msvcr120: Add fmax implementation.

dlls/msvcr120/msvcr120.spec

@@ -2164,9 +2164,9 @@
 @ stub fma
 @ stub fmaf
 @ stub fmal
-@ stub fmax
-@ stub fmaxf
-@ stub fmaxl
+@ cdecl fmax(double double) MSVCR120_fmax
+@ cdecl fmaxf(float float) MSVCR120_fmaxf
+@ cdecl fmaxl(double double) MSVCR120_fmax
 @ stub fmin
 @ stub fminf
 @ stub fminl

dlls/msvcr120_app/msvcr120_app.spec

@@ -1833,9 +1833,9 @@
 @ stub fma
 @ stub fmaf
 @ stub fmal
-@ stub fmax
-@ stub fmaxf
-@ stub fmaxl
+@ cdecl fmax(double double) msvcr120.fmax
+@ cdecl fmaxf(float float) msvcr120.fmaxf
+@ cdecl fmaxl(double double) msvcr120.fmaxl
 @ stub fmin
 @ stub fminf
 @ stub fminl

dlls/msvcrt/math.c

@@ -2605,3 +2605,31 @@ short CDECL MSVCR120__ldtest(LDOUBLE *x)
 {
     return MSVCR120__dclass(*x);
 }
+
+/*********************************************************************
+ *      fmaxf (MSVCR120.@)
+ */
+float CDECL MSVCR120_fmaxf(float x, float y)
+{
+    if(isnanf(x))
+        return y;
+    if(isnanf(y))
+        return x;
+    if(x==0 && y==0)
+        return signbit(x) ? y : x;
+    return x<y ? y : x;
+}
+
+/*********************************************************************
+ *      fmax (MSVCR120.@)
+ */
+double CDECL MSVCR120_fmax(double x, double y)
+{
+    if(isnan(x))
+        return y;
+    if(isnan(y))
+        return x;
+    if(x==0 && y==0)
+        return signbit(x) ? y : x;
+    return x<y ? y : x;
+}