[rfa] cris-tdep.c: Frame unwind and dummy call fixes

[Andrew Cagney]

> This is a rather monstrous patch, but the CRIS port hadn't been updated in quite a while, and a lot of the stuff that had changed since the last update was interconnected.  In addition to this, I got rid of support for a now obsolete ABI (instead of updating it) which further increases the size of the patch.
> 
> Andrew, it was a while since I contributed to the tree, and my public key has most likely changed since then.  Should I send you a new identity.pub?

Try it, if it doesn't work let me know.

> I realize this patch comes very close to branching for 6.1, so if it fits badly with the 6.1 schedule for whatever reason then I'm prepared to wait until later.

You're the maintainer so feel free to commit (I held the branch date 
over by a day so if its done in the next few hours you'll avoid the need 
to double commit :-).

The intent of the branch commit rules is to accommodate exactly your 
situtation:

http://sources.redhat.com/gdb/current/onlinedocs/gdbint_15.html#SEC133
Pragmatics: Provided updates are restricted to non-core functionality 
there is little chance that a broken change will be fatal. This means 
that changes such as adding a new architectures or (within reason) 
support for a new host are considered acceptable.

Its changes to GDB's core that don't go down so well.

Andrew