softfloat: Fix the default qNAN for target-ppc

Currently float128_default_nan() returns 0xFFFF800000000000 in the
higher double word, but it should return 0x7FFF800000000000 which
is the correct higher double word for default qNAN on PowerPC.

Signed-off-by: Bharata B Rao <bharata@linux.vnet.ibm.com>
Signed-off-by: Nikunj A Dadhania <nikunj@linux.vnet.ibm.com>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>

fpu/softfloat-specialize.h

@@ -185,7 +185,7 @@ float128 float128_default_nan(float_status *status)
         r.high = LIT64(0x7FFF7FFFFFFFFFFF);
     } else {
         r.low = LIT64(0x0000000000000000);
-#if defined(TARGET_S390X)
+#if defined(TARGET_S390X) || defined(TARGET_PPC)
         r.high = LIT64(0x7FFF800000000000);
 #else
         r.high = LIT64(0xFFFF800000000000);