[OpenWrt-Devel] Comprehensive ipv4 and ipv6 unaligned access, patch for ar71xx

David Woodhouse dwmw2 at infradead.org
Mon Apr 23 22:49:47 CEST 2012

On Mon, 2012-04-23 at 22:32 +0200, Gert Doering wrote:
> On Mon, Apr 23, 2012 at 07:43:36PM +0100, David Woodhouse wrote:
> > It's documented here: http://wiki.openwrt.org/doc/uci/dhcp6c
> Oh!  This is tremendously cool!  Thanks to everyone who made that
> possible.
> Nit to pick: calling this "sla_len" is a bit weird.  Especially if
> you don't know whether the ISP will assign a /48, /56 or /60, this
> sounds a bit awkward to ensure /64 on an interface - or am I
> misreading this?

No, I think you have it right. I've just routed individual /64s to my
ADSL lines before, and I see just one of them advertised in DHCPv6 PD.
Not quite sure why they aren't all listed.

I added a fresh /60 to my line and saw it advertised, but then I had to
tweak the the sla_len to match before anything more would work.

It doesn't make sense - we ought to be able to say "give the first /64
to the wired lan", "give the second to wireless", etc. Then if we get
separate /64s or a larger delegation it should just dish it out
accordingly. We shouldn't have to calculate and preconfigure sla_len.

> > We should make sure it's enabled by default.
> That, and IPv6 for PPP (for PPP-enabled WAN interfaces).

Definitely. There's almost no chance that trying IPv6CP would cause a
problem; you'll just get a ProtRej from 20th century servers. It's just
the same as CCP with servers that don't want it.

-------------- next part --------------
A non-text attachment was scrubbed...
Name: smime.p7s
Type: application/x-pkcs7-signature
Size: 5818 bytes
Desc: not available
URL: <http://lists.openwrt.org/pipermail/openwrt-devel/attachments/20120423/31eab102/attachment.bin>

More information about the openwrt-devel mailing list