(usagi-users 00438) Re: RFC3041: privacy extenstion for global IPv6 addresses, plans for implementing?

[Peter Bieringer <pb@xxxxxxxxxxxx>]

At 03:03 27.04.2001, YOSHIFUJI Hideaki / 
=?iso-2022-jp?B?GyRCNUhGIzFRTEAbKEI=?= wrote:
> In article <5.1.0.14.0.20010426232100.00b03de8@xxxxxxxxxxxxxxxxx> (at Thu, 
> 26 Apr 2001 23:26:58 +0200), Peter Bieringer <pb@xxxxxxxxxxxx> says:
>
> > are there plans at USAGI to implement (optional) RFC3041 for aggregatable
> > global unicast addresses? E.g. via kernel compile option and proc 
> controllable?
>
> Personally, I'd like to implement it soon... after finishing backport ...

Great.

> Anyway, if you have a plan, of course, you can write it for us. :-)

Hmm, don't believe that I have the experience to do that.

There are also some implementation issues which have to clarified first:
1) use of stable storage
if yes: how to put it from user space into kernel

2) how to generate the first random value on boot (no stable storage) or on 
setup (stable storage)

3) which proc controls should be available for that?
 * enabling per interface
 * read out current random value or next random value for userspace programs
 * additional or overwrite a preconfigured static one
   * use of static one for incoming connections and random one for 
outgoing connections (source address binding, depending on destination 
[internet | site], if no site-local addresses are used).

> > BTW: I plan to implement RFC3041 calculation for demonstration and check
> > issues into "ipv6calc". Can someone point me to a C example which does a
> > simple MD5sum of a string like
> >       char* md5sum_of_string(char*) ?
>
> Please look at icmpv6_ni_group() in kernel/linux24/icmp.c.

Thanks.

        Peter






> --
> yoshfuji