On Wed, 28 Feb 2024, Jakub Jelinek wrote:
> Hi!
>
> The following testcases are miscompiled, because graphite ignores boolean,
> enumerated or _BitInt comparisons, rewrites the code as if the comparisons
> were always true or always false.
>
> The INTEGER_TYPE checks were initially added in r6-2239 but at that point
> it was both in add_conditions_to_domain and in parameter_index_in_region.
> Later on the check was also added to stmt_simple_for_scop_p, and finally
> r8-3931 changed the stmt_simple_for_scop_p check to INTEGRAL_TYPE_P
> and turned the parameter_index_in_region -> assign_parameter_index_in_region
> into INTEGRAL_TYPE_P assertion, but the add_conditions_to_domain check
> for INTEGER_TYPE remained.
>
> The following patch uses INTEGRAL_TYPE_P to complete the change.
>
> Bootstrapped/regtested on x86_64-linux and i686-linux (--with-isl only
> on the former though), ok for trunk?
OK. As said this should probably be an assert, but I won't bother you
with that (and the possible fallout).
Thanks,
Richard.
> 2024-02-28 Jakub Jelinek <ja...@redhat.com>
>
> PR tree-optimization/114041
> * graphite-sese-to-poly.cc (add_conditions_to_domain): Check for
> INTEGRAL_TYPE_P check rather than INTEGER_TYPE.
>
> * gcc.dg/graphite/run-id-pr114041-1.c: New test.
> * gcc.dg/graphite/run-id-pr114041-2.c: New test.
>
> --- gcc/graphite-sese-to-poly.cc.jj 2024-01-03 11:51:29.136764430 +0100
> +++ gcc/graphite-sese-to-poly.cc 2024-02-27 19:35:07.668304435 +0100
> @@ -391,8 +391,9 @@ add_conditions_to_domain (poly_bb_p pbb)
> {
> case GIMPLE_COND:
> {
> - /* Don't constrain on anything else than INTEGER_TYPE. */
> - if (TREE_CODE (TREE_TYPE (gimple_cond_lhs (stmt))) != INTEGER_TYPE)
> + /* Don't constrain on anything else than INTEGRAL_TYPE_P. */
> + tree cmp_type = TREE_TYPE (gimple_cond_lhs (stmt));
> + if (!INTEGRAL_TYPE_P (cmp_type))
> break;
>
> gcond *cond_stmt = as_a <gcond *> (stmt);
> --- gcc/testsuite/gcc.dg/graphite/run-id-pr114041-1.c.jj 2024-02-27
> 18:42:26.864025806 +0100
> +++ gcc/testsuite/gcc.dg/graphite/run-id-pr114041-1.c 2024-02-27
> 18:43:07.310466262 +0100
> @@ -0,0 +1,23 @@
> +/* PR tree-optimization/114041 */
> +/* { dg-require-effective-target bitint } */
> +/* { dg-options "-O -fgraphite-identity" } */
> +
> +unsigned a[24], b[24];
> +
> +__attribute__((noipa)) unsigned
> +foo (unsigned _BitInt(8) x)
> +{
> + for (int i = 0; i < 24; ++i)
> + a[i] = i;
> + unsigned e = __builtin_stdc_bit_ceil (x);
> + for (int i = 0; i < 24; ++i)
> + b[i] = i;
> + return e;
> +}
> +
> +int
> +main ()
> +{
> + if (foo (0) != 1)
> + __builtin_abort ();
> +}
> --- gcc/testsuite/gcc.dg/graphite/run-id-pr114041-2.c.jj 2024-02-27
> 19:36:02.373547881 +0100
> +++ gcc/testsuite/gcc.dg/graphite/run-id-pr114041-2.c 2024-02-27
> 19:36:22.515269333 +0100
> @@ -0,0 +1,27 @@
> +/* PR tree-optimization/114041 */
> +/* { dg-options "-O -fgraphite-identity" } */
> +
> +unsigned a[24], b[24];
> +enum E { E0 = 0, E1 = 1, E42 = 42, E56 = 56 };
> +
> +__attribute__((noipa)) unsigned
> +foo (enum E x)
> +{
> + for (int i = 0; i < 24; ++i)
> + a[i] = i;
> + unsigned e;
> + if (x >= E42)
> + e = __builtin_clz ((unsigned) x);
> + else
> + e = 42;
> + for (int i = 0; i < 24; ++i)
> + b[i] = i;
> + return e;
> +}
> +
> +int
> +main ()
> +{
> + if (foo (E1) != 42 || foo (E56) != __SIZEOF_INT__ * __CHAR_BIT__ - 6)
> + __builtin_abort ();
> +}
>
> Jakub
>
>
--
Richard Biener <rguent...@suse.de>
SUSE Software Solutions Germany GmbH,
Frankenstrasse 146, 90461 Nuernberg, Germany;
GF: Ivo Totev, Andrew McDonald, Werner Knoblich; (HRB 36809, AG Nuernberg)
