On Mon, Apr 14, 2014 at 8:19 PM, Thomas Graf <tg...@redhat.com> wrote: > On 04/14/2014 08:54 AM, Pravin Shelar wrote: >> >> On Wed, Apr 9, 2014 at 4:05 PM, Thomas Graf <tg...@redhat.com> wrote: >>> >>> I believe skb_get_hash() can be 0 and that would result in mistaking any >>> empty slot to be a cached entry and we would only look at the first flow >>> mask. >> >> >> software skb_get_hash() returns zero for error case which should be >> rare. Thats why I avoided special check for zero hash for every cache >> lookup. > > > I agree but in those rare cases the lookup will be incorrect. With > an appropriate unlikely() hint to the compiler the additional cost is > minimal.
I am not sure why lookup would be incorrect. If cached mask is not correct then lookup fails and it does fallback to complete mask list search anyways. _______________________________________________ dev mailing list dev@openvswitch.org http://openvswitch.org/mailman/listinfo/dev
