Linux-RDMA Archive on lore.kernel.org
 help / color / Atom feed
* [PATCH rdma-next 0/2] Two fixes to -next
@ 2021-04-18 13:49 Leon Romanovsky
  2021-04-18 13:49 ` [PATCH mlx5-next 1/2] IB/mlx5: Set right RoCE l3 type and roce version while deleting GID Leon Romanovsky
                   ` (2 more replies)
  0 siblings, 3 replies; 4+ messages in thread
From: Leon Romanovsky @ 2021-04-18 13:49 UTC (permalink / raw)
  To: Doug Ledford, Jason Gunthorpe
  Cc: Leon Romanovsky, Jakub Kicinski, linux-rdma, Maor Gottlieb,
	netdev, Parav Pandit, Saeed Mahameed

From: Leon Romanovsky <leonro@nvidia.com>

The two fixes are targeted to the -next. Maor's change fixes DM code
that was accepted in this cycle and Parav's change doesn't qualify
urgency of -rc8.

Thanks

Maor Gottlieb (1):
  RDMA/mlx5: Fix type assignment for ICM DM

Parav Pandit (1):
  IB/mlx5: Set right RoCE l3 type and roce version while deleting GID

 drivers/infiniband/hw/mlx5/dm.c               | 23 +++++++++++--------
 drivers/infiniband/hw/mlx5/main.c             |  8 +++----
 .../net/ethernet/mellanox/mlx5/core/lib/gid.c |  4 ++--
 3 files changed, 19 insertions(+), 16 deletions(-)

-- 
2.30.2


^ permalink raw reply	[flat|nested] 4+ messages in thread

* [PATCH mlx5-next 1/2] IB/mlx5: Set right RoCE l3 type and roce version while deleting GID
  2021-04-18 13:49 [PATCH rdma-next 0/2] Two fixes to -next Leon Romanovsky
@ 2021-04-18 13:49 ` Leon Romanovsky
  2021-04-18 13:49 ` [PATCH rdma-next 2/2] RDMA/mlx5: Fix type assignment for ICM DM Leon Romanovsky
  2021-04-20 19:10 ` [PATCH rdma-next 0/2] Two fixes to -next Jason Gunthorpe
  2 siblings, 0 replies; 4+ messages in thread
From: Leon Romanovsky @ 2021-04-18 13:49 UTC (permalink / raw)
  To: Doug Ledford, Jason Gunthorpe
  Cc: Parav Pandit, Jakub Kicinski, linux-rdma, netdev, Saeed Mahameed

From: Parav Pandit <parav@nvidia.com>

Currently when GID is deleted, it zero out all the fields of the RoCE
address in the SET_ROCE_ADDRESS command for a specified index.

roce_version = 0 means RoCEv1 in the SET_ROCE_ADDRESS command.

This assumes that device has RoCEv1 always enabled which is not always
correct. For example Subfunction does not support RoCEv1.

Due to this assumption a previously added RoCEv2 GID is always deleted as
RoCEv1 GID. This results in a below syndrome.

mlx5_core.sf mlx5_core.sf.4: mlx5_cmd_check:777:(pid 4256): SET_ROCE_ADDRESS(0x761) op_mod(0x0) failed, status bad parameter(0x3), syndrome (0x12822d)

Hence set the right RoCE version during GID deletion provided by the
core.

Signed-off-by: Parav Pandit <parav@nvidia.com>
Signed-off-by: Leon Romanovsky <leonro@nvidia.com>
---
 drivers/infiniband/hw/mlx5/main.c                 | 8 ++++----
 drivers/net/ethernet/mellanox/mlx5/core/lib/gid.c | 4 ++--
 2 files changed, 6 insertions(+), 6 deletions(-)

diff --git a/drivers/infiniband/hw/mlx5/main.c b/drivers/infiniband/hw/mlx5/main.c
index 7ea6137f8d12..6d1dd09a4388 100644
--- a/drivers/infiniband/hw/mlx5/main.c
+++ b/drivers/infiniband/hw/mlx5/main.c
@@ -555,15 +555,15 @@ static int set_roce_addr(struct mlx5_ib_dev *dev, u32 port_num,
 			 unsigned int index, const union ib_gid *gid,
 			 const struct ib_gid_attr *attr)
 {
-	enum ib_gid_type gid_type = IB_GID_TYPE_ROCE;
+	enum ib_gid_type gid_type;
 	u16 vlan_id = 0xffff;
 	u8 roce_version = 0;
 	u8 roce_l3_type = 0;
 	u8 mac[ETH_ALEN];
 	int ret;
 
+	gid_type = attr->gid_type;
 	if (gid) {
-		gid_type = attr->gid_type;
 		ret = rdma_read_gid_l2_fields(attr, &vlan_id, &mac[0]);
 		if (ret)
 			return ret;
@@ -575,7 +575,7 @@ static int set_roce_addr(struct mlx5_ib_dev *dev, u32 port_num,
 		break;
 	case IB_GID_TYPE_ROCE_UDP_ENCAP:
 		roce_version = MLX5_ROCE_VERSION_2;
-		if (ipv6_addr_v4mapped((void *)gid))
+		if (gid && ipv6_addr_v4mapped((void *)gid))
 			roce_l3_type = MLX5_ROCE_L3_TYPE_IPV4;
 		else
 			roce_l3_type = MLX5_ROCE_L3_TYPE_IPV6;
@@ -602,7 +602,7 @@ static int mlx5_ib_del_gid(const struct ib_gid_attr *attr,
 			   __always_unused void **context)
 {
 	return set_roce_addr(to_mdev(attr->device), attr->port_num,
-			     attr->index, NULL, NULL);
+			     attr->index, NULL, attr);
 }
 
 __be16 mlx5_get_roce_udp_sport_min(const struct mlx5_ib_dev *dev,
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/lib/gid.c b/drivers/net/ethernet/mellanox/mlx5/core/lib/gid.c
index a68738c8f4bc..215fa8bdce48 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/lib/gid.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/lib/gid.c
@@ -142,10 +142,10 @@ int mlx5_core_roce_gid_set(struct mlx5_core_dev *dev, unsigned int index,
 		}
 
 		ether_addr_copy(addr_mac, mac);
-		MLX5_SET_RA(in_addr, roce_version, roce_version);
-		MLX5_SET_RA(in_addr, roce_l3_type, roce_l3_type);
 		memcpy(addr_l3_addr, gid, gidsz);
 	}
+	MLX5_SET_RA(in_addr, roce_version, roce_version);
+	MLX5_SET_RA(in_addr, roce_l3_type, roce_l3_type);
 
 	if (MLX5_CAP_GEN(dev, num_vhca_ports) > 0)
 		MLX5_SET(set_roce_address_in, in, vhca_port_num, port_num);
-- 
2.30.2


^ permalink raw reply	[flat|nested] 4+ messages in thread

* [PATCH rdma-next 2/2] RDMA/mlx5: Fix type assignment for ICM DM
  2021-04-18 13:49 [PATCH rdma-next 0/2] Two fixes to -next Leon Romanovsky
  2021-04-18 13:49 ` [PATCH mlx5-next 1/2] IB/mlx5: Set right RoCE l3 type and roce version while deleting GID Leon Romanovsky
@ 2021-04-18 13:49 ` Leon Romanovsky
  2021-04-20 19:10 ` [PATCH rdma-next 0/2] Two fixes to -next Jason Gunthorpe
  2 siblings, 0 replies; 4+ messages in thread
From: Leon Romanovsky @ 2021-04-18 13:49 UTC (permalink / raw)
  To: Doug Ledford, Jason Gunthorpe; +Cc: Maor Gottlieb, linux-rdma

From: Maor Gottlieb <maorg@nvidia.com>

We should hold the UAPI DM type in the base struct and not
the internal mlx5 type. Fix it.

Fixes: 9905fb65119f ("RDMA/mlx5: Re-organize the DM code")
Signed-off-by: Maor Gottlieb <maorg@nvidia.com>
Signed-off-by: Leon Romanovsky <leonro@nvidia.com>
---
 drivers/infiniband/hw/mlx5/dm.c | 23 +++++++++++++----------
 1 file changed, 13 insertions(+), 10 deletions(-)

diff --git a/drivers/infiniband/hw/mlx5/dm.c b/drivers/infiniband/hw/mlx5/dm.c
index 235aad6beacb..094bf85589db 100644
--- a/drivers/infiniband/hw/mlx5/dm.c
+++ b/drivers/infiniband/hw/mlx5/dm.c
@@ -331,12 +331,20 @@ static struct ib_dm *handle_alloc_dm_memic(struct ib_ucontext *ctx,
 	return ERR_PTR(err);
 }
 
+static enum mlx5_sw_icm_type get_icm_type(int uapi_type)
+{
+	return uapi_type == MLX5_IB_UAPI_DM_TYPE_STEERING_SW_ICM ?
+		       MLX5_SW_ICM_TYPE_STEERING :
+		       MLX5_SW_ICM_TYPE_HEADER_MODIFY;
+}
+
 static struct ib_dm *handle_alloc_dm_sw_icm(struct ib_ucontext *ctx,
 					    struct ib_dm_alloc_attr *attr,
 					    struct uverbs_attr_bundle *attrs,
 					    int type)
 {
 	struct mlx5_core_dev *dev = to_mdev(ctx->device)->mdev;
+	enum mlx5_sw_icm_type icm_type = get_icm_type(type);
 	struct mlx5_ib_dm_icm *dm;
 	u64 act_size;
 	int err;
@@ -368,7 +376,7 @@ static struct ib_dm *handle_alloc_dm_sw_icm(struct ib_ucontext *ctx,
 	act_size = roundup_pow_of_two(act_size);
 
 	dm->base.size = act_size;
-	err = mlx5_dm_sw_icm_alloc(dev, type, act_size, attr->alignment,
+	err = mlx5_dm_sw_icm_alloc(dev, icm_type, act_size, attr->alignment,
 				   to_mucontext(ctx)->devx_uid,
 				   &dm->base.dev_addr, &dm->obj_id);
 	if (err)
@@ -377,7 +385,7 @@ static struct ib_dm *handle_alloc_dm_sw_icm(struct ib_ucontext *ctx,
 	err = uverbs_copy_to(attrs, MLX5_IB_ATTR_ALLOC_DM_RESP_START_OFFSET,
 			     &dm->base.dev_addr, sizeof(dm->base.dev_addr));
 	if (err) {
-		mlx5_dm_sw_icm_dealloc(dev, type, dm->base.size,
+		mlx5_dm_sw_icm_dealloc(dev, icm_type, dm->base.size,
 				       to_mucontext(ctx)->devx_uid,
 				       dm->base.dev_addr, dm->obj_id);
 		goto free;
@@ -409,11 +417,9 @@ struct ib_dm *mlx5_ib_alloc_dm(struct ib_device *ibdev,
 	case MLX5_IB_UAPI_DM_TYPE_MEMIC:
 		return handle_alloc_dm_memic(context, attr, attrs);
 	case MLX5_IB_UAPI_DM_TYPE_STEERING_SW_ICM:
-		return handle_alloc_dm_sw_icm(context, attr, attrs,
-					     MLX5_SW_ICM_TYPE_STEERING);
+		return handle_alloc_dm_sw_icm(context, attr, attrs, type);
 	case MLX5_IB_UAPI_DM_TYPE_HEADER_MODIFY_SW_ICM:
-		return handle_alloc_dm_sw_icm(context, attr, attrs,
-					      MLX5_SW_ICM_TYPE_HEADER_MODIFY);
+		return handle_alloc_dm_sw_icm(context, attr, attrs, type);
 	default:
 		return ERR_PTR(-EOPNOTSUPP);
 	}
@@ -441,10 +447,7 @@ static void mlx5_dm_memic_dealloc(struct mlx5_ib_dm_memic *dm)
 static int mlx5_dm_icm_dealloc(struct mlx5_ib_ucontext *ctx,
 			       struct mlx5_ib_dm_icm *dm)
 {
-	enum mlx5_sw_icm_type type =
-		dm->base.type == MLX5_IB_UAPI_DM_TYPE_STEERING_SW_ICM ?
-			MLX5_SW_ICM_TYPE_STEERING :
-			MLX5_SW_ICM_TYPE_HEADER_MODIFY;
+	enum mlx5_sw_icm_type type = get_icm_type(dm->base.type);
 	struct mlx5_core_dev *dev = to_mdev(dm->base.ibdm.device)->mdev;
 	int err;
 
-- 
2.30.2


^ permalink raw reply	[flat|nested] 4+ messages in thread

* Re: [PATCH rdma-next 0/2] Two fixes to -next
  2021-04-18 13:49 [PATCH rdma-next 0/2] Two fixes to -next Leon Romanovsky
  2021-04-18 13:49 ` [PATCH mlx5-next 1/2] IB/mlx5: Set right RoCE l3 type and roce version while deleting GID Leon Romanovsky
  2021-04-18 13:49 ` [PATCH rdma-next 2/2] RDMA/mlx5: Fix type assignment for ICM DM Leon Romanovsky
@ 2021-04-20 19:10 ` Jason Gunthorpe
  2 siblings, 0 replies; 4+ messages in thread
From: Jason Gunthorpe @ 2021-04-20 19:10 UTC (permalink / raw)
  To: Leon Romanovsky
  Cc: Doug Ledford, Leon Romanovsky, Jakub Kicinski, linux-rdma,
	Maor Gottlieb, netdev, Parav Pandit, Saeed Mahameed

On Sun, Apr 18, 2021 at 04:49:38PM +0300, Leon Romanovsky wrote:
> From: Leon Romanovsky <leonro@nvidia.com>
> 
> The two fixes are targeted to the -next. Maor's change fixes DM code
> that was accepted in this cycle and Parav's change doesn't qualify
> urgency of -rc8.
> 
> Thanks
> 
> Maor Gottlieb (1):
>   RDMA/mlx5: Fix type assignment for ICM DM
> 
> Parav Pandit (1):
>   IB/mlx5: Set right RoCE l3 type and roce version while deleting GID

Applied to for-next, thanks

Jason

^ permalink raw reply	[flat|nested] 4+ messages in thread

end of thread, back to index

Thread overview: 4+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2021-04-18 13:49 [PATCH rdma-next 0/2] Two fixes to -next Leon Romanovsky
2021-04-18 13:49 ` [PATCH mlx5-next 1/2] IB/mlx5: Set right RoCE l3 type and roce version while deleting GID Leon Romanovsky
2021-04-18 13:49 ` [PATCH rdma-next 2/2] RDMA/mlx5: Fix type assignment for ICM DM Leon Romanovsky
2021-04-20 19:10 ` [PATCH rdma-next 0/2] Two fixes to -next Jason Gunthorpe

Linux-RDMA Archive on lore.kernel.org

Archives are clonable:
	git clone --mirror https://lore.kernel.org/linux-rdma/0 linux-rdma/git/0.git

	# If you have public-inbox 1.1+ installed, you may
	# initialize and index your mirror using the following commands:
	public-inbox-init -V2 linux-rdma linux-rdma/ https://lore.kernel.org/linux-rdma \
		linux-rdma@vger.kernel.org
	public-inbox-index linux-rdma

Example config snippet for mirrors

Newsgroup available over NNTP:
	nntp://nntp.lore.kernel.org/org.kernel.vger.linux-rdma


AGPL code for this site: git clone https://public-inbox.org/public-inbox.git