On Fri, 11 Dec 2020, Jakub Jelinek wrote:
> On Fri, Dec 11, 2020 at 11:34:01AM +0100, Richard Biener wrote:
> > Shouldn't we simply use gen_int_mode?
>
> You're right, that is shorter. So like this?
Yes, OK.
> BTW, looking at gen_int_mode, wouldn't it be better to add an overload
> for HOST_WIDE_INT first argument so that it wouldn't need to go through
> poly_int? Or would that then be ambiguous with int, unsigned etc.
> arguments?
Not sure but marshalling as poly-int64 should be cheap.
Richard.
> 2020-12-10 Jakub Jelinek <ja...@redhat.com>
>
> PR rtl-optimization/98229
> * optabs.c (expand_doubleword_mod): Canonicalize op1 and
> 1 - INTVAL (op1) as word_mode constants when used in
> word_mode arithmetics.
>
> * gcc.c-torture/compile/pr98229.c: New test.
>
> --- gcc/optabs.c.jj 2020-12-11 00:30:41.892272376 +0100
> +++ gcc/optabs.c 2020-12-11 11:51:15.902855312 +0100
> @@ -1081,7 +1081,8 @@ expand_doubleword_mod (machine_mode mode
> return NULL_RTX;
> }
> }
> - rtx remainder = expand_divmod (1, TRUNC_MOD_EXPR, word_mode, sum, op1,
> + rtx remainder = expand_divmod (1, TRUNC_MOD_EXPR, word_mode, sum,
> + gen_int_mode (INTVAL (op1), word_mode),
> NULL_RTX, 1, OPTAB_DIRECT);
> if (remainder == NULL_RTX)
> return NULL_RTX;
> @@ -1099,7 +1100,8 @@ expand_doubleword_mod (machine_mode mode
> return NULL_RTX;
> }
> mask = expand_simple_binop (word_mode, AND, mask,
> - GEN_INT (1 - INTVAL (op1)),
> + gen_int_mode (1 - INTVAL (op1),
> + word_mode),
> NULL_RTX, 1, OPTAB_DIRECT);
> if (mask == NULL_RTX)
> return NULL_RTX;
> --- gcc/testsuite/gcc.c-torture/compile/pr98229.c.jj 2020-12-11
> 11:49:34.518993120 +0100
> +++ gcc/testsuite/gcc.c-torture/compile/pr98229.c 2020-12-11
> 11:49:34.517993131 +0100
> @@ -0,0 +1,7 @@
> +/* PR rtl-optimization/98229 */
> +
> +unsigned long long
> +foo (unsigned long long x)
> +{
> + return x % ~0U;
> +}
>
> Jakub
>
>
--
Richard Biener <rguent...@suse.de>
SUSE Software Solutions Germany GmbH, Maxfeldstrasse 5, 90409 Nuernberg,
Germany; GF: Felix Imendörffer; HRB 36809 (AG Nuernberg)
