Add a specific function for truncating float-to-int conversions

1 files changed, 3 insertions, 8 deletions

diff --git a/Alc/ALu.c b/Alc/ALu.c
index 177509b6..0c930a29 100644
--- a/Alc/ALu.c
+++ b/Alc/ALu.c
@@ -352,25 +352,20 @@ void aluSelectPostProcess(ALCdevice *device)
  */
 void BsincPrepare(const ALuint increment, BsincState *state, const BSincTable *table)
 {
-    ALfloat sf;
-    ALsizei si;
+    ALfloat sf = 0.0f;
+    ALsizei si = BSINC_SCALE_COUNT-1;
 
     if(increment > FRACTIONONE)
     {
         sf = (ALfloat)FRACTIONONE / increment;
         sf = maxf(0.0f, (BSINC_SCALE_COUNT-1) * (sf-table->scaleBase) * table->scaleRange);
-        si = fastf2i(sf);
+        si = float2int(sf);
         /* The interpolation factor is fit to this diagonally-symmetric curve
          * to reduce the transition ripple caused by interpolating different
          * scales of the sinc function.
          */
         sf = 1.0f - cosf(asinf(sf - si));
     }
-    else
-    {
-        sf = 0.0f;
-        si = BSINC_SCALE_COUNT - 1;
-    }
 
     state->sf = sf;
     state->m = table->m[si];