if (eattr & FLOAT)
{
double fval = *(double *)&eval;
-
- if (fval >= 1)
- {
- warn("value clamped to +1.");
- eval = 0x7fffff;
- }
- else if (fval <= -1)
- {
- warn("value clamped to -1.");
- eval = 0x800000;
- }
- else
- {
- // Convert fraction to 24 bits fixed point with sign and rounding
- // Yeah, that cast to int32_t has to be there because casting
- // a float to unsigned int is "undefined" according to the C
- // standard. Which most compilers seem to do the sensible thing
- // and just cast the f**king value properly, except gcc 4.x.x
- // for arm (tested on raspbian).
- // Thanks, C and gcc! Thanks for making me waste a few hours \o/
- eval = 0;//!!! FIX !!! (uint32_t)(int32_t)round(fval*(1 << 23));
- }
+ eval = DoubleToDSPFloat(fval);
}
else
{
else
{
// In L: we deposit stuff to both X: and Y: instead
- // We will be a bit lazy and require that there is a 2nd value in the same source line.
- // (Motorola's assembler can parse 12-digit hex values, which we can't do at the moment)
- // This of course requires to parse 2 values in one pass.
- // If there isn't another value in this line, assume X: value is 0.
+ // We will be a bit lazy and require that there is a 2nd value
+ // in the same source line. (Motorola's assembler can parse
+ // 12-digit hex values, which we can't do at the moment) This
+ // of course requires to parse 2 values in one pass. If there
+ // isn't another value in this line, assume X: value is 0.
int secondword = 0;
uint32_t evaly;
l_parse_loop:
if (eattr & FLOAT)
{
float fval = *(float *)&eval;
- if (fval >= 1)
- {
- warn("value clamped to +1.");
- eval = 0x7fffff;
- }
- else if (fval <= -1)
- {
- warn("value clamped to -1.");
- eval = 0x800000;
- }
- else
- {
- // Convert fraction to 24 bits fixed point with sign and rounding
- // Yeah, that cast to int32_t has to be there because casting
- // a float to unsigned int is "undefined" according to the C
- // standard. Which most compilers seem to do the sensible thing
- // and just cast the f**king value properly, except gcc 4.x.x
- // for arm (tested on raspbian).
- // Thanks, C and gcc! Thanks for making me waste a few hours \o/
- eval = 0;//!!! FIX !!! (uint32_t)(int32_t)round(fval*(1 << 23));
- }
+ eval = DoubleToDSPFloat(fval);
}
else
{
}
}
+
goto comma;
}
#include <float.h>
#include <math.h>
#include <stdio.h>
+#include "error.h"
//
// Check for IEEE-754 conformance (C99 compilers should be OK here)
}
+//
+// Convert a double to a DSP56001 style fixed point float.
+// Seems to be 23 bits of float value with 1 bit (MSB) for the sign.
+//
+uint32_t DoubleToDSPFloat(double d)
+{
+ if (d >= 1)
+ {
+ warn("DSP value clamped to +1.");
+ return 0x7FFFFF;
+ }
+ else if (d <= -1)
+ {
+ warn("DSP value clamped to -1.");
+ return 0x800000;
+ }
+
+ return trunc(round(ldexp(d, 23)));
+}
+
+
//
// Convert a host native floating point number to a fixed point number.
//