https://gcc.gnu.org/bugzilla/show_bug.cgi?id=91606
Richard Biener <rguenth at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
CC| |jason at gcc dot gnu.org
--- Comment #10 from Richard Biener <rguenth at gcc dot gnu.org> ---
So I think we can recover what the C++ FE tries to do (assign alias-set zero
to the whole PMF struct) by really treating it as a fat pointer. That
restores using alias-set zero for the _31->__pfn accesses and fixes the
testcase.
Index: gcc/cp/decl.c
===================================================================
--- gcc/cp/decl.c (revision 276396)
+++ gcc/cp/decl.c (working copy)
@@ -9554,10 +9554,12 @@ build_ptrmemfunc_type (tree type)
TYPE_PTRMEMFUNC_FLAG (t) = 1;
field = build_decl (input_location, FIELD_DECL, pfn_identifier, type);
+ DECL_NONADDRESSABLE_P (field) = 1;
fields = field;
field = build_decl (input_location, FIELD_DECL, delta_identifier,
delta_type_node);
+ DECL_NONADDRESSABLE_P (field) = 1;
DECL_CHAIN (field) = fields;
fields = field;
this tells the middle-end that there are no accesses to those fields
via pointers to the field but they always appear via pointers to
the containing record (or containing records). I think that's true
for what the C++ FE emits and for what a user can do with a PMF.
Otherwise I fear what the C++ FE does in it's get_alias_set langhook
doesn't work by design :/
I will test the above.
