[PATCH] Signed zero for {max,min}val intrinsics

[PATCH] Signed zero for {max,min}val intrinsics

[Janne Blomqvist]

The Fortran standard specifies (e.g. F2018 7.4.3.2) that intrinsic
procedures shall treat positive and negative real zero as equivalent,
unless it is explicitly specified otherwise.  For {max,min}val there
is no such explicit mention.  Thus, remove code to handle signed
zeros.

Regtested on x86_64-pc-linux-gnu, Ok for trunk?

2018-08-22  Janne Blomqvist  <blomqvist.janne@gmail.com>

	* trans-intrinsic.c (gfc_conv_intrinsic_minmaxval): Delete
	HONOR_SIGNED_ZEROS checks.
---
 gcc/fortran/trans-intrinsic.c | 26 ++++++--------------------
 1 file changed, 6 insertions(+), 20 deletions(-)

diff --git a/gcc/fortran/trans-intrinsic.c b/gcc/fortran/trans-intrinsic.c
index 387cf80b921..b2cea93742a 100644
--- a/gcc/fortran/trans-intrinsic.c
+++ b/gcc/fortran/trans-intrinsic.c
@@ -5511,22 +5511,10 @@ gfc_conv_intrinsic_minmaxval (gfc_se * se, gfc_expr * expr, enum tree_code op)
     {
       /* MIN_EXPR/MAX_EXPR has unspecified behavior with NaNs or
 	 signed zeros.  */
-      if (HONOR_SIGNED_ZEROS (DECL_MODE (limit)))
-	{
-	  tmp = fold_build2_loc (input_location, op, logical_type_node,
-				 arrayse.expr, limit);
-	  ifbody = build2_v (MODIFY_EXPR, limit, arrayse.expr);
-	  tmp = build3_v (COND_EXPR, tmp, ifbody,
-			  build_empty_stmt (input_location));
-	  gfc_add_expr_to_block (&block2, tmp);
-	}
-      else
-	{
-	  tmp = fold_build2_loc (input_location,
-				 op == GT_EXPR ? MAX_EXPR : MIN_EXPR,
-				 type, arrayse.expr, limit);
-	  gfc_add_modify (&block2, limit, tmp);
-	}
+      tmp = fold_build2_loc (input_location,
+			     op == GT_EXPR ? MAX_EXPR : MIN_EXPR,
+			     type, arrayse.expr, limit);
+      gfc_add_modify (&block2, limit, tmp);
     }
 
   if (fast)
@@ -5535,8 +5523,7 @@ gfc_conv_intrinsic_minmaxval (gfc_se * se, gfc_expr * expr, enum tree_code op)
 
       /* MIN_EXPR/MAX_EXPR has unspecified behavior with NaNs or
 	 signed zeros.  */
-      if (HONOR_NANS (DECL_MODE (limit))
-	  || HONOR_SIGNED_ZEROS (DECL_MODE (limit)))
+      if (HONOR_NANS (DECL_MODE (limit)))
 	{
 	  tmp = fold_build2_loc (input_location, op, logical_type_node,
 				 arrayse.expr, limit);
@@ -5598,8 +5585,7 @@ gfc_conv_intrinsic_minmaxval (gfc_se * se, gfc_expr * expr, enum tree_code op)
 
       /* MIN_EXPR/MAX_EXPR has unspecified behavior with NaNs or
 	 signed zeros.  */
-      if (HONOR_NANS (DECL_MODE (limit))
-	  || HONOR_SIGNED_ZEROS (DECL_MODE (limit)))
+      if (HONOR_NANS (DECL_MODE (limit)))
 	{
 	  tmp = fold_build2_loc (input_location, op, logical_type_node,
 				 arrayse.expr, limit);
-- 
2.17.1

Re: [PATCH] Signed zero for {max,min}val intrinsics

[Jerry DeLisle]

On 08/22/2018 01:06 PM, Janne Blomqvist wrote:
> The Fortran standard specifies (e.g. F2018 7.4.3.2) that intrinsic
> procedures shall treat positive and negative real zero as equivalent,
> unless it is explicitly specified otherwise.  For {max,min}val there
> is no such explicit mention.  Thus, remove code to handle signed
> zeros.
> 
> Regtested on x86_64-pc-linux-gnu, Ok for trunk?

Looks good, OK.

Thanks,

Jerry

---snip---