Re: CVS commit: src/sys/netinet6

[Taylor R Campbell <campbell+netbsd-source-changes-d%mumble.net@localhost>]

> Date: Wed, 27 Dec 2017 10:31:25 +0900
> From: Ryota Ozaki <ozaki-r%netbsd.org@localhost>
> 
> On Wed, Dec 27, 2017 at 10:27 AM, Taylor R Campbell
> <campbell+netbsd-source-changes-d%mumble.net@localhost> wrote:
> > Can you add this to the collection?
> >
> > #ifdef NET_MPSAFE
> > #define SOFTNET_INTERLOCK_IF_NET_MPSAFE (&softnet_lock)
> > #else
> > #define SOFTNET_INTERLOCK_IF_MPSAFE     NULL
> > #endif
> 
> Oh I meant if not NET_MPSAFE, nd6_dad_stoptimer can be called with or without
> softnet_lock. So the macro doesn't work.

So push it into the callers: nd6_dad_stop, nd6_dad_duplicated?  That's
why I mentioned...

> >> > Can you just pass the interlock -- softnet_lock or NULL, depending on
> >> > NET_MPSAFE -- as an argument to nd6_dad_stoptimer?  It looks like it
> >> > will trickle up into the call stack a little bit, but not all that
> >> > far, unless .dom_if_link_state_change and .dom_if_down ever hold
> >> > softnet lock.

...the .dom_if_link_state_change and .dom_if_down callbacks, which
where as far as I looked back in the call graph.