https://gcc.gnu.org/bugzilla/show_bug.cgi?id=115538
Bug ID: 115538
Summary: wrong-code with SLP vectorization of conversions
Product: gcc
Version: 15.0
Status: UNCONFIRMED
Severity: normal
Priority: P3
Component: tree-optimization
Assignee: unassigned at gcc dot gnu.org
Reporter: rguenth at gcc dot gnu.org
Target Milestone: ---
When using SLP vectorizable_conversion does
/* Supportable by target? */
switch (modifier)
{
case NONE:
...
/* Check if the intermediate type can hold OP0's range.
When converting from float to integer this is not necessary
because values that do not fit the (smaller) target type are
unspecified anyway. */
if (demotion && float_expr_p)
{
wide_int op_min_value, op_max_value;
if (!vect_get_range_info (op0, &op_min_value, &op_max_value))
break;
if (cvt_type == NULL_TREE
|| (wi::min_precision (op_max_value, SIGNED)
> TYPE_PRECISION (cvt_type))
|| (wi::min_precision (op_min_value, SIGNED)
> TYPE_PRECISION (cvt_type)))
continue;
but that uses op0 from the representative (or error_mark_node in case of
a VEC_PERM_EXPR SLP operand).
No testcase yet.
