> -----Original Message-----
> From: linux-rdma-owner@xxxxxxxxxxxxxxx <linux-rdma-
> owner@xxxxxxxxxxxxxxx> On Behalf Of Leon Romanovsky
> Sent: Tuesday, April 2, 2019 8:50 AM
> To: Doug Ledford <dledford@xxxxxxxxxx>; Jason Gunthorpe
> <jgg@xxxxxxxxxxxx>
> Cc: Leon Romanovsky <leonro@xxxxxxxxxxxx>; RDMA mailing list <linux-
> rdma@xxxxxxxxxxxxxxx>
> Subject: [PATCH rdma-next v2] RDMA/nldev: Return device protocol
>
> From: Leon Romanovsky <leonro@xxxxxxxxxxxx>
>
> Reuse existing RDMA_NLDEV_ATTR_LINK_TYPE to give ability for stable
> names UDEV rule create Ib device stable names based on link type protocol.
> The assumption that devices like mlx4 with duality in their link type under
> one IB device struct won't be allowed in the future.
>
> Signed-off-by: Leon Romanovsky <leonro@xxxxxxxxxxxx>
> ---
> v1 -> v2:
> * Created special netlink attribute to return device protocol.
> v0 -> v1:
> * Added special return type for RXE to return same information as it
> was provided by link_set while RXE link is created.
> ---
> drivers/infiniband/core/nldev.c | 26 ++++++++++++++++++++++++++
> include/uapi/rdma/rdma_netlink.h | 4 ++++
> 2 files changed, 30 insertions(+)
>
> diff --git a/drivers/infiniband/core/nldev.c b/drivers/infiniband/core/nldev.c
> index 854f47d78d8e..1cd599a098f6 100644
> --- a/drivers/infiniband/core/nldev.c
> +++ b/drivers/infiniband/core/nldev.c
> @@ -118,6 +118,8 @@ static const struct nla_policy
> nldev_policy[RDMA_NLDEV_ATTR_MAX] = {
> .len = RDMA_NLDEV_ATTR_ENTRY_STRLEN
> },
> [RDMA_NLDEV_SYS_ATTR_NETNS_MODE] = { .type = NLA_U8 },
> [RDMA_NLDEV_NET_NS_FD] = { .type = NLA_U32 },
> + [RDMA_NLDEV_ATTR_DEV_PROTOCOL] = { .type =
> NLA_NUL_STRING,
> + .len = RDMA_NLDEV_ATTR_ENTRY_STRLEN
> },
> };
>
> static int put_driver_name_print_type(struct sk_buff *msg, const char
> *name, @@ -200,6 +202,7 @@ static int fill_nldev_handle(struct sk_buff
> *msg, struct ib_device *device) static int fill_dev_info(struct sk_buff *msg,
> struct ib_device *device) {
> char fw[IB_FW_VERSION_NAME_MAX];
> + u8 port;
>
> if (fill_nldev_handle(msg, device))
> return -EMSGSIZE;
> @@ -228,6 +231,29 @@ static int fill_dev_info(struct sk_buff *msg, struct
> ib_device *device)
> return -EMSGSIZE;
> if (nla_put_u8(msg, RDMA_NLDEV_ATTR_DEV_NODE_TYPE, device-
> >node_type))
> return -EMSGSIZE;
> +
> + /*
> + * Link type is determined on first port and mlx4 device
> + * which can potentialy have two different link type for the same
> + * IB device is considered as better to be avoided in the future,
> + */
> + port = rdma_start_port(device);
> + if (rdma_protocol_ib(device, port) &&
> + nla_put_string(msg, RDMA_NLDEV_ATTR_DEV_PROTOCOL, "ib"))
> + return -EMSGSIZE;
> + if (rdma_protocol_iwarp(device, port) &&
> + nla_put_string(msg, RDMA_NLDEV_ATTR_DEV_PROTOCOL, "iw"))
> + return -EMSGSIZE;
> + if (rdma_protocol_roce(device, port) &&
> + nla_put_string(msg, RDMA_NLDEV_ATTR_DEV_PROTOCOL, "roce"))
> + return -EMSGSIZE;
> + if (rdma_protocol_usnic(device, port) &&
> + nla_put_string(msg, RDMA_NLDEV_ATTR_DEV_PROTOCOL,
> "usnic"))
> + return -EMSGSIZE;
> + if (rdma_cap_opa_mad(device, port) &&
> + nla_put_string(msg, RDMA_NLDEV_ATTR_DEV_PROTOCOL, "opa"))
> + return -EMSGSIZE;
> +
OPA definition is,
#define RDMA_CORE_PORT_INTEL_OPA (RDMA_CORE_PORT_IBA_IB \
| RDMA_CORE_CAP_OPA_MAD)
So for OPA two attributes will be set, first IB followed by OPA.
I am not sure, it should be just 'ib'. CAP should be treated differently?
> return 0;
> }
>
> diff --git a/include/uapi/rdma/rdma_netlink.h
> b/include/uapi/rdma/rdma_netlink.h
> index 09b0ad8ea9d4..d4a5b7bcd51a 100644
> --- a/include/uapi/rdma/rdma_netlink.h
> +++ b/include/uapi/rdma/rdma_netlink.h
> @@ -484,6 +484,10 @@ enum rdma_nldev_attr {
> * File descriptor handle of the net namespace object
> */
> RDMA_NLDEV_NET_NS_FD, /* u32 */
> + /*
> + * Device protocol, e.g. ib, iw, usnic, roce and opa
> + */
> + RDMA_NLDEV_ATTR_DEV_PROTOCOL, /* string */
>
Thanks. This is what I was asking initially to differentiate two properties.
> /*
> * Always the end
> --
> 2.20.1
