math: fix expm1f overflow threshold

the threshold was wrong so expm1f overflowed to inf a bit too early
and on most targets uint32_t compare is faster than float compare so
use that.

this also fixes sinhf incorrectly returning nan for some values where
the internal expm1f overflowed.

diff --git a/src/math/expm1f.c b/src/math/expm1f.c
index 297e0b4..09a41af 100644 (file)
--- a/src/math/expm1f.c
+++ b/src/math/expm1f.c
@@ -16,7 +16,6 @@
 #include "libm.h"
 
 static const float
-o_threshold = 8.8721679688e+01, /* 0x42b17180 */
 ln2_hi      = 6.9313812256e-01, /* 0x3f317180 */
 ln2_lo      = 9.0580006145e-06, /* 0x3717f7d1 */
 invln2      = 1.4426950216e+00, /* 0x3fb8aa3b */
@@ -41,7 +40,7 @@ float expm1f(float x)
                        return x;
                if (sign)
                        return -1;
-               if (x > o_threshold) {
+               if (hx > 0x42b17217) { /* x > log(FLT_MAX) */
                        x *= 0x1p127f;
                        return x;
                }