Segfault on O2 optimisation for test case for PR18689

[Tobias Schlüter]

Paul Thomas wrote:
> In an earlier posting, I indicated that PR18689 could be fixed at the same 
> time as PR16939.  This problem involves dummy arguments being character 
> pointer arrays.  However, I have hit a FAIL on  the test program, when the 
> optimisation is O2 or better.
...
> This is specific to actual arguments.  I guess the scalarisation loop is 
> broken in some way?

I tried your patch together with Jakub's fix for PR15080, but the
scalarization still goes wrong.  So it looks like you found a different bug
somewhere, the .original dump looks fine.  The bug is very fragile: addding
debug code makes the testcase pass.  Maybe we're dealing with a
misoptimization for a change.

(Paul sent me his patch for the CHARACTER issues off-list)

- Tobi