On Wed, Feb 08, 2017 at 12:43:40AM +0000, Parav Pandit wrote:
@@ -2096,6 +2114,15 @@ struct ib_device { struct ib_rwq_ind_table_init_attr *init_attr, struct ib_udata *udata); int (*destroy_rwq_ind_table)(struct ib_rwq_ind_table *wq_ind_table); + /* rdma netdev operations */ + struct net_device *(*alloc_rdma_netdev)( + struct ib_device *device, + u8 port_num, + enum rdma_netdev_t type, + const char *name, + unsigned char name_assign_type, + void (*setup)(struct net_device *)); + void (*free_rdma_netdev)(struct net_device *netdev); struct ib_dma_mapping_ops *dma_ops; struct module *owner;As its clear from the cover letter and from the request to place this in drivers/infiniband/ulp, Instead of increasing the ib_dev structure further, Can you change the code to make use of ib_register_client() and friend functions to register vnic as ULP. (similar to other ULP such as uverbs, srp, ipoib). This will also allow you get to get notified for removing the vnic device when underlying rdma device gets removed. Based on the property that gets exposed by the ibdev, vnic driver filters whether it needs to load its vnic to specific device or not. This way modules are isolated between core and ULP little better. Would it work for you?
HFI_VNIC driver is using ib_register_client() and friend fucntions. Below patch in this series does that.
[RFC v3 08/11] IB/hfi-vnic: VNIC Ethernet Management Agent (VEMA) function Niranjana
-- To unsubscribe from this list: send the line "unsubscribe linux-rdma" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
