--- /dev/null
+
+#include <stdint.h>
+#include "platform.h"
+#include "specialize.h"
+
+/*----------------------------------------------------------------------------
+| Returns the result of converting the canonical NaN `a' to the double-
+| precision floating-point format.
+*----------------------------------------------------------------------------*/
+
+uint_fast64_t softfloat_commonNaNToF64UI( struct commonNaN a )
+{
+
+ return
+ (uint_fast64_t) a.sign<<63 | UINT64_C( 0x7FF8000000000000 )
+ | a.v64>>12;
+
+}
+