From 45bded2c216da6010184ac5ebe88c27f73439009 Mon Sep 17 00:00:00 2001
From: Kamal Heib <kamalh@mellanox.com>
Date: Wed, 18 Jan 2017 14:10:32 +0200
Subject: IB/mlx5: Verify that Q counters are supported

Make sure that the Q counters are supported by the FW before trying
to allocate/deallocte them, this will avoid driver load failure when
they aren't supported by the FW.

Fixes: 0837e86a7a34 ('IB/mlx5: Add per port counters')
Cc: <stable@vger.kernel.org> # v4.7+
Signed-off-by: Kamal Heib <kamalh@mellanox.com>
Reviewed-by: Mark Bloch <markb@mellanox.com>
Signed-off-by: Leon Romanovsky <leon@kernel.org>
Signed-off-by: Doug Ledford <dledford@redhat.com>
---
 drivers/infiniband/hw/mlx5/main.c | 14 +++++++++-----
 1 file changed, 9 insertions(+), 5 deletions(-)

(limited to 'drivers/infiniband/hw/mlx5/main.c')

diff --git a/drivers/infiniband/hw/mlx5/main.c b/drivers/infiniband/hw/mlx5/main.c
index a191b9327b0c..0187f1d7234a 100644
--- a/drivers/infiniband/hw/mlx5/main.c
+++ b/drivers/infiniband/hw/mlx5/main.c
@@ -3313,9 +3313,11 @@ static void *mlx5_ib_add(struct mlx5_core_dev *mdev)
 	if (err)
 		goto err_rsrc;
 
-	err = mlx5_ib_alloc_q_counters(dev);
-	if (err)
-		goto err_odp;
+	if (MLX5_CAP_GEN(dev->mdev, max_qp_cnt)) {
+		err = mlx5_ib_alloc_q_counters(dev);
+		if (err)
+			goto err_odp;
+	}
 
 	dev->mdev->priv.uar = mlx5_get_uars_page(dev->mdev);
 	if (!dev->mdev->priv.uar)
@@ -3364,7 +3366,8 @@ err_uar_page:
 	mlx5_put_uars_page(dev->mdev, dev->mdev->priv.uar);
 
 err_q_cnt:
-	mlx5_ib_dealloc_q_counters(dev);
+	if (MLX5_CAP_GEN(dev->mdev, max_qp_cnt))
+		mlx5_ib_dealloc_q_counters(dev);
 
 err_odp:
 	mlx5_ib_odp_remove_one(dev);
@@ -3397,7 +3400,8 @@ static void mlx5_ib_remove(struct mlx5_core_dev *mdev, void *context)
 	mlx5_free_bfreg(dev->mdev, &dev->fp_bfreg);
 	mlx5_free_bfreg(dev->mdev, &dev->bfreg);
 	mlx5_put_uars_page(dev->mdev, mdev->priv.uar);
-	mlx5_ib_dealloc_q_counters(dev);
+	if (MLX5_CAP_GEN(dev->mdev, max_qp_cnt))
+		mlx5_ib_dealloc_q_counters(dev);
 	destroy_umrc_res(dev);
 	mlx5_ib_odp_remove_one(dev);
 	destroy_dev_resources(&dev->devr);
-- 
cgit v1.2.3


From c43f1112c068f3b4b20a0a9d461c341d9caeb376 Mon Sep 17 00:00:00 2001
From: Maor Gottlieb <maorg@mellanox.com>
Date: Wed, 18 Jan 2017 14:10:33 +0200
Subject: IB/mlx5: Add additional checks before processing MADs

Check the has_smi bit in vport context and class version of MADs
before allowing MADs processing to take place.
MAD_IFC SMI commands can be executed only if smi bit is set.

Fixes: e126ba97dba9 ('mlx5: Add driver for Mellanox Connect-IB adapters')
Signed-off-by: Maor Gottlieb <maorg@mellanox.com>
Signed-off-by: Parvi Kaustubhi <parvik@mellanox.com>
Reviewed-by: Eli Cohen <eli@mellanox.com>
Signed-off-by: Leon Romanovsky <leon@kernel.org>
Signed-off-by: Doug Ledford <dledford@redhat.com>
---
 drivers/infiniband/hw/mlx5/mad.c  | 12 ++++++++++++
 drivers/infiniband/hw/mlx5/main.c | 33 +++++++++++++++++++++++++++++++++
 include/linux/mlx5/driver.h       |  1 +
 3 files changed, 46 insertions(+)

(limited to 'drivers/infiniband/hw/mlx5/main.c')

diff --git a/drivers/infiniband/hw/mlx5/mad.c b/drivers/infiniband/hw/mlx5/mad.c
index 39e58489dcc2..af962e7fdc3a 100644
--- a/drivers/infiniband/hw/mlx5/mad.c
+++ b/drivers/infiniband/hw/mlx5/mad.c
@@ -42,12 +42,24 @@ enum {
 	MLX5_IB_VENDOR_CLASS2 = 0xa
 };
 
+static bool can_do_mad_ifc(struct mlx5_ib_dev *dev, u8 port_num,
+			   struct ib_mad *in_mad)
+{
+	if (in_mad->mad_hdr.mgmt_class != IB_MGMT_CLASS_SUBN_LID_ROUTED &&
+	    in_mad->mad_hdr.mgmt_class != IB_MGMT_CLASS_SUBN_DIRECTED_ROUTE)
+		return true;
+	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)
 {
 	u8 op_modifier = 0;
 
+	if (!can_do_mad_ifc(dev, port, (struct ib_mad *)in_mad))
+		return -EPERM;
+
 	/* Key check traps can't be generated unless we have in_wc to
 	 * tell us where to send the trap.
 	 */
diff --git a/drivers/infiniband/hw/mlx5/main.c b/drivers/infiniband/hw/mlx5/main.c
index 0187f1d7234a..1dea4073d83f 100644
--- a/drivers/infiniband/hw/mlx5/main.c
+++ b/drivers/infiniband/hw/mlx5/main.c
@@ -2533,6 +2533,35 @@ static void mlx5_ib_event(struct mlx5_core_dev *dev, void *context,
 		ibdev->ib_active = false;
 }
 
+static int set_has_smi_cap(struct mlx5_ib_dev *dev)
+{
+	struct mlx5_hca_vport_context vport_ctx;
+	int err;
+	int port;
+
+	for (port = 1; port <= MLX5_CAP_GEN(dev->mdev, num_ports); port++) {
+		dev->mdev->port_caps[port - 1].has_smi = false;
+		if (MLX5_CAP_GEN(dev->mdev, port_type) ==
+		    MLX5_CAP_PORT_TYPE_IB) {
+			if (MLX5_CAP_GEN(dev->mdev, ib_virt)) {
+				err = mlx5_query_hca_vport_context(dev->mdev, 0,
+								   port, 0,
+								   &vport_ctx);
+				if (err) {
+					mlx5_ib_err(dev, "query_hca_vport_context for port=%d failed %d\n",
+						    port, err);
+					return err;
+				}
+				dev->mdev->port_caps[port - 1].has_smi =
+					vport_ctx.has_smi;
+			} else {
+				dev->mdev->port_caps[port - 1].has_smi = true;
+			}
+		}
+	}
+	return 0;
+}
+
 static void get_ext_port_caps(struct mlx5_ib_dev *dev)
 {
 	int port;
@@ -2557,6 +2586,10 @@ static int get_port_caps(struct mlx5_ib_dev *dev)
 	if (!dprops)
 		goto out;
 
+	err = set_has_smi_cap(dev);
+	if (err)
+		goto out;
+
 	err = mlx5_ib_query_device(&dev->ib_dev, dprops, &uhw);
 	if (err) {
 		mlx5_ib_warn(dev, "query_device failed %d\n", err);
diff --git a/include/linux/mlx5/driver.h b/include/linux/mlx5/driver.h
index 3a309f6a4a15..b8d69aeb1784 100644
--- a/include/linux/mlx5/driver.h
+++ b/include/linux/mlx5/driver.h
@@ -289,6 +289,7 @@ struct mlx5_port_caps {
 	int	gid_table_len;
 	int	pkey_table_len;
 	u8	ext_port_cap;
+	bool	has_smi;
 };
 
 struct mlx5_cmd_mailbox {
-- 
cgit v1.2.3


From ed88451e1f2d400fd6a743d0a481631cf9f97550 Mon Sep 17 00:00:00 2001
From: Majd Dibbiny <majd@mellanox.com>
Date: Wed, 18 Jan 2017 14:10:35 +0200
Subject: IB/mlx5: Assign DSCP for R-RoCE QPs Address Path

For Routable RoCE QPs, the DSCP should be set in the QP's
address path.

The DSCP's value is derived from the traffic class.

Fixes: 2811ba51b049 ("IB/mlx5: Add RoCE fields to Address Vector")
Cc: Achiad Shochat <achiad@mellanox.com>
Signed-off-by: Majd Dibbiny <majd@mellanox.com>
Reviewed-by: Moni Shoua <monis@mellanox.com>
Signed-off-by: Leon Romanovsky <leon@kernel.org>
Reviewed-by: Yuval Shaia <yuval.shaia@oracle.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
---
 drivers/infiniband/hw/mlx5/main.c    | 21 +++++++++++++++++++++
 drivers/infiniband/hw/mlx5/mlx5_ib.h |  2 ++
 drivers/infiniband/hw/mlx5/qp.c      |  7 +++++++
 3 files changed, 30 insertions(+)

(limited to 'drivers/infiniband/hw/mlx5/main.c')

diff --git a/drivers/infiniband/hw/mlx5/main.c b/drivers/infiniband/hw/mlx5/main.c
index 1dea4073d83f..6a81f0273f45 100644
--- a/drivers/infiniband/hw/mlx5/main.c
+++ b/drivers/infiniband/hw/mlx5/main.c
@@ -325,6 +325,27 @@ __be16 mlx5_get_roce_udp_sport(struct mlx5_ib_dev *dev, u8 port_num,
 	return cpu_to_be16(MLX5_CAP_ROCE(dev->mdev, r_roce_min_src_udp_port));
 }
 
+int mlx5_get_roce_gid_type(struct mlx5_ib_dev *dev, u8 port_num,
+			   int index, enum ib_gid_type *gid_type)
+{
+	struct ib_gid_attr attr;
+	union ib_gid gid;
+	int ret;
+
+	ret = ib_get_cached_gid(&dev->ib_dev, port_num, index, &gid, &attr);
+	if (ret)
+		return ret;
+
+	if (!attr.ndev)
+		return -ENODEV;
+
+	dev_put(attr.ndev);
+
+	*gid_type = attr.gid_type;
+
+	return 0;
+}
+
 static int mlx5_use_mad_ifc(struct mlx5_ib_dev *dev)
 {
 	if (MLX5_CAP_GEN(dev->mdev, port_type) == MLX5_CAP_PORT_TYPE_IB)
diff --git a/drivers/infiniband/hw/mlx5/mlx5_ib.h b/drivers/infiniband/hw/mlx5/mlx5_ib.h
index e1a4b93dce6b..dda01d7e8847 100644
--- a/drivers/infiniband/hw/mlx5/mlx5_ib.h
+++ b/drivers/infiniband/hw/mlx5/mlx5_ib.h
@@ -872,6 +872,8 @@ int mlx5_ib_set_vf_guid(struct ib_device *device, int vf, u8 port,
 
 __be16 mlx5_get_roce_udp_sport(struct mlx5_ib_dev *dev, u8 port_num,
 			       int index);
+int mlx5_get_roce_gid_type(struct mlx5_ib_dev *dev, u8 port_num,
+			   int index, enum ib_gid_type *gid_type);
 
 /* GSI QP helper functions */
 struct ib_qp *mlx5_ib_gsi_create_qp(struct ib_pd *pd,
diff --git a/drivers/infiniband/hw/mlx5/qp.c b/drivers/infiniband/hw/mlx5/qp.c
index 9af9b953ce30..e22d9572ae8f 100644
--- a/drivers/infiniband/hw/mlx5/qp.c
+++ b/drivers/infiniband/hw/mlx5/qp.c
@@ -2198,6 +2198,7 @@ static int mlx5_set_path(struct mlx5_ib_dev *dev, struct mlx5_ib_qp *qp,
 {
 	enum rdma_link_layer ll = rdma_port_get_link_layer(&dev->ib_dev, port);
 	int err;
+	enum ib_gid_type gid_type;
 
 	if (attr_mask & IB_QP_PKEY_INDEX)
 		path->pkey_index = cpu_to_be16(alt ? attr->alt_pkey_index :
@@ -2216,10 +2217,16 @@ static int mlx5_set_path(struct mlx5_ib_dev *dev, struct mlx5_ib_qp *qp,
 	if (ll == IB_LINK_LAYER_ETHERNET) {
 		if (!(ah->ah_flags & IB_AH_GRH))
 			return -EINVAL;
+		err = mlx5_get_roce_gid_type(dev, port, ah->grh.sgid_index,
+					     &gid_type);
+		if (err)
+			return err;
 		memcpy(path->rmac, ah->dmac, sizeof(ah->dmac));
 		path->udp_sport = mlx5_get_roce_udp_sport(dev, port,
 							  ah->grh.sgid_index);
 		path->dci_cfi_prio_sl = (ah->sl & 0x7) << 4;
+		if (gid_type == IB_GID_TYPE_ROCE_UDP_ENCAP)
+			path->ecn_dscp = (ah->grh.traffic_class >> 2) & 0x3f;
 	} else {
 		path->fl_free_ar = (path_flags & MLX5_PATH_FLAG_FL) ? 0x80 : 0;
 		path->fl_free_ar |=
-- 
cgit v1.2.3


From 5abb0da9cd438483d80517b1d5f0a62a2f818426 Mon Sep 17 00:00:00 2001
From: Leon Romanovsky <leonro@mellanox.com>
Date: Wed, 18 Jan 2017 14:10:36 +0200
Subject: IB/mlx5: Remove deprecated module parameter

Commit 9603b61de1ee ("mlx5: Move pci device handling from mlx5_ib
to mlx5_core") moved prof_sel module parameter from mlx5_ib to mlx5_core
and marked it as deprecated in 2014.

Three years after deprecation, it is time to remove the deprecated
module parameter.

Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
Reviewed-by: Jack Morgenstein <jackm@mellanox.com>
Signed-off-by: Leon Romanovsky <leon@kernel.org>
Reviewed-by: Yuval Shaia <yuval.shaia@oracle.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
---
 drivers/infiniband/hw/mlx5/main.c | 7 -------
 1 file changed, 7 deletions(-)

(limited to 'drivers/infiniband/hw/mlx5/main.c')

diff --git a/drivers/infiniband/hw/mlx5/main.c b/drivers/infiniband/hw/mlx5/main.c
index 6a81f0273f45..fc02f5f05b38 100644
--- a/drivers/infiniband/hw/mlx5/main.c
+++ b/drivers/infiniband/hw/mlx5/main.c
@@ -64,10 +64,6 @@ MODULE_DESCRIPTION("Mellanox Connect-IB HCA IB driver");
 MODULE_LICENSE("Dual BSD/GPL");
 MODULE_VERSION(DRIVER_VERSION);
 
-static int deprecated_prof_sel = 2;
-module_param_named(prof_sel, deprecated_prof_sel, int, 0444);
-MODULE_PARM_DESC(prof_sel, "profile selector. Deprecated here. Moved to module mlx5_core");
-
 static char mlx5_version[] =
 	DRIVER_NAME ": Mellanox Connect-IB Infiniband driver v"
 	DRIVER_VERSION " (" DRIVER_RELDATE ")\n";
@@ -3479,9 +3475,6 @@ static int __init mlx5_ib_init(void)
 {
 	int err;
 
-	if (deprecated_prof_sel != 2)
-		pr_warn("prof_sel is deprecated for mlx5_ib, set it for mlx5_core\n");
-
 	err = mlx5_register_interface(&mlx5_ib_interface);
 
 	return err;
-- 
cgit v1.2.3


From 2ac693f9953a2787f942abcbea5d1dc22a310932 Mon Sep 17 00:00:00 2001
From: Moses Reuben <mosesr@mellanox.com>
Date: Wed, 18 Jan 2017 14:59:50 +0200
Subject: IB/mlx5: Add flow tag support

Set flow tag in flow table entry, when IB_FLOW_SPEC_ACTION_TAG
is part of the flow specifications.

Flow tag doesn't support multicast flows, so it's passing to
hardware only when used.

Signed-off-by: Moses Reuben <mosesr@mellanox.com>
Signed-off-by: Leon Romanovsky <leon@kernel.org>
Signed-off-by: Doug Ledford <dledford@redhat.com>
---
 drivers/infiniband/hw/mlx5/main.c | 26 +++++++++++++++++++++++---
 1 file changed, 23 insertions(+), 3 deletions(-)

(limited to 'drivers/infiniband/hw/mlx5/main.c')

diff --git a/drivers/infiniband/hw/mlx5/main.c b/drivers/infiniband/hw/mlx5/main.c
index fc02f5f05b38..2e85f73c3235 100644
--- a/drivers/infiniband/hw/mlx5/main.c
+++ b/drivers/infiniband/hw/mlx5/main.c
@@ -1669,6 +1669,7 @@ static void set_tos(void *outer_c, void *outer_v, u8 mask, u8 val)
 #define LAST_IPV6_FIELD traffic_class
 #define LAST_TCP_UDP_FIELD src_port
 #define LAST_TUNNEL_FIELD tunnel_id
+#define LAST_FLOW_TAG_FIELD tag_id
 
 /* Field is the last supported field */
 #define FIELDS_NOT_SUPPORTED(filter, field)\
@@ -1679,7 +1680,7 @@ static void set_tos(void *outer_c, void *outer_v, u8 mask, u8 val)
 		   sizeof(filter.field))
 
 static int parse_flow_attr(u32 *match_c, u32 *match_v,
-			   const union ib_flow_spec *ib_spec)
+			   const union ib_flow_spec *ib_spec, u32 *tag_id)
 {
 	void *misc_params_c = MLX5_ADDR_OF(fte_match_param, match_c,
 					   misc_parameters);
@@ -1871,6 +1872,15 @@ static int parse_flow_attr(u32 *match_c, u32 *match_v,
 		MLX5_SET(fte_match_set_misc, misc_params_v, vxlan_vni,
 			 ntohl(ib_spec->tunnel.val.tunnel_id));
 		break;
+	case IB_FLOW_SPEC_ACTION_TAG:
+		if (FIELDS_NOT_SUPPORTED(ib_spec->flow_tag,
+					 LAST_FLOW_TAG_FIELD))
+			return -EOPNOTSUPP;
+		if (ib_spec->flow_tag.tag_id >= BIT(24))
+			return -EINVAL;
+
+		*tag_id = ib_spec->flow_tag.tag_id;
+		break;
 	default:
 		return -EINVAL;
 	}
@@ -2054,6 +2064,7 @@ static struct mlx5_ib_flow_handler *create_flow_rule(struct mlx5_ib_dev *dev,
 	struct mlx5_flow_spec *spec;
 	const void *ib_flow = (const void *)flow_attr + sizeof(*flow_attr);
 	unsigned int spec_index;
+	u32 flow_tag = MLX5_FS_DEFAULT_FLOW_TAG;
 	int err = 0;
 
 	if (!is_valid_attr(flow_attr))
@@ -2070,7 +2081,7 @@ static struct mlx5_ib_flow_handler *create_flow_rule(struct mlx5_ib_dev *dev,
 
 	for (spec_index = 0; spec_index < flow_attr->num_of_specs; spec_index++) {
 		err = parse_flow_attr(spec->match_criteria,
-				      spec->match_value, ib_flow);
+				      spec->match_value, ib_flow, &flow_tag);
 		if (err < 0)
 			goto free;
 
@@ -2080,7 +2091,16 @@ static struct mlx5_ib_flow_handler *create_flow_rule(struct mlx5_ib_dev *dev,
 	spec->match_criteria_enable = get_match_criteria_enable(spec->match_criteria);
 	flow_act.action = dst ? MLX5_FLOW_CONTEXT_ACTION_FWD_DEST :
 		MLX5_FLOW_CONTEXT_ACTION_FWD_NEXT_PRIO;
-	flow_act.flow_tag = MLX5_FS_DEFAULT_FLOW_TAG;
+
+	if (flow_tag != MLX5_FS_DEFAULT_FLOW_TAG &&
+	    (flow_attr->type == IB_FLOW_ATTR_ALL_DEFAULT ||
+	     flow_attr->type == IB_FLOW_ATTR_MC_DEFAULT)) {
+		mlx5_ib_warn(dev, "Flow tag %u and attribute type %x isn't allowed in leftovers\n",
+			     flow_tag, flow_attr->type);
+		err = -EINVAL;
+		goto free;
+	}
+	flow_act.flow_tag = flow_tag;
 	handler->rule = mlx5_add_flow_rules(ft, spec,
 					    &flow_act,
 					    dst, 1);
-- 
cgit v1.2.3


From 1ffd3a26f890222e49c4f4782763e2063026b611 Mon Sep 17 00:00:00 2001
From: Leon Romanovsky <leonro@mellanox.com>
Date: Wed, 18 Jan 2017 14:59:51 +0200
Subject: IB/mlx5: Replace ENOTSUPP usage with EOPNOTSUPP

Flow steering is supposed to return EOPNOTSUPP error
for unsupported fields and not ENOTSUPP error.

Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
Signed-off-by: Leon Romanovsky <leon@kernel.org>
Signed-off-by: Doug Ledford <dledford@redhat.com>
---
 drivers/infiniband/hw/mlx5/main.c | 12 ++++++------
 1 file changed, 6 insertions(+), 6 deletions(-)

(limited to 'drivers/infiniband/hw/mlx5/main.c')

diff --git a/drivers/infiniband/hw/mlx5/main.c b/drivers/infiniband/hw/mlx5/main.c
index 2e85f73c3235..0c12c1d6735d 100644
--- a/drivers/infiniband/hw/mlx5/main.c
+++ b/drivers/infiniband/hw/mlx5/main.c
@@ -1704,7 +1704,7 @@ static int parse_flow_attr(u32 *match_c, u32 *match_v,
 	switch (ib_spec->type & ~IB_FLOW_SPEC_INNER) {
 	case IB_FLOW_SPEC_ETH:
 		if (FIELDS_NOT_SUPPORTED(ib_spec->eth.mask, LAST_ETH_FIELD))
-			return -ENOTSUPP;
+			return -EOPNOTSUPP;
 
 		ether_addr_copy(MLX5_ADDR_OF(fte_match_set_lyr_2_4, headers_c,
 					     dmac_47_16),
@@ -1752,7 +1752,7 @@ static int parse_flow_attr(u32 *match_c, u32 *match_v,
 		break;
 	case IB_FLOW_SPEC_IPV4:
 		if (FIELDS_NOT_SUPPORTED(ib_spec->ipv4.mask, LAST_IPV4_FIELD))
-			return -ENOTSUPP;
+			return -EOPNOTSUPP;
 
 		MLX5_SET(fte_match_set_lyr_2_4, headers_c,
 			 ethertype, 0xffff);
@@ -1784,7 +1784,7 @@ static int parse_flow_attr(u32 *match_c, u32 *match_v,
 		break;
 	case IB_FLOW_SPEC_IPV6:
 		if (FIELDS_NOT_SUPPORTED(ib_spec->ipv6.mask, LAST_IPV6_FIELD))
-			return -ENOTSUPP;
+			return -EOPNOTSUPP;
 
 		MLX5_SET(fte_match_set_lyr_2_4, headers_c,
 			 ethertype, 0xffff);
@@ -1825,7 +1825,7 @@ static int parse_flow_attr(u32 *match_c, u32 *match_v,
 	case IB_FLOW_SPEC_TCP:
 		if (FIELDS_NOT_SUPPORTED(ib_spec->tcp_udp.mask,
 					 LAST_TCP_UDP_FIELD))
-			return -ENOTSUPP;
+			return -EOPNOTSUPP;
 
 		MLX5_SET(fte_match_set_lyr_2_4, headers_c, ip_protocol,
 			 0xff);
@@ -1845,7 +1845,7 @@ static int parse_flow_attr(u32 *match_c, u32 *match_v,
 	case IB_FLOW_SPEC_UDP:
 		if (FIELDS_NOT_SUPPORTED(ib_spec->tcp_udp.mask,
 					 LAST_TCP_UDP_FIELD))
-			return -ENOTSUPP;
+			return -EOPNOTSUPP;
 
 		MLX5_SET(fte_match_set_lyr_2_4, headers_c, ip_protocol,
 			 0xff);
@@ -1865,7 +1865,7 @@ static int parse_flow_attr(u32 *match_c, u32 *match_v,
 	case IB_FLOW_SPEC_VXLAN_TUNNEL:
 		if (FIELDS_NOT_SUPPORTED(ib_spec->tunnel.mask,
 					 LAST_TUNNEL_FIELD))
-			return -ENOTSUPP;
+			return -EOPNOTSUPP;
 
 		MLX5_SET(fte_match_set_misc, misc_params_c, vxlan_vni,
 			 ntohl(ib_spec->tunnel.mask.tunnel_id));
-- 
cgit v1.2.3


From 7c16f47779498650e9f11a395f8d63accedf35a3 Mon Sep 17 00:00:00 2001
From: Kamal Heib <kamalh@mellanox.com>
Date: Wed, 18 Jan 2017 15:25:09 +0200
Subject: IB/mlx5: Expose Q counters groups only if they are supported by FW

This patch modify the Q counters implementation, so each one of the
three Q counters groups will be exposed by the driver only if they are
supported by the firmware.

Signed-off-by: Kamal Heib <kamalh@mellanox.com>
Reviewed-by: Mark Bloch <markb@mellanox.com>
Signed-off-by: Leon Romanovsky <leon@kernel.org>
Signed-off-by: Doug Ledford <dledford@redhat.com>
---
 drivers/infiniband/hw/mlx5/main.c    | 158 ++++++++++++++++++++++++++---------
 drivers/infiniband/hw/mlx5/mlx5_ib.h |   9 +-
 drivers/infiniband/hw/mlx5/qp.c      |   4 +-
 3 files changed, 128 insertions(+), 43 deletions(-)

(limited to 'drivers/infiniband/hw/mlx5/main.c')

diff --git a/drivers/infiniband/hw/mlx5/main.c b/drivers/infiniband/hw/mlx5/main.c
index 0c12c1d6735d..f8fe98d22965 100644
--- a/drivers/infiniband/hw/mlx5/main.c
+++ b/drivers/infiniband/hw/mlx5/main.c
@@ -3072,13 +3072,102 @@ static void mlx5_disable_eth(struct mlx5_ib_dev *dev)
 		mlx5_nic_vport_disable_roce(dev->mdev);
 }
 
+struct mlx5_ib_q_counter {
+	const char *name;
+	size_t offset;
+};
+
+#define INIT_Q_COUNTER(_name)		\
+	{ .name = #_name, .offset = MLX5_BYTE_OFF(query_q_counter_out, _name)}
+
+static const struct mlx5_ib_q_counter basic_q_cnts[] = {
+	INIT_Q_COUNTER(rx_write_requests),
+	INIT_Q_COUNTER(rx_read_requests),
+	INIT_Q_COUNTER(rx_atomic_requests),
+	INIT_Q_COUNTER(out_of_buffer),
+};
+
+static const struct mlx5_ib_q_counter out_of_seq_q_cnts[] = {
+	INIT_Q_COUNTER(out_of_sequence),
+};
+
+static const struct mlx5_ib_q_counter retrans_q_cnts[] = {
+	INIT_Q_COUNTER(duplicate_request),
+	INIT_Q_COUNTER(rnr_nak_retry_err),
+	INIT_Q_COUNTER(packet_seq_err),
+	INIT_Q_COUNTER(implied_nak_seq_err),
+	INIT_Q_COUNTER(local_ack_timeout_err),
+};
+
 static void mlx5_ib_dealloc_q_counters(struct mlx5_ib_dev *dev)
 {
 	unsigned int i;
 
-	for (i = 0; i < dev->num_ports; i++)
+	for (i = 0; i < dev->num_ports; i++) {
 		mlx5_core_dealloc_q_counter(dev->mdev,
-					    dev->port[i].q_cnt_id);
+					    dev->port[i].q_cnts.set_id);
+		kfree(dev->port[i].q_cnts.names);
+		kfree(dev->port[i].q_cnts.offsets);
+	}
+}
+
+static int __mlx5_ib_alloc_q_counters(struct mlx5_ib_dev *dev,
+				      const char ***names,
+				      size_t **offsets,
+				      u32 *num)
+{
+	u32 num_counters;
+
+	num_counters = ARRAY_SIZE(basic_q_cnts);
+
+	if (MLX5_CAP_GEN(dev->mdev, out_of_seq_cnt))
+		num_counters += ARRAY_SIZE(out_of_seq_q_cnts);
+
+	if (MLX5_CAP_GEN(dev->mdev, retransmission_q_counters))
+		num_counters += ARRAY_SIZE(retrans_q_cnts);
+
+	*names = kcalloc(num_counters, sizeof(**names), GFP_KERNEL);
+	if (!*names)
+		return -ENOMEM;
+
+	*offsets = kcalloc(num_counters, sizeof(**offsets), GFP_KERNEL);
+	if (!*offsets)
+		goto err_names;
+
+	*num = num_counters;
+
+	return 0;
+
+err_names:
+	kfree(*names);
+	return -ENOMEM;
+}
+
+static void mlx5_ib_fill_q_counters(struct mlx5_ib_dev *dev,
+				    const char **names,
+				    size_t *offsets)
+{
+	int i;
+	int j = 0;
+
+	for (i = 0; i < ARRAY_SIZE(basic_q_cnts); i++, j++) {
+		names[j] = basic_q_cnts[i].name;
+		offsets[j] = basic_q_cnts[i].offset;
+	}
+
+	if (MLX5_CAP_GEN(dev->mdev, out_of_seq_cnt)) {
+		for (i = 0; i < ARRAY_SIZE(out_of_seq_q_cnts); i++, j++) {
+			names[j] = out_of_seq_q_cnts[i].name;
+			offsets[j] = out_of_seq_q_cnts[i].offset;
+		}
+	}
+
+	if (MLX5_CAP_GEN(dev->mdev, retransmission_q_counters)) {
+		for (i = 0; i < ARRAY_SIZE(retrans_q_cnts); i++, j++) {
+			names[j] = retrans_q_cnts[i].name;
+			offsets[j] = retrans_q_cnts[i].offset;
+		}
+	}
 }
 
 static int mlx5_ib_alloc_q_counters(struct mlx5_ib_dev *dev)
@@ -3087,14 +3176,26 @@ static int mlx5_ib_alloc_q_counters(struct mlx5_ib_dev *dev)
 	int ret;
 
 	for (i = 0; i < dev->num_ports; i++) {
+		struct mlx5_ib_port *port = &dev->port[i];
+
 		ret = mlx5_core_alloc_q_counter(dev->mdev,
-						&dev->port[i].q_cnt_id);
+						&port->q_cnts.set_id);
 		if (ret) {
 			mlx5_ib_warn(dev,
 				     "couldn't allocate queue counter for port %d, err %d\n",
 				     i + 1, ret);
 			goto dealloc_counters;
 		}
+
+		ret = __mlx5_ib_alloc_q_counters(dev,
+						 &port->q_cnts.names,
+						 &port->q_cnts.offsets,
+						 &port->q_cnts.num_counters);
+		if (ret)
+			goto dealloc_counters;
+
+		mlx5_ib_fill_q_counters(dev, port->q_cnts.names,
+					port->q_cnts.offsets);
 	}
 
 	return 0;
@@ -3102,62 +3203,39 @@ static int mlx5_ib_alloc_q_counters(struct mlx5_ib_dev *dev)
 dealloc_counters:
 	while (--i >= 0)
 		mlx5_core_dealloc_q_counter(dev->mdev,
-					    dev->port[i].q_cnt_id);
+					    dev->port[i].q_cnts.set_id);
 
 	return ret;
 }
 
-static const char * const names[] = {
-	"rx_write_requests",
-	"rx_read_requests",
-	"rx_atomic_requests",
-	"out_of_buffer",
-	"out_of_sequence",
-	"duplicate_request",
-	"rnr_nak_retry_err",
-	"packet_seq_err",
-	"implied_nak_seq_err",
-	"local_ack_timeout_err",
-};
-
-static const size_t stats_offsets[] = {
-	MLX5_BYTE_OFF(query_q_counter_out, rx_write_requests),
-	MLX5_BYTE_OFF(query_q_counter_out, rx_read_requests),
-	MLX5_BYTE_OFF(query_q_counter_out, rx_atomic_requests),
-	MLX5_BYTE_OFF(query_q_counter_out, out_of_buffer),
-	MLX5_BYTE_OFF(query_q_counter_out, out_of_sequence),
-	MLX5_BYTE_OFF(query_q_counter_out, duplicate_request),
-	MLX5_BYTE_OFF(query_q_counter_out, rnr_nak_retry_err),
-	MLX5_BYTE_OFF(query_q_counter_out, packet_seq_err),
-	MLX5_BYTE_OFF(query_q_counter_out, implied_nak_seq_err),
-	MLX5_BYTE_OFF(query_q_counter_out, local_ack_timeout_err),
-};
-
 static struct rdma_hw_stats *mlx5_ib_alloc_hw_stats(struct ib_device *ibdev,
 						    u8 port_num)
 {
-	BUILD_BUG_ON(ARRAY_SIZE(names) != ARRAY_SIZE(stats_offsets));
+	struct mlx5_ib_dev *dev = to_mdev(ibdev);
+	struct mlx5_ib_port *port = &dev->port[port_num - 1];
 
 	/* We support only per port stats */
 	if (port_num == 0)
 		return NULL;
 
-	return rdma_alloc_hw_stats_struct(names, ARRAY_SIZE(names),
+	return rdma_alloc_hw_stats_struct(port->q_cnts.names,
+					  port->q_cnts.num_counters,
 					  RDMA_HW_STATS_DEFAULT_LIFESPAN);
 }
 
 static int mlx5_ib_get_hw_stats(struct ib_device *ibdev,
 				struct rdma_hw_stats *stats,
-				u8 port, int index)
+				u8 port_num, int index)
 {
 	struct mlx5_ib_dev *dev = to_mdev(ibdev);
+	struct mlx5_ib_port *port = &dev->port[port_num - 1];
 	int outlen = MLX5_ST_SZ_BYTES(query_q_counter_out);
 	void *out;
 	__be32 val;
 	int ret;
 	int i;
 
-	if (!port || !stats)
+	if (!stats)
 		return -ENOSYS;
 
 	out = mlx5_vzalloc(outlen);
@@ -3165,18 +3243,19 @@ static int mlx5_ib_get_hw_stats(struct ib_device *ibdev,
 		return -ENOMEM;
 
 	ret = mlx5_core_query_q_counter(dev->mdev,
-					dev->port[port - 1].q_cnt_id, 0,
+					port->q_cnts.set_id, 0,
 					out, outlen);
 	if (ret)
 		goto free;
 
-	for (i = 0; i < ARRAY_SIZE(names); i++) {
-		val = *(__be32 *)(out + stats_offsets[i]);
+	for (i = 0; i < port->q_cnts.num_counters; i++) {
+		val = *(__be32 *)(out + port->q_cnts.offsets[i]);
 		stats->value[i] = (u64)be32_to_cpu(val);
 	}
+
 free:
 	kvfree(out);
-	return ARRAY_SIZE(names);
+	return port->q_cnts.num_counters;
 }
 
 static void *mlx5_ib_add(struct mlx5_core_dev *mdev)
@@ -3328,8 +3407,7 @@ static void *mlx5_ib_add(struct mlx5_core_dev *mdev)
 			(1ull << IB_USER_VERBS_CMD_DEALLOC_MW);
 	}
 
-	if (MLX5_CAP_GEN(dev->mdev, out_of_seq_cnt) &&
-	    MLX5_CAP_GEN(dev->mdev, retransmission_q_counters)) {
+	if (MLX5_CAP_GEN(dev->mdev, max_qp_cnt)) {
 		dev->ib_dev.get_hw_stats	= mlx5_ib_get_hw_stats;
 		dev->ib_dev.alloc_hw_stats	= mlx5_ib_alloc_hw_stats;
 	}
diff --git a/drivers/infiniband/hw/mlx5/mlx5_ib.h b/drivers/infiniband/hw/mlx5/mlx5_ib.h
index dda01d7e8847..66090835de27 100644
--- a/drivers/infiniband/hw/mlx5/mlx5_ib.h
+++ b/drivers/infiniband/hw/mlx5/mlx5_ib.h
@@ -579,8 +579,15 @@ struct mlx5_ib_resources {
 	struct mutex	mutex;
 };
 
+struct mlx5_ib_q_counters {
+	const char **names;
+	size_t *offsets;
+	u32 num_counters;
+	u16 set_id;
+};
+
 struct mlx5_ib_port {
-	u16 q_cnt_id;
+	struct mlx5_ib_q_counters q_cnts;
 };
 
 struct mlx5_roce {
diff --git a/drivers/infiniband/hw/mlx5/qp.c b/drivers/infiniband/hw/mlx5/qp.c
index e22d9572ae8f..5c7d655655bb 100644
--- a/drivers/infiniband/hw/mlx5/qp.c
+++ b/drivers/infiniband/hw/mlx5/qp.c
@@ -2784,7 +2784,7 @@ static int __mlx5_ib_modify_qp(struct ib_qp *ibqp,
 			       qp->port) - 1;
 		mibport = &dev->port[port_num];
 		context->qp_counter_set_usr_page |=
-			cpu_to_be32((u32)(mibport->q_cnt_id) << 24);
+			cpu_to_be32((u32)(mibport->q_cnts.set_id) << 24);
 	}
 
 	if (!ibqp->uobject && cur_state == IB_QPS_RESET && new_state == IB_QPS_INIT)
@@ -2812,7 +2812,7 @@ static int __mlx5_ib_modify_qp(struct ib_qp *ibqp,
 
 		raw_qp_param.operation = op;
 		if (cur_state == IB_QPS_RESET && new_state == IB_QPS_INIT) {
-			raw_qp_param.rq_q_ctr_id = mibport->q_cnt_id;
+			raw_qp_param.rq_q_ctr_id = mibport->q_cnts.set_id;
 			raw_qp_param.set_mask |= MLX5_RAW_QP_MOD_SET_RQ_Q_CTR_ID;
 		}
 
-- 
cgit v1.2.3


From e8161334403ed34a9214bcce517d50bbe3e02b97 Mon Sep 17 00:00:00 2001
From: Noa Osherovich <noaos@mellanox.com>
Date: Wed, 18 Jan 2017 15:40:01 +0200
Subject: IB/mlx5: Expose vlan offloads capabilities

Check device's capabilities and report which raw packet capabilities
are supported.

Signed-off-by: Noa Osherovich <noaos@mellanox.com>
Reviewed-by: Maor Gottlieb <maorg@mellanox.com>
Reviewed-by: Yishai Hadas <yishaih@mellanox.com>
Signed-off-by: Leon Romanovsky <leon@kernel.org>
Signed-off-by: Doug Ledford <dledford@redhat.com>
---
 drivers/infiniband/hw/mlx5/main.c | 14 ++++++++++++--
 1 file changed, 12 insertions(+), 2 deletions(-)

(limited to 'drivers/infiniband/hw/mlx5/main.c')

diff --git a/drivers/infiniband/hw/mlx5/main.c b/drivers/infiniband/hw/mlx5/main.c
index f8fe98d22965..fe37da2be26f 100644
--- a/drivers/infiniband/hw/mlx5/main.c
+++ b/drivers/infiniband/hw/mlx5/main.c
@@ -581,8 +581,15 @@ static int mlx5_ib_query_device(struct ib_device *ibdev,
 		props->device_cap_flags |= IB_DEVICE_BLOCK_MULTICAST_LOOPBACK;
 
 	if (MLX5_CAP_GEN(dev->mdev, eth_net_offloads)) {
-		if (MLX5_CAP_ETH(mdev, csum_cap))
+		if (MLX5_CAP_ETH(mdev, csum_cap)) {
+			/* Legacy bit to support old userspace libraries */
 			props->device_cap_flags |= IB_DEVICE_RAW_IP_CSUM;
+			props->raw_packet_caps |= IB_RAW_PACKET_CAP_IP_CSUM;
+		}
+
+		if (MLX5_CAP_ETH(dev->mdev, vlan_cap))
+			props->raw_packet_caps |=
+				IB_RAW_PACKET_CAP_CVLAN_STRIPPING;
 
 		if (field_avail(typeof(resp), tso_caps, uhw->outlen)) {
 			max_tso = MLX5_CAP_ETH(mdev, max_lso_cap);
@@ -621,8 +628,11 @@ static int mlx5_ib_query_device(struct ib_device *ibdev,
 	}
 
 	if (MLX5_CAP_GEN(dev->mdev, eth_net_offloads) &&
-	    MLX5_CAP_ETH(dev->mdev, scatter_fcs))
+	    MLX5_CAP_ETH(dev->mdev, scatter_fcs)) {
+		/* Legacy bit to support old userspace libraries */
 		props->device_cap_flags |= IB_DEVICE_RAW_SCATTER_FCS;
+		props->raw_packet_caps |= IB_RAW_PACKET_CAP_SCATTER_FCS;
+	}
 
 	if (mlx5_get_flow_namespace(dev->mdev, MLX5_FLOW_NAMESPACE_BYPASS))
 		props->device_cap_flags |= IB_DEVICE_MANAGED_FLOW_STEERING;
-- 
cgit v1.2.3


From 81713d3788d2e6bc005f15ee1c59d0eb06050a6b Mon Sep 17 00:00:00 2001
From: Artemy Kovalyov <artemyko@mellanox.com>
Date: Wed, 18 Jan 2017 16:58:11 +0200
Subject: IB/mlx5: Add implicit MR support

Add implicit MR, covering entire user address space.
The MR is implemented as an indirect KSM MR consisting of
1GB direct MRs.
Pages and direct MRs are added/removed to MR by ODP.

Signed-off-by: Artemy Kovalyov <artemyko@mellanox.com>
Signed-off-by: Leon Romanovsky <leon@kernel.org>
Signed-off-by: Doug Ledford <dledford@redhat.com>
---
 drivers/infiniband/hw/mlx5/main.c    |   2 +
 drivers/infiniband/hw/mlx5/mlx5_ib.h |  20 +-
 drivers/infiniband/hw/mlx5/mr.c      |  33 ++-
 drivers/infiniband/hw/mlx5/odp.c     | 505 ++++++++++++++++++++++++++++++++---
 include/linux/mlx5/driver.h          |   2 +
 5 files changed, 513 insertions(+), 49 deletions(-)

(limited to 'drivers/infiniband/hw/mlx5/main.c')

diff --git a/drivers/infiniband/hw/mlx5/main.c b/drivers/infiniband/hw/mlx5/main.c
index fe37da2be26f..eb8719ca500e 100644
--- a/drivers/infiniband/hw/mlx5/main.c
+++ b/drivers/infiniband/hw/mlx5/main.c
@@ -3583,6 +3583,8 @@ static int __init mlx5_ib_init(void)
 {
 	int err;
 
+	mlx5_ib_odp_init();
+
 	err = mlx5_register_interface(&mlx5_ib_interface);
 
 	return err;
diff --git a/drivers/infiniband/hw/mlx5/mlx5_ib.h b/drivers/infiniband/hw/mlx5/mlx5_ib.h
index efc44de3c7d7..3cd064b5f0bf 100644
--- a/drivers/infiniband/hw/mlx5/mlx5_ib.h
+++ b/drivers/infiniband/hw/mlx5/mlx5_ib.h
@@ -202,6 +202,7 @@ struct mlx5_ib_flow_db {
 #define MLX5_IB_UPD_XLT_ADDR	      BIT(3)
 #define MLX5_IB_UPD_XLT_PD	      BIT(4)
 #define MLX5_IB_UPD_XLT_ACCESS	      BIT(5)
+#define MLX5_IB_UPD_XLT_INDIRECT      BIT(6)
 
 /* Private QP creation flags to be passed in ib_qp_init_attr.create_flags.
  *
@@ -503,6 +504,10 @@ struct mlx5_ib_mr {
 	int			live;
 	void			*descs_alloc;
 	int			access_flags; /* Needed for rereg MR */
+
+	struct mlx5_ib_mr      *parent;
+	atomic_t		num_leaf_free;
+	wait_queue_head_t       q_leaf_free;
 };
 
 struct mlx5_ib_mw {
@@ -637,6 +642,7 @@ struct mlx5_ib_dev {
 	 * being used by a page fault handler.
 	 */
 	struct srcu_struct      mr_srcu;
+	u32			null_mkey;
 #endif
 	struct mlx5_ib_flow_db	flow_db;
 	/* protect resources needed as part of reset flow */
@@ -789,6 +795,9 @@ struct ib_mw *mlx5_ib_alloc_mw(struct ib_pd *pd, enum ib_mw_type type,
 int mlx5_ib_dealloc_mw(struct ib_mw *mw);
 int mlx5_ib_update_xlt(struct mlx5_ib_mr *mr, u64 idx, int npages,
 		       int page_shift, int flags);
+struct mlx5_ib_mr *mlx5_ib_alloc_implicit_mr(struct mlx5_ib_pd *pd,
+					     int access_flags);
+void mlx5_ib_free_implicit_mr(struct mlx5_ib_mr *mr);
 int mlx5_ib_rereg_user_mr(struct ib_mr *ib_mr, int flags, u64 start,
 			  u64 length, u64 virt_addr, int access_flags,
 			  struct ib_pd *pd, struct ib_udata *udata);
@@ -868,6 +877,9 @@ int __init mlx5_ib_odp_init(void);
 void mlx5_ib_odp_cleanup(void);
 void mlx5_ib_invalidate_range(struct ib_umem *umem, unsigned long start,
 			      unsigned long end);
+void mlx5_odp_init_mr_cache_entry(struct mlx5_cache_ent *ent);
+void mlx5_odp_populate_klm(struct mlx5_klm *pklm, size_t offset,
+			   size_t nentries, struct mlx5_ib_mr *mr, int flags);
 #else /* CONFIG_INFINIBAND_ON_DEMAND_PAGING */
 static inline void mlx5_ib_internal_fill_odp_caps(struct mlx5_ib_dev *dev)
 {
@@ -875,9 +887,13 @@ static inline void mlx5_ib_internal_fill_odp_caps(struct mlx5_ib_dev *dev)
 }
 
 static inline int mlx5_ib_odp_init_one(struct mlx5_ib_dev *ibdev) { return 0; }
-static inline void mlx5_ib_odp_remove_one(struct mlx5_ib_dev *ibdev)	{}
+static inline void mlx5_ib_odp_remove_one(struct mlx5_ib_dev *ibdev)	    {}
 static inline int mlx5_ib_odp_init(void) { return 0; }
-static inline void mlx5_ib_odp_cleanup(void)				{}
+static inline void mlx5_ib_odp_cleanup(void)				    {}
+static inline void mlx5_odp_init_mr_cache_entry(struct mlx5_cache_ent *ent) {}
+static inline void mlx5_odp_populate_klm(struct mlx5_klm *pklm, size_t offset,
+					 size_t nentries, struct mlx5_ib_mr *mr,
+					 int flags) {}
 
 #endif /* CONFIG_INFINIBAND_ON_DEMAND_PAGING */
 
diff --git a/drivers/infiniband/hw/mlx5/mr.c b/drivers/infiniband/hw/mlx5/mr.c
index 8f5b94d483e4..3c1f483d003f 100644
--- a/drivers/infiniband/hw/mlx5/mr.c
+++ b/drivers/infiniband/hw/mlx5/mr.c
@@ -469,7 +469,7 @@ struct mlx5_ib_mr *mlx5_mr_cache_alloc(struct mlx5_ib_dev *dev, int entry)
 			spin_unlock_irq(&ent->lock);
 
 			err = add_keys(dev, entry, 1);
-			if (err)
+			if (err && err != -EAGAIN)
 				return ERR_PTR(err);
 
 			wait_for_completion(&ent->compl);
@@ -669,8 +669,10 @@ int mlx5_mr_cache_init(struct mlx5_ib_dev *dev)
 		INIT_DELAYED_WORK(&ent->dwork, delayed_cache_work_func);
 		queue_work(cache->wq, &ent->work);
 
-		if (i > MAX_UMR_CACHE_ENTRY)
+		if (i > MAX_UMR_CACHE_ENTRY) {
+			mlx5_odp_init_mr_cache_entry(ent);
 			continue;
+		}
 
 		if (!use_umr(dev, ent->order))
 			continue;
@@ -935,6 +937,10 @@ static inline int populate_xlt(struct mlx5_ib_mr *mr, int idx, int npages,
 {
 	struct mlx5_ib_dev *dev = mr->dev;
 	struct ib_umem *umem = mr->umem;
+	if (flags & MLX5_IB_UPD_XLT_INDIRECT) {
+		mlx5_odp_populate_klm(xlt, idx, npages, mr, flags);
+		return npages;
+	}
 
 	npages = min_t(size_t, npages, ib_umem_num_pages(umem) - idx);
 
@@ -968,7 +974,9 @@ int mlx5_ib_update_xlt(struct mlx5_ib_mr *mr, u64 idx, int npages,
 	struct mlx5_umr_wr wr;
 	struct ib_sge sg;
 	int err = 0;
-	int desc_size = sizeof(struct mlx5_mtt);
+	int desc_size = (flags & MLX5_IB_UPD_XLT_INDIRECT)
+			       ? sizeof(struct mlx5_klm)
+			       : sizeof(struct mlx5_mtt);
 	const int page_align = MLX5_UMR_MTT_ALIGNMENT / desc_size;
 	const int page_mask = page_align - 1;
 	size_t pages_mapped = 0;
@@ -1186,6 +1194,18 @@ struct ib_mr *mlx5_ib_reg_user_mr(struct ib_pd *pd, u64 start, u64 length,
 
 	mlx5_ib_dbg(dev, "start 0x%llx, virt_addr 0x%llx, length 0x%llx, access_flags 0x%x\n",
 		    start, virt_addr, length, access_flags);
+
+#ifdef CONFIG_INFINIBAND_ON_DEMAND_PAGING
+	if (!start && length == U64_MAX) {
+		if (!(access_flags & IB_ACCESS_ON_DEMAND) ||
+		    !(dev->odp_caps.general_caps & IB_ODP_SUPPORT_IMPLICIT))
+			return ERR_PTR(-EINVAL);
+
+		mr = mlx5_ib_alloc_implicit_mr(to_mpd(pd), access_flags);
+		return &mr->ibmr;
+	}
+#endif
+
 	err = mr_umem_get(pd, start, length, access_flags, &umem, &npages,
 			   &page_shift, &ncont, &order);
 
@@ -1471,8 +1491,11 @@ int mlx5_ib_dereg_mr(struct ib_mr *ibmr)
 		/* Wait for all running page-fault handlers to finish. */
 		synchronize_srcu(&dev->mr_srcu);
 		/* Destroy all page mappings */
-		mlx5_ib_invalidate_range(umem, ib_umem_start(umem),
-					 ib_umem_end(umem));
+		if (umem->odp_data->page_list)
+			mlx5_ib_invalidate_range(umem, ib_umem_start(umem),
+						 ib_umem_end(umem));
+		else
+			mlx5_ib_free_implicit_mr(mr);
 		/*
 		 * We kill the umem before the MR for ODP,
 		 * so that there will not be any invalidations in
diff --git a/drivers/infiniband/hw/mlx5/odp.c b/drivers/infiniband/hw/mlx5/odp.c
index e5bc267aca73..d7b12f0750e2 100644
--- a/drivers/infiniband/hw/mlx5/odp.c
+++ b/drivers/infiniband/hw/mlx5/odp.c
@@ -34,6 +34,7 @@
 #include <rdma/ib_umem_odp.h>
 
 #include "mlx5_ib.h"
+#include "cmd.h"
 
 #define MAX_PREFETCH_LEN (4*1024*1024U)
 
@@ -41,6 +42,140 @@
  * a pagefault. */
 #define MMU_NOTIFIER_TIMEOUT 1000
 
+#define MLX5_IMR_MTT_BITS (30 - PAGE_SHIFT)
+#define MLX5_IMR_MTT_SHIFT (MLX5_IMR_MTT_BITS + PAGE_SHIFT)
+#define MLX5_IMR_MTT_ENTRIES BIT_ULL(MLX5_IMR_MTT_BITS)
+#define MLX5_IMR_MTT_SIZE BIT_ULL(MLX5_IMR_MTT_SHIFT)
+#define MLX5_IMR_MTT_MASK (~(MLX5_IMR_MTT_SIZE - 1))
+
+#define MLX5_KSM_PAGE_SHIFT MLX5_IMR_MTT_SHIFT
+
+static u64 mlx5_imr_ksm_entries;
+
+static int check_parent(struct ib_umem_odp *odp,
+			       struct mlx5_ib_mr *parent)
+{
+	struct mlx5_ib_mr *mr = odp->private;
+
+	return mr && mr->parent == parent;
+}
+
+static struct ib_umem_odp *odp_next(struct ib_umem_odp *odp)
+{
+	struct mlx5_ib_mr *mr = odp->private, *parent = mr->parent;
+	struct ib_ucontext *ctx = odp->umem->context;
+	struct rb_node *rb;
+
+	down_read(&ctx->umem_rwsem);
+	while (1) {
+		rb = rb_next(&odp->interval_tree.rb);
+		if (!rb)
+			goto not_found;
+		odp = rb_entry(rb, struct ib_umem_odp, interval_tree.rb);
+		if (check_parent(odp, parent))
+			goto end;
+	}
+not_found:
+	odp = NULL;
+end:
+	up_read(&ctx->umem_rwsem);
+	return odp;
+}
+
+static struct ib_umem_odp *odp_lookup(struct ib_ucontext *ctx,
+				      u64 start, u64 length,
+				      struct mlx5_ib_mr *parent)
+{
+	struct ib_umem_odp *odp;
+	struct rb_node *rb;
+
+	down_read(&ctx->umem_rwsem);
+	odp = rbt_ib_umem_lookup(&ctx->umem_tree, start, length);
+	if (!odp)
+		goto end;
+
+	while (1) {
+		if (check_parent(odp, parent))
+			goto end;
+		rb = rb_next(&odp->interval_tree.rb);
+		if (!rb)
+			goto not_found;
+		odp = rb_entry(rb, struct ib_umem_odp, interval_tree.rb);
+		if (ib_umem_start(odp->umem) > start + length)
+			goto not_found;
+	}
+not_found:
+	odp = NULL;
+end:
+	up_read(&ctx->umem_rwsem);
+	return odp;
+}
+
+void mlx5_odp_populate_klm(struct mlx5_klm *pklm, size_t offset,
+			   size_t nentries, struct mlx5_ib_mr *mr, int flags)
+{
+	struct ib_pd *pd = mr->ibmr.pd;
+	struct ib_ucontext *ctx = pd->uobject->context;
+	struct mlx5_ib_dev *dev = to_mdev(pd->device);
+	struct ib_umem_odp *odp;
+	unsigned long va;
+	int i;
+
+	if (flags & MLX5_IB_UPD_XLT_ZAP) {
+		for (i = 0; i < nentries; i++, pklm++) {
+			pklm->bcount = cpu_to_be32(MLX5_IMR_MTT_SIZE);
+			pklm->key = cpu_to_be32(dev->null_mkey);
+			pklm->va = 0;
+		}
+		return;
+	}
+
+	odp = odp_lookup(ctx, offset * MLX5_IMR_MTT_SIZE,
+			     nentries * MLX5_IMR_MTT_SIZE, mr);
+
+	for (i = 0; i < nentries; i++, pklm++) {
+		pklm->bcount = cpu_to_be32(MLX5_IMR_MTT_SIZE);
+		va = (offset + i) * MLX5_IMR_MTT_SIZE;
+		if (odp && odp->umem->address == va) {
+			struct mlx5_ib_mr *mtt = odp->private;
+
+			pklm->key = cpu_to_be32(mtt->ibmr.lkey);
+			odp = odp_next(odp);
+		} else {
+			pklm->key = cpu_to_be32(dev->null_mkey);
+		}
+		mlx5_ib_dbg(dev, "[%d] va %lx key %x\n",
+			    i, va, be32_to_cpu(pklm->key));
+	}
+}
+
+static void mr_leaf_free_action(struct work_struct *work)
+{
+	struct ib_umem_odp *odp = container_of(work, struct ib_umem_odp, work);
+	int idx = ib_umem_start(odp->umem) >> MLX5_IMR_MTT_SHIFT;
+	struct mlx5_ib_mr *mr = odp->private, *imr = mr->parent;
+
+	mr->parent = NULL;
+	synchronize_srcu(&mr->dev->mr_srcu);
+
+	if (!READ_ONCE(odp->dying)) {
+		mr->parent = imr;
+		if (atomic_dec_and_test(&imr->num_leaf_free))
+			wake_up(&imr->q_leaf_free);
+		return;
+	}
+
+	ib_umem_release(odp->umem);
+	if (imr->live)
+		mlx5_ib_update_xlt(imr, idx, 1, 0,
+				   MLX5_IB_UPD_XLT_INDIRECT |
+				   MLX5_IB_UPD_XLT_ATOMIC);
+	mlx5_mr_cache_free(mr->dev, mr);
+
+	if (atomic_dec_and_test(&imr->num_leaf_free))
+		wake_up(&imr->q_leaf_free);
+}
+
 void mlx5_ib_invalidate_range(struct ib_umem *umem, unsigned long start,
 			      unsigned long end)
 {
@@ -111,6 +246,13 @@ void mlx5_ib_invalidate_range(struct ib_umem *umem, unsigned long start,
 	 */
 
 	ib_umem_odp_unmap_dma_pages(umem, start, end);
+
+	if (unlikely(!umem->npages && mr->parent &&
+		     !umem->odp_data->dying)) {
+		WRITE_ONCE(umem->odp_data->dying, 1);
+		atomic_inc(&mr->parent->num_leaf_free);
+		schedule_work(&umem->odp_data->work);
+	}
 }
 
 void mlx5_ib_internal_fill_odp_caps(struct mlx5_ib_dev *dev)
@@ -147,6 +289,11 @@ void mlx5_ib_internal_fill_odp_caps(struct mlx5_ib_dev *dev)
 	if (MLX5_CAP_ODP(dev->mdev, rc_odp_caps.atomic))
 		caps->per_transport_caps.rc_odp_caps |= IB_ODP_SUPPORT_ATOMIC;
 
+	if (MLX5_CAP_GEN(dev->mdev, fixed_buffer_size) &&
+	    MLX5_CAP_GEN(dev->mdev, null_mkey) &&
+	    MLX5_CAP_GEN(dev->mdev, umr_extended_translation_offset))
+		caps->general_caps |= IB_ODP_SUPPORT_IMPLICIT;
+
 	return;
 }
 
@@ -184,6 +331,197 @@ static void mlx5_ib_page_fault_resume(struct mlx5_ib_dev *dev,
 			    wq_num);
 }
 
+static struct mlx5_ib_mr *implicit_mr_alloc(struct ib_pd *pd,
+					    struct ib_umem *umem,
+					    bool ksm, int access_flags)
+{
+	struct mlx5_ib_dev *dev = to_mdev(pd->device);
+	struct mlx5_ib_mr *mr;
+	int err;
+
+	mr = mlx5_mr_cache_alloc(dev, ksm ? MLX5_IMR_KSM_CACHE_ENTRY :
+					    MLX5_IMR_MTT_CACHE_ENTRY);
+
+	if (IS_ERR(mr))
+		return mr;
+
+	mr->ibmr.pd = pd;
+
+	mr->dev = dev;
+	mr->access_flags = access_flags;
+	mr->mmkey.iova = 0;
+	mr->umem = umem;
+
+	if (ksm) {
+		err = mlx5_ib_update_xlt(mr, 0,
+					 mlx5_imr_ksm_entries,
+					 MLX5_KSM_PAGE_SHIFT,
+					 MLX5_IB_UPD_XLT_INDIRECT |
+					 MLX5_IB_UPD_XLT_ZAP |
+					 MLX5_IB_UPD_XLT_ENABLE);
+
+	} else {
+		err = mlx5_ib_update_xlt(mr, 0,
+					 MLX5_IMR_MTT_ENTRIES,
+					 PAGE_SHIFT,
+					 MLX5_IB_UPD_XLT_ZAP |
+					 MLX5_IB_UPD_XLT_ENABLE |
+					 MLX5_IB_UPD_XLT_ATOMIC);
+	}
+
+	if (err)
+		goto fail;
+
+	mr->ibmr.lkey = mr->mmkey.key;
+	mr->ibmr.rkey = mr->mmkey.key;
+
+	mr->live = 1;
+
+	mlx5_ib_dbg(dev, "key %x dev %p mr %p\n",
+		    mr->mmkey.key, dev->mdev, mr);
+
+	return mr;
+
+fail:
+	mlx5_ib_err(dev, "Failed to register MKEY %d\n", err);
+	mlx5_mr_cache_free(dev, mr);
+
+	return ERR_PTR(err);
+}
+
+static struct ib_umem_odp *implicit_mr_get_data(struct mlx5_ib_mr *mr,
+						u64 io_virt, size_t bcnt)
+{
+	struct ib_ucontext *ctx = mr->ibmr.pd->uobject->context;
+	struct mlx5_ib_dev *dev = to_mdev(mr->ibmr.pd->device);
+	struct ib_umem_odp *odp, *result = NULL;
+	u64 addr = io_virt & MLX5_IMR_MTT_MASK;
+	int nentries = 0, start_idx = 0, ret;
+	struct mlx5_ib_mr *mtt;
+	struct ib_umem *umem;
+
+	mutex_lock(&mr->umem->odp_data->umem_mutex);
+	odp = odp_lookup(ctx, addr, 1, mr);
+
+	mlx5_ib_dbg(dev, "io_virt:%llx bcnt:%zx addr:%llx odp:%p\n",
+		    io_virt, bcnt, addr, odp);
+
+next_mr:
+	if (likely(odp)) {
+		if (nentries)
+			nentries++;
+	} else {
+		umem = ib_alloc_odp_umem(ctx, addr, MLX5_IMR_MTT_SIZE);
+		if (IS_ERR(umem)) {
+			mutex_unlock(&mr->umem->odp_data->umem_mutex);
+			return ERR_CAST(umem);
+		}
+
+		mtt = implicit_mr_alloc(mr->ibmr.pd, umem, 0, mr->access_flags);
+		if (IS_ERR(mtt)) {
+			mutex_unlock(&mr->umem->odp_data->umem_mutex);
+			ib_umem_release(umem);
+			return ERR_CAST(mtt);
+		}
+
+		odp = umem->odp_data;
+		odp->private = mtt;
+		mtt->umem = umem;
+		mtt->mmkey.iova = addr;
+		mtt->parent = mr;
+		INIT_WORK(&odp->work, mr_leaf_free_action);
+
+		if (!nentries)
+			start_idx = addr >> MLX5_IMR_MTT_SHIFT;
+		nentries++;
+	}
+
+	odp->dying = 0;
+
+	/* Return first odp if region not covered by single one */
+	if (likely(!result))
+		result = odp;
+
+	addr += MLX5_IMR_MTT_SIZE;
+	if (unlikely(addr < io_virt + bcnt)) {
+		odp = odp_next(odp);
+		if (odp && odp->umem->address != addr)
+			odp = NULL;
+		goto next_mr;
+	}
+
+	if (unlikely(nentries)) {
+		ret = mlx5_ib_update_xlt(mr, start_idx, nentries, 0,
+					 MLX5_IB_UPD_XLT_INDIRECT |
+					 MLX5_IB_UPD_XLT_ATOMIC);
+		if (ret) {
+			mlx5_ib_err(dev, "Failed to update PAS\n");
+			result = ERR_PTR(ret);
+		}
+	}
+
+	mutex_unlock(&mr->umem->odp_data->umem_mutex);
+	return result;
+}
+
+struct mlx5_ib_mr *mlx5_ib_alloc_implicit_mr(struct mlx5_ib_pd *pd,
+					     int access_flags)
+{
+	struct ib_ucontext *ctx = pd->ibpd.uobject->context;
+	struct mlx5_ib_mr *imr;
+	struct ib_umem *umem;
+
+	umem = ib_umem_get(ctx, 0, 0, IB_ACCESS_ON_DEMAND, 0);
+	if (IS_ERR(umem))
+		return ERR_CAST(umem);
+
+	imr = implicit_mr_alloc(&pd->ibpd, umem, 1, access_flags);
+	if (IS_ERR(imr)) {
+		ib_umem_release(umem);
+		return ERR_CAST(imr);
+	}
+
+	imr->umem = umem;
+	init_waitqueue_head(&imr->q_leaf_free);
+	atomic_set(&imr->num_leaf_free, 0);
+
+	return imr;
+}
+
+static int mr_leaf_free(struct ib_umem *umem, u64 start,
+			u64 end, void *cookie)
+{
+	struct mlx5_ib_mr *mr = umem->odp_data->private, *imr = cookie;
+
+	if (mr->parent != imr)
+		return 0;
+
+	ib_umem_odp_unmap_dma_pages(umem,
+				    ib_umem_start(umem),
+				    ib_umem_end(umem));
+
+	if (umem->odp_data->dying)
+		return 0;
+
+	WRITE_ONCE(umem->odp_data->dying, 1);
+	atomic_inc(&imr->num_leaf_free);
+	schedule_work(&umem->odp_data->work);
+
+	return 0;
+}
+
+void mlx5_ib_free_implicit_mr(struct mlx5_ib_mr *imr)
+{
+	struct ib_ucontext *ctx = imr->ibmr.pd->uobject->context;
+
+	down_read(&ctx->umem_rwsem);
+	rbt_ib_umem_for_each_in_range(&ctx->umem_tree, 0, ULLONG_MAX,
+				      mr_leaf_free, imr);
+	up_read(&ctx->umem_rwsem);
+
+	wait_event(imr->q_leaf_free, !atomic_read(&imr->num_leaf_free));
+}
+
 /*
  * Handle a single data segment in a page-fault WQE or RDMA region.
  *
@@ -195,47 +533,43 @@ static void mlx5_ib_page_fault_resume(struct mlx5_ib_dev *dev,
  * -EFAULT when there's an error mapping the requested pages. The caller will
  *  abort the page fault handling.
  */
-static int pagefault_single_data_segment(struct mlx5_ib_dev *mib_dev,
+static int pagefault_single_data_segment(struct mlx5_ib_dev *dev,
 					 u32 key, u64 io_virt, size_t bcnt,
 					 u32 *bytes_committed,
 					 u32 *bytes_mapped)
 {
 	int srcu_key;
-	unsigned int current_seq;
+	unsigned int current_seq = 0;
 	u64 start_idx;
 	int npages = 0, ret = 0;
 	struct mlx5_ib_mr *mr;
 	u64 access_mask = ODP_READ_ALLOWED_BIT;
+	struct ib_umem_odp *odp;
+	int implicit = 0;
+	size_t size;
 
-	srcu_key = srcu_read_lock(&mib_dev->mr_srcu);
-	mr = mlx5_ib_odp_find_mr_lkey(mib_dev, key);
+	srcu_key = srcu_read_lock(&dev->mr_srcu);
+	mr = mlx5_ib_odp_find_mr_lkey(dev, key);
 	/*
 	 * If we didn't find the MR, it means the MR was closed while we were
 	 * handling the ODP event. In this case we return -EFAULT so that the
 	 * QP will be closed.
 	 */
 	if (!mr || !mr->ibmr.pd) {
-		pr_err("Failed to find relevant mr for lkey=0x%06x, probably the MR was destroyed\n",
-		       key);
+		mlx5_ib_dbg(dev, "Failed to find relevant mr for lkey=0x%06x, probably the MR was destroyed\n",
+			    key);
 		ret = -EFAULT;
 		goto srcu_unlock;
 	}
 	if (!mr->umem->odp_data) {
-		pr_debug("skipping non ODP MR (lkey=0x%06x) in page fault handler.\n",
-			 key);
+		mlx5_ib_dbg(dev, "skipping non ODP MR (lkey=0x%06x) in page fault handler.\n",
+			    key);
 		if (bytes_mapped)
 			*bytes_mapped +=
 				(bcnt - *bytes_committed);
 		goto srcu_unlock;
 	}
 
-	current_seq = ACCESS_ONCE(mr->umem->odp_data->notifiers_seq);
-	/*
-	 * Ensure the sequence number is valid for some time before we call
-	 * gup.
-	 */
-	smp_rmb();
-
 	/*
 	 * Avoid branches - this code will perform correctly
 	 * in all iterations (in iteration 2 and above,
@@ -244,63 +578,109 @@ static int pagefault_single_data_segment(struct mlx5_ib_dev *mib_dev,
 	io_virt += *bytes_committed;
 	bcnt -= *bytes_committed;
 
+	if (!mr->umem->odp_data->page_list) {
+		odp = implicit_mr_get_data(mr, io_virt, bcnt);
+
+		if (IS_ERR(odp)) {
+			ret = PTR_ERR(odp);
+			goto srcu_unlock;
+		}
+		mr = odp->private;
+		implicit = 1;
+
+	} else {
+		odp = mr->umem->odp_data;
+	}
+
+next_mr:
+	current_seq = READ_ONCE(odp->notifiers_seq);
+	/*
+	 * Ensure the sequence number is valid for some time before we call
+	 * gup.
+	 */
+	smp_rmb();
+
+	size = min_t(size_t, bcnt, ib_umem_end(odp->umem) - io_virt);
 	start_idx = (io_virt - (mr->mmkey.iova & PAGE_MASK)) >> PAGE_SHIFT;
 
 	if (mr->umem->writable)
 		access_mask |= ODP_WRITE_ALLOWED_BIT;
-	npages = ib_umem_odp_map_dma_pages(mr->umem, io_virt, bcnt,
-					   access_mask, current_seq);
-	if (npages < 0) {
-		ret = npages;
+
+	ret = ib_umem_odp_map_dma_pages(mr->umem, io_virt, size,
+					access_mask, current_seq);
+
+	if (ret < 0)
 		goto srcu_unlock;
-	}
 
-	if (npages > 0) {
-		mutex_lock(&mr->umem->odp_data->umem_mutex);
+	if (ret > 0) {
+		int np = ret;
+
+		mutex_lock(&odp->umem_mutex);
 		if (!ib_umem_mmu_notifier_retry(mr->umem, current_seq)) {
 			/*
 			 * No need to check whether the MTTs really belong to
 			 * this MR, since ib_umem_odp_map_dma_pages already
 			 * checks this.
 			 */
-			ret = mlx5_ib_update_xlt(mr, start_idx, npages,
+			ret = mlx5_ib_update_xlt(mr, start_idx, np,
 						 PAGE_SHIFT,
 						 MLX5_IB_UPD_XLT_ATOMIC);
 		} else {
 			ret = -EAGAIN;
 		}
-		mutex_unlock(&mr->umem->odp_data->umem_mutex);
+		mutex_unlock(&odp->umem_mutex);
 		if (ret < 0) {
 			if (ret != -EAGAIN)
-				pr_err("Failed to update mkey page tables\n");
+				mlx5_ib_err(dev, "Failed to update mkey page tables\n");
 			goto srcu_unlock;
 		}
 
 		if (bytes_mapped) {
-			u32 new_mappings = npages * PAGE_SIZE -
+			u32 new_mappings = np * PAGE_SIZE -
 				(io_virt - round_down(io_virt, PAGE_SIZE));
-			*bytes_mapped += min_t(u32, new_mappings, bcnt);
+			*bytes_mapped += min_t(u32, new_mappings, size);
 		}
+
+		npages += np;
+	}
+
+	bcnt -= size;
+	if (unlikely(bcnt)) {
+		struct ib_umem_odp *next;
+
+		io_virt += size;
+		next = odp_next(odp);
+		if (unlikely(!next || next->umem->address != io_virt)) {
+			mlx5_ib_dbg(dev, "next implicit leaf removed at 0x%llx. got %p\n",
+				    io_virt, next);
+			ret = -EAGAIN;
+			goto srcu_unlock_no_wait;
+		}
+		odp = next;
+		mr = odp->private;
+		goto next_mr;
 	}
 
 srcu_unlock:
 	if (ret == -EAGAIN) {
-		if (!mr->umem->odp_data->dying) {
-			struct ib_umem_odp *odp_data = mr->umem->odp_data;
+		if (implicit || !odp->dying) {
 			unsigned long timeout =
 				msecs_to_jiffies(MMU_NOTIFIER_TIMEOUT);
 
 			if (!wait_for_completion_timeout(
-					&odp_data->notifier_completion,
+					&odp->notifier_completion,
 					timeout)) {
-				pr_warn("timeout waiting for mmu notifier completion\n");
+				mlx5_ib_warn(dev, "timeout waiting for mmu notifier. seq %d against %d\n",
+					     current_seq, odp->notifiers_seq);
 			}
 		} else {
 			/* The MR is being killed, kill the QP as well. */
 			ret = -EFAULT;
 		}
 	}
-	srcu_read_unlock(&mib_dev->mr_srcu, srcu_key);
+
+srcu_unlock_no_wait:
+	srcu_read_unlock(&dev->mr_srcu, srcu_key);
 	*bytes_committed = 0;
 	return ret ? ret : npages;
 }
@@ -618,8 +998,8 @@ static void mlx5_ib_mr_wqe_pfault_handler(struct mlx5_ib_dev *dev,
 		goto resolve_page_fault;
 	} else if (ret < 0 || total_wqe_bytes > bytes_mapped) {
 		if (ret != -ENOENT)
-			mlx5_ib_err(dev, "Error getting user pages for page fault. Error: %d\n",
-				    ret);
+			mlx5_ib_err(dev, "PAGE FAULT error: %d. QP 0x%x. type: 0x%x\n",
+				    ret, pfault->wqe.wq_num, pfault->type);
 		goto resolve_page_fault;
 	}
 
@@ -627,7 +1007,7 @@ static void mlx5_ib_mr_wqe_pfault_handler(struct mlx5_ib_dev *dev,
 resolve_page_fault:
 	mlx5_ib_page_fault_resume(dev, pfault, resume_with_error);
 	mlx5_ib_dbg(dev, "PAGE FAULT completed. QP 0x%x resume_with_error=%d, type: 0x%x\n",
-		    pfault->token, resume_with_error,
+		    pfault->wqe.wq_num, resume_with_error,
 		    pfault->type);
 	free_page((unsigned long)buffer);
 }
@@ -700,10 +1080,9 @@ static void mlx5_ib_mr_rdma_pfault_handler(struct mlx5_ib_dev *dev,
 		ret = pagefault_single_data_segment(dev, rkey, address,
 						    prefetch_len,
 						    &bytes_committed, NULL);
-		if (ret < 0) {
+		if (ret < 0 && ret != -EAGAIN) {
 			mlx5_ib_warn(dev, "Prefetch failed. ret: %d, QP 0x%x, address: 0x%.16llx, length = 0x%.16x\n",
-				     ret, pfault->token, address,
-				     prefetch_len);
+				     ret, pfault->token, address, prefetch_len);
 		}
 	}
 }
@@ -728,19 +1107,61 @@ void mlx5_ib_pfault(struct mlx5_core_dev *mdev, void *context,
 	}
 }
 
-int mlx5_ib_odp_init_one(struct mlx5_ib_dev *ibdev)
+void mlx5_odp_init_mr_cache_entry(struct mlx5_cache_ent *ent)
+{
+	if (!(ent->dev->odp_caps.general_caps & IB_ODP_SUPPORT_IMPLICIT))
+		return;
+
+	switch (ent->order - 2) {
+	case MLX5_IMR_MTT_CACHE_ENTRY:
+		ent->page = PAGE_SHIFT;
+		ent->xlt = MLX5_IMR_MTT_ENTRIES *
+			   sizeof(struct mlx5_mtt) /
+			   MLX5_IB_UMR_OCTOWORD;
+		ent->access_mode = MLX5_MKC_ACCESS_MODE_MTT;
+		ent->limit = 0;
+		break;
+
+	case MLX5_IMR_KSM_CACHE_ENTRY:
+		ent->page = MLX5_KSM_PAGE_SHIFT;
+		ent->xlt = mlx5_imr_ksm_entries *
+			   sizeof(struct mlx5_klm) /
+			   MLX5_IB_UMR_OCTOWORD;
+		ent->access_mode = MLX5_MKC_ACCESS_MODE_KSM;
+		ent->limit = 0;
+		break;
+	}
+}
+
+int mlx5_ib_odp_init_one(struct mlx5_ib_dev *dev)
 {
 	int ret;
 
-	ret = init_srcu_struct(&ibdev->mr_srcu);
+	ret = init_srcu_struct(&dev->mr_srcu);
 	if (ret)
 		return ret;
 
+	if (dev->odp_caps.general_caps & IB_ODP_SUPPORT_IMPLICIT) {
+		ret = mlx5_cmd_null_mkey(dev->mdev, &dev->null_mkey);
+		if (ret) {
+			mlx5_ib_err(dev, "Error getting null_mkey %d\n", ret);
+			return ret;
+		}
+	}
+
 	return 0;
 }
 
-void mlx5_ib_odp_remove_one(struct mlx5_ib_dev *ibdev)
+void mlx5_ib_odp_remove_one(struct mlx5_ib_dev *dev)
+{
+	cleanup_srcu_struct(&dev->mr_srcu);
+}
+
+int mlx5_ib_odp_init(void)
 {
-	cleanup_srcu_struct(&ibdev->mr_srcu);
+	mlx5_imr_ksm_entries = BIT_ULL(get_order(TASK_SIZE) -
+				       MLX5_IMR_MTT_BITS);
+
+	return 0;
 }
 
diff --git a/include/linux/mlx5/driver.h b/include/linux/mlx5/driver.h
index 2534b8a0fd7b..886ff2b00500 100644
--- a/include/linux/mlx5/driver.h
+++ b/include/linux/mlx5/driver.h
@@ -1053,6 +1053,8 @@ enum {
 
 enum {
 	MAX_UMR_CACHE_ENTRY = 20,
+	MLX5_IMR_MTT_CACHE_ENTRY,
+	MLX5_IMR_KSM_CACHE_ENTRY,
 	MAX_MR_CACHE_ENTRIES
 };
 
-- 
cgit v1.2.3


From 72cd57178f3447b612f136c7a215418b5a20bf3d Mon Sep 17 00:00:00 2001
From: Or Gerlitz <ogerlitz@mellanox.com>
Date: Tue, 24 Jan 2017 13:02:36 +0200
Subject: IB/mlx5: Support raw packet protocol

Mark support for the new raw packet protocol on Eth ports.

Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Reviewed-by: Matan Barak <matanb@mellanox.com>
Signed-off-by: Leon Romanovsky <leon@kernel.org>
Signed-off-by: Doug Ledford <dledford@redhat.com>
---
 drivers/infiniband/hw/mlx5/main.c | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)

(limited to 'drivers/infiniband/hw/mlx5/main.c')

diff --git a/drivers/infiniband/hw/mlx5/main.c b/drivers/infiniband/hw/mlx5/main.c
index eb8719ca500e..c79a5c9f1f9a 100644
--- a/drivers/infiniband/hw/mlx5/main.c
+++ b/drivers/infiniband/hw/mlx5/main.c
@@ -2938,11 +2938,13 @@ static u32 get_core_cap_flags(struct ib_device *ibdev)
 	if (ll == IB_LINK_LAYER_INFINIBAND)
 		return RDMA_CORE_PORT_IBA_IB;
 
+	ret = RDMA_CORE_PORT_RAW_PACKET;
+
 	if (!(l3_type_cap & MLX5_ROCE_L3_TYPE_IPV4_CAP))
-		return 0;
+		return ret;
 
 	if (!(l3_type_cap & MLX5_ROCE_L3_TYPE_IPV6_CAP))
-		return 0;
+		return ret;
 
 	if (roce_version_cap & MLX5_ROCE_VERSION_1_CAP)
 		ret |= RDMA_CORE_PORT_IBA_ROCE;
-- 
cgit v1.2.3


From c4550c63b363914071d63a36e5bedcfe22057940 Mon Sep 17 00:00:00 2001
From: Or Gerlitz <ogerlitz@mellanox.com>
Date: Tue, 24 Jan 2017 13:02:39 +0200
Subject: IB: Query ports via the core instead of direct into the driver

Change the drivers to call ib_query_port in their get port
immutable handler instead of their own query port handler.

Doing this required to set the core cap flags of this device
before the ib_query_port call is made, since the IB core might
need these caps to serve the port query.

Drivers are ensured by the IB core that the port attributes passed
to the port query verb implementation are zero, and hence we
removed the zeroing from the drivers.

This patch doesn't add any new functionality.

Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Reviewed-by: Matan Barak <matanb@mellanox.com>
Signed-off-by: Leon Romanovsky <leon@kernel.org>
Reviewed-by: Steve Wise <swise@opengridcomputing.com>
Acked-by: Adit Ranadive <aditr@vmware.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
---
 drivers/infiniband/hw/cxgb3/iwch_provider.c     |  7 ++++---
 drivers/infiniband/hw/cxgb4/provider.c          |  8 ++++----
 drivers/infiniband/hw/hfi1/verbs.c              |  1 +
 drivers/infiniband/hw/hns/hns_roce_main.c       |  7 ++++---
 drivers/infiniband/hw/i40iw/i40iw_verbs.c       |  8 ++++----
 drivers/infiniband/hw/mlx4/alias_GUID.c         |  1 +
 drivers/infiniband/hw/mlx4/main.c               | 18 +++++++++---------
 drivers/infiniband/hw/mlx4/sysfs.c              |  1 +
 drivers/infiniband/hw/mlx5/mad.c                |  2 +-
 drivers/infiniband/hw/mlx5/main.c               | 11 +++++++----
 drivers/infiniband/hw/mthca/mthca_provider.c    |  9 +++++----
 drivers/infiniband/hw/nes/nes_verbs.c           |  5 +++--
 drivers/infiniband/hw/ocrdma/ocrdma_main.c      |  9 +++++----
 drivers/infiniband/hw/ocrdma/ocrdma_verbs.c     |  1 +
 drivers/infiniband/hw/qedr/verbs.c              |  9 +++++----
 drivers/infiniband/hw/qib/qib_verbs.c           |  1 +
 drivers/infiniband/hw/usnic/usnic_ib_main.c     |  5 +++--
 drivers/infiniband/hw/usnic/usnic_ib_verbs.c    |  2 +-
 drivers/infiniband/hw/vmw_pvrdma/pvrdma_main.c  |  5 +++--
 drivers/infiniband/hw/vmw_pvrdma/pvrdma_verbs.c |  4 ++--
 drivers/infiniband/sw/rdmavt/vt.c               |  7 ++++---
 drivers/infiniband/sw/rxe/rxe_verbs.c           |  6 ++++--
 22 files changed, 73 insertions(+), 54 deletions(-)

(limited to 'drivers/infiniband/hw/mlx5/main.c')

diff --git a/drivers/infiniband/hw/cxgb3/iwch_provider.c b/drivers/infiniband/hw/cxgb3/iwch_provider.c
index 9d5fe1853da4..9e39252a570a 100644
--- a/drivers/infiniband/hw/cxgb3/iwch_provider.c
+++ b/drivers/infiniband/hw/cxgb3/iwch_provider.c
@@ -1133,7 +1133,7 @@ static int iwch_query_port(struct ib_device *ibdev,
 	dev = to_iwch_dev(ibdev);
 	netdev = dev->rdev.port_info.lldevs[port-1];
 
-	memset(props, 0, sizeof(struct ib_port_attr));
+	/* props being zeroed by the caller, avoid zeroing it here */
 	props->max_mtu = IB_MTU_4096;
 	if (netdev->mtu >= 4096)
 		props->active_mtu = IB_MTU_4096;
@@ -1338,13 +1338,14 @@ static int iwch_port_immutable(struct ib_device *ibdev, u8 port_num,
 	struct ib_port_attr attr;
 	int err;
 
-	err = iwch_query_port(ibdev, port_num, &attr);
+	immutable->core_cap_flags = RDMA_CORE_PORT_IWARP;
+
+	err = ib_query_port(ibdev, port_num, &attr);
 	if (err)
 		return err;
 
 	immutable->pkey_tbl_len = attr.pkey_tbl_len;
 	immutable->gid_tbl_len = attr.gid_tbl_len;
-	immutable->core_cap_flags = RDMA_CORE_PORT_IWARP;
 
 	return 0;
 }
diff --git a/drivers/infiniband/hw/cxgb4/provider.c b/drivers/infiniband/hw/cxgb4/provider.c
index 49b51b7e0fd7..5b00b50a484f 100644
--- a/drivers/infiniband/hw/cxgb4/provider.c
+++ b/drivers/infiniband/hw/cxgb4/provider.c
@@ -358,8 +358,7 @@ static int c4iw_query_port(struct ib_device *ibdev, u8 port,
 
 	dev = to_c4iw_dev(ibdev);
 	netdev = dev->rdev.lldi.ports[port-1];
-
-	memset(props, 0, sizeof(struct ib_port_attr));
+	/* props being zeroed by the caller, avoid zeroing it here */
 	props->max_mtu = IB_MTU_4096;
 	if (netdev->mtu >= 4096)
 		props->active_mtu = IB_MTU_4096;
@@ -505,13 +504,14 @@ static int c4iw_port_immutable(struct ib_device *ibdev, u8 port_num,
 	struct ib_port_attr attr;
 	int err;
 
-	err = c4iw_query_port(ibdev, port_num, &attr);
+	immutable->core_cap_flags = RDMA_CORE_PORT_IWARP;
+
+	err = ib_query_port(ibdev, port_num, &attr);
 	if (err)
 		return err;
 
 	immutable->pkey_tbl_len = attr.pkey_tbl_len;
 	immutable->gid_tbl_len = attr.gid_tbl_len;
-	immutable->core_cap_flags = RDMA_CORE_PORT_IWARP;
 
 	return 0;
 }
diff --git a/drivers/infiniband/hw/hfi1/verbs.c b/drivers/infiniband/hw/hfi1/verbs.c
index 95ed4d6da510..f2d4e0428725 100644
--- a/drivers/infiniband/hw/hfi1/verbs.c
+++ b/drivers/infiniband/hw/hfi1/verbs.c
@@ -1384,6 +1384,7 @@ static int query_port(struct rvt_dev_info *rdi, u8 port_num,
 	struct hfi1_pportdata *ppd = &dd->pport[port_num - 1];
 	u16 lid = ppd->lid;
 
+	/* props being zeroed by the caller, avoid zeroing it here */
 	props->lid = lid ? lid : 0;
 	props->lmc = ppd->lmc;
 	/* OPA logical states match IB logical states */
diff --git a/drivers/infiniband/hw/hns/hns_roce_main.c b/drivers/infiniband/hw/hns/hns_roce_main.c
index 4953d9cb83a7..7be8158f15ee 100644
--- a/drivers/infiniband/hw/hns/hns_roce_main.c
+++ b/drivers/infiniband/hw/hns/hns_roce_main.c
@@ -249,7 +249,7 @@ static int hns_roce_query_port(struct ib_device *ib_dev, u8 port_num,
 	assert(port_num > 0);
 	port = port_num - 1;
 
-	memset(props, 0, sizeof(*props));
+	/* props being zeroed by the caller, avoid zeroing it here */
 
 	props->max_mtu = hr_dev->caps.max_mtu;
 	props->gid_tbl_len = hr_dev->caps.gid_table_len[port];
@@ -400,14 +400,15 @@ static int hns_roce_port_immutable(struct ib_device *ib_dev, u8 port_num,
 	struct ib_port_attr attr;
 	int ret;
 
-	ret = hns_roce_query_port(ib_dev, port_num, &attr);
+	immutable->core_cap_flags = RDMA_CORE_PORT_IBA_ROCE;
+
+	ret = ib_query_port(ib_dev, port_num, &attr);
 	if (ret)
 		return ret;
 
 	immutable->pkey_tbl_len = attr.pkey_tbl_len;
 	immutable->gid_tbl_len = attr.gid_tbl_len;
 
-	immutable->core_cap_flags = RDMA_CORE_PORT_IBA_ROCE;
 	immutable->max_mad_size = IB_MGMT_MAD_SIZE;
 
 	return 0;
diff --git a/drivers/infiniband/hw/i40iw/i40iw_verbs.c b/drivers/infiniband/hw/i40iw/i40iw_verbs.c
index 29e97df9e1a7..3e830486417b 100644
--- a/drivers/infiniband/hw/i40iw/i40iw_verbs.c
+++ b/drivers/infiniband/hw/i40iw/i40iw_verbs.c
@@ -97,8 +97,7 @@ static int i40iw_query_port(struct ib_device *ibdev,
 	struct i40iw_device *iwdev = to_iwdev(ibdev);
 	struct net_device *netdev = iwdev->netdev;
 
-	memset(props, 0, sizeof(*props));
-
+	/* props being zeroed by the caller, avoid zeroing it here */
 	props->max_mtu = IB_MTU_4096;
 	if (netdev->mtu >= 4096)
 		props->active_mtu = IB_MTU_4096;
@@ -2506,14 +2505,15 @@ static int i40iw_port_immutable(struct ib_device *ibdev, u8 port_num,
 	struct ib_port_attr attr;
 	int err;
 
-	err = i40iw_query_port(ibdev, port_num, &attr);
+	immutable->core_cap_flags = RDMA_CORE_PORT_IWARP;
+
+	err = ib_query_port(ibdev, port_num, &attr);
 
 	if (err)
 		return err;
 
 	immutable->pkey_tbl_len = attr.pkey_tbl_len;
 	immutable->gid_tbl_len = attr.gid_tbl_len;
-	immutable->core_cap_flags = RDMA_CORE_PORT_IWARP;
 
 	return 0;
 }
diff --git a/drivers/infiniband/hw/mlx4/alias_GUID.c b/drivers/infiniband/hw/mlx4/alias_GUID.c
index 06020c54db20..ea24230ea0d4 100644
--- a/drivers/infiniband/hw/mlx4/alias_GUID.c
+++ b/drivers/infiniband/hw/mlx4/alias_GUID.c
@@ -499,6 +499,7 @@ static int set_guid_rec(struct ib_device *ibdev,
 	struct list_head *head =
 		&dev->sriov.alias_guid.ports_guid[port - 1].cb_list;
 
+	memset(&attr, 0, sizeof(attr));
 	err = __mlx4_ib_query_port(ibdev, port, &attr, 1);
 	if (err) {
 		pr_debug("mlx4_ib_query_port failed (err: %d), port: %d\n",
diff --git a/drivers/infiniband/hw/mlx4/main.c b/drivers/infiniband/hw/mlx4/main.c
index e0c1ef7696e8..ba6af84cc236 100644
--- a/drivers/infiniband/hw/mlx4/main.c
+++ b/drivers/infiniband/hw/mlx4/main.c
@@ -741,7 +741,7 @@ int __mlx4_ib_query_port(struct ib_device *ibdev, u8 port,
 {
 	int err;
 
-	memset(props, 0, sizeof *props);
+	/* props being zeroed by the caller, avoid zeroing it here */
 
 	err = mlx4_ib_port_link_layer(ibdev, port) == IB_LINK_LAYER_INFINIBAND ?
 		ib_link_query_port(ibdev, port, props, netw_view) :
@@ -1014,7 +1014,7 @@ static int mlx4_ib_modify_port(struct ib_device *ibdev, u8 port, int mask,
 
 	mutex_lock(&mdev->cap_mask_mutex);
 
-	err = mlx4_ib_query_port(ibdev, port, &attr);
+	err = ib_query_port(ibdev, port, &attr);
 	if (err)
 		goto out;
 
@@ -2537,13 +2537,6 @@ static int mlx4_port_immutable(struct ib_device *ibdev, u8 port_num,
 	struct mlx4_ib_dev *mdev = to_mdev(ibdev);
 	int err;
 
-	err = mlx4_ib_query_port(ibdev, port_num, &attr);
-	if (err)
-		return err;
-
-	immutable->pkey_tbl_len = attr.pkey_tbl_len;
-	immutable->gid_tbl_len = attr.gid_tbl_len;
-
 	if (mlx4_ib_port_link_layer(ibdev, port_num) == IB_LINK_LAYER_INFINIBAND) {
 		immutable->core_cap_flags = RDMA_CORE_PORT_IBA_IB;
 		immutable->max_mad_size = IB_MGMT_MAD_SIZE;
@@ -2559,6 +2552,13 @@ static int mlx4_port_immutable(struct ib_device *ibdev, u8 port_num,
 			immutable->max_mad_size = IB_MGMT_MAD_SIZE;
 	}
 
+	err = ib_query_port(ibdev, port_num, &attr);
+	if (err)
+		return err;
+
+	immutable->pkey_tbl_len = attr.pkey_tbl_len;
+	immutable->gid_tbl_len = attr.gid_tbl_len;
+
 	return 0;
 }
 
diff --git a/drivers/infiniband/hw/mlx4/sysfs.c b/drivers/infiniband/hw/mlx4/sysfs.c
index 69fb5ba94d0f..0ba5ba7540c8 100644
--- a/drivers/infiniband/hw/mlx4/sysfs.c
+++ b/drivers/infiniband/hw/mlx4/sysfs.c
@@ -226,6 +226,7 @@ static int add_port_entries(struct mlx4_ib_dev *device, int port_num)
 	int ret = 0 ;
 	struct ib_port_attr attr;
 
+	memset(&attr, 0, sizeof(attr));
 	/* get the physical gid and pkey table sizes.*/
 	ret = __mlx4_ib_query_port(&device->ib_dev, port_num, &attr, 1);
 	if (ret)
diff --git a/drivers/infiniband/hw/mlx5/mad.c b/drivers/infiniband/hw/mlx5/mad.c
index af962e7fdc3a..8dacb49eabd9 100644
--- a/drivers/infiniband/hw/mlx5/mad.c
+++ b/drivers/infiniband/hw/mlx5/mad.c
@@ -527,7 +527,7 @@ int mlx5_query_mad_ifc_port(struct ib_device *ibdev, u8 port,
 	if (!in_mad || !out_mad)
 		goto out;
 
-	memset(props, 0, sizeof(*props));
+	/* props being zeroed by the caller, avoid zeroing it here */
 
 	init_query_mad(in_mad);
 	in_mad->attr_id  = IB_SMP_ATTR_PORT_INFO;
diff --git a/drivers/infiniband/hw/mlx5/main.c b/drivers/infiniband/hw/mlx5/main.c
index c79a5c9f1f9a..8fd35ebb13e3 100644
--- a/drivers/infiniband/hw/mlx5/main.c
+++ b/drivers/infiniband/hw/mlx5/main.c
@@ -170,7 +170,7 @@ static int mlx5_query_port_roce(struct ib_device *device, u8 port_num,
 	enum ib_mtu ndev_ib_mtu;
 	u16 qkey_viol_cntr;
 
-	memset(props, 0, sizeof(*props));
+	/* props being zeroed by the caller, avoid zeroing it here */
 
 	props->port_cap_flags  |= IB_PORT_CM_SUP;
 	props->port_cap_flags  |= IB_PORT_IP_BASED_GIDS;
@@ -857,7 +857,7 @@ static int mlx5_query_hca_port(struct ib_device *ibdev, u8 port,
 		goto out;
 	}
 
-	memset(props, 0, sizeof(*props));
+	/* props being zeroed by the caller, avoid zeroing it here */
 
 	err = mlx5_query_hca_vport_context(mdev, 0, port, 0, rep);
 	if (err)
@@ -1005,7 +1005,7 @@ static int mlx5_ib_modify_port(struct ib_device *ibdev, u8 port, int mask,
 
 	mutex_lock(&dev->cap_mask_mutex);
 
-	err = mlx5_ib_query_port(ibdev, port, &attr);
+	err = ib_query_port(ibdev, port, &attr);
 	if (err)
 		goto out;
 
@@ -2644,6 +2644,7 @@ static int get_port_caps(struct mlx5_ib_dev *dev)
 	}
 
 	for (port = 1; port <= MLX5_CAP_GEN(dev->mdev, num_ports); port++) {
+		memset(pprops, 0, sizeof(*pprops));
 		err = mlx5_ib_query_port(&dev->ib_dev, port, pprops);
 		if (err) {
 			mlx5_ib_warn(dev, "query_port %d failed %d\n",
@@ -2963,7 +2964,9 @@ static int mlx5_port_immutable(struct ib_device *ibdev, u8 port_num,
 	enum rdma_link_layer ll = mlx5_ib_port_link_layer(ibdev, port_num);
 	int err;
 
-	err = mlx5_ib_query_port(ibdev, port_num, &attr);
+	immutable->core_cap_flags = get_core_cap_flags(ibdev);
+
+	err = ib_query_port(ibdev, port_num, &attr);
 	if (err)
 		return err;
 
diff --git a/drivers/infiniband/hw/mthca/mthca_provider.c b/drivers/infiniband/hw/mthca/mthca_provider.c
index d31708742ba5..ce163184e742 100644
--- a/drivers/infiniband/hw/mthca/mthca_provider.c
+++ b/drivers/infiniband/hw/mthca/mthca_provider.c
@@ -146,7 +146,7 @@ static int mthca_query_port(struct ib_device *ibdev,
 	if (!in_mad || !out_mad)
 		goto out;
 
-	memset(props, 0, sizeof *props);
+	/* props being zeroed by the caller, avoid zeroing it here */
 
 	init_query_mad(in_mad);
 	in_mad->attr_id  = IB_SMP_ATTR_PORT_INFO;
@@ -212,7 +212,7 @@ static int mthca_modify_port(struct ib_device *ibdev,
 	if (mutex_lock_interruptible(&to_mdev(ibdev)->cap_mask_mutex))
 		return -ERESTARTSYS;
 
-	err = mthca_query_port(ibdev, port, &attr);
+	err = ib_query_port(ibdev, port, &attr);
 	if (err)
 		goto out;
 
@@ -1166,13 +1166,14 @@ static int mthca_port_immutable(struct ib_device *ibdev, u8 port_num,
 	struct ib_port_attr attr;
 	int err;
 
-	err = mthca_query_port(ibdev, port_num, &attr);
+	immutable->core_cap_flags = RDMA_CORE_PORT_IBA_IB;
+
+	err = ib_query_port(ibdev, port_num, &attr);
 	if (err)
 		return err;
 
 	immutable->pkey_tbl_len = attr.pkey_tbl_len;
 	immutable->gid_tbl_len = attr.gid_tbl_len;
-	immutable->core_cap_flags = RDMA_CORE_PORT_IBA_IB;
 	immutable->max_mad_size = IB_MGMT_MAD_SIZE;
 
 	return 0;
diff --git a/drivers/infiniband/hw/nes/nes_verbs.c b/drivers/infiniband/hw/nes/nes_verbs.c
index aff9fb14768b..b7179f4ac3a6 100644
--- a/drivers/infiniband/hw/nes/nes_verbs.c
+++ b/drivers/infiniband/hw/nes/nes_verbs.c
@@ -475,7 +475,7 @@ static int nes_query_port(struct ib_device *ibdev, u8 port, struct ib_port_attr
 	struct nes_vnic *nesvnic = to_nesvnic(ibdev);
 	struct net_device *netdev = nesvnic->netdev;
 
-	memset(props, 0, sizeof(*props));
+	/* props being zeroed by the caller, avoid zeroing it here */
 
 	props->max_mtu = IB_MTU_4096;
 
@@ -3670,13 +3670,14 @@ static int nes_port_immutable(struct ib_device *ibdev, u8 port_num,
 	struct ib_port_attr attr;
 	int err;
 
+	immutable->core_cap_flags = RDMA_CORE_PORT_IWARP;
+
 	err = nes_query_port(ibdev, port_num, &attr);
 	if (err)
 		return err;
 
 	immutable->pkey_tbl_len = attr.pkey_tbl_len;
 	immutable->gid_tbl_len = attr.gid_tbl_len;
-	immutable->core_cap_flags = RDMA_CORE_PORT_IWARP;
 
 	return 0;
 }
diff --git a/drivers/infiniband/hw/ocrdma/ocrdma_main.c b/drivers/infiniband/hw/ocrdma/ocrdma_main.c
index 896071502739..3e43bdc81e7a 100644
--- a/drivers/infiniband/hw/ocrdma/ocrdma_main.c
+++ b/drivers/infiniband/hw/ocrdma/ocrdma_main.c
@@ -93,15 +93,16 @@ static int ocrdma_port_immutable(struct ib_device *ibdev, u8 port_num,
 	int err;
 
 	dev = get_ocrdma_dev(ibdev);
-	err = ocrdma_query_port(ibdev, port_num, &attr);
+	immutable->core_cap_flags = RDMA_CORE_PORT_IBA_ROCE;
+	if (ocrdma_is_udp_encap_supported(dev))
+		immutable->core_cap_flags |= RDMA_CORE_CAP_PROT_ROCE_UDP_ENCAP;
+
+	err = ib_query_port(ibdev, port_num, &attr);
 	if (err)
 		return err;
 
 	immutable->pkey_tbl_len = attr.pkey_tbl_len;
 	immutable->gid_tbl_len = attr.gid_tbl_len;
-	immutable->core_cap_flags = RDMA_CORE_PORT_IBA_ROCE;
-	if (ocrdma_is_udp_encap_supported(dev))
-		immutable->core_cap_flags |= RDMA_CORE_CAP_PROT_ROCE_UDP_ENCAP;
 	immutable->max_mad_size = IB_MGMT_MAD_SIZE;
 
 	return 0;
diff --git a/drivers/infiniband/hw/ocrdma/ocrdma_verbs.c b/drivers/infiniband/hw/ocrdma/ocrdma_verbs.c
index 6af44f8db3d5..013d15c7e593 100644
--- a/drivers/infiniband/hw/ocrdma/ocrdma_verbs.c
+++ b/drivers/infiniband/hw/ocrdma/ocrdma_verbs.c
@@ -210,6 +210,7 @@ int ocrdma_query_port(struct ib_device *ibdev,
 	struct ocrdma_dev *dev;
 	struct net_device *netdev;
 
+	/* props being zeroed by the caller, avoid zeroing it here */
 	dev = get_ocrdma_dev(ibdev);
 	if (port > 1) {
 		pr_err("%s(%d) invalid_port=0x%x\n", __func__,
diff --git a/drivers/infiniband/hw/qedr/verbs.c b/drivers/infiniband/hw/qedr/verbs.c
index 57c8de208077..91a16d38d1fc 100644
--- a/drivers/infiniband/hw/qedr/verbs.c
+++ b/drivers/infiniband/hw/qedr/verbs.c
@@ -238,8 +238,8 @@ int qedr_query_port(struct ib_device *ibdev, u8 port, struct ib_port_attr *attr)
 	}
 
 	rdma_port = dev->ops->rdma_query_port(dev->rdma_ctx);
-	memset(attr, 0, sizeof(*attr));
 
+	/* *attr being zeroed by the caller, avoid zeroing it here */
 	if (rdma_port->port_state == QED_RDMA_PORT_UP) {
 		attr->state = IB_PORT_ACTIVE;
 		attr->phys_state = 5;
@@ -3549,14 +3549,15 @@ int qedr_port_immutable(struct ib_device *ibdev, u8 port_num,
 	struct ib_port_attr attr;
 	int err;
 
-	err = qedr_query_port(ibdev, port_num, &attr);
+	immutable->core_cap_flags = RDMA_CORE_PORT_IBA_ROCE |
+				    RDMA_CORE_PORT_IBA_ROCE_UDP_ENCAP;
+
+	err = ib_query_port(ibdev, port_num, &attr);
 	if (err)
 		return err;
 
 	immutable->pkey_tbl_len = attr.pkey_tbl_len;
 	immutable->gid_tbl_len = attr.gid_tbl_len;
-	immutable->core_cap_flags = RDMA_CORE_PORT_IBA_ROCE |
-				    RDMA_CORE_PORT_IBA_ROCE_UDP_ENCAP;
 	immutable->max_mad_size = IB_MGMT_MAD_SIZE;
 
 	return 0;
diff --git a/drivers/infiniband/hw/qib/qib_verbs.c b/drivers/infiniband/hw/qib/qib_verbs.c
index 4b54c0ddd08a..9161574601f6 100644
--- a/drivers/infiniband/hw/qib/qib_verbs.c
+++ b/drivers/infiniband/hw/qib/qib_verbs.c
@@ -1303,6 +1303,7 @@ static int qib_query_port(struct rvt_dev_info *rdi, u8 port_num,
 	enum ib_mtu mtu;
 	u16 lid = ppd->lid;
 
+	/* props being zeroed by the caller, avoid zeroing it here */
 	props->lid = lid ? lid : be16_to_cpu(IB_LID_PERMISSIVE);
 	props->lmc = ppd->lmc;
 	props->state = dd->f_iblink_state(ppd->lastibcstat);
diff --git a/drivers/infiniband/hw/usnic/usnic_ib_main.c b/drivers/infiniband/hw/usnic/usnic_ib_main.c
index dde0b23f6559..4f5a45db08e1 100644
--- a/drivers/infiniband/hw/usnic/usnic_ib_main.c
+++ b/drivers/infiniband/hw/usnic/usnic_ib_main.c
@@ -321,11 +321,12 @@ static int usnic_port_immutable(struct ib_device *ibdev, u8 port_num,
 	struct ib_port_attr attr;
 	int err;
 
-	err = usnic_ib_query_port(ibdev, port_num, &attr);
+	immutable->core_cap_flags = RDMA_CORE_PORT_USNIC;
+
+	err = ib_query_port(ibdev, port_num, &attr);
 	if (err)
 		return err;
 
-	immutable->core_cap_flags = RDMA_CORE_PORT_USNIC;
 	immutable->pkey_tbl_len = attr.pkey_tbl_len;
 	immutable->gid_tbl_len = attr.gid_tbl_len;
 
diff --git a/drivers/infiniband/hw/usnic/usnic_ib_verbs.c b/drivers/infiniband/hw/usnic/usnic_ib_verbs.c
index 74819a7951e2..0ba274ff7be6 100644
--- a/drivers/infiniband/hw/usnic/usnic_ib_verbs.c
+++ b/drivers/infiniband/hw/usnic/usnic_ib_verbs.c
@@ -330,7 +330,7 @@ int usnic_ib_query_port(struct ib_device *ibdev, u8 port,
 
 	mutex_lock(&us_ibdev->usdev_lock);
 	__ethtool_get_link_ksettings(us_ibdev->netdev, &cmd);
-	memset(props, 0, sizeof(*props));
+	/* props being zeroed by the caller, avoid zeroing it here */
 
 	props->lid = 0;
 	props->lmc = 1;
diff --git a/drivers/infiniband/hw/vmw_pvrdma/pvrdma_main.c b/drivers/infiniband/hw/vmw_pvrdma/pvrdma_main.c
index 231a1ce1f4be..b8b0081de478 100644
--- a/drivers/infiniband/hw/vmw_pvrdma/pvrdma_main.c
+++ b/drivers/infiniband/hw/vmw_pvrdma/pvrdma_main.c
@@ -132,13 +132,14 @@ static int pvrdma_port_immutable(struct ib_device *ibdev, u8 port_num,
 	struct ib_port_attr attr;
 	int err;
 
-	err = pvrdma_query_port(ibdev, port_num, &attr);
+	immutable->core_cap_flags = RDMA_CORE_PORT_IBA_ROCE;
+
+	err = ib_query_port(ibdev, port_num, &attr);
 	if (err)
 		return err;
 
 	immutable->pkey_tbl_len = attr.pkey_tbl_len;
 	immutable->gid_tbl_len = attr.gid_tbl_len;
-	immutable->core_cap_flags = RDMA_CORE_PORT_IBA_ROCE;
 	immutable->max_mad_size = IB_MGMT_MAD_SIZE;
 	return 0;
 }
diff --git a/drivers/infiniband/hw/vmw_pvrdma/pvrdma_verbs.c b/drivers/infiniband/hw/vmw_pvrdma/pvrdma_verbs.c
index 54891370d18a..b3a982be8006 100644
--- a/drivers/infiniband/hw/vmw_pvrdma/pvrdma_verbs.c
+++ b/drivers/infiniband/hw/vmw_pvrdma/pvrdma_verbs.c
@@ -135,7 +135,7 @@ int pvrdma_query_port(struct ib_device *ibdev, u8 port,
 		return err;
 	}
 
-	memset(props, 0, sizeof(*props));
+	/* props being zeroed by the caller, avoid zeroing it here */
 
 	props->state = pvrdma_port_state_to_ib(resp->attrs.state);
 	props->max_mtu = pvrdma_mtu_to_ib(resp->attrs.max_mtu);
@@ -275,7 +275,7 @@ int pvrdma_modify_port(struct ib_device *ibdev, u8 port, int mask,
 	}
 
 	mutex_lock(&vdev->port_mutex);
-	ret = pvrdma_query_port(ibdev, port, &attr);
+	ret = ib_query_port(ibdev, port, &attr);
 	if (ret)
 		goto out;
 
diff --git a/drivers/infiniband/sw/rdmavt/vt.c b/drivers/infiniband/sw/rdmavt/vt.c
index d430c2f7cec4..1165639a914b 100644
--- a/drivers/infiniband/sw/rdmavt/vt.c
+++ b/drivers/infiniband/sw/rdmavt/vt.c
@@ -165,7 +165,7 @@ static int rvt_query_port(struct ib_device *ibdev, u8 port_num,
 		return -EINVAL;
 
 	rvp = rdi->ports[port_index];
-	memset(props, 0, sizeof(*props));
+	/* props being zeroed by the caller, avoid zeroing it here */
 	props->sm_lid = rvp->sm_lid;
 	props->sm_sl = rvp->sm_sl;
 	props->port_cap_flags = rvp->port_cap_flags;
@@ -326,13 +326,14 @@ static int rvt_get_port_immutable(struct ib_device *ibdev, u8 port_num,
 	if (port_index < 0)
 		return -EINVAL;
 
-	err = rvt_query_port(ibdev, port_num, &attr);
+	immutable->core_cap_flags = rdi->dparms.core_cap_flags;
+
+	err = ib_query_port(ibdev, port_num, &attr);
 	if (err)
 		return err;
 
 	immutable->pkey_tbl_len = attr.pkey_tbl_len;
 	immutable->gid_tbl_len = attr.gid_tbl_len;
-	immutable->core_cap_flags = rdi->dparms.core_cap_flags;
 	immutable->max_mad_size = rdi->dparms.max_mad_size;
 
 	return 0;
diff --git a/drivers/infiniband/sw/rxe/rxe_verbs.c b/drivers/infiniband/sw/rxe/rxe_verbs.c
index beb7021ff18a..371ef3bac8d4 100644
--- a/drivers/infiniband/sw/rxe/rxe_verbs.c
+++ b/drivers/infiniband/sw/rxe/rxe_verbs.c
@@ -86,6 +86,7 @@ static int rxe_query_port(struct ib_device *dev,
 
 	port = &rxe->port;
 
+	/* *attr being zeroed by the caller, avoid zeroing it here */
 	*attr = port->attr;
 
 	mutex_lock(&rxe->usdev_lock);
@@ -261,13 +262,14 @@ static int rxe_port_immutable(struct ib_device *dev, u8 port_num,
 	int err;
 	struct ib_port_attr attr;
 
-	err = rxe_query_port(dev, port_num, &attr);
+	immutable->core_cap_flags = RDMA_CORE_PORT_IBA_ROCE_UDP_ENCAP;
+
+	err = ib_query_port(dev, port_num, &attr);
 	if (err)
 		return err;
 
 	immutable->pkey_tbl_len = attr.pkey_tbl_len;
 	immutable->gid_tbl_len = attr.gid_tbl_len;
-	immutable->core_cap_flags = RDMA_CORE_PORT_IBA_ROCE_UDP_ENCAP;
 	immutable->max_mad_size = IB_MGMT_MAD_SIZE;
 
 	return 0;
-- 
cgit v1.2.3


From cdbe33d0f82d68ff74f05502a4c26e65ec7e90bb Mon Sep 17 00:00:00 2001
From: Eli Cohen <eli@mellanox.com>
Date: Tue, 14 Feb 2017 07:25:38 +0200
Subject: IB/mlx5: Fix configuration of port capabilities

When the "ib_virt" cap is set, configuration of port capabilities need
to be done through mlx5_core_modify_hca_vport_context.
Since modify_hca_vport_context accepts mask and value, there is no need
to read the port capabilities and calculate the new cap values so we
avoid the mutex when ib_virt is set.

Signed-off-by: Eli Cohen <eli@mellanox.com>
Reviewed-by: Matan Barak <matanb@mellanox.com>
Signed-off-by: Leon Romanovsky <leon@kernel.org>
Signed-off-by: Doug Ledford <dledford@redhat.com>
---
 drivers/infiniband/hw/mlx5/main.c | 35 +++++++++++++++++++++++++++++++++++
 1 file changed, 35 insertions(+)

(limited to 'drivers/infiniband/hw/mlx5/main.c')

diff --git a/drivers/infiniband/hw/mlx5/main.c b/drivers/infiniband/hw/mlx5/main.c
index 8fd35ebb13e3..4a043cf35b9a 100644
--- a/drivers/infiniband/hw/mlx5/main.c
+++ b/drivers/infiniband/hw/mlx5/main.c
@@ -995,6 +995,31 @@ static int mlx5_ib_modify_device(struct ib_device *ibdev, int mask,
 	return err;
 }
 
+static int set_port_caps_atomic(struct mlx5_ib_dev *dev, u8 port_num, u32 mask,
+				u32 value)
+{
+	struct mlx5_hca_vport_context ctx = {};
+	int err;
+
+	err = mlx5_query_hca_vport_context(dev->mdev, 0,
+					   port_num, 0, &ctx);
+	if (err)
+		return err;
+
+	if (~ctx.cap_mask1_perm & mask) {
+		mlx5_ib_warn(dev, "trying to change bitmask 0x%X but change supported 0x%X\n",
+			     mask, ctx.cap_mask1_perm);
+		return -EINVAL;
+	}
+
+	ctx.cap_mask1 = value;
+	ctx.cap_mask1_perm = mask;
+	err = mlx5_core_modify_hca_vport_context(dev->mdev, 0,
+						 port_num, 0, &ctx);
+
+	return err;
+}
+
 static int mlx5_ib_modify_port(struct ib_device *ibdev, u8 port, int mask,
 			       struct ib_port_modify *props)
 {
@@ -1002,6 +1027,16 @@ static int mlx5_ib_modify_port(struct ib_device *ibdev, u8 port, int mask,
 	struct ib_port_attr attr;
 	u32 tmp;
 	int err;
+	u32 change_mask;
+	u32 value;
+	bool is_ib = (mlx5_ib_port_link_layer(ibdev, port) ==
+		      IB_LINK_LAYER_INFINIBAND);
+
+	if (MLX5_CAP_GEN(dev->mdev, ib_virt) && is_ib) {
+		change_mask = props->clr_port_cap_mask | props->set_port_cap_mask;
+		value = ~props->clr_port_cap_mask | props->set_port_cap_mask;
+		return set_port_caps_atomic(dev, port, change_mask, value);
+	}
 
 	mutex_lock(&dev->cap_mask_mutex);
 
-- 
cgit v1.2.3