Re: (usagi-users 00311) Re: USAGI IPv6 patches
On Sat, 17 Mar 2001, YOSHIFUJI Hideaki / [iso-2022-jp] 吉藤英明 wrote:
> In article <[🔎] 15026.31457.76487.253648@pizda.ninka.net> (at Fri, 16 Mar 2001 12:43:13 -0800 (PST)), "David S. Miller" <davem@redhat.com> says:
>
> > I really am furious that people have made such a stink about the USAGI
> > ipv6 patches, claiming that our ipv6 in the main kernel is "too broken
> > to use". I am going to take such USAGI ipv6 propaganda with a grain
> > of salt in the future.
>
> It, ipv6 code in linux is rather broken, is true,
> especially in addrconf/neighbor discovery logic.
>
> And, I do not think current Linux behavior of ipv4-mapped / bind(2) is
> (seriously) broken - it is RFC compliance (while linux has some typo /
> bugs; I'll send you later) - while we want to change it to another what
> is also RFC compliance (i,e, disabling v4-mapped is NOT acceptable).
> DOUBLE_BIND is such a thing but still in experimental (and it is an
> optionable feature in our code).
In my opinion, the work should be prioritized at this point.
I think you can assume that everyone who enables IPv6 at the moment is
looking for experimental code. Changing how it works if it's clearly
"right" shouldn't be avoided too much.
I'd like the most to get bind(2) operate properly. Practically, this is a
requirement if you want to run IPv4 and IPv6 services both on the same
box. I'm in for the OpenBSD approach -- brokenness and insecurities should
not be tolerated, or at least it should not be enabled by default.
If some tasks were to be pushed in ASAP, this would IMO definitely be one.
The whole IPv6 userspace pretty much hangs on it.
Personally, I can live with some bugs in source address selection,
neighour discovery etc. for a couple of months. But wanting to offer
(experimentally) both IPv4 and IPv6 services, this is a _requirement_.
I'm sure some, especially people who use BSD too, as I do, feel the same
way.
--
Pekka Savola "Tell me of difficulties surmounted,
Netcore Oy not those you stumble over and fall"
Systems. Networks. Security. -- Robert Jordan: A Crown of Swords
Reply to: