On Fri, Sep 15, 2023 at 02:37:57PM -0400, Kamal Heib wrote:
> This patch adds support for reporting the ipoib net device for a given
> RDMA device by calling ib_get_net_dev_by_params() when filling the
> port's info.
>
> $ rdma link show mlx5_0/1
> link mlx5_0/1 subnet_prefix fe80:0000:0000:0000 lid 66 sm_lid 3 lmc 0
> state ACTIVE physical_state LINK_UP netdev ibp196s0f0
>
> Signed-off-by: Kamal Heib <kheib@xxxxxxxxxx>
> ---
> v1: Check namespace and query pkey.
> ---
> drivers/infiniband/core/nldev.c | 27 ++++++++++++++++++++++++++-
> 1 file changed, 26 insertions(+), 1 deletion(-)
>
> diff --git a/drivers/infiniband/core/nldev.c b/drivers/infiniband/core/nldev.c
> index d5d3e4f0de77..f3fa8143cdc7 100644
> --- a/drivers/infiniband/core/nldev.c
> +++ b/drivers/infiniband/core/nldev.c
> @@ -308,10 +308,12 @@ static int fill_port_info(struct sk_buff *msg,
> struct ib_device *device, u32 port,
> const struct net *net)
> {
> + struct net_device *ipoib_netdev = NULL;
> struct net_device *netdev = NULL;
> struct ib_port_attr attr;
> - int ret;
> u64 cap_flags = 0;
> + u16 pkey;
> + int ret;
>
> if (fill_nldev_handle(msg, device))
> return -EMSGSIZE;
> @@ -340,6 +342,26 @@ static int fill_port_info(struct sk_buff *msg,
> return -EMSGSIZE;
> if (nla_put_u8(msg, RDMA_NLDEV_ATTR_LMC, attr.lmc))
> return -EMSGSIZE;
> +
> + ret = ib_query_pkey(device, port, 0, &pkey);
> + if (ret)
> + goto out;
That doesn't really make any sense..
The whole idea there is only one ipoib device doesn't even make sense.
I still don't like this at all.
> +
> + ipoib_netdev = ib_get_net_dev_by_params(device, port,
> + pkey,
> + NULL, NULL);
> + if (ipoib_netdev && net_eq(dev_net(ipoib_netdev), net)) {
> + ret = nla_put_u32(msg,
> + RDMA_NLDEV_ATTR_NDEV_INDEX,
> + ipoib_netdev->ifindex);
> + if (ret)
> + goto out;
> + ret = nla_put_string(msg,
> + RDMA_NLDEV_ATTR_NDEV_NAME,
> + ipoib_netdev->name);
> + if (ret)
> + goto out;
and I don't think the existing netlink attributes should be redefined
to be about ipoib
Jason
