Fix multiple_of_p behavior with NOP_EXPR
Commit Message
We were passing down the original type to recursive invocations
of multiple_of_p for say (int)(unsigned * unsigned).
Bootstrap and regtest pending on x86_64-unknown-linux-gnu.
2022-01-24 Richard Biener <rguenther@suse.de>
* fold-const.cc (multiple_of_p): Pass the correct type of
the expression to the recursive invocation of multiple_of_p
for conversions and use CASE_CONVERT.
---
gcc/fold-const.cc | 6 +++---
1 file changed, 3 insertions(+), 3 deletions(-)
Comments
On 1/24/2022 7:46 AM, Richard Biener via Gcc-patches wrote:
> We were passing down the original type to recursive invocations
> of multiple_of_p for say (int)(unsigned * unsigned).
>
> Bootstrap and regtest pending on x86_64-unknown-linux-gnu.
>
> 2022-01-24 Richard Biener <rguenther@suse.de>
>
> * fold-const.cc (multiple_of_p): Pass the correct type of
> the expression to the recursive invocation of multiple_of_p
> for conversions and use CASE_CONVERT.
OK by me. Your call on gcc-12 vs gcc-13 timing.
jeff
@@ -14149,14 +14149,14 @@ multiple_of_p (tree type, const_tree top, const_tree bottom)
}
return 0;
- case NOP_EXPR:
+ CASE_CONVERT:
/* Can't handle conversions from non-integral or wider integral type. */
if ((TREE_CODE (TREE_TYPE (TREE_OPERAND (top, 0))) != INTEGER_TYPE)
|| (TYPE_PRECISION (type)
< TYPE_PRECISION (TREE_TYPE (TREE_OPERAND (top, 0)))))
return 0;
-
- /* fall through */
+ return multiple_of_p (TREE_TYPE (TREE_OPERAND (top, 0)),
+ TREE_OPERAND (top, 0), bottom);
case SAVE_EXPR:
return multiple_of_p (type, TREE_OPERAND (top, 0), bottom);