glsl: lower mediump temporaries to 16 bits except structures (v2)

[mesa.git] / src / mesa / main / macros.h

diff --git a/src/mesa/main/macros.h b/src/mesa/main/macros.h
index 0608650aeb4844f2b0b6cde706de88f31a8f461e..bd07fa71bc662e2a1e52384df969053e80aa4c40 100644 (file)
--- a/src/mesa/main/macros.h
+++ b/src/mesa/main/macros.h
@@ -33,7 +33,10 @@
 
 #include "util/macros.h"
 #include "util/u_math.h"
-#include "imports.h"
+#include "util/rounding.h"
+#include "util/compiler.h"
+#include "main/glheader.h"
+#include "mesa_private.h"
 
 
 /**
@@ -131,12 +134,12 @@ extern GLfloat _mesa_ubyte_to_float_color_tab[256];
 #define INT_TO_USHORT(i)   ((i) < 0 ? 0 : ((GLushort) ((i) >> 15)))
 #define UINT_TO_USHORT(i)  ((i) < 0 ? 0 : ((GLushort) ((i) >> 16)))
 #define UNCLAMPED_FLOAT_TO_USHORT(us, f)  \
-        us = ( (GLushort) F_TO_I( CLAMP((f), 0.0F, 1.0F) * 65535.0F) )
+        us = ( (GLushort) _mesa_lroundevenf( CLAMP((f), 0.0F, 1.0F) * 65535.0F) )
 #define CLAMPED_FLOAT_TO_USHORT(us, f)  \
-        us = ( (GLushort) F_TO_I( (f) * 65535.0F) )
+        us = ( (GLushort) _mesa_lroundevenf( (f) * 65535.0F) )
 
 #define UNCLAMPED_FLOAT_TO_SHORT(s, f)  \
-        s = ( (GLshort) F_TO_I( CLAMP((f), -1.0F, 1.0F) * 32767.0F) )
+        s = ( (GLshort) _mesa_lroundevenf( CLAMP((f), -1.0F, 1.0F) * 32767.0F) )
 
 /***
  *** UNCLAMPED_FLOAT_TO_UBYTE: clamp float to [0,1] and map to ubyte in [0,255]
@@ -167,9 +170,9 @@ extern GLfloat _mesa_ubyte_to_float_color_tab[256];
         } while (0)
 #else
 #define UNCLAMPED_FLOAT_TO_UBYTE(ub, f) \
-       ub = ((GLubyte) F_TO_I(CLAMP((f), 0.0F, 1.0F) * 255.0F))
+       ub = ((GLubyte) _mesa_lroundevenf(CLAMP((f), 0.0F, 1.0F) * 255.0F))
 #define CLAMPED_FLOAT_TO_UBYTE(ub, f) \
-       ub = ((GLubyte) F_TO_I((f) * 255.0F))
+       ub = ((GLubyte) _mesa_lroundevenf((f) * 255.0F))
 #endif
 
 static fi_type UINT_AS_UNION(GLuint u)
@@ -193,6 +196,30 @@ static inline fi_type FLOAT_AS_UNION(GLfloat f)
    return tmp;
 }
 
+static inline uint64_t DOUBLE_AS_UINT64(double d)
+{
+   union {
+      double d;
+      uint64_t u64;
+   } tmp;
+   tmp.d = d;
+   return tmp.u64;
+}
+
+static inline double UINT64_AS_DOUBLE(uint64_t u)
+{
+   union {
+      double d;
+      uint64_t u64;
+   } tmp;
+   tmp.u64 = u;
+   return tmp.d;
+}
+
+/* First sign-extend x, then return uint32_t. */
+#define INT_AS_UINT(x) ((uint32_t)((int32_t)(x)))
+#define FLOAT_AS_UINT(x) (FLOAT_AS_UNION(x).u)
+
 /**
  * Convert a floating point value to an unsigned fixed point value.
  *
@@ -277,14 +304,14 @@ COPY_4UBV(GLubyte dst[4], const GLubyte src[4])
 }
 
 /** Copy \p SZ elements into a 4-element vector */
-#define COPY_SZ_4V(DST, SZ, SRC)  \
-do {                              \
-   switch (SZ) {                  \
-   case 4: (DST)[3] = (SRC)[3];   \
-   case 3: (DST)[2] = (SRC)[2];   \
-   case 2: (DST)[1] = (SRC)[1];   \
-   case 1: (DST)[0] = (SRC)[0];   \
-   }                              \
+#define COPY_SZ_4V(DST, SZ, SRC)                  \
+do {                                              \
+   switch (SZ) {                                  \
+   case 4: (DST)[3] = (SRC)[3]; /* fallthrough */ \
+   case 3: (DST)[2] = (SRC)[2]; /* fallthrough */ \
+   case 2: (DST)[1] = (SRC)[1]; /* fallthrough */ \
+   case 1: (DST)[0] = (SRC)[0]; /* fallthrough */ \
+   }                                              \
 } while(0)
 
 /** Copy \p SZ elements into a homegeneous (4-element) vector, giving
@@ -659,62 +686,12 @@ INTERP_4F(GLfloat t, GLfloat dst[4], const GLfloat out[4], const GLfloat in[4])
 
 
 
-/** Clamp X to [MIN,MAX] */
-#define CLAMP( X, MIN, MAX )  ( (X)<(MIN) ? (MIN) : ((X)>(MAX) ? (MAX) : (X)) )
-
-/** Minimum of two values: */
-#define MIN2( A, B )   ( (A)<(B) ? (A) : (B) )
-
-/** Maximum of two values: */
-#define MAX2( A, B )   ( (A)>(B) ? (A) : (B) )
-
-/** Minimum and maximum of three values: */
-#define MIN3( A, B, C ) ((A) < (B) ? MIN2(A, C) : MIN2(B, C))
-#define MAX3( A, B, C ) ((A) > (B) ? MAX2(A, C) : MAX2(B, C))
-
 static inline unsigned
 minify(unsigned value, unsigned levels)
 {
     return MAX2(1, value >> levels);
 }
 
-/**
- * Return true if the given value is a power of two.
- *
- * Note that this considers 0 a power of two.
- */
-static inline bool
-is_power_of_two(unsigned value)
-{
-   return (value & (value - 1)) == 0;
-}
-
-/**
- * Align a value up to an alignment value
- *
- * If \c value is not already aligned to the requested alignment value, it
- * will be rounded up.
- *
- * \param value  Value to be rounded
- * \param alignment  Alignment value to be used.  This must be a power of two.
- *
- * \sa ROUND_DOWN_TO()
- */
-#define ALIGN(value, alignment)  (((value) + (alignment) - 1) & ~((alignment) - 1))
-
-/**
- * Align a value down to an alignment value
- *
- * If \c value is not already aligned to the requested alignment value, it
- * will be rounded down.
- *
- * \param value  Value to be rounded
- * \param alignment  Alignment value to be used.  This must be a power of two.
- *
- * \sa ALIGN()
- */
-#define ROUND_DOWN_TO(value, alignment) ((value) & ~(alignment - 1))
-
 
 /** Cross product of two 3-element vectors */
 static inline void
@@ -811,4 +788,13 @@ DIFFERENT_SIGNS(GLfloat x, GLfloat y)
 /* Stringify */
 #define STRINGIFY(x) #x
 
+/*
+ * For GL_ARB_vertex_buffer_object we need to treat vertex array pointers
+ * as offsets into buffer stores.  Since the vertex array pointer and
+ * buffer store pointer are both pointers and we need to add them, we use
+ * this macro.
+ * Both pointers/offsets are expressed in bytes.
+ */
+#define ADD_POINTERS(A, B)  ( (GLubyte *) (A) + (uintptr_t) (B) )
+
 #endif