https://gcc.gnu.org/bugzilla/show_bug.cgi?id=90501
Richard Biener <rguenth at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Status|UNCONFIRMED |ASSIGNED
Last reconfirmed| |2019-05-16
Assignee|unassigned at gcc dot gnu.org |rguenth at gcc dot
gnu.org
Ever confirmed|0 |1
--- Comment #6 from Richard Biener <rguenth at gcc dot gnu.org> ---
OK, so the early inliner inlines
result = fun (); [return slot optimization]
which results in result becoming address-taken after inlining:
static tree
declare_return_variable (copy_body_data *id, tree return_slot, tree
modify_dest,
basic_block entry_bb)
{
...
/* If there was a return slot, then the return value is the
dereferenced address of that object. */
if (return_slot)
{
/* The front end shouldn't have used both return_slot and
a modify expression. */
gcc_assert (!modify_dest);
if (DECL_BY_REFERENCE (result))
{
tree return_slot_addr = build_fold_addr_expr (return_slot);
STRIP_USELESS_TYPE_CONVERSION (return_slot_addr);
/* We are going to construct *&return_slot and we can't do that
for variables believed to be not addressable.
FIXME: This check possibly can match, because values returned
via return slot optimization are not believed to have address
taken by alias analysis. */
gcc_assert (TREE_CODE (return_slot) != SSA_NAME);
var = return_slot_addr;
note this bug was present before my changes with -fno-early-inlining.
Interesting we never hit this for other languages. I guess most of the
time the addresses end up in places we do not require TREE_ADDRESSABLE
to be set (in dereferences).
