Re: CVS commit: src/sys

[Robert Elz <kre%munnari.OZ.AU@localhost>]

    Date:        Sat, 25 Nov 2017 18:11:22 +0900 (JST)
    From:        Ryo ONODERA <ryo%tetera.org@localhost>
    Message-ID:  <20171125.181122.805047996391646417.ryo%tetera.org@localhost>

  | After the following commit, my IPv6 network does not work at all.
  | What information is required to investigate my problem?

The ATF tests (the network tests) have all started failing as well.
I have started looking to see if I can see which actual mod broke things.


While here, I am not sure I agree with the reason for the changes (even
if they can be made to work) - and regardless of what ancient systems
used to do.

Sending to a detached address should fail (regardless of from where
the packet is sent - including the system that owns the addr) - that
allows simple testing that all of the addresses are functioning properly.

And I think I remember saying before that tentative addresses are not
even close to the same thing - a detached addr is one which does not
work (because its network link does not work).   A tentative addr is
one which does work, but which we have just not proved is not a dplicate
(but 99% of the time - or more - is just fine.)

Sending to/from detached addresses should be prevented.   Sending to/from
tentative addresses should be allowed, always - until they are detected
as invalid, at which point, they are simply removed (or made detached,
or whatever) after which they can no longer be used.

kre