http://gcc.gnu.org/bugzilla/show_bug.cgi?id=51782
Martin Jambor <jamborm at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Status|NEW |ASSIGNED
AssignedTo|unassigned at gcc dot |jamborm at gcc dot gnu.org
|gnu.org |
--- Comment #15 from Martin Jambor <jamborm at gcc dot gnu.org> 2012-02-17
13:05:02 UTC ---
In addition to mistakes corrected by the patch in comment #12,
build_ref_for_offset indeed also does not care about address spaces of
MEM_REFs it is creating. Thus, mine.
Perhaps we should even add a check to a verifier to catch situations
when a MEM_REF's zero argument has a TREE_TYPE^2 with a different
address space than the MEM_REF itself?