* [PATCH] tree.cc: Use useless_type_conversion_p in tree_builtin_call_types_compatible_p while in gimple form [PR105253]
@ 2022-04-13 13:28 Jakub Jelinek
2022-04-13 13:40 ` Richard Biener
0 siblings, 1 reply; 2+ messages in thread
From: Jakub Jelinek @ 2022-04-13 13:28 UTC (permalink / raw)
To: Richard Biener; +Cc: gcc-patches
Hi!
tree_builtin_call_types_compatible_p uses TYPE_MAIN_VARIANT comparisons
or tree_nop_conversion_p to ensure a builtin has correct GENERIC arguments.
Unfortunately this regressed when get_call_combined_fn is called during
GIMPLE optimizations. E.g. when number_of_iterations_popcount is called,
it doesn't ensure TYPE_MAIN_VARIABLE compatible argument type, it picks
__builtin_popcount{,l,ll} based just on types' precision and doesn't
fold_convert the arg to the right type. We are in GIMPLE, such conversions
are useless...
So, either we'd need to fix number_of_iterations_popcount to add casts
and inspect anything else that creates CALL_EXPRs late, or we can
in tree_builtin_call_types_compatible_p just use the GIMPLE type
comparisons (useless_type_conversion_p) when we are in GIMPLE form and
the TYPE_MAIN_VARIANT comparison or tree_nop_conversion_p test otherwise.
I think especially this late in stage4 the latter seems safer to me.
Bootstrapped/regtested on x86_64-linux and i686-linux, ok for trunk?
2022-04-13 Jakub Jelinek <jakub@redhat.com>
PR middle-end/105253
* tree.cc (tree_builtin_call_types_compatible_p): If PROP_gimple,
use useless_type_conversion_p checks instead of TYPE_MAIN_VARIANT
comparisons or tree_nop_conversion_p checks.
* gcc.target/i386/pr105253.c: New test.
--- gcc/tree.cc.jj 2022-04-13 09:44:05.021939106 +0200
+++ gcc/tree.cc 2022-04-13 11:43:25.756747436 +0200
@@ -8427,8 +8427,12 @@ tree_builtin_call_types_compatible_p (co
if (tree decl = builtin_decl_explicit (DECL_FUNCTION_CODE (fndecl)))
fndecl = decl;
- if (TYPE_MAIN_VARIANT (TREE_TYPE (call))
- != TYPE_MAIN_VARIANT (TREE_TYPE (TREE_TYPE (fndecl))))
+ bool gimple_form = (cfun && (cfun->curr_properties & PROP_gimple)) != 0;
+ if (gimple_form
+ ? !useless_type_conversion_p (TREE_TYPE (call),
+ TREE_TYPE (TREE_TYPE (fndecl)))
+ : (TYPE_MAIN_VARIANT (TREE_TYPE (call))
+ != TYPE_MAIN_VARIANT (TREE_TYPE (TREE_TYPE (fndecl)))))
return false;
tree targs = TYPE_ARG_TYPES (TREE_TYPE (fndecl));
@@ -8440,12 +8444,15 @@ tree_builtin_call_types_compatible_p (co
return true;
tree arg = CALL_EXPR_ARG (call, i);
tree type = TREE_VALUE (targs);
- if (TYPE_MAIN_VARIANT (type) != TYPE_MAIN_VARIANT (TREE_TYPE (arg)))
+ if (gimple_form
+ ? !useless_type_conversion_p (type, TREE_TYPE (arg))
+ : TYPE_MAIN_VARIANT (type) != TYPE_MAIN_VARIANT (TREE_TYPE (arg)))
{
/* For pointer arguments be more forgiving, e.g. due to
FILE * vs. fileptr_type_node, or say char * vs. const char *
differences etc. */
- if (POINTER_TYPE_P (type)
+ if (!gimple_form
+ && POINTER_TYPE_P (type)
&& POINTER_TYPE_P (TREE_TYPE (arg))
&& tree_nop_conversion_p (type, TREE_TYPE (arg)))
continue;
@@ -8457,8 +8464,11 @@ tree_builtin_call_types_compatible_p (co
&& INTEGRAL_TYPE_P (TREE_TYPE (arg))
&& !TYPE_UNSIGNED (TREE_TYPE (arg))
&& targetm.calls.promote_prototypes (TREE_TYPE (fndecl))
- && tree_nop_conversion_p (integer_type_node,
- TREE_TYPE (arg)))
+ && (gimple_form
+ ? useless_type_conversion_p (integer_type_node,
+ TREE_TYPE (arg))
+ : tree_nop_conversion_p (integer_type_node,
+ TREE_TYPE (arg))))
continue;
return false;
}
--- gcc/testsuite/gcc.target/i386/pr105253.c.jj 2022-04-13 11:47:52.460016521 +0200
+++ gcc/testsuite/gcc.target/i386/pr105253.c 2022-04-13 11:48:23.085588161 +0200
@@ -0,0 +1,17 @@
+/* PR middle-end/105253 */
+/* { dg-do compile } */
+/* { dg-options "-O2 -mno-popcnt" } */
+/* { dg-final { scan-assembler-not "__popcount\[sd]i2" } } */
+
+int
+foo (unsigned long long *p)
+{
+ int i, cnt = 0;
+ unsigned long long elem;
+ for (i = 0; i < 4; i++)
+ {
+ for (elem = p[i]; elem; cnt++)
+ elem &= elem - 1;
+ }
+ return cnt;
+}
Jakub
^ permalink raw reply [flat|nested] 2+ messages in thread
* Re: [PATCH] tree.cc: Use useless_type_conversion_p in tree_builtin_call_types_compatible_p while in gimple form [PR105253]
2022-04-13 13:28 [PATCH] tree.cc: Use useless_type_conversion_p in tree_builtin_call_types_compatible_p while in gimple form [PR105253] Jakub Jelinek
@ 2022-04-13 13:40 ` Richard Biener
0 siblings, 0 replies; 2+ messages in thread
From: Richard Biener @ 2022-04-13 13:40 UTC (permalink / raw)
To: Jakub Jelinek; +Cc: gcc-patches
On Wed, 13 Apr 2022, Jakub Jelinek wrote:
> Hi!
>
> tree_builtin_call_types_compatible_p uses TYPE_MAIN_VARIANT comparisons
> or tree_nop_conversion_p to ensure a builtin has correct GENERIC arguments.
> Unfortunately this regressed when get_call_combined_fn is called during
> GIMPLE optimizations. E.g. when number_of_iterations_popcount is called,
> it doesn't ensure TYPE_MAIN_VARIABLE compatible argument type, it picks
> __builtin_popcount{,l,ll} based just on types' precision and doesn't
> fold_convert the arg to the right type. We are in GIMPLE, such conversions
> are useless...
> So, either we'd need to fix number_of_iterations_popcount to add casts
> and inspect anything else that creates CALL_EXPRs late, or we can
> in tree_builtin_call_types_compatible_p just use the GIMPLE type
> comparisons (useless_type_conversion_p) when we are in GIMPLE form and
> the TYPE_MAIN_VARIANT comparison or tree_nop_conversion_p test otherwise.
>
> I think especially this late in stage4 the latter seems safer to me.
Agreed.
> Bootstrapped/regtested on x86_64-linux and i686-linux, ok for trunk?
OK.
Thanks,
Richard.
> 2022-04-13 Jakub Jelinek <jakub@redhat.com>
>
> PR middle-end/105253
> * tree.cc (tree_builtin_call_types_compatible_p): If PROP_gimple,
> use useless_type_conversion_p checks instead of TYPE_MAIN_VARIANT
> comparisons or tree_nop_conversion_p checks.
>
> * gcc.target/i386/pr105253.c: New test.
>
> --- gcc/tree.cc.jj 2022-04-13 09:44:05.021939106 +0200
> +++ gcc/tree.cc 2022-04-13 11:43:25.756747436 +0200
> @@ -8427,8 +8427,12 @@ tree_builtin_call_types_compatible_p (co
> if (tree decl = builtin_decl_explicit (DECL_FUNCTION_CODE (fndecl)))
> fndecl = decl;
>
> - if (TYPE_MAIN_VARIANT (TREE_TYPE (call))
> - != TYPE_MAIN_VARIANT (TREE_TYPE (TREE_TYPE (fndecl))))
> + bool gimple_form = (cfun && (cfun->curr_properties & PROP_gimple)) != 0;
> + if (gimple_form
> + ? !useless_type_conversion_p (TREE_TYPE (call),
> + TREE_TYPE (TREE_TYPE (fndecl)))
> + : (TYPE_MAIN_VARIANT (TREE_TYPE (call))
> + != TYPE_MAIN_VARIANT (TREE_TYPE (TREE_TYPE (fndecl)))))
> return false;
>
> tree targs = TYPE_ARG_TYPES (TREE_TYPE (fndecl));
> @@ -8440,12 +8444,15 @@ tree_builtin_call_types_compatible_p (co
> return true;
> tree arg = CALL_EXPR_ARG (call, i);
> tree type = TREE_VALUE (targs);
> - if (TYPE_MAIN_VARIANT (type) != TYPE_MAIN_VARIANT (TREE_TYPE (arg)))
> + if (gimple_form
> + ? !useless_type_conversion_p (type, TREE_TYPE (arg))
> + : TYPE_MAIN_VARIANT (type) != TYPE_MAIN_VARIANT (TREE_TYPE (arg)))
> {
> /* For pointer arguments be more forgiving, e.g. due to
> FILE * vs. fileptr_type_node, or say char * vs. const char *
> differences etc. */
> - if (POINTER_TYPE_P (type)
> + if (!gimple_form
> + && POINTER_TYPE_P (type)
> && POINTER_TYPE_P (TREE_TYPE (arg))
> && tree_nop_conversion_p (type, TREE_TYPE (arg)))
> continue;
> @@ -8457,8 +8464,11 @@ tree_builtin_call_types_compatible_p (co
> && INTEGRAL_TYPE_P (TREE_TYPE (arg))
> && !TYPE_UNSIGNED (TREE_TYPE (arg))
> && targetm.calls.promote_prototypes (TREE_TYPE (fndecl))
> - && tree_nop_conversion_p (integer_type_node,
> - TREE_TYPE (arg)))
> + && (gimple_form
> + ? useless_type_conversion_p (integer_type_node,
> + TREE_TYPE (arg))
> + : tree_nop_conversion_p (integer_type_node,
> + TREE_TYPE (arg))))
> continue;
> return false;
> }
> --- gcc/testsuite/gcc.target/i386/pr105253.c.jj 2022-04-13 11:47:52.460016521 +0200
> +++ gcc/testsuite/gcc.target/i386/pr105253.c 2022-04-13 11:48:23.085588161 +0200
> @@ -0,0 +1,17 @@
> +/* PR middle-end/105253 */
> +/* { dg-do compile } */
> +/* { dg-options "-O2 -mno-popcnt" } */
> +/* { dg-final { scan-assembler-not "__popcount\[sd]i2" } } */
> +
> +int
> +foo (unsigned long long *p)
> +{
> + int i, cnt = 0;
> + unsigned long long elem;
> + for (i = 0; i < 4; i++)
> + {
> + for (elem = p[i]; elem; cnt++)
> + elem &= elem - 1;
> + }
> + return cnt;
> +}
>
> Jakub
>
>
--
Richard Biener <rguenther@suse.de>
SUSE Software Solutions Germany GmbH, Maxfeldstrasse 5, 90409 Nuernberg,
Germany; GF: Ivo Totev; HRB 36809 (AG Nuernberg)
^ permalink raw reply [flat|nested] 2+ messages in thread
end of thread, other threads:[~2022-04-13 13:40 UTC | newest]
Thread overview: 2+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2022-04-13 13:28 [PATCH] tree.cc: Use useless_type_conversion_p in tree_builtin_call_types_compatible_p while in gimple form [PR105253] Jakub Jelinek
2022-04-13 13:40 ` Richard Biener
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).