On 07/09/18 03:27, Eric Dumazet wrote: > On 09/06/2018 07:26 AM, Edward Cree wrote: >> Signed-off-by: Edward Cree <ec...@solarflare.com> > Lack of changelog here ? > > I do not know what is a good packet. The comment on netif_receive_skb_list() defines this as "skbs for which netif_receive_skb() would have returned %NET_RX_SUCCESS". But I shall put that into the changelog as well.
> You are adding a lot of conditional expressions, that cpu > will mispredict quite often. I don't see an alternative, since this is needed by patch #4 and I daresay there are other drivers that will also want to get NET_RX_SUCCESS-like information (possibly from regular receives as well as GRO; I'm not quite sure why sfc only cares about the latter). > Typical micro benchmarks wont really notice. Any suggestions on how to construct a test that will?
