/*
* Mesa 3-D graphics library
- * Version: 7.5
*
* Copyright (C) 1999-2008 Brian Paul All Rights Reserved.
*
* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
* OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
* FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
- * BRIAN PAUL BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
- * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
- * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
*/
* these casts generate warnings.
* The following union typedef is used to solve that.
*/
-typedef union { GLfloat f; GLint i; } fi_type;
+typedef union { GLfloat f; GLint i; GLuint u; } fi_type;
#define DEG2RAD (M_PI/180.0)
-/***
- *** SQRTF: single-precision square root
- ***/
-#if 0 /* _mesa_sqrtf() not accurate enough - temporarily disabled */
-# define SQRTF(X) _mesa_sqrtf(X)
-#else
-# define SQRTF(X) (float) sqrt((float) (X))
-#endif
-
-
-/***
- *** INV_SQRTF: single-precision inverse square root
- ***/
-#if 0
-#define INV_SQRTF(X) _mesa_inv_sqrt(X)
-#else
-#define INV_SQRTF(X) (1.0F / SQRTF(X)) /* this is faster on a P4 */
-#endif
-
-
/**
* \name Work-arounds for platforms that lack C99 math functions
*/
#define asinf(f) ((float) asin(f))
#define atan2f(x,y) ((float) atan2(x,y))
#define atanf(f) ((float) atan(f))
-#define cielf(f) ((float) ciel(f))
+#define ceilf(f) ((float) ceil(f))
#define cosf(f) ((float) cos(f))
#define coshf(f) ((float) cosh(f))
#define expf(f) ((float) exp(f))
static inline float atanhf(float x) { return (logf(1.0f + x) - logf(1.0f - x)) / 2.0f; }
static inline int isblank(int ch) { return ch == ' ' || ch == '\t'; }
#define strtoll(p, e, b) _strtoi64(p, e, b)
+#define strcasecmp(s1, s2) _stricmp(s1, s2)
#endif
/*@}*/
+
+/*
+ * signbit() is a macro on Linux. Not available on Windows.
+ */
+#ifndef signbit
+#define signbit(x) ((x) < 0.0f)
+#endif
+
+
+/** single-precision inverse square root */
+static inline float
+INV_SQRTF(float x)
+{
+ /* XXX we could try Quake's fast inverse square root function here */
+ return 1.0F / sqrtf(x);
+}
+
+
/***
*** LOG2: Log base 2 of float
***/
-#ifdef USE_IEEE
-#if 0
-/* This is pretty fast, but not accurate enough (only 2 fractional bits).
- * Based on code from http://www.stereopsis.com/log2.html
- */
static inline GLfloat LOG2(GLfloat x)
{
+#ifdef USE_IEEE
+#if 0
+ /* This is pretty fast, but not accurate enough (only 2 fractional bits).
+ * Based on code from http://www.stereopsis.com/log2.html
+ */
const GLfloat y = x * x * x * x;
const GLuint ix = *((GLuint *) &y);
const GLuint exp = (ix >> 23) & 0xFF;
const GLint log2 = ((GLint) exp) - 127;
return (GLfloat) log2 * (1.0 / 4.0); /* 4, because of x^4 above */
-}
#endif
-/* Pretty fast, and accurate.
- * Based on code from http://www.flipcode.com/totd/
- */
-static inline GLfloat LOG2(GLfloat val)
-{
+ /* Pretty fast, and accurate.
+ * Based on code from http://www.flipcode.com/totd/
+ */
fi_type num;
GLint log_2;
- num.f = val;
+ num.f = x;
log_2 = ((num.i >> 23) & 255) - 128;
num.i &= ~(255 << 23);
num.i += 127 << 23;
num.f = ((-1.0f/3) * num.f + 2) * num.f - 2.0f/3;
return num.f + log_2;
-}
#else
-/*
- * NOTE: log_base_2(x) = log(x) / log(2)
- * NOTE: 1.442695 = 1/log(2).
- */
-#define LOG2(x) ((GLfloat) (log(x) * 1.442695F))
+ /*
+ * NOTE: log_base_2(x) = log(x) / log(2)
+ * NOTE: 1.442695 = 1/log(2).
+ */
+ return (GLfloat) (log(x) * 1.442695F);
#endif
+}
+
/***
#define IS_INF_OR_NAN(x) (!isfinite(x))
#elif defined(finite)
#define IS_INF_OR_NAN(x) (!finite(x))
-#elif defined(__VMS)
-#define IS_INF_OR_NAN(x) (!finite(x))
#elif defined(__STDC_VERSION__) && __STDC_VERSION__ >= 199901L
#define IS_INF_OR_NAN(x) (!isfinite(x))
#else
#endif
-/***
- *** IS_NEGATIVE: test if float is negative
- ***/
-#if defined(USE_IEEE)
-static inline int GET_FLOAT_BITS( float x )
-{
- fi_type fi;
- fi.f = x;
- return fi.i;
-}
-#define IS_NEGATIVE(x) (GET_FLOAT_BITS(x) < 0)
-#else
-#define IS_NEGATIVE(x) (x < 0.0F)
-#endif
-
-
-/***
- *** DIFFERENT_SIGNS: test if two floats have opposite signs
- ***/
-#if defined(USE_IEEE)
-#define DIFFERENT_SIGNS(x,y) ((GET_FLOAT_BITS(x) ^ GET_FLOAT_BITS(y)) & (1<<31))
-#else
-/* Could just use (x*y<0) except for the flatshading requirements.
- * Maybe there's a better way?
- */
-#define DIFFERENT_SIGNS(x,y) ((x) * (y) <= 0.0F && (x) - (y) != 0.0F)
-#endif
-
-
/***
*** CEILF: ceiling of float
*** FLOORF: floor of float
* Convert float to int using a fast method. The rounding mode may vary.
* XXX We could use an x86-64/SSE2 version here.
*/
-#if defined(USE_X86_ASM) && defined(__GNUC__) && defined(__i386__)
static inline int F_TO_I(float f)
{
+#if defined(USE_X86_ASM) && defined(__GNUC__) && defined(__i386__)
int r;
__asm__ ("fistpl %0" : "=m" (r) : "t" (f) : "st");
return r;
-}
#elif defined(USE_X86_ASM) && defined(_MSC_VER)
-static inline int F_TO_I(float f)
-{
int r;
_asm {
fld f
fistp r
}
return r;
-}
-#elif defined(__WATCOMC__) && defined(__386__)
-long F_TO_I(float f);
-#pragma aux iround = \
- "push eax" \
- "fistp dword ptr [esp]" \
- "pop eax" \
- parm [8087] \
- value [eax] \
- modify exact [eax];
#else
-#define F_TO_I(f) IROUND(f)
+ return IROUND(f);
#endif
+}
-/***
- *** IFLOOR: return (as an integer) floor of float
- ***/
-#if defined(USE_X86_ASM) && defined(__GNUC__) && defined(__i386__)
-/*
- * IEEE floor for computers that round to nearest or even.
- * 'f' must be between -4194304 and 4194303.
- * This floor operation is done by "(iround(f + .5) + iround(f - .5)) >> 1",
- * but uses some IEEE specific tricks for better speed.
- * Contributed by Josh Vanderhoof
- */
-static inline int ifloor(float f)
+/** Return (as an integer) floor of float */
+static inline int IFLOOR(float f)
{
+#if defined(USE_X86_ASM) && defined(__GNUC__) && defined(__i386__)
+ /*
+ * IEEE floor for computers that round to nearest or even.
+ * 'f' must be between -4194304 and 4194303.
+ * This floor operation is done by "(iround(f + .5) + iround(f - .5)) >> 1",
+ * but uses some IEEE specific tricks for better speed.
+ * Contributed by Josh Vanderhoof
+ */
int ai, bi;
double af, bf;
af = (3 << 22) + 0.5 + (double)f;
__asm__ ("fstps %0" : "=m" (ai) : "t" (af) : "st");
__asm__ ("fstps %0" : "=m" (bi) : "t" (bf) : "st");
return (ai - bi) >> 1;
-}
-#define IFLOOR(x) ifloor(x)
#elif defined(USE_IEEE)
-static inline int ifloor(float f)
-{
int ai, bi;
double af, bf;
fi_type u;
-
af = (3 << 22) + 0.5 + (double)f;
bf = (3 << 22) + 0.5 - (double)f;
u.f = (float) af; ai = u.i;
u.f = (float) bf; bi = u.i;
return (ai - bi) >> 1;
-}
-#define IFLOOR(x) ifloor(x)
#else
-static inline int ifloor(float f)
-{
int i = IROUND(f);
return (i > f) ? i - 1 : i;
-}
-#define IFLOOR(x) ifloor(x)
#endif
+}
-/***
- *** ICEIL: return (as an integer) ceiling of float
- ***/
-#if defined(USE_X86_ASM) && defined(__GNUC__) && defined(__i386__)
-/*
- * IEEE ceil for computers that round to nearest or even.
- * 'f' must be between -4194304 and 4194303.
- * This ceil operation is done by "(iround(f + .5) + iround(f - .5) + 1) >> 1",
- * but uses some IEEE specific tricks for better speed.
- * Contributed by Josh Vanderhoof
- */
-static inline int iceil(float f)
+/** Return (as an integer) ceiling of float */
+static inline int ICEIL(float f)
{
+#if defined(USE_X86_ASM) && defined(__GNUC__) && defined(__i386__)
+ /*
+ * IEEE ceil for computers that round to nearest or even.
+ * 'f' must be between -4194304 and 4194303.
+ * This ceil operation is done by "(iround(f + .5) + iround(f - .5) + 1) >> 1",
+ * but uses some IEEE specific tricks for better speed.
+ * Contributed by Josh Vanderhoof
+ */
int ai, bi;
double af, bf;
af = (3 << 22) + 0.5 + (double)f;
__asm__ ("fstps %0" : "=m" (ai) : "t" (af) : "st");
__asm__ ("fstps %0" : "=m" (bi) : "t" (bf) : "st");
return (ai - bi + 1) >> 1;
-}
-#define ICEIL(x) iceil(x)
#elif defined(USE_IEEE)
-static inline int iceil(float f)
-{
int ai, bi;
double af, bf;
fi_type u;
u.f = (float) af; ai = u.i;
u.f = (float) bf; bi = u.i;
return (ai - bi + 1) >> 1;
-}
-#define ICEIL(x) iceil(x)
#else
-static inline int iceil(float f)
-{
int i = IROUND(f);
return (i < f) ? i + 1 : i;
-}
-#define ICEIL(x) iceil(x)
#endif
+}
/**
extern void *
_mesa_realloc( void *oldBuffer, size_t oldSize, size_t newSize );
-extern void
-_mesa_memset16( unsigned short *dst, unsigned short val, size_t n );
-
-extern double
-_mesa_sqrtd(double x);
-
-extern float
-_mesa_sqrtf(float x);
-
-extern float
-_mesa_inv_sqrtf(float x);
-
-extern void
-_mesa_init_sqrt_table(void);
-
#ifndef FFS_DEFINED
#define FFS_DEFINED 1
#ifdef __GNUC__
-
-#if defined(__MINGW32__) || defined(__CYGWIN__) || defined(ANDROID) || defined(__APPLE__)
#define ffs __builtin_ffs
#define ffsll __builtin_ffsll
-#endif
-
#else
-
extern int ffs(int i);
extern int ffsll(long long int i);
-
#endif /*__ GNUC__ */
#endif /* FFS_DEFINED */
_mesa_bitcount_64(uint64_t n);
#endif
+/**
+ * Find the last (most significant) bit set in a word.
+ *
+ * Essentially ffs() in the reverse direction.
+ */
+static inline unsigned int
+_mesa_fls(unsigned int n)
+{
+#if defined(__GNUC__) && ((__GNUC__ * 100 + __GNUC_MINOR__) >= 304)
+ return n == 0 ? 0 : 32 - __builtin_clz(n);
+#else
+ unsigned int v = 1;
+
+ if (n == 0)
+ return 0;
+
+ while (n >>= 1)
+ v++;
+
+ return v;
+#endif
+}
+
+extern int
+_mesa_round_to_even(float val);
extern GLhalfARB
_mesa_float_to_half(float f);
projects / mesa.git / blobdiff