Mon, Apr 20, 2020 at 09:02:58PM CEST, dsahern@xxxxxxxxx wrote: >On 4/20/20 12:56 PM, Maor Gottlieb wrote: >> >> On 4/20/2020 9:48 PM, Jiri Pirko wrote: >>> Mon, Apr 20, 2020 at 08:04:01PM CEST, dsahern@xxxxxxxxx wrote: >>>> On 4/20/20 12:01 PM, Jiri Pirko wrote: >>>>> Generic ndo with lag-specific arg? Odd. Plus, there is a small chance >>>>> this is ever going to be used for other master. And if so, could be >>>>> very >>>>> easily renamed then... >>>> core code should be generic, not specific and renamed at a later date >>>> when a second use case arises. >>> Yeah, I guess we just have to agree to disagree :) >> >> So I am remaining with the flags. Any suggestion for better name for the >> enum? Should I move master_xmit_get_slave from lag.h to netdevice.h? >>> > >IMHO, yes, that is a better place. > >generic ndo name and implementation. >type specific flag as needed. > >This is consistent with net_device and ndo - both generic concepts - >with specifics relegated to flags (e.g., IFF_*) Why there is need for flags? Why a single bool can't do as I suggested? Do you see any usecase for another flag?
