This is the mail archive of the gcc-patches@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

[patch RFC] Add -fno-finite-math-only to some gcc.dg tests


Hi,

There are new failures on sh4-*-*

gcc.dg/builtins-43.c scan-tree-dump-times link_error 0
gcc.dg/builtins-44.c scan-tree-dump-times link_error 0
gcc.dg/builtins-45.c scan-tree-dump-times link_error 0
gcc.dg/pr28796-2.c execution test
gcc.dg/unordered-3.c (test for excess errors)

which start to fail somewhere between revision 117947 and
revision 118015.  It's because sh4*-*-* without -mieee
defaults to -ffinite-math-only and it looks that some generic
change made the above tests fail with -ffinite-math-only.
I've confirmed that these tests now fail for i686-pc-linux-gnu
on the trunk with -ffinite-math-only.  It seems that the new
behavior makes sense because these tests include the built-in
NaNs and Infs which may be ignored with -ffinite-math-only.

The attached patch adds -fno-fininte-math-only to dg-options
of these tests.  It's regtested on i686-pc-linux-gnu with no
new failures.

Regards,
	kaz
--
:ADDPATCH testsuite:

2006-10-26  Kaz Kojima  <kkojima@gcc.gnu.org>

	* gcc.dg/builtins-43.c: Add -fno-finite-math-only.
	* gcc.dg/builtins-44.c: Likewise.
	* gcc.dg/builtins-45.c: Likewise.
	* gcc.dg/pr28796-2.c: Likewise.
	* gcc.dg//unordered-3.c: Likewise.	

diff -uprN ORIG/trunk/gcc/testsuite/gcc.dg/builtins-43.c LOCAL/trunk/gcc/testsuite/gcc.dg/builtins-43.c
--- ORIG/trunk/gcc/testsuite/gcc.dg/builtins-43.c	2005-10-29 06:49:27.000000000 +0900
+++ LOCAL/trunk/gcc/testsuite/gcc.dg/builtins-43.c	2006-10-26 08:11:32.000000000 +0900
@@ -1,5 +1,5 @@
 /* { dg-do compile } */
-/* { dg-options "-O1 -fno-trapping-math -fdump-tree-gimple -fdump-tree-optimized" } */
+/* { dg-options "-O1 -fno-trapping-math -fno-finite-math-only -fdump-tree-gimple -fdump-tree-optimized" } */
   
 extern void f(int);
 extern void link_error ();
diff -uprN ORIG/trunk/gcc/testsuite/gcc.dg/builtins-44.c LOCAL/trunk/gcc/testsuite/gcc.dg/builtins-44.c
--- ORIG/trunk/gcc/testsuite/gcc.dg/builtins-44.c	2005-10-29 06:49:26.000000000 +0900
+++ LOCAL/trunk/gcc/testsuite/gcc.dg/builtins-44.c	2006-10-26 08:11:57.000000000 +0900
@@ -1,5 +1,5 @@
 /* { dg-do compile } */
-/* { dg-options "-O1 -fno-trapping-math -fdump-tree-optimized" } */
+/* { dg-options "-O1 -fno-trapping-math -fno-finite-math-only -fdump-tree-optimized" } */
   
 extern void f(int);
 extern void link_error ();
diff -uprN ORIG/trunk/gcc/testsuite/gcc.dg/builtins-45.c LOCAL/trunk/gcc/testsuite/gcc.dg/builtins-45.c
--- ORIG/trunk/gcc/testsuite/gcc.dg/builtins-45.c	2005-10-29 06:49:28.000000000 +0900
+++ LOCAL/trunk/gcc/testsuite/gcc.dg/builtins-45.c	2006-10-26 08:12:22.000000000 +0900
@@ -1,5 +1,5 @@
 /* { dg-do compile } */
-/* { dg-options "-O1 -fno-trapping-math -fdump-tree-optimized" } */
+/* { dg-options "-O1 -fno-trapping-math -fno-finite-math-only -fdump-tree-optimized" } */
   
 extern void f(int);
 extern void link_error ();
diff -uprN ORIG/trunk/gcc/testsuite/gcc.dg/pr28796-2.c LOCAL/trunk/gcc/testsuite/gcc.dg/pr28796-2.c
--- ORIG/trunk/gcc/testsuite/gcc.dg/pr28796-2.c	2006-10-25 09:15:58.000000000 +0900
+++ LOCAL/trunk/gcc/testsuite/gcc.dg/pr28796-2.c	2006-10-26 08:13:52.000000000 +0900
@@ -1,5 +1,5 @@
 /* { dg-do run } */
-/* { dg-options "-O2 -funsafe-math-optimizations" } */
+/* { dg-options "-O2 -funsafe-math-optimizations -fno-finite-math-only" } */
 
 extern void abort (void);
 
diff -uprN ORIG/trunk/gcc/testsuite/gcc.dg/unordered-3.c LOCAL/trunk/gcc/testsuite/gcc.dg/unordered-3.c
--- ORIG/trunk/gcc/testsuite/gcc.dg/unordered-3.c	2005-10-29 06:49:28.000000000 +0900
+++ LOCAL/trunk/gcc/testsuite/gcc.dg/unordered-3.c	2006-10-26 08:13:10.000000000 +0900
@@ -1,4 +1,5 @@
 /* { dg-do link } */
+/* { dg-options "-fno-finite-math-only" }  */
 
 void link_error ();
 


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]