On 23/05/2019 18:25, Jakub Kicinski wrote: > Whether it's on you to fix this is debatable :) Since you're diving > into actions and adding support for shared ones, I'd say it's time to > rectify the situation. > > Let's look at it this way - if you fix the RTM_GETACTION you will > necessarily add the cookie and all the other stuff you need in your > upcoming driver :) Yes, but then I don't have an upstream user for RTM_GETACTION offload! I'd need to teach an existing driver to look up counters by action cookie, which would probably mean a hashtable, a bunch of locking and lifetime rules... not something I'm comfortable doing in someone else's driver that I don't know the innards of and probably don't have hardware to test on.
I'll put together an RFC patch, anyway, and maybe some other driver maintainer (hint, hint!) will follow up with a user ;-) Should it be a new entry in enum tc_setup_type (TC_SETUP_ACTION), or in enum tc_fl_command (TC_CLSFLOWER_GETACTION)? I'd imagine the former as we want this to (a) be generic beyond just flower and perhaps also (b) support creating and deleting actions as well (not just dumping them). -Ed
