Fix ldbl-96 scalblnl underflowing results (bug 17803).

The ldbl-96 implementation of scalblnl (used for x86_64 and ia64) uses
a condition k <= -63 to determine when a standard underflowing result
tiny*__copysignl(tiny,x) should be returned.  However, that condition
corresponds to values with exponent -16446 or less, and in the case of
-16446, the correct result for round-to-nearest depends on whether the
value is exactly 0x1p-16446 (half the least subnormal) or more than
that.  This patch fixes the bug by changing the condition to k <= -64
and accordingly adjusting the exponent by 64 not 63 when converting to
a normal value.

Tested for x86_64.

	[BZ #17803]
	* sysdeps/ieee754/ldbl-96/s_scalblnl.c (twom63): Rename to
	twom64.  Adjust value to 0x1p-64L.
	(__scalblnl): Only return standard underflowing result for K <=
	-64 not K <= -63; adjust exponent for underflowing result by 64
	not 63.
	* math/libm-test.inc (scalbn_test_data): Add more tests.
	(scalbln_test_data): Likewise.

1 files changed, 4 insertions, 4 deletions

diff --git a/sysdeps/ieee754/ldbl-96/s_scalblnl.c b/sysdeps/ieee754/ldbl-96/s_scalblnl.c
index 5c2f38eafd..457e999c6c 100644
--- a/sysdeps/ieee754/ldbl-96/s_scalblnl.c
+++ b/sysdeps/ieee754/ldbl-96/s_scalblnl.c
@@ -26,7 +26,7 @@
 
 static const long double
 two63   =  0x1p63L,
-twom63  =  1.08420217248550443400e-19,
+twom64  =  0x1p-64L,
 huge   = 1.0e+4900L,
 tiny   = 1.0e-4900L;
 
@@ -52,9 +52,9 @@ __scalblnl (long double x, long int n)
 	k = k+n;
 	if (__builtin_expect(k > 0, 1))		/* normal result */
 	    {SET_LDOUBLE_EXP(x,(es&0x8000)|k); return x;}
-	if (k <= -63)
+	if (k <= -64)
 	    return tiny*__copysignl(tiny,x); 	/*underflow*/
-	k += 63;				/* subnormal result */
+	k += 64;				/* subnormal result */
 	SET_LDOUBLE_EXP(x,(es&0x8000)|k);
-	return x*twom63;
+	return x*twom64;
 }