On Thu, Jan 12, 2017 at 01:16:22PM -0700, Jason Gunthorpe wrote:
> On Thu, Jan 12, 2017 at 09:35:08PM +0200, Leon Romanovsky wrote:
>
> > > > In datagram mode, the IB UD (Unreliable Datagram) transport is used
> > > > so the MTU of the interface is equal to the IB L2 MTU minus the
> > > > IPoIB encapsulation header. Any request to change the MTU value
> > > > above the maximum range will change the MTU to the max allowed, but
> > > > will not show any warning message. An ipoib_warn is issued in such
> > > > cases, letting the user know that even though the value is legal,
> > > > it can't be currently applied.
>
> How does RC mode work then?

RC mode doesn't have limitation on MTU size and it allows messages
larger than IB link-layer MTU.

>
> > Second, I have a different view from you on the issue. User configured
> > some value, which is not correct for IPoIB. In ideal world (without legacy),
> > we were supposed to return error to him with proper message, but in our
> > case (legacy applications) we can't (we tried and it broke some legacy
> > ifcongfig, if I remember well). So it leaves us with one available
> > option is to warn user about improper value.
>
> This is a legitimate configuration though, the user could have a mixed
> MTU fabric and rely on path MTU discovery, so sets the MTU to the
> largest.
>
> AFAIK the way to handle such configurations is with a multicast route that
> has a reduced mtu.

Multicast traffic is sent in datagram mode hence the limit will imply.

Maybe, I went too far by calling it error, but IMHO absence of indication to
user of decreased MTU is not good either.

>
> Jason