2014-08-28 11:30, Ananyev, Konstantin: > From: Thomas Monjalon > > 2014-08-28 09:01, Wu, Jingjing: > > > I have another question: > > > If we use the way 'rx_classification_filter_ctl' works, the specific > > > structures defined in rte_i40e.h will be visible in user's application, > > > such as testpmd. > > > I know I shouldn't make commands linked with i40e like what I did before. > > > But will the i40e specific structures become visible be acceptable? > > > > I think testpmd should be limited to generic API. > > So it wouldn't be acceptable to be dependent of i40e files. > > But having some specific i40e tests in examples or app/test is OK. > > > > Probably I didn't get you right:
Indeed ;) > Are you suggesting to have a new clone of testpmd for any new device > we are going to support? No. I say there shouldn't be any PMD dependency on testpmd. It means we should use only generic API. > That seems like too much hassle to me. > Plus what to do if someone would like to test configuration with two > different devices involved: ixgbe and i40e for example? ixgbe and i40e features should use the same generic API for flow director. > I suggest we keep one testpmd for all devices we support. > Of course we'll probably have to make some rework to avoid > if (strncmp(drv_name, "xxx") spread all over it. > We need to find some better way to discover/setup HW specific features. Agreed -- Thomas
