3 files changed, 54 insertions, 41 deletions

diff --git a/apps/codecs/lib/asm_arm.h b/apps/codecs/lib/asm_arm.h
index 629e47b3bd..54ce4b0d98 100644
--- a/apps/codecs/lib/asm_arm.h
+++ b/apps/codecs/lib/asm_arm.h
@@ -16,9 +16,7 @@
  ********************************************************************/
 #ifdef CPU_ARM
 
-#if !defined(_V_WIDE_MATH) && !defined(_LOW_ACCURACY_)
-#define _V_WIDE_MATH
-
+#define INCL_OPTIMIZED_MULT32
 #if ARM_ARCH >= 6
 static inline int32_t MULT32(int32_t x, int32_t y) {
   int32_t hi;
@@ -37,10 +35,12 @@ static inline int32_t MULT32(int32_t x, int32_t y) {
 }
 #endif
 
+#define INCL_OPTIMIZED_MULT31
 static inline int32_t MULT31(int32_t x, int32_t y) {
   return MULT32(x,y)<<1;
 }
 
+#define INCL_OPTIMIZED_MULT31_SHIFT15
 static inline int32_t MULT31_SHIFT15(int32_t x, int32_t y) {
   int32_t lo,hi;
   asm volatile("smull   %0, %1, %2, %3\n\t"
@@ -52,6 +52,7 @@ static inline int32_t MULT31_SHIFT15(int32_t x, int32_t y) {
   return(hi);
 }
 
+#define INCL_OPTIMIZED_MULT31_SHIFT16
 static inline int32_t MULT31_SHIFT16(int32_t x, int32_t y) {
   int32_t lo,hi;
   asm volatile("smull   %0, %1, %2, %3\n\t"
@@ -63,6 +64,7 @@ static inline int32_t MULT31_SHIFT16(int32_t x, int32_t y) {
   return(hi);
 }
 
+#define INCL_OPTIMIZED_XPROD32
 #define XPROD32(a, b, t, v, x, y) \
 { \
   int32_t l; \
@@ -75,6 +77,8 @@ static inline int32_t MULT31_SHIFT16(int32_t x, int32_t y) {
       : "r" ((a)), "r" ((b)), "r" ((t)), "r" ((v)) ); \
 }
 
+#define INCL_OPTIMIZED_XPROD31_R
+#define INCL_OPTIMIZED_XNPROD31_R
 #if ARM_ARCH >= 6
 /* These may yield slightly different result from the macros below
    because only the high 32 bits of the multiplications are accumulated while
@@ -134,6 +138,7 @@ static inline int32_t MULT31_SHIFT16(int32_t x, int32_t y) {
 }
 #endif
 
+#define INCL_OPTIMIZED_XPROD31
 static inline void XPROD31(int32_t  a, int32_t  b,
                            int32_t  t, int32_t  v,
                            int32_t *x, int32_t *y)
@@ -144,6 +149,7 @@ static inline void XPROD31(int32_t  a, int32_t  b,
   *y = _y1;
 }
 
+#define INCL_OPTIMIZED_XNPROD31
 static inline void XNPROD31(int32_t  a, int32_t  b,
                             int32_t  t, int32_t  v,
                             int32_t *x, int32_t *y)
@@ -261,7 +267,6 @@ void vect_mult_bw(int32_t *data, int32_t *window, int n)
 
 #endif
 
-#endif
 /* not used anymore */
 /*
 #ifndef _V_CLIP_MATH
@@ -282,11 +287,6 @@ static inline int32_t CLIP_TO_15(int32_t x) {
 
 #endif
 */
-#ifndef _V_LSP_MATH_ASM
-#define _V_LSP_MATH_ASM
-
 
-
-#endif
 #endif
 
diff --git a/apps/codecs/lib/asm_mcf5249.h b/apps/codecs/lib/asm_mcf5249.h
index 5fb3cff94a..88d439631d 100644
--- a/apps/codecs/lib/asm_mcf5249.h
+++ b/apps/codecs/lib/asm_mcf5249.h
@@ -21,9 +21,7 @@
 
 #if defined(CPU_COLDFIRE)
 
-#ifndef _V_WIDE_MATH
-#define _V_WIDE_MATH
-
+#define INCL_OPTIMIZED_MULT32
 static inline int32_t MULT32(int32_t x, int32_t y) {
 
   asm volatile ("mac.l %[x], %[y], %%acc0;"    /* multiply & shift  */
@@ -35,6 +33,7 @@ static inline int32_t MULT32(int32_t x, int32_t y) {
   return x;
 }
 
+#define INCL_OPTIMIZED_MULT31
 static inline int32_t MULT31(int32_t x, int32_t y) {
   asm volatile ("mac.l %[x], %[y], %%acc0;" /* multiply */
                 "movclr.l %%acc0, %[x];"    /* move and clear */
@@ -44,6 +43,7 @@ static inline int32_t MULT31(int32_t x, int32_t y) {
   return x;
 }
 
+#define INCL_OPTIMIZED_MULT31_SHIFT15
 static inline int32_t MULT31_SHIFT15(int32_t x, int32_t y) {
   int32_t r;
 
@@ -61,6 +61,7 @@ static inline int32_t MULT31_SHIFT15(int32_t x, int32_t y) {
   return r;
 }
 
+#define INCL_OPTIMIZED_MULT31_SHIFT16
 static inline int32_t MULT31_SHIFT16(int32_t x, int32_t y) {
   int32_t r;
 
@@ -76,6 +77,7 @@ static inline int32_t MULT31_SHIFT16(int32_t x, int32_t y) {
   return x;
 }
 
+#define INCL_OPTIMIZED_XPROD31
 static inline
 void XPROD31(int32_t  a, int32_t  b,
              int32_t  t, int32_t  v,
@@ -95,6 +97,7 @@ void XPROD31(int32_t  a, int32_t  b,
                 : "cc", "memory");
 }
 
+#define INCL_OPTIMIZED_XNPROD31
 static inline
 void XNPROD31(int32_t  a, int32_t  b,
               int32_t  t, int32_t  v,
@@ -114,15 +117,11 @@ void XNPROD31(int32_t  a, int32_t  b,
                 : "cc", "memory");
 }
 
-#if 0    /* canonical Tremor definition */
-#define XPROD32(_a, _b, _t, _v, _x, _y)         \
-  { (_x)=MULT32(_a,_t)+MULT32(_b,_v);           \
-    (_y)=MULT32(_b,_t)-MULT32(_a,_v); }
-#endif
 
 /* this could lose the LSB by overflow, but i don't think it'll ever happen.
    if anyone think they can hear a bug caused by this, please try the above
    version. */
+#define INCL_OPTIMIZED_XPROD32
 #define XPROD32(_a, _b, _t, _v, _x, _y)     \
   asm volatile ("mac.l %[a], %[t], %%acc0;" \
                 "mac.l %[b], %[v], %%acc0;" \
@@ -137,6 +136,7 @@ void XNPROD31(int32_t  a, int32_t  b,
                   [t] "r" (_t), [v] "r" (_v) \
                 : "cc");
 
+#define INCL_OPTIMIZED_XPROD31_R
 #define XPROD31_R(_a, _b, _t, _v, _x, _y)   \
   asm volatile ("mac.l %[a], %[t], %%acc0;" \
                 "mac.l %[b], %[v], %%acc0;" \
@@ -149,6 +149,7 @@ void XNPROD31(int32_t  a, int32_t  b,
                   [t] "r" (_t), [v] "r" (_v) \
                 : "cc");
 
+#define INCL_OPTIMIZED_XNPROD31_R
 #define XNPROD31_R(_a, _b, _t, _v, _x, _y)  \
   asm volatile ("mac.l %[a], %[t], %%acc0;" \
                 "msac.l %[b], %[v], %%acc0;" \
@@ -336,7 +337,6 @@ void vect_mult_bw(int32_t *data, int32_t *window, int n)
 
 #endif
 
-#endif
 /* not used anymore */
 /*
 #ifndef _V_CLIP_MATH
diff --git a/apps/codecs/lib/codeclib_misc.h b/apps/codecs/lib/codeclib_misc.h
index f3b1805e26..08be93716f 100644
--- a/apps/codecs/lib/codeclib_misc.h
+++ b/apps/codecs/lib/codeclib_misc.h
@@ -15,32 +15,15 @@
 
  ********************************************************************/
 
-//#include "config-tremor.h"
-
-#ifndef _V_RANDOM_H_
-#define _V_RANDOM_H_
-//#include "ivorbiscodec.h"
-//#include "os_types.h"
+#ifndef _CODECLIB_MISC_H_
+#define _CODECLIB_MISC_H_
 
+#include <stdint.h>
 #include "asm_arm.h"
 #include "asm_mcf5249.h"
 
-
-/* Some prototypes that were not defined elsewhere */
-//void *_vorbis_block_alloc(vorbis_block *vb,long bytes);
-//void _vorbis_block_ripcord(vorbis_block *vb);
-//extern int _ilog(unsigned int v);
-
-#ifndef _V_WIDE_MATH
-#define _V_WIDE_MATH
-
-#ifndef ROCKBOX
-#include <inttypes.h>
-#endif /* ROCKBOX */
-
 #ifndef  _LOW_ACCURACY_
 /* 64 bit multiply */
-/* #include <sys/types.h> */
 
 #ifdef ROCKBOX_LITTLE_ENDIAN
 union magic {
@@ -60,29 +43,43 @@ union magic {
 };
 #endif
 
+#ifndef INCL_OPTIMIZED_MULT32
+#define INCL_OPTIMIZED_MULT32
 static inline int32_t MULT32(int32_t x, int32_t y) {
   union magic magic;
   magic.whole = (int64_t)x * y;
   return magic.halves.hi;
 }
+#endif
 
+#ifndef INCL_OPTIMIZED_MULT31
+#define INCL_OPTIMIZED_MULT31
 static inline int32_t MULT31(int32_t x, int32_t y) {
   return MULT32(x,y)<<1;
 }
+#endif
 
+#ifndef INCL_OPTIMIZED_MULT31_SHIFT15
+#define INCL_OPTIMIZED_MULT31_SHIFT15
 static inline int32_t MULT31_SHIFT15(int32_t x, int32_t y) {
   union magic magic;
   magic.whole  = (int64_t)x * y;
   return ((uint32_t)(magic.halves.lo)>>15) | ((magic.halves.hi)<<17);
 }
+#endif
 
+#ifndef INCL_OPTIMIZED_MULT31_SHIFT16
+#define INCL_OPTIMIZED_MULT31_SHIFT16
 static inline int32_t MULT31_SHIFT16(int32_t x, int32_t y) {
   union magic magic;
   magic.whole  = (int64_t)x * y;
   return ((uint32_t)(magic.halves.lo)>>16) | ((magic.halves.hi)<<16);
 }
+#endif
 
 #else
+/* Rockbox: unused */
+#if 0
 /* 32 bit multiply, more portable but less accurate */
 
 /*
@@ -110,6 +107,7 @@ static inline int32_t MULT31_SHIFT15(int32_t x, int32_t y) {
   return (x >> 6) * y;  /* y preshifted >>9 */
 }
 #endif
+#endif
 
 /*
  * The XPROD functions are meant to optimize the cross products found all
@@ -121,13 +119,17 @@ static inline int32_t MULT31_SHIFT15(int32_t x, int32_t y) {
  * macros.
  */
 
+#ifndef INCL_OPTIMIZED_XPROD32
+#define INCL_OPTIMIZED_XPROD32
 /* replaced XPROD32 with a macro to avoid memory reference
    _x, _y are the results (must be l-values) */
 #define XPROD32(_a, _b, _t, _v, _x, _y)     \
   { (_x)=MULT32(_a,_t)+MULT32(_b,_v);       \
     (_y)=MULT32(_b,_t)-MULT32(_a,_v); }
+#endif
 
-
+/* Rockbox: Unused */
+/*
 #ifdef __i386__
 
 #define XPROD31(_a, _b, _t, _v, _x, _y)     \
@@ -138,7 +140,10 @@ static inline int32_t MULT31_SHIFT15(int32_t x, int32_t y) {
     *(_y)=MULT31(_b,_t)+MULT31(_a,_v); }
 
 #else
+*/
 
+#ifndef INCL_OPTIMIZED_XPROD31
+#define INCL_OPTIMIZED_XPROD31
 static inline void XPROD31(int32_t a, int32_t b,
                       int32_t t, int32_t v,
                       int32_t *x, int32_t *y)
@@ -146,7 +151,10 @@ static inline void XPROD31(int32_t a, int32_t b,
   *x = MULT31(a, t) + MULT31(b, v);
   *y = MULT31(b, t) - MULT31(a, v);
 }
+#endif
 
+#ifndef INCL_OPTIMIZED_XNPROD31
+#define INCL_OPTIMIZED_XNPROD31
 static inline void XNPROD31(int32_t a, int32_t b,
                        int32_t  t, int32_t  v,
                        int32_t *x, int32_t *y)
@@ -155,19 +163,25 @@ static inline void XNPROD31(int32_t a, int32_t b,
   *y = MULT31(b, t) + MULT31(a, v);
 }
 #endif
+/*#endif*/
 
+#ifndef INCL_OPTIMIZED_XPROD31_R
+#define INCL_OPTIMIZED_XPROD31_R
 #define XPROD31_R(_a, _b, _t, _v, _x, _y)\
 {\
   _x = MULT31(_a, _t) + MULT31(_b, _v);\
   _y = MULT31(_b, _t) - MULT31(_a, _v);\
 }
+#endif
 
+#ifndef INCL_OPTIMIZED_XNPROD31_R
+#define INCL_OPTIMIZED_XNPROD31_R
 #define XNPROD31_R(_a, _b, _t, _v, _x, _y)\
 {\
   _x = MULT31(_a, _t) - MULT31(_b, _v);\
   _y = MULT31(_b, _t) + MULT31(_a, _v);\
 }
-
+#endif
 
 #ifndef _V_VECT_OPS
 #define _V_VECT_OPS
@@ -213,7 +227,6 @@ void vect_mult_bw(int32_t *data, int32_t *window, int n)
 }
 #endif
 
-#endif
 /* not used anymore */
 /*
 #ifndef _V_CLIP_MATH