On Thu, Apr 15, 2021 at 10:44 AM Walsh, Conor <conor.wa...@intel.com> wrote:
>
> <snip>
>
> > >
> > > The IPv6 subnets used were not within the 2001:200::/48 subnet
> >
> > {{32, 1, 2, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0}, 48, 0},
> > {{32, 1, 2, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0}, 48, 1},
> > etc...
> >
> > Err... all those subnets were exactly == 2001:200::/48.
> > Is the issue that the example needs *separate* subnets taken from
> > within 2001:200::/48 ?
>
> Hi David,
> As it is ATM any IP with that range will match for all the rules for the 8
> ports and the application cannot differentiate between them and it just
> returns the traffic to sender.
> After this change the ports can be individually matched using the smaller /64
> ranges for each port which are still within the given /48 subnet range
> (RFC5180).
Ok, thanks for confirming, could you update the commitlog to make it clear?
>
> >
> >
> > > Changed to 2001:200:0:{0-7}::/64 where 0-7 is the port ID
> > >
> > > Fixes: 37afe381bde4 ("examples/l3fwd: use reserved IP addresses")
> >
> > And this looks like a Cc: sta...@dpdk.org candidate.
>
> I can send a v8 and CC stable if you think that’s needed.
Yes please.
I was waiting to look at the rest of the series, I have some comments
that will require a v8 (from my pov).
--
David Marchand
