Merge branch 'mlx5-next' of git://git.kernel.org/pub/scm/linux/kernel/git/mellanox/linux

Merge mlx5-next shared branched into net-next,

From Bodong Wang:
1) Introduction of ECPF (Embedded CPU Physical Function), and low level
bits for mlx5 SmartNic capabilities support.
2) Vport enumeration refactoring that affect mlx5_ib and mlx5_core

From Aya Levin,
3) Add support for 50Gbps per lane link modes in the Port Type and Speed
register (PTYS)
4) Refactor low level query functions for PTYS register
5) Add support for 50Gbps per lane link modes to mlx5_ib

Note: due to a change in API in mlx5/core and a later patch from net-next,
a fixup was squashed with this merge commit that replaces FDB_UPLINK_VPORT
with MLX5_VPORT_UPLINK which exists only in upstream net-next.

Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>

8 files changed, 156 insertions, 116 deletions

diff --git a/drivers/infiniband/hw/mlx5/cmd.c b/drivers/infiniband/hw/mlx5/cmd.c
index 356bccc715ee..6bcc63aaa50b 100644
--- a/drivers/infiniband/hw/mlx5/cmd.c
+++ b/drivers/infiniband/hw/mlx5/cmd.c
@@ -345,3 +345,40 @@ int mlx5_cmd_alloc_q_counter(struct mlx5_core_dev *dev, u16 *counter_id,
 				       counter_set_id);
 	return err;
 }
+
+int mlx5_cmd_mad_ifc(struct mlx5_core_dev *dev, const void *inb, void *outb,
+		     u16 opmod, u8 port)
+{
+	int outlen = MLX5_ST_SZ_BYTES(mad_ifc_out);
+	int inlen = MLX5_ST_SZ_BYTES(mad_ifc_in);
+	int err = -ENOMEM;
+	void *data;
+	void *resp;
+	u32 *out;
+	u32 *in;
+
+	in = kzalloc(inlen, GFP_KERNEL);
+	out = kzalloc(outlen, GFP_KERNEL);
+	if (!in || !out)
+		goto out;
+
+	MLX5_SET(mad_ifc_in, in, opcode, MLX5_CMD_OP_MAD_IFC);
+	MLX5_SET(mad_ifc_in, in, op_mod, opmod);
+	MLX5_SET(mad_ifc_in, in, port, port);
+
+	data = MLX5_ADDR_OF(mad_ifc_in, in, mad);
+	memcpy(data, inb, MLX5_FLD_SZ_BYTES(mad_ifc_in, mad));
+
+	err = mlx5_cmd_exec(dev, in, inlen, out, outlen);
+	if (err)
+		goto out;
+
+	resp = MLX5_ADDR_OF(mad_ifc_out, out, response_mad_packet);
+	memcpy(outb, resp,
+	       MLX5_FLD_SZ_BYTES(mad_ifc_out, response_mad_packet));
+
+out:
+	kfree(out);
+	kfree(in);
+	return err;
+}
diff --git a/drivers/infiniband/hw/mlx5/cmd.h b/drivers/infiniband/hw/mlx5/cmd.h
index 1e76dc67a369..923a7b93f507 100644
--- a/drivers/infiniband/hw/mlx5/cmd.h
+++ b/drivers/infiniband/hw/mlx5/cmd.h
@@ -63,4 +63,6 @@ int mlx5_cmd_xrcd_alloc(struct mlx5_core_dev *dev, u32 *xrcdn, u16 uid);
 int mlx5_cmd_xrcd_dealloc(struct mlx5_core_dev *dev, u32 xrcdn, u16 uid);
 int mlx5_cmd_alloc_q_counter(struct mlx5_core_dev *dev, u16 *counter_id,
 			     u16 uid);
+int mlx5_cmd_mad_ifc(struct mlx5_core_dev *dev, const void *inb, void *outb,
+		     u16 opmod, u8 port);
 #endif /* MLX5_IB_CMD_H */
diff --git a/drivers/infiniband/hw/mlx5/ib_rep.c b/drivers/infiniband/hw/mlx5/ib_rep.c
index 46a9ddc8ca56..99cae9a10195 100644
--- a/drivers/infiniband/hw/mlx5/ib_rep.c
+++ b/drivers/infiniband/hw/mlx5/ib_rep.c
@@ -3,10 +3,11 @@
  * Copyright (c) 2018 Mellanox Technologies. All rights reserved.
  */
 
+#include <linux/mlx5/vport.h>
 #include "ib_rep.h"
 #include "srq.h"
 
-static const struct mlx5_ib_profile rep_profile = {
+static const struct mlx5_ib_profile vf_rep_profile = {
 	STAGE_CREATE(MLX5_IB_STAGE_INIT,
 		     mlx5_ib_stage_init_init,
 		     mlx5_ib_stage_init_cleanup),
@@ -46,30 +47,16 @@ static const struct mlx5_ib_profile rep_profile = {
 };
 
 static int
-mlx5_ib_nic_rep_load(struct mlx5_core_dev *dev, struct mlx5_eswitch_rep *rep)
-{
-	struct mlx5_ib_dev *ibdev;
-
-	ibdev = mlx5_ib_rep_to_dev(rep);
-	if (!__mlx5_ib_add(ibdev, ibdev->profile))
-		return -EINVAL;
-	return 0;
-}
-
-static void
-mlx5_ib_nic_rep_unload(struct mlx5_eswitch_rep *rep)
-{
-	struct mlx5_ib_dev *ibdev;
-
-	ibdev = mlx5_ib_rep_to_dev(rep);
-	__mlx5_ib_remove(ibdev, ibdev->profile, MLX5_IB_STAGE_MAX);
-}
-
-static int
 mlx5_ib_vport_rep_load(struct mlx5_core_dev *dev, struct mlx5_eswitch_rep *rep)
 {
+	const struct mlx5_ib_profile *profile;
 	struct mlx5_ib_dev *ibdev;
 
+	if (rep->vport == MLX5_VPORT_UPLINK)
+		profile = &uplink_rep_profile;
+	else
+		profile = &vf_rep_profile;
+
 	ibdev = (struct mlx5_ib_dev *)ib_alloc_device(sizeof(*ibdev));
 	if (!ibdev)
 		return -ENOMEM;
@@ -78,7 +65,7 @@ mlx5_ib_vport_rep_load(struct mlx5_core_dev *dev, struct mlx5_eswitch_rep *rep)
 	ibdev->mdev = dev;
 	ibdev->num_ports = max(MLX5_CAP_GEN(dev, num_ports),
 			       MLX5_CAP_GEN(dev, num_vhca_ports));
-	if (!__mlx5_ib_add(ibdev, &rep_profile))
+	if (!__mlx5_ib_add(ibdev, profile))
 		return -EINVAL;
 
 	rep->rep_if[REP_IB].priv = ibdev;
@@ -105,15 +92,14 @@ static void *mlx5_ib_vport_get_proto_dev(struct mlx5_eswitch_rep *rep)
 	return mlx5_ib_rep_to_dev(rep);
 }
 
-static void mlx5_ib_rep_register_vf_vports(struct mlx5_ib_dev *dev)
+void mlx5_ib_register_vport_reps(struct mlx5_core_dev *mdev)
 {
-	struct mlx5_eswitch *esw   = dev->mdev->priv.eswitch;
-	int total_vfs = MLX5_TOTAL_VPORTS(dev->mdev);
+	struct mlx5_eswitch *esw = mdev->priv.eswitch;
+	int total_vports = MLX5_TOTAL_VPORTS(mdev);
+	struct mlx5_eswitch_rep_if rep_if = {};
 	int vport;
 
-	for (vport = 1; vport < total_vfs; vport++) {
-		struct mlx5_eswitch_rep_if rep_if = {};
-
+	for (vport = 0; vport < total_vports; vport++) {
 		rep_if.load = mlx5_ib_vport_rep_load;
 		rep_if.unload = mlx5_ib_vport_rep_unload;
 		rep_if.get_proto_dev = mlx5_ib_vport_get_proto_dev;
@@ -121,39 +107,16 @@ static void mlx5_ib_rep_register_vf_vports(struct mlx5_ib_dev *dev)
 	}
 }
 
-static void mlx5_ib_rep_unregister_vf_vports(struct mlx5_ib_dev *dev)
+void mlx5_ib_unregister_vport_reps(struct mlx5_core_dev *mdev)
 {
-	struct mlx5_eswitch *esw   = dev->mdev->priv.eswitch;
-	int total_vfs = MLX5_TOTAL_VPORTS(dev->mdev);
+	struct mlx5_eswitch *esw = mdev->priv.eswitch;
+	int total_vports = MLX5_TOTAL_VPORTS(mdev);
 	int vport;
 
-	for (vport = 1; vport < total_vfs; vport++)
+	for (vport = total_vports - 1; vport >= 0; vport--)
 		mlx5_eswitch_unregister_vport_rep(esw, vport, REP_IB);
 }
 
-void mlx5_ib_register_vport_reps(struct mlx5_ib_dev *dev)
-{
-	struct mlx5_eswitch *esw = dev->mdev->priv.eswitch;
-	struct mlx5_eswitch_rep_if rep_if = {};
-
-	rep_if.load = mlx5_ib_nic_rep_load;
-	rep_if.unload = mlx5_ib_nic_rep_unload;
-	rep_if.get_proto_dev = mlx5_ib_vport_get_proto_dev;
-	rep_if.priv = dev;
-
-	mlx5_eswitch_register_vport_rep(esw, 0, &rep_if, REP_IB);
-
-	mlx5_ib_rep_register_vf_vports(dev);
-}
-
-void mlx5_ib_unregister_vport_reps(struct mlx5_ib_dev *dev)
-{
-	struct mlx5_eswitch *esw   = dev->mdev->priv.eswitch;
-
-	mlx5_ib_rep_unregister_vf_vports(dev); /* VFs vports */
-	mlx5_eswitch_unregister_vport_rep(esw, 0, REP_IB); /* UPLINK PF*/
-}
-
 u8 mlx5_ib_eswitch_mode(struct mlx5_eswitch *esw)
 {
 	return mlx5_eswitch_mode(esw);
diff --git a/drivers/infiniband/hw/mlx5/ib_rep.h b/drivers/infiniband/hw/mlx5/ib_rep.h
index 2ba73636a2fb..798d41e61fb4 100644
--- a/drivers/infiniband/hw/mlx5/ib_rep.h
+++ b/drivers/infiniband/hw/mlx5/ib_rep.h
@@ -10,14 +10,16 @@
 #include "mlx5_ib.h"
 
 #ifdef CONFIG_MLX5_ESWITCH
+extern const struct mlx5_ib_profile uplink_rep_profile;
+
 u8 mlx5_ib_eswitch_mode(struct mlx5_eswitch *esw);
 struct mlx5_ib_dev *mlx5_ib_get_rep_ibdev(struct mlx5_eswitch *esw,
 					  int vport_index);
 struct mlx5_ib_dev *mlx5_ib_get_uplink_ibdev(struct mlx5_eswitch *esw);
 struct mlx5_eswitch_rep *mlx5_ib_vport_rep(struct mlx5_eswitch *esw,
 					   int vport_index);
-void mlx5_ib_register_vport_reps(struct mlx5_ib_dev *dev);
-void mlx5_ib_unregister_vport_reps(struct mlx5_ib_dev *dev);
+void mlx5_ib_register_vport_reps(struct mlx5_core_dev *mdev);
+void mlx5_ib_unregister_vport_reps(struct mlx5_core_dev *mdev);
 int create_flow_rule_vport_sq(struct mlx5_ib_dev *dev,
 			      struct mlx5_ib_sq *sq);
 struct net_device *mlx5_ib_get_rep_netdev(struct mlx5_eswitch *esw,
@@ -48,8 +50,8 @@ struct mlx5_eswitch_rep *mlx5_ib_vport_rep(struct mlx5_eswitch *esw,
 	return NULL;
 }
 
-static inline void mlx5_ib_register_vport_reps(struct mlx5_ib_dev *dev) {}
-static inline void mlx5_ib_unregister_vport_reps(struct mlx5_ib_dev *dev) {}
+static inline void mlx5_ib_register_vport_reps(struct mlx5_core_dev *mdev) {}
+static inline void mlx5_ib_unregister_vport_reps(struct mlx5_core_dev *mdev) {}
 static inline int create_flow_rule_vport_sq(struct mlx5_ib_dev *dev,
 					    struct mlx5_ib_sq *sq)
 {
diff --git a/drivers/infiniband/hw/mlx5/mad.c b/drivers/infiniband/hw/mlx5/mad.c
index 558638468edb..6c529e6f3a01 100644
--- a/drivers/infiniband/hw/mlx5/mad.c
+++ b/drivers/infiniband/hw/mlx5/mad.c
@@ -36,6 +36,7 @@
 #include <rdma/ib_smi.h>
 #include <rdma/ib_pma.h>
 #include "mlx5_ib.h"
+#include "cmd.h"
 
 enum {
 	MLX5_IB_VENDOR_CLASS1 = 0x9,
@@ -51,9 +52,10 @@ static bool can_do_mad_ifc(struct mlx5_ib_dev *dev, u8 port_num,
 	return dev->mdev->port_caps[port_num - 1].has_smi;
 }
 
-int mlx5_MAD_IFC(struct mlx5_ib_dev *dev, int ignore_mkey, int ignore_bkey,
-		 u8 port, const struct ib_wc *in_wc, const struct ib_grh *in_grh,
-		 const void *in_mad, void *response_mad)
+static int mlx5_MAD_IFC(struct mlx5_ib_dev *dev, int ignore_mkey,
+			int ignore_bkey, u8 port, const struct ib_wc *in_wc,
+			const struct ib_grh *in_grh, const void *in_mad,
+			void *response_mad)
 {
 	u8 op_modifier = 0;
 
@@ -68,7 +70,8 @@ int mlx5_MAD_IFC(struct mlx5_ib_dev *dev, int ignore_mkey, int ignore_bkey,
 	if (ignore_bkey || !in_wc)
 		op_modifier |= 0x2;
 
-	return mlx5_core_mad_ifc(dev->mdev, in_mad, response_mad, op_modifier, port);
+	return mlx5_cmd_mad_ifc(dev->mdev, in_mad, response_mad, op_modifier,
+				port);
 }
 
 static int process_mad(struct ib_device *ibdev, int mad_flags, u8 port_num,
diff --git a/drivers/infiniband/hw/mlx5/main.c b/drivers/infiniband/hw/mlx5/main.c
index 94fe253d4956..581ae11e2fc9 100644
--- a/drivers/infiniband/hw/mlx5/main.c
+++ b/drivers/infiniband/hw/mlx5/main.c
@@ -331,8 +331,8 @@ out:
 	spin_unlock(&port->mp.mpi_lock);
 }
 
-static int translate_eth_proto_oper(u32 eth_proto_oper, u8 *active_speed,
-				    u8 *active_width)
+static int translate_eth_legacy_proto_oper(u32 eth_proto_oper, u8 *active_speed,
+					   u8 *active_width)
 {
 	switch (eth_proto_oper) {
 	case MLX5E_PROT_MASK(MLX5E_1000BASE_CX_SGMII):
@@ -389,10 +389,66 @@ static int translate_eth_proto_oper(u32 eth_proto_oper, u8 *active_speed,
 	return 0;
 }
 
+static int translate_eth_ext_proto_oper(u32 eth_proto_oper, u8 *active_speed,
+					u8 *active_width)
+{
+	switch (eth_proto_oper) {
+	case MLX5E_PROT_MASK(MLX5E_SGMII_100M):
+	case MLX5E_PROT_MASK(MLX5E_1000BASE_X_SGMII):
+		*active_width = IB_WIDTH_1X;
+		*active_speed = IB_SPEED_SDR;
+		break;
+	case MLX5E_PROT_MASK(MLX5E_5GBASE_R):
+		*active_width = IB_WIDTH_1X;
+		*active_speed = IB_SPEED_DDR;
+		break;
+	case MLX5E_PROT_MASK(MLX5E_10GBASE_XFI_XAUI_1):
+		*active_width = IB_WIDTH_1X;
+		*active_speed = IB_SPEED_QDR;
+		break;
+	case MLX5E_PROT_MASK(MLX5E_40GBASE_XLAUI_4_XLPPI_4):
+		*active_width = IB_WIDTH_4X;
+		*active_speed = IB_SPEED_QDR;
+		break;
+	case MLX5E_PROT_MASK(MLX5E_25GAUI_1_25GBASE_CR_KR):
+		*active_width = IB_WIDTH_1X;
+		*active_speed = IB_SPEED_EDR;
+		break;
+	case MLX5E_PROT_MASK(MLX5E_50GAUI_2_LAUI_2_50GBASE_CR2_KR2):
+	case MLX5E_PROT_MASK(MLX5E_50GAUI_1_LAUI_1_50GBASE_CR_KR):
+		*active_width = IB_WIDTH_1X;
+		*active_speed = IB_SPEED_HDR;
+		break;
+	case MLX5E_PROT_MASK(MLX5E_100GAUI_2_100GBASE_CR2_KR2):
+		*active_width = IB_WIDTH_2X;
+		*active_speed = IB_SPEED_HDR;
+		break;
+	case MLX5E_PROT_MASK(MLX5E_200GAUI_4_200GBASE_CR4_KR4):
+		*active_width = IB_WIDTH_4X;
+		*active_speed = IB_SPEED_HDR;
+		break;
+	default:
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static int translate_eth_proto_oper(u32 eth_proto_oper, u8 *active_speed,
+				    u8 *active_width, bool ext)
+{
+	return ext ?
+		translate_eth_ext_proto_oper(eth_proto_oper, active_speed,
+					     active_width) :
+		translate_eth_legacy_proto_oper(eth_proto_oper, active_speed,
+						active_width);
+}
+
 static int mlx5_query_port_roce(struct ib_device *device, u8 port_num,
 				struct ib_port_attr *props)
 {
 	struct mlx5_ib_dev *dev = to_mdev(device);
+	u32 out[MLX5_ST_SZ_DW(ptys_reg)] = {0};
 	struct mlx5_core_dev *mdev;
 	struct net_device *ndev, *upper;
 	enum ib_mtu ndev_ib_mtu;
@@ -400,6 +456,7 @@ static int mlx5_query_port_roce(struct ib_device *device, u8 port_num,
 	u16 qkey_viol_cntr;
 	u32 eth_prot_oper;
 	u8 mdev_port_num;
+	bool ext;
 	int err;
 
 	mdev = mlx5_ib_get_native_port_mdev(dev, port_num, &mdev_port_num);
@@ -416,16 +473,18 @@ static int mlx5_query_port_roce(struct ib_device *device, u8 port_num,
 	/* Possible bad flows are checked before filling out props so in case
 	 * of an error it will still be zeroed out.
 	 */
-	err = mlx5_query_port_eth_proto_oper(mdev, &eth_prot_oper,
-					     mdev_port_num);
+	err = mlx5_query_port_ptys(mdev, out, sizeof(out), MLX5_PTYS_EN,
+				   mdev_port_num);
 	if (err)
 		goto out;
+	ext = MLX5_CAP_PCAM_FEATURE(dev->mdev, ptys_extended_ethernet);
+	eth_prot_oper = MLX5_GET_ETH_PROTO(ptys_reg, out, ext, eth_proto_oper);
 
 	props->active_width     = IB_WIDTH_4X;
 	props->active_speed     = IB_SPEED_QDR;
 
 	translate_eth_proto_oper(eth_prot_oper, &props->active_speed,
-				 &props->active_width);
+				 &props->active_width, ext);
 
 	props->port_cap_flags |= IB_PORT_CM_SUP;
 	props->ip_gids = true;
@@ -6386,7 +6445,7 @@ static const struct mlx5_ib_profile pf_profile = {
 		     mlx5_ib_stage_delay_drop_cleanup),
 };
 
-static const struct mlx5_ib_profile nic_rep_profile = {
+const struct mlx5_ib_profile uplink_rep_profile = {
 	STAGE_CREATE(MLX5_IB_STAGE_INIT,
 		     mlx5_ib_stage_init_init,
 		     mlx5_ib_stage_init_cleanup),
@@ -6479,6 +6538,12 @@ static void *mlx5_ib_add(struct mlx5_core_dev *mdev)
 
 	printk_once(KERN_INFO "%s", mlx5_version);
 
+	if (MLX5_ESWITCH_MANAGER(mdev) &&
+	    mlx5_ib_eswitch_mode(mdev->priv.eswitch) == SRIOV_OFFLOADS) {
+		mlx5_ib_register_vport_reps(mdev);
+		return mdev;
+	}
+
 	port_type_cap = MLX5_CAP_GEN(mdev, port_type);
 	ll = mlx5_port_type_cap_to_rdma_ll(port_type_cap);
 
@@ -6493,14 +6558,6 @@ static void *mlx5_ib_add(struct mlx5_core_dev *mdev)
 	dev->num_ports = max(MLX5_CAP_GEN(mdev, num_ports),
 			     MLX5_CAP_GEN(mdev, num_vhca_ports));
 
-	if (MLX5_ESWITCH_MANAGER(mdev) &&
-	    mlx5_ib_eswitch_mode(mdev->priv.eswitch) == SRIOV_OFFLOADS) {
-		dev->rep = mlx5_ib_vport_rep(mdev->priv.eswitch, 0);
-		dev->profile = &nic_rep_profile;
-		mlx5_ib_register_vport_reps(dev);
-		return dev;
-	}
-
 	return __mlx5_ib_add(dev, &pf_profile);
 }
 
@@ -6509,6 +6566,11 @@ static void mlx5_ib_remove(struct mlx5_core_dev *mdev, void *context)
 	struct mlx5_ib_multiport_info *mpi;
 	struct mlx5_ib_dev *dev;
 
+	if (MLX5_ESWITCH_MANAGER(mdev) && context == mdev) {
+		mlx5_ib_unregister_vport_reps(mdev);
+		return;
+	}
+
 	if (mlx5_core_is_mp_slave(mdev)) {
 		mpi = context;
 		mutex_lock(&mlx5_ib_multiport_mutex);
@@ -6520,10 +6582,7 @@ static void mlx5_ib_remove(struct mlx5_core_dev *mdev, void *context)
 	}
 
 	dev = context;
-	if (dev->profile == &nic_rep_profile)
-		mlx5_ib_unregister_vport_reps(dev);
-	else
-		__mlx5_ib_remove(dev, dev->profile, MLX5_IB_STAGE_MAX);
+	__mlx5_ib_remove(dev, dev->profile, MLX5_IB_STAGE_MAX);
 
 	ib_dealloc_device((struct ib_device *)dev);
 }
diff --git a/drivers/infiniband/hw/mlx5/mlx5_ib.h b/drivers/infiniband/hw/mlx5/mlx5_ib.h
index b06d3b1efea8..eedba0d2ec4b 100644
--- a/drivers/infiniband/hw/mlx5/mlx5_ib.h
+++ b/drivers/infiniband/hw/mlx5/mlx5_ib.h
@@ -587,6 +587,7 @@ struct mlx5_ib_mr {
 	struct mlx5_ib_mr      *parent;
 	atomic_t		num_leaf_free;
 	wait_queue_head_t       q_leaf_free;
+	struct mlx5_async_work  cb_work;
 };
 
 struct mlx5_ib_mw {
@@ -944,6 +945,7 @@ struct mlx5_ib_dev {
 	struct mlx5_memic	memic;
 	u16			devx_whitelist_uid;
 	struct mlx5_srq_table   srq_table;
+	struct mlx5_async_ctx   async_ctx;
 };
 
 static inline struct mlx5_ib_cq *to_mibcq(struct mlx5_core_cq *mcq)
@@ -1038,9 +1040,6 @@ void mlx5_ib_db_unmap_user(struct mlx5_ib_ucontext *context, struct mlx5_db *db)
 void __mlx5_ib_cq_clean(struct mlx5_ib_cq *cq, u32 qpn, struct mlx5_ib_srq *srq);
 void mlx5_ib_cq_clean(struct mlx5_ib_cq *cq, u32 qpn, struct mlx5_ib_srq *srq);
 void mlx5_ib_free_srq_wqe(struct mlx5_ib_srq *srq, int wqe_index);
-int mlx5_MAD_IFC(struct mlx5_ib_dev *dev, int ignore_mkey, int ignore_bkey,
-		 u8 port, const struct ib_wc *in_wc, const struct ib_grh *in_grh,
-		 const void *in_mad, void *response_mad);
 struct ib_ah *mlx5_ib_create_ah(struct ib_pd *pd, struct rdma_ah_attr *ah_attr,
 				u32 flags, struct ib_udata *udata);
 int mlx5_ib_query_ah(struct ib_ah *ibah, struct rdma_ah_attr *ah_attr);
diff --git a/drivers/infiniband/hw/mlx5/mr.c b/drivers/infiniband/hw/mlx5/mr.c
index fd6ea1f75085..bf2b6ea23851 100644
--- a/drivers/infiniband/hw/mlx5/mr.c
+++ b/drivers/infiniband/hw/mlx5/mr.c
@@ -123,9 +123,10 @@ static void update_odp_mr(struct mlx5_ib_mr *mr)
 }
 #endif
 
-static void reg_mr_callback(int status, void *context)
+static void reg_mr_callback(int status, struct mlx5_async_work *context)
 {
-	struct mlx5_ib_mr *mr = context;
+	struct mlx5_ib_mr *mr =
+		container_of(context, struct mlx5_ib_mr, cb_work);
 	struct mlx5_ib_dev *dev = mr->dev;
 	struct mlx5_mr_cache *cache = &dev->cache;
 	int c = order2idx(dev, mr->order);
@@ -216,9 +217,9 @@ static int add_keys(struct mlx5_ib_dev *dev, int c, int num)
 		ent->pending++;
 		spin_unlock_irq(&ent->lock);
 		err = mlx5_core_create_mkey_cb(dev->mdev, &mr->mmkey,
-					       in, inlen,
+					       &dev->async_ctx, in, inlen,
 					       mr->out, sizeof(mr->out),
-					       reg_mr_callback, mr);
+					       reg_mr_callback, &mr->cb_work);
 		if (err) {
 			spin_lock_irq(&ent->lock);
 			ent->pending--;
@@ -679,6 +680,7 @@ int mlx5_mr_cache_init(struct mlx5_ib_dev *dev)
 		return -ENOMEM;
 	}
 
+	mlx5_cmd_init_async_ctx(dev->mdev, &dev->async_ctx);
 	timer_setup(&dev->delay_timer, delay_time_func, 0);
 	for (i = 0; i < MAX_MR_CACHE_ENTRIES; i++) {
 		ent = &cache->ent[i];
@@ -725,33 +727,6 @@ int mlx5_mr_cache_init(struct mlx5_ib_dev *dev)
 	return 0;
 }
 
-static void wait_for_async_commands(struct mlx5_ib_dev *dev)
-{
-	struct mlx5_mr_cache *cache = &dev->cache;
-	struct mlx5_cache_ent *ent;
-	int total = 0;
-	int i;
-	int j;
-
-	for (i = 0; i < MAX_MR_CACHE_ENTRIES; i++) {
-		ent = &cache->ent[i];
-		for (j = 0 ; j < 1000; j++) {
-			if (!ent->pending)
-				break;
-			msleep(50);
-		}
-	}
-	for (i = 0; i < MAX_MR_CACHE_ENTRIES; i++) {
-		ent = &cache->ent[i];
-		total += ent->pending;
-	}
-
-	if (total)
-		mlx5_ib_warn(dev, "aborted while there are %d pending mr requests\n", total);
-	else
-		mlx5_ib_warn(dev, "done with all pending requests\n");
-}
-
 int mlx5_mr_cache_cleanup(struct mlx5_ib_dev *dev)
 {
 	int i;
@@ -763,12 +738,12 @@ int mlx5_mr_cache_cleanup(struct mlx5_ib_dev *dev)
 	flush_workqueue(dev->cache.wq);
 
 	mlx5_mr_cache_debugfs_cleanup(dev);
+	mlx5_cmd_cleanup_async_ctx(&dev->async_ctx);
 
 	for (i = 0; i < MAX_MR_CACHE_ENTRIES; i++)
 		clean_keys(dev, i);
 
 	destroy_workqueue(dev->cache.wq);
-	wait_for_async_commands(dev);
 	del_timer_sync(&dev->delay_timer);
 
 	return 0;