[PR86438] avoid too-long shift in test

The test fell back to long long and long when __int128 is not
available, but it assumed sizeof(long) < sizeof(long long) because of
a shift count that would be out of range for a long long if their
widths are the same.  Fixed by splitting it up into two shifts.


for  gcc/testsuite/ChangeLog

	PR rtl-optimization/86438
	* gcc.dg/torture/pr86438.c: Split up too-wide shift.

From-SVN: r270388

gcc/testsuite/ChangeLog

+2019-04-16  Alexandre Oliva <aoliva@redhat.com>
+
+	PR rtl-optimization/86438
+	* gcc.dg/torture/pr86438.c: Split up too-wide shift.
+
 2019-04-16  Jakub Jelinek  <jakub@redhat.com>
 
 	PR target/90096

gcc/testsuite/gcc.dg/torture/pr86438.c

@@ -24,6 +24,6 @@ main (void)
   u64 d = (g ? 5 : 4);
   u32 f = __builtin_sub_overflow_p (d, (u128) d, (u64) 0);
   u128 x = g + f + d;
-  check (x >> (sizeof (u64) * __CHAR_BIT__), x);
+  check ((x >> 1) >> (sizeof (u64) * __CHAR_BIT__ - 1), x);
   return 0;
 }