diff mbox series

match.pd: Fix up (X & Y) CMP 0 -> X CMP2 ~Y simplifications [PR104499]

Message ID 20220211185829.GU2646553@tucnak
State New
Headers show
Series match.pd: Fix up (X & Y) CMP 0 -> X CMP2 ~Y simplifications [PR104499] | expand

Commit Message

Jakub Jelinek Feb. 11, 2022, 6:58 p.m. UTC 
Hi!

The following testcase ICEs on x86_64-linux, because match.pd emits
there a NOP_EXPR cast from int*8 vector type with BLKmode to
unsigned*8 vector type with BLKmode and vec-lowering isn't prepared
to handle such casts.

Fixed by using VIEW_CONVERT_EXPR instead.

Bootstrapped/regtested on powerpc64le-linux, ok for trunk?

2022-02-11  Jakub Jelinek  <jakub@redhat.com>

	PR tree-optimization/104499
	* match.pd ((X & Y) CMP 0 -> X CMP2 ~Y): Use view_convert instead
	of convert.

	* gcc.c-torture/compile/pr104499.c: New test.


	Jakub

Comments

Richard Biener Feb. 11, 2022, 7:05 p.m. UTC | #1 
> Am 11.02.2022 um 20:00 schrieb Jakub Jelinek via Gcc-patches <gcc-patches@gcc.gnu.org>:
> 
> Hi!
> 
> The following testcase ICEs on x86_64-linux, because match.pd emits
> there a NOP_EXPR cast from int*8 vector type with BLKmode to
> unsigned*8 vector type with BLKmode and vec-lowering isn't prepared
> to handle such casts.
> 
> Fixed by using VIEW_CONVERT_EXPR instead.
> 
> Bootstrapped/regtested on powerpc64le-linux, ok for trunk?

Ok

> 2022-02-11  Jakub Jelinek  <jakub@redhat.com>
> 
>    PR tree-optimization/104499
>    * match.pd ((X & Y) CMP 0 -> X CMP2 ~Y): Use view_convert instead
>    of convert.
> 
>    * gcc.c-torture/compile/pr104499.c: New test.
> 
> --- gcc/match.pd.jj    2022-02-11 11:20:24.272393155 +0100
> +++ gcc/match.pd    2022-02-11 13:45:02.612636452 +0100
> @@ -5778,7 +5778,7 @@ (define_operator_list SYNC_FETCH_AND_AND
>       (if (TYPE_UNSIGNED (TREE_TYPE (@1)))
>        (icmp @0 { csts; })
>        (with { tree utype = unsigned_type_for (TREE_TYPE (@1)); }
> -     (icmp (convert:utype @0) { csts; }))))))))
> +     (icmp (view_convert:utype @0) { csts; }))))))))
> 
> /* When one argument is a constant, overflow detection can be simplified.
>    Currently restricted to single use so as not to interfere too much with
> --- gcc/testsuite/gcc.c-torture/compile/pr104499.c.jj    2022-02-11 13:47:03.819963924 +0100
> +++ gcc/testsuite/gcc.c-torture/compile/pr104499.c    2022-02-11 13:46:44.421231605 +0100
> @@ -0,0 +1,11 @@
> +/* PR tree-optimization/104499 */
> +
> +typedef int __attribute__((__vector_size__ (8 * sizeof (int)))) V;
> +
> +V v;
> +
> +void
> +foo (void)
> +{
> +  v = ((1 | v) != 1);
> +}
> 
>    Jakub
>
diff mbox series

Patch

--- gcc/match.pd.jj	2022-02-11 11:20:24.272393155 +0100
+++ gcc/match.pd	2022-02-11 13:45:02.612636452 +0100
@@ -5778,7 +5778,7 @@  (define_operator_list SYNC_FETCH_AND_AND
       (if (TYPE_UNSIGNED (TREE_TYPE (@1)))
        (icmp @0 { csts; })
        (with { tree utype = unsigned_type_for (TREE_TYPE (@1)); }
-	 (icmp (convert:utype @0) { csts; }))))))))
+	 (icmp (view_convert:utype @0) { csts; }))))))))
 
 /* When one argument is a constant, overflow detection can be simplified.
    Currently restricted to single use so as not to interfere too much with
--- gcc/testsuite/gcc.c-torture/compile/pr104499.c.jj	2022-02-11 13:47:03.819963924 +0100
+++ gcc/testsuite/gcc.c-torture/compile/pr104499.c	2022-02-11 13:46:44.421231605 +0100
@@ -0,0 +1,11 @@ 
+/* PR tree-optimization/104499 */
+
+typedef int __attribute__((__vector_size__ (8 * sizeof (int)))) V;
+
+V v;
+
+void
+foo (void)
+{
+  v = ((1 | v) != 1);
+}