[openib-general] [PATCH] IPoIB CM Experimental support

[Michael S. Tsirkin]

> Reading a little more:
> 
>  > +	/* Simple heuristic: dev->mtu > 2K ==> connected mode */
> 
> I'm not sure this is such a good idea.  I think it's setting a trap
> for people if we have magic behavior -- eg just imagine the questions
> if changing the MTU makes multicast stop working.

I know. Still, this only happens if you enable CM. Maybe it will help
to mention this in the comment in KConfig? Log a message as well?

What do you think?

I have a notion that once this code is upstream we can work on
ways to teach kernel about net devices where MTU changes dynamically.
Or possibly, some tricks with icmp can make it work.

-- 
MST