Re: [lwip-devel] IPv6 L2TP PPP review

[Ivan Delamer]

Hi Sylvain,

> By the way, I struggled a little bit setting UP a static IPv6 global
> scope address and ended up with the following which I am not very 
> happy
> with:
>
> netif.ip6_addr[1].addr[0] = PP_HTONL(0x20010000);
> netif.ip6_addr[1].addr[1] = PP_HTONL(0x00000000);
> netif.ip6_addr[1].addr[2] = PP_HTONL(0x00000000);
> netif.ip6_addr[1].addr[3] = PP_HTONL(0x00000002);
> netif_ip6_addr_set_state(&netif, 1, 
> IP6_ADDR_TENTATIVE|IP6_ADDR_PREFERRED);
> netif_create_ip6_linklocal_address(&netif, 1);
>
>
> Using IP6_ADDR_VALID instead of IP6_ADDR_TENTATIVE didn't work and I
> don't know why from looking at the source code, do you have a clue ?

The initial state you should set for your address depends on whether 
you have DAD enabled or not.

If DAD is enabled (usual case), you should set it to 
IP6_ADDR_TENTATIVE.

If DAD is disabled, you should set it to IP6_ADDR_PREFERRED.

I think that in a PPP link DAD is actually not required, and probably 
wouldn't work because it requires multicast support. So I would set the 
state to IP6_ADDR_PREFERRED and it should work.

I'm not sure why IP6_ADDR_VALID wouldn't work, although generally you 
want it to be either PREFERRED or DEPRECATED, which is VALID with an 
extra flag.

What are you seeing that makes you say "it didn't work"?

Cheers
Ivan


> Date: Mon, 2 Mar 2015 21:35:00 +0100
> From: Sylvain Rochet <address@hidden>
> To: lwip-devel <address@hidden>
> Subject: Re: [lwip-devel] IPv6 L2TP PPP review
> Message-ID: <address@hidden>
> Content-Type: text/plain; charset="utf-8"
>
> Hello Ivan,
>
>
> On Mon, Mar 02, 2015 at 01:16:55PM -0700, Ivan Delamer wrote:
> > Hello Sylvain,
> >
> > I think your code is correct. I use something very similar for UDP
> > connections, only using netconns.
>
> Thank you :)
>
>
> > In general once you are working with UDP or TCP, it is really quite
> > similar for IPv4 or IPv6, because you are one layer over that. It is
> > just address binding/comparison that needs to be done appropriately.
> >
> > The API is not completely safe and requires care from the programmer.
>
> I saw that, there is casts everywhere meaning functions 
> calls/prototypes
> coherency is not checked at all.
>
>
> > I think we have it as a task to improve ipX_addr_t to include a
> > version field. We just wanted to initially implement IPv6 in a way
> > that was easy to turn it off, and the resulting stack was exactly
> > the same as the old IPv4-only stack. We will have to improve it
> > though...
>
> I agree it is a bit disappointing, this is why I wasn't sure, I am 
> more
> used to sockaddr, sockaddr_in, sockaddr_in6, sockaddr_storage, 
> "russian
> dolls" structures.
>
>
> By the way, I struggled a little bit setting UP a static IPv6 global
> scope address and ended up with the following which I am not very 
> happy
> with:
>
> netif.ip6_addr[1].addr[0] = PP_HTONL(0x20010000);
> netif.ip6_addr[1].addr[1] = PP_HTONL(0x00000000);
> netif.ip6_addr[1].addr[2] = PP_HTONL(0x00000000);
> netif.ip6_addr[1].addr[3] = PP_HTONL(0x00000002);
> netif_ip6_addr_set_state(&netif, 1, 
> IP6_ADDR_TENTATIVE|IP6_ADDR_PREFERRED);
> netif_create_ip6_linklocal_address(&netif, 1);
>
>
> Using IP6_ADDR_VALID instead of IP6_ADDR_TENTATIVE didn't work and I
> don't know why from looking at the source code, do you have a clue ?
>
>
> Cheers,
> Sylvain