On Wed, Jun 13, 2018 at 09:03:16AM +0200, Hannes Reinecke wrote: > On Wed, 13 Jun 2018 09:16:56 +0300 > Moni Shoua <monis@xxxxxxxxxxxx> wrote: > > > > > > > Good point. > > > (I hate the sysfs interface anyway.) > > > Will be looking into it. > > > > > Sure, go ahead but remember that bonding driver still uses the sysfs > > interface and probably for a good reason. So, consider this when you > > make plans. > > > > ??? > Not sure what this got to do with it... > > Anyway, current plan is to hook into iproute2 'rdma' tool; we already > have: > > ./rdma/rdma link show > 1/1: rxe0/1: state ACTIVE physical_state LINK_UP netdev eth1 > 2/1: rxe1/1: state ACTIVE physical_state LINK_UP netdev eth2 > > Plan is to add an interface > > rdma link add <dev/port> netdev <ifname> > rdma link del <dev/port> netdev <ifname> It should be modeled after 'ip link add' rdma link add link NET_DEV # The physical ethernet device to create rxe on name NAME # Name of the RDMA device, port is always 1 index INDEX # RDMA device INDEX # (optional type rxe # Specify this is a RXE device (vs SIW someday) <any future rxe specific options> Maybe call 'link' 'netdev' for rdma-tool, but the other options are needed > to cover the functionality of the current sysfs interface. > D'accord? Yes please, Bernard has wanted this for soft iwarp as well. I would then want to deprecate the module parmater very quickly. The sysfs can stick around, but maybe could print a deprecation warning. Also, 'type rxe' should trigger module auto-loading of the rxe module. Thanks, Jason -- 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
