Add floor32, ceil32, round32 compatibility functions for burnings renderer.

To avoid changing burnings now those functions have no IRRLICHT_FAST_MATH anymore, 
there's a new header irrMathFastCompat.h which has ..._fast functions doing the old behavior.
With the troubles they have documented.
I changed burnings to use those functions throughout.
Or as much as possible... Burnings probably also uses classes like SColor which also have functions 
using those, but I don't plan  to adapt them.
Maybe IRRLICHT_FAST_MATH should be a flag exlusive to burnings in the future, I don't think it makes 
much sense otherwise anymore (it often expects 32-bit asm).

git-svn-id: svn://svn.code.sf.net/p/irrlicht/code/trunk@6012 dfc29bdd-3216-0410-991c-e03cc46cb475

source/Irrlicht/irrMathFastCompat.h

@@ -0,0 +1,126 @@
+// This file is part of the "Irrlicht Engine".
+// For conditions of distribution and use, see copyright notice in irrlicht.h
+
+#ifndef __IRR_FAST_MATH_COMPAT_H_INCLUDED__
+#define __IRR_FAST_MATH_COMPAT_H_INCLUDED__
+
+#include "irrMath.h"
+
+namespace irr
+{
+namespace core
+{
+
+
+// IRRLICHT_FAST_MATH functions which I wanted to kick out because they return
+// wrong results. But last time I proposed that I've been asked to keep them for 
+// Burnings software renderer. So to avoid changing that accidentally or messing up
+// it's speed I'll keep them around, but only as internal header.
+// They should not be used otherwise any longer.
+
+	// Some examples for unexpected results when using this with IRRLICHT_FAST_MATH:
+	// Input 1, expected 1, got 0
+	// Input 3, expected 3, got 2
+	// Input -1.40129846e-45, expected -1, got 0
+	REALINLINE s32 floor32_fast(f32 x)
+	{
+#ifdef IRRLICHT_FAST_MATH
+		const f32 h = 0.5f;
+
+		s32 t;
+
+#if defined(_MSC_VER) && !defined(_WIN64)
+		__asm
+		{
+			fld	x
+			fsub	h
+			fistp	t
+		}
+#elif defined(__GNUC__)
+		__asm__ __volatile__ (
+			"fsub %2 \n\t"
+			"fistpl %0"
+			: "=m" (t)
+			: "t" (x), "f" (h)
+			: "st"
+			);
+#else
+		return (s32) floorf ( x );
+#endif
+		return t;
+#else // no fast math
+		return (s32) floorf ( x );
+#endif
+	}
+	
+	// Some examples for unexpected results when using this with IRRLICHT_FAST_MATH:
+	// Input 1.40129846e-45, expected 1, got 0
+	// Input -1, expected -1, got 0
+	// Input -3, expected -3, got -2
+	REALINLINE s32 ceil32_fast ( f32 x )
+	{
+#ifdef IRRLICHT_FAST_MATH
+		const f32 h = 0.5f;
+
+		s32 t;
+
+#if defined(_MSC_VER) && !defined(_WIN64)
+		__asm
+		{
+			fld	x
+			fadd	h
+			fistp	t
+		}
+#elif defined(__GNUC__)
+		__asm__ __volatile__ (
+			"fadd %2 \n\t"
+			"fistpl %0 \n\t"
+			: "=m"(t)
+			: "t"(x), "f"(h)
+			: "st"
+			);
+#else
+		return (s32) ceilf ( x );
+#endif
+		return t;
+#else // not fast math
+		return (s32) ceilf ( x );
+#endif
+	}	
+	
+	// Some examples for unexpected results when using this with IRRLICHT_FAST_MATH:
+	// Input 0.5, expected 1, got 0
+	// Input 2.5, expected 3, got 2
+	// Input -1.40129846e-45, expected -nan(ind), got -inf
+	// Input -2.80259693e-45, expected -nan(ind), got -inf	
+	REALINLINE s32 round32_fast(f32 x)
+	{
+#if defined(IRRLICHT_FAST_MATH)
+		s32 t;
+
+#if defined(_MSC_VER) && !defined(_WIN64)
+		__asm
+		{
+			fld   x
+			fistp t
+		}
+#elif defined(__GNUC__)
+		__asm__ __volatile__ (
+			"fistpl %0 \n\t"
+			: "=m"(t)
+			: "t"(x)
+			: "st"
+			);
+#else
+		return (s32) round_(x);
+#endif
+		return t;
+#else // no fast math
+		return (s32) round_(x);
+#endif
+	}	
+	
+} // end namespace core
+} // end namespace irr	
+
+#endif // __IRR_FAST_MATH_COMPAT_H_INCLUDED__