> On 25 Jan 2021, at 10:43, Alexandr Nedvedicky
> <[email protected]> wrote:
>
> hello,
>
> On Fri, Jan 22, 2021 at 05:32:47PM +1000, David Gwynne wrote:
>> I tried this diff, and it broke the ability to use dynamic addresses.
>> ie, the following rules should work:
>>
>> pass in on gre52 inet proto icmp route-to (gre49:peer)
>> pass in on vmx0 inet proto icmp route-to (gre:peer)
>
> I see, I did not know those should work.
We are suffering a bit from not having a high level design :/
>>
>> however, other forms of dynamic interface addresses should fail. or do
>> we want to support route-to if0:broadcast too?
>
> I can't think of any valid reason why 'ifp0:broadcast' should work. this
> seems to be poor hack to work around some awkward glitch. I would prefer
> to disable this option now. We can always add it later, when we will
> understand the true purpose.
Agreed.
Cheers,
dlg
>
>
> thanks and
> regards
> sashan
