/*
* 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.
*/
#include "compiler.h"
#include "glheader.h"
-
+#include "errors.h"
#ifdef __cplusplus
extern "C" {
/** Memory macros */
/*@{*/
-/** Allocate \p BYTES bytes */
-#define MALLOC(BYTES) malloc(BYTES)
-/** Allocate and zero \p BYTES bytes */
-#define CALLOC(BYTES) calloc(1, BYTES)
/** Allocate a structure of type \p T */
#define MALLOC_STRUCT(T) (struct T *) malloc(sizeof(struct T))
/** Allocate and zero a structure of type \p T */
#define CALLOC_STRUCT(T) (struct T *) calloc(1, sizeof(struct T))
-/** Free memory */
-#define FREE(PTR) free(PTR)
/*@}*/
* 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))
#endif
#if defined(_MSC_VER)
-static INLINE float truncf(float x) { return x < 0.0f ? ceilf(x) : floorf(x); }
-static INLINE float exp2f(float x) { return powf(2.0f, x); }
-static INLINE float log2f(float x) { return logf(x) * 1.442695041f; }
-static INLINE float asinhf(float x) { return logf(x + sqrtf(x * x + 1.0f)); }
-static INLINE float acoshf(float x) { return logf(x + sqrtf(x * x - 1.0f)); }
-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'; }
+#if _MSC_VER < 1800 /* Not req'd on VS2013 and above */
+static inline float truncf(float x) { return x < 0.0f ? ceilf(x) : floorf(x); }
+static inline float exp2f(float x) { return powf(2.0f, x); }
+static inline float log2f(float x) { return logf(x) * 1.442695041f; }
+static inline float asinhf(float x) { return logf(x + sqrtf(x * x + 1.0f)); }
+static inline float acoshf(float x) { return logf(x + sqrtf(x * x - 1.0f)); }
+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)
+#endif /* _MSC_VER < 1800 */
+#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)
+static inline GLfloat LOG2(GLfloat x)
{
+#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))
-#endif
+
/***
*** IS_INF_OR_NAN: test if float is infinite or NaN
***/
-#ifdef USE_IEEE
-static INLINE int IS_INF_OR_NAN( float x )
-{
- fi_type tmp;
- tmp.f = x;
- return !(int)((unsigned int)((tmp.i & 0x7fffffff)-0x7f800000) >> 31);
-}
-#elif defined(isfinite)
+#if defined(isfinite)
#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
#endif
-/***
- *** IROUND: return (as an integer) float rounded to nearest integer
- ***/
-#if defined(USE_X86_ASM) && defined(__GNUC__) && defined(__i386__)
-static INLINE int iround(float f)
+/**
+ * Convert float to int by rounding to nearest integer, away from zero.
+ */
+static inline int IROUND(float f)
{
+ return (int) ((f >= 0.0F) ? (f + 0.5F) : (f - 0.5F));
+}
+
+
+/**
+ * Convert float to int64 by rounding to nearest integer.
+ */
+static inline GLint64 IROUND64(float f)
+{
+ return (GLint64) ((f >= 0.0F) ? (f + 0.5F) : (f - 0.5F));
+}
+
+
+/**
+ * Convert positive float to int by rounding to nearest integer.
+ */
+static inline int IROUND_POS(float f)
+{
+ assert(f >= 0.0F);
+ return (int) (f + 0.5F);
+}
+
+#ifdef __x86_64__
+# include <xmmintrin.h>
+#endif
+
+/**
+ * Convert float to int using a fast method. The rounding mode may vary.
+ */
+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;
-}
-#define IROUND(x) iround(x)
#elif defined(USE_X86_ASM) && defined(_MSC_VER)
-static INLINE int iround(float f)
-{
int r;
_asm {
fld f
fistp r
}
return r;
-}
-#define IROUND(x) iround(x)
-#elif defined(__WATCOMC__) && defined(__386__)
-long iround(float f);
-#pragma aux iround = \
- "push eax" \
- "fistp dword ptr [esp]" \
- "pop eax" \
- parm [8087] \
- value [eax] \
- modify exact [eax];
-#define IROUND(x) iround(x)
+#elif defined(__x86_64__)
+ return _mm_cvt_ss2si(_mm_load_ss(&f));
#else
-#define IROUND(f) ((int) (((f) >= 0.0F) ? ((f) + 0.5F) : ((f) - 0.5F)))
-#endif
-
-#define IROUND64(f) ((GLint64) (((f) >= 0.0F) ? ((f) + 0.5F) : ((f) - 0.5F)))
-
-/***
- *** IROUND_POS: return (as an integer) positive float rounded to nearest int
- ***/
-#ifdef DEBUG
-#define IROUND_POS(f) (assert((f) >= 0.0F), IROUND(f))
-#else
-#define IROUND_POS(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)
-{
+#else
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)
-{
+#else
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
+}
/**
* Is x a power of two?
*/
-static INLINE int
+static inline int
_mesa_is_pow_two(int x)
{
return !(x & (x - 1));
* results would be different depending on optimization
* level used for build.
*/
-static INLINE int32_t
+static inline int32_t
_mesa_next_pow_two_32(uint32_t x)
{
-#if defined(__GNUC__) && \
- ((__GNUC__ == 3 && __GNUC_MINOR__ >= 4) || __GNUC__ >= 4)
+#ifdef HAVE___BUILTIN_CLZ
uint32_t y = (x != 1);
return (1 + y) << ((__builtin_clz(x - y) ^ 31) );
#else
#endif
}
-static INLINE int64_t
+static inline int64_t
_mesa_next_pow_two_64(uint64_t x)
{
-#if defined(__GNUC__) && \
- ((__GNUC__ == 3 && __GNUC_MINOR__ >= 4) || __GNUC__ >= 4)
+#ifdef HAVE___BUILTIN_CLZLL
uint64_t y = (x != 1);
- if (sizeof(x) == sizeof(long))
- return (1 + y) << ((__builtin_clzl(x - y) ^ 63));
- else
- return (1 + y) << ((__builtin_clzll(x - y) ^ 63));
+ STATIC_ASSERT(sizeof(x) == sizeof(long long));
+ return (1 + y) << ((__builtin_clzll(x - y) ^ 63));
#else
x--;
x |= x >> 1;
/*
* Returns the floor form of binary logarithm for a 32-bit integer.
*/
-static INLINE GLuint
+static inline GLuint
_mesa_logbase2(GLuint n)
{
-#if defined(__GNUC__) && \
- ((__GNUC__ == 3 && __GNUC_MINOR__ >= 4) || __GNUC__ >= 4)
+#ifdef HAVE___BUILTIN_CLZ
return (31 - __builtin_clz(n | 1));
#else
GLuint pos = 0;
/**
* Return 1 if this is a little endian machine, 0 if big endian.
*/
-static INLINE GLboolean
+static inline GLboolean
_mesa_little_endian(void)
{
const GLuint ui = 1; /* intentionally not static */
extern void
_mesa_exec_free( void *addr );
-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);
-
-#ifdef __GNUC__
-
-#if defined(__MINGW32__) || defined(ANDROID)
+#ifndef FFS_DEFINED
+#define FFS_DEFINED 1
+#ifdef HAVE___BUILTIN_FFS
#define ffs __builtin_ffs
+#else
+extern int ffs(int i);
+#endif
+
+#ifdef HAVE___BUILTIN_FFSLL
#define ffsll __builtin_ffsll
+#else
+extern int ffsll(long long int i);
#endif
+#endif /* FFS_DEFINED */
-#define _mesa_ffs(i) ffs(i)
-#define _mesa_ffsll(i) ffsll(i)
-#if ((_GNUC__ == 3 && __GNUC_MINOR__ >= 4) || __GNUC__ >= 4)
+#ifdef HAVE___BUILTIN_POPCOUNT
#define _mesa_bitcount(i) __builtin_popcount(i)
#else
extern unsigned int
_mesa_bitcount(unsigned int n);
#endif
+#ifdef HAVE___BUILTIN_POPCOUNTLL
+#define _mesa_bitcount_64(i) __builtin_popcountll(i)
#else
-extern int
-_mesa_ffs(int32_t i);
+extern unsigned int
+_mesa_bitcount_64(uint64_t n);
+#endif
-extern int
-_mesa_ffsll(int64_t i);
+/**
+ * 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)
+{
+#ifdef HAVE___BUILTIN_CLZ
+ return n == 0 ? 0 : 32 - __builtin_clz(n);
+#else
+ unsigned int v = 1;
-extern unsigned int
-_mesa_bitcount(unsigned int n);
+ 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);
extern float
_mesa_half_to_float(GLhalfARB h);
-
-extern void *
-_mesa_bsearch( const void *key, const void *base, size_t nmemb, size_t size,
- int (*compar)(const void *, const void *) );
-
-extern char *
-_mesa_getenv( const char *var );
+static inline bool
+_mesa_half_is_negative(GLhalfARB h)
+{
+ return h & 0x8000;
+}
extern char *
_mesa_strdup( const char *s );
extern int
_mesa_snprintf( char *str, size_t size, const char *fmt, ... ) PRINTFLIKE(3, 4);
-struct gl_context;
-
-extern void
-_mesa_warning( struct gl_context *gc, const char *fmtString, ... ) PRINTFLIKE(2, 3);
-
-extern void
-_mesa_problem( const struct gl_context *ctx, const char *fmtString, ... ) PRINTFLIKE(2, 3);
-
-extern void
-_mesa_error( struct gl_context *ctx, GLenum error, const char *fmtString, ... ) PRINTFLIKE(3, 4);
-
-extern void
-_mesa_debug( const struct gl_context *ctx, const char *fmtString, ... ) PRINTFLIKE(2, 3);
+extern int
+_mesa_vsnprintf(char *str, size_t size, const char *fmt, va_list arg);
#if defined(_MSC_VER) && !defined(snprintf)
projects / mesa.git / blobdiff