initializer_constant_valid_p_1 attempts to handle narrowing
differences and sums but fails to handle when the overall
value looks like
VIEW_CONVERT_EXPR<long long int>(NON_LVALUE_EXPR <v>
- VEC_COND_EXPR < { 0, 0 } == { 0, 0 } , { -1, -1 } , { 0, 0 } > )
where endtype is scalar integer but value is a vector type.
In this particular case all is good and we recurse since
two vector lanes is more than 64bits of long long. But still
it compares apples and oranges.
Fixed by appropriately also requiring the type of the
value to be scalar integral.
* varasm.cc (initializer_constant_valid_p_1): Also
constrain the type of value to be scalar integral
before dispatching to narrowing_initializer_constant_valid_p.
if (cache && cache[0] == value)
return cache[1];
if (! INTEGRAL_TYPE_P (endtype)
+ || ! INTEGRAL_TYPE_P (TREE_TYPE (value))
|| TYPE_PRECISION (endtype) >= TYPE_PRECISION (TREE_TYPE (value)))
{
tree ncache[4] = { NULL_TREE, NULL_TREE, NULL_TREE, NULL_TREE };
if (cache && cache[0] == value)
return cache[1];
if (! INTEGRAL_TYPE_P (endtype)
+ || ! INTEGRAL_TYPE_P (TREE_TYPE (value))
|| TYPE_PRECISION (endtype) >= TYPE_PRECISION (TREE_TYPE (value)))
{
tree ncache[4] = { NULL_TREE, NULL_TREE, NULL_TREE, NULL_TREE };