[PATCH] gimple: Ignore *0 = {CLOBBER} in path isolation [PR96722]

[PATCH] gimple: Ignore *0 = {CLOBBER} in path isolation [PR96722]

[Jakub Jelinek]

Hi!

Clobbers of MEM_REF with NULL address are just fancy nops, something we just
ignore and don't emit any code for it (ditto for other clobbers), they just
mark end of life on something, so we shouldn't infer from those that there
is some UB.

Bootstrapped/regtested on x86_64-linux and i686-linux, ok for trunk?

2020-08-24  Jakub Jelinek  <jakub@redhat.com>

	PR tree-optimization/96722
	* gimple.c (infer_nonnull_range): Formatting fix.
	(infer_nonnull_range_by_dereference): Return false for clobber stmts.

	* g++.dg/opt/pr96722.C: New test.

--- gcc/gimple.c.jj	2020-08-03 22:54:51.417531677 +0200
+++ gcc/gimple.c	2020-08-24 13:23:22.082312349 +0200
@@ -2917,8 +2917,8 @@ check_loadstore (gimple *, tree op, tree
 bool
 infer_nonnull_range (gimple *stmt, tree op)
 {
-  return infer_nonnull_range_by_dereference (stmt, op)
-    || infer_nonnull_range_by_attribute (stmt, op);
+  return (infer_nonnull_range_by_dereference (stmt, op)
+	  || infer_nonnull_range_by_attribute (stmt, op));
 }
 
 /* Return true if OP can be inferred to be non-NULL after STMT
@@ -2930,7 +2930,8 @@ infer_nonnull_range_by_dereference (gimp
      non-NULL if -fdelete-null-pointer-checks is enabled.  */
   if (!flag_delete_null_pointer_checks
       || !POINTER_TYPE_P (TREE_TYPE (op))
-      || gimple_code (stmt) == GIMPLE_ASM)
+      || gimple_code (stmt) == GIMPLE_ASM
+      || gimple_clobber_p (stmt))
     return false;
 
   if (walk_stmt_load_store_ops (stmt, (void *)op,
--- gcc/testsuite/g++.dg/opt/pr96722.C.jj	2020-08-24 13:24:45.357132323 +0200
+++ gcc/testsuite/g++.dg/opt/pr96722.C	2020-08-24 13:25:06.224836626 +0200
@@ -0,0 +1,20 @@
+// PR tree-optimization/96722
+// { dg-do run }
+// { dg-options "-O2" }
+
+struct S { int s; ~S () {} };
+
+void
+foo (S *a)
+{
+  if (a)
+    return;
+  a->~S ();
+}
+
+int
+main ()
+{
+  S s;
+  foo (&s);
+}

	Jakub

Re: [PATCH] gimple: Ignore *0 = {CLOBBER} in path isolation [PR96722]

[Richard Biener]

On Mon, 24 Aug 2020, Jakub Jelinek wrote:

> Hi!
> 
> Clobbers of MEM_REF with NULL address are just fancy nops, something we just
> ignore and don't emit any code for it (ditto for other clobbers), they just
> mark end of life on something, so we shouldn't infer from those that there
> is some UB.
> 
> Bootstrapped/regtested on x86_64-linux and i686-linux, ok for trunk?

OK.

Richard.

> 2020-08-24  Jakub Jelinek  <jakub@redhat.com>
> 
> 	PR tree-optimization/96722
> 	* gimple.c (infer_nonnull_range): Formatting fix.
> 	(infer_nonnull_range_by_dereference): Return false for clobber stmts.
> 
> 	* g++.dg/opt/pr96722.C: New test.
> 
> --- gcc/gimple.c.jj	2020-08-03 22:54:51.417531677 +0200
> +++ gcc/gimple.c	2020-08-24 13:23:22.082312349 +0200
> @@ -2917,8 +2917,8 @@ check_loadstore (gimple *, tree op, tree
>  bool
>  infer_nonnull_range (gimple *stmt, tree op)
>  {
> -  return infer_nonnull_range_by_dereference (stmt, op)
> -    || infer_nonnull_range_by_attribute (stmt, op);
> +  return (infer_nonnull_range_by_dereference (stmt, op)
> +	  || infer_nonnull_range_by_attribute (stmt, op));
>  }
>  
>  /* Return true if OP can be inferred to be non-NULL after STMT
> @@ -2930,7 +2930,8 @@ infer_nonnull_range_by_dereference (gimp
>       non-NULL if -fdelete-null-pointer-checks is enabled.  */
>    if (!flag_delete_null_pointer_checks
>        || !POINTER_TYPE_P (TREE_TYPE (op))
> -      || gimple_code (stmt) == GIMPLE_ASM)
> +      || gimple_code (stmt) == GIMPLE_ASM
> +      || gimple_clobber_p (stmt))
>      return false;
>  
>    if (walk_stmt_load_store_ops (stmt, (void *)op,
> --- gcc/testsuite/g++.dg/opt/pr96722.C.jj	2020-08-24 13:24:45.357132323 +0200
> +++ gcc/testsuite/g++.dg/opt/pr96722.C	2020-08-24 13:25:06.224836626 +0200
> @@ -0,0 +1,20 @@
> +// PR tree-optimization/96722
> +// { dg-do run }
> +// { dg-options "-O2" }
> +
> +struct S { int s; ~S () {} };
> +
> +void
> +foo (S *a)
> +{
> +  if (a)
> +    return;
> +  a->~S ();
> +}
> +
> +int
> +main ()
> +{
> +  S s;
> +  foo (&s);
> +}
> 
> 	Jakub
> 
> 

-- 
Richard Biener <rguenther@suse.de>
SUSE Software Solutions Germany GmbH, Maxfeldstrasse 5, 90409 Nuernberg,
Germany; GF: Felix Imendörffer; HRB 36809 (AG Nuernberg)