Compile pr37870.c with -mlong-double-80
authorH.J. Lu <hongjiu.lu@intel.com>
Wed, 8 Jul 2015 19:35:43 +0000 (19:35 +0000)
committerH.J. Lu <hjl@gcc.gnu.org>
Wed, 8 Jul 2015 19:35:43 +0000 (12:35 -0700)
On x86, the "long double" type is platforma specific, which may be the
same as double or __float128.  Since pr37870.c requires 80-bit floating
point type, it should be compiled with -mlong-double-80.

* gcc.target/i386/pr37870.c (dg-options): Add -mlong-double-80.

From-SVN: r225578

gcc/testsuite/ChangeLog
gcc/testsuite/gcc.target/i386/pr37870.c

index 9a3a93e29b81ef6a604a6e20e8b4c8c4deb41756..3fa2511c40fcbcb42c20e76953d55b1b75fe1dc9 100644 (file)
@@ -1,3 +1,7 @@
+2015-07-08  H.J. Lu  <hongjiu.lu@intel.com>
+
+       * gcc.target/i386/pr37870.c (dg-options): Add -mlong-double-80.
+
 2015-07-08  H.J. Lu  <hongjiu.lu@intel.com>
 
        * gcc.target/i386/readeflags-1.c (abort): New prototype.
index 19cfb2058653692fa71b8532b349f67cc8aa575a..bc78ed7c77da466f173e9d94c6a9a83d72a16c6e 100644 (file)
@@ -1,6 +1,6 @@
 /* PR middle-end/37870 */
 /* { dg-do run } */
-/* { dg-options "-O2" } */
+/* { dg-options "-O2 -mlong-double-80" } */
 
 unsigned int
 foo (long double x)