[PATCH, testsuite]: Use __builtin_nanq in float12-cmp-invalid.c

Uros Bizjak ubizjak@gmail.com
Fri Jun 24 14:57:00 GMT 2016


2016-06-24  Uros Bizjak  <ubizjak@gmail.com>

    * gcc.dg/torture/float128-cmp-invalid.c (main): Use __builtin_nanq.

Tested on x86_64-linux-gnu {,-m32}  and committed to mainline.

Uros.

Index: torture/float128-cmp-invalid.c
===================================================================
--- torture/float128-cmp-invalid.c      (revision 237757)
+++ torture/float128-cmp-invalid.c      (working copy)
@@ -9,7 +9,7 @@
 int
 main (void)
 {
-  volatile __float128 a = __builtin_nan (""), b = 0;
+  volatile __float128 a = __builtin_nanq (""), b = 0.0q;
   volatile int r = a < b;
   if (!fetestexcept (FE_INVALID))
     abort ();



More information about the Gcc-patches mailing list