Hi Flavio, Just one correction I noticed while implementing your feedback. I've replied inline.
Thanks, Cian > -----Original Message----- > From: Ferriter, Cian > Sent: Thursday 24 June 2021 12:54 > To: Flavio Leitner <[email protected]> > Cc: [email protected]; [email protected] > Subject: RE: [ovs-dev] [v13 05/12] dpif-netdev: Add command to switch dpif > implementation. > > Hi Flavio, > > Thanks for the review. My responses are inline. > > Cian > > > -----Original Message----- > > From: Flavio Leitner <[email protected]> > > Sent: Wednesday 23 June 2021 19:18 > > To: Ferriter, Cian <[email protected]> > > Cc: [email protected]; [email protected] > > Subject: Re: [ovs-dev] [v13 05/12] dpif-netdev: Add command to switch dpif > > implementation. > > > > On Thu, Jun 17, 2021 at 05:18:18PM +0100, Cian Ferriter wrote: > > > From: Harry van Haaren <[email protected]> > > > <snip most of the diff> > > > + > > > + if (!avx512f_available || !bmi2_available) { > > > + return -ENOTSUP; > > > > Also the callers of the probe function only cares about true or > > false, so there is no need to return errno here, then the new > > include can also be removed. > > > > I'll simplify the return and remove the errno include. > This return -ENOTSUP is used in dpif_netdev_impl_set() to select the correct error message to show the user on error. static const char *error_description[2] = { "Unknown DPIF implementation", "CPU doesn't support the required instruction for", }; I'm going to leave in. I hope that makes sense. > > > + } > > > + > > > + return 0; > > > +} > > > + > > > int32_t > > > dp_netdev_input_outer_avx512(struct dp_netdev_pmd_thread *pmd, > > > struct dp_packet_batch *packets, <snip the rest of the diff> _______________________________________________ dev mailing list [email protected] https://mail.openvswitch.org/mailman/listinfo/ovs-dev
