> > > If it's the latter, and you can't somehow avoid "accidentally" > > > sending on the wrong interface, then I agree with the change for > > > nce_cmp_ll_addr -- you need to check all of the interfaces. > > > > I wonder if there's still a risk that we might erroneously think it's a > > duplicate if the nominated multicast interface changes between the time we > > send the packet and the time we receive it. Maybe we're still best off > > checking across the group. > > I suppose that's possible in theory. It'd take some careful timing to > get it to happen.
Yes, but the resulting message is known to be a call generator, so I'm reluctant to leave any known edge cases that lead to it spuriously being output. So, I'm slightly inclined to just add the dl_mp == NULL check, and leave the checks I added in place. Any objection? -- meem
