Re: [gomp] Fix sharing-1.c and sharing-1.C

[Richard Henderson <rth at redhat dot com>]

On Tue, Oct 18, 2005 at 03:59:52PM -0400, Diego Novillo wrote:
> Well, we do abort here regardless of whether we default to shared or 
> firstprivate.  &x outside of 'omp parallel' is always going to be 
> different from &x inside.

Well, not *always*.  We could always pass by reference.

> Are you saying they're supposed to be the same?

My guess so far had been that they needn't have been, thus the
copy in/out semantics I've been using.  But I admit that I'm
not 100% sure...

FWIW, icc 9.0 passes this test case.  One can guess, then, that
we can't use copy-in/out if the variable is TREE_ADDRESSABLE.


r~