quota;
	dev->quotas.mtt			= func_cap->mtt_quota;
	dev->caps.num_qps		= 1 << hca_param->log_num_qps;
	dev->caps.num_srqs		= 1 << hca_param->log_num_srqs;
	dev->caps.num_cqs		= 1 << hca_param->log_num_cqs;
	dev->caps.num_mpts		= 1 << hca_param->log_mpt_sz;
	dev->caps.num_eqs		= func_cap->max_eq;
	dev->caps.reserved_eqs		= func_cap->reserved_eq;
	dev->caps.reserved_lkey		= func_cap->reserved_lkey;
	dev->caps.num_pds               = MLX4_NUM_PDS;
	dev->caps.num_mgms              = 0;
	dev->caps.num_amgms             = 0;

	if (dev->caps.num_ports > MLX4_MAX_PORTS) {
		mlx4_err(dev, "HCA has %d ports, but we only support %d, aborting\n",
			 dev->caps.num_ports, MLX4_MAX_PORTS);
		err = -ENODEV;
		goto free_mem;
	}

	mlx4_replace_zero_macs(dev);

	err = mlx4_slave_special_qp_cap(dev);
	if (err) {
		mlx4_err(dev, "Set special QP caps failed. aborting\n");
		goto free_mem;
	}

	if (dev->caps.uar_page_size * (dev->caps.num_uars -
				       dev->caps.reserved_uars) >
				       pci_resource_len(dev->persist->pdev,
							2)) {
		mlx4_err(dev, "HCA reported UAR region size of 0x%x bigger than PCI resource 2 size of 0x%llx, aborting\n",
			 dev->caps.uar_page_size * dev->caps.num_uars,
			 (unsigned long long)
			 pci_resource_len(dev->persist->pdev, 2));
		err = -ENOMEM;
		goto err_mem;
	}

	if (hca_param->dev_cap_enabled & MLX4_DEV_CAP_64B_EQE_ENABLED) {
		dev->caps.eqe_size   = 64;
		dev->caps.eqe_factor = 1;
	} else {
		dev->caps.eqe_size   = 32;
		dev->caps.eqe_factor = 0;
	}

	if (hca_param->dev_cap_enabled & MLX4_DEV_CAP_64B_CQE_ENABLED) {
		dev->caps.cqe_size   = 64;
		dev->caps.userspace_caps |= MLX4_USER_DEV_CAP_LARGE_CQE;
	} else {
		dev->caps.cqe_size   = 32;
	}

	if (hca_param->dev_cap_enabled & MLX4_DEV_CAP_EQE_STRIDE_ENABLED) {
		dev->caps.eqe_size = hca_param->eqe_size;
		dev->caps.eqe_factor = 0;
	}

	if (hca_param->dev_cap_enabled & MLX4_DEV_CAP_CQE_STRIDE_ENABLED) {
		dev->caps.cqe_size = hca_param->cqe_size;
		/* User still need to know when CQE > 32B */
		dev->caps.userspace_caps |= MLX4_USER_DEV_CAP_LARGE_CQE;
	}

	dev->caps.flags2 &= ~MLX4_DEV_CAP_FLAG2_TS;
	mlx4_warn(dev, "Timestamping is not supported in slave mode\n");

	dev->caps.flags2 &= ~MLX4_DEV_CAP_FLAG2_USER_MAC_EN;
	mlx4_dbg(dev, "User MAC FW update is not supported in slave mode\n");

	slave_adjust_steering_mode(dev, dev_cap, hca_param);
	mlx4_dbg(dev, "RSS support for IP fragments is %s\n",
		 hca_param->rss_ip_frags ? "on" : "off");

	if (func_cap->extra_flags & MLX4_QUERY_FUNC_FLAGS_BF_RES_QP &&
	    dev->caps.bf_reg_size)
		dev->caps.alloc_res_qp_mask |= MLX4_RESERVE_ETH_BF_QP;

	if (func_cap->extra_flags & MLX4_QUERY_FUNC_FLAGS_A0_RES_QP)
		dev->caps.alloc_res_qp_mask |= MLX4_RESERVE_A0_QP;

err_mem:
	if (err)
		mlx4_slave_destroy_special_qp_cap(dev);
free_mem:
	kfree(hca_param);
	kfree(func_cap);
	kfree(dev_cap);
	return err;
}

static void mlx4_request_modules(struct mlx4_dev *dev)
{
	int port;
	int has_ib_port = false;
	int has_eth_port = false;
#define EN_DRV_NAME	"mlx4_en"
#define IB_DRV_NAME	"mlx4_ib"

	for (port = 1; port <= dev->caps.num_ports; port++) {
		if (dev->caps.port_type[port] == MLX4_PORT_TYPE_IB)
			has_ib_port = true;
		else if (dev->caps.port_type[port] == MLX4_PORT_TYPE_ETH)
			has_eth_port = true;
	}

	if (has_eth_port)
		request_module_nowait(EN_DRV_NAME);
	if (has_ib_port || (dev->caps.flags & MLX4_DEV_CAP_FLAG_IBOE))
		request_module_nowait(IB_DRV_NAME);
}

/*
 * Change the port configuration of the device.
 * Every user of this function must hold the port mutex.
 */
int mlx4_change_port_types(struct mlx4_dev *dev,
			   enum mlx4_port_type *port_types)
{
	int err = 0;
	int change = 0;
	int port;

	for (port = 0; port <  dev->caps.num_ports; port++) {
		/* Change the port type only if the new type is different
		 * from the current, and not set to Auto */
		if (port_types[port] != dev->caps.port_type[port + 1])
			change = 1;
	}
	if (change) {
		mlx4_unregister_device(dev);
		for (port = 1; port <= dev->caps.num_ports; port++) {
			mlx4_CLOSE_PORT(dev, port);
			dev->caps.port_type[port] = port_types[port - 1];
			err = mlx4_SET_PORT(dev, port, -1);
			if (err) {
				mlx4_err(dev, "Failed to set port %d, aborting\n",
					 port);
				goto out;
			}
		}
		mlx4_set_port_mask(dev);
		err = mlx4_register_device(dev);
		if (err) {
			mlx4_err(dev, "Failed to register device\n");
			goto out;
		}
		mlx4_request_modules(dev);
	}

out:
	return err;
}

static ssize_t show_port_type(struct device *dev,
			      struct device_attribute *attr,
			      char *buf)
{
	struct mlx4_port_info *info = container_of(attr, struct mlx4_port_info,
						   port_attr);
	struct mlx4_dev *mdev = info->dev;
	char type[8];

	sprintf(type, "%s",
		(mdev->caps.port_type[info->port] == MLX4_PORT_TYPE_IB) ?
		"ib" : "eth");
	if (mdev->caps.possible_type[info->port] == MLX4_PORT_TYPE_AUTO)
		sprintf(buf, "auto (%s)\n", type);
	else
		sprintf(buf, "%s\n", type);

	return strlen(buf);
}

static int __set_port_type(struct mlx4_port_info *info,
			   enum mlx4_port_type port_type)
{
	struct mlx4_dev *mdev = info->dev;
	struct mlx4_priv *priv = mlx4_priv(mdev);
	enum mlx4_port_type types[MLX4_MAX_PORTS];
	enum mlx4_port_type new_types[MLX4_MAX_PORTS];
	int i;
	int err = 0;

	if ((port_type & mdev->caps.supported_type[info->port]) != port_type) {
		mlx4_err(mdev,
			 "Requested port type for port %d is not supported on this HCA\n",
			 info->port);
		return -EOPNOTSUPP;
	}

	mlx4_stop_sense(mdev);
	mutex_lock(&priv->port_mutex);
	info->tmp_type = port_type;

	/* Possible type is always the one that was delivered */
	mdev->caps.possible_type[info->port] = info->tmp_type;

	for (i = 0; i < mdev->caps.num_ports; i++) {
		types[i] = priv->port[i+1].tmp_type ? priv->port[i+1].tmp_type :
					mdev->caps.possible_type[i+1];
		if (types[i] == MLX4_PORT_TYPE_AUTO)
			types[i] = mdev->caps.port_type[i+1];
	}

	if (!(mdev->caps.flags & MLX4_DEV_CAP_FLAG_DPDP) &&
	    !(mdev->caps.flags & MLX4_DEV_CAP_FLAG_SENSE_SUPPORT)) {
		for (i = 1; i <= mdev->caps.num_ports; i++) {
			if (mdev->caps.possible_type