codeclib/libtremor: Clean up duplication of inline funcs, constify the ones in codeclib and copy over the slightly faster MULT31_SHIFT15 from tremor for cf.

git-svn-id: svn://svn.rockbox.org/rockbox/trunk@30578 a1c6a512-1295-4272-9138-f99709370657

3 files changed, 14 insertions, 14 deletions

diff --git a/apps/codecs/lib/asm_arm.h b/apps/codecs/lib/asm_arm.h
index 54ce4b0..8e5d0e6 100644
--- a/apps/codecs/lib/asm_arm.h
+++ b/apps/codecs/lib/asm_arm.h
@@ -166,7 +166,7 @@ static inline void XNPROD31(int32_t  a, int32_t  b,
 
 /* asm versions of vector operations for block.c, window.c */
 static inline
-void vect_add(int32_t *x, int32_t *y, int n)
+void vect_add(int32_t *x, const int32_t *y, int n)
 {
   while (n>=4) {
     asm volatile ("ldmia %[x], {r0, r1, r2, r3};"
@@ -190,7 +190,7 @@ void vect_add(int32_t *x, int32_t *y, int n)
 }
 
 static inline
-void vect_copy(int32_t *x, int32_t *y, int n)
+void vect_copy(int32_t *x, const int32_t *y, int n)
 {
   while (n>=4) {
     asm volatile ("ldmia %[y]!, {r0, r1, r2, r3};"
@@ -208,7 +208,7 @@ void vect_copy(int32_t *x, int32_t *y, int n)
 }
 
 static inline
-void vect_mult_fw(int32_t *data, int32_t *window, int n)
+void vect_mult_fw(int32_t *data, const int32_t *window, int n)
 {
   while (n>=4) {
     asm volatile ("ldmia %[d], {r0, r1, r2, r3};"
@@ -237,7 +237,7 @@ void vect_mult_fw(int32_t *data, int32_t *window, int n)
 }
 
 static inline
-void vect_mult_bw(int32_t *data, int32_t *window, int n)
+void vect_mult_bw(int32_t *data, const int32_t *window, int n)
 {
   while (n>=4) {
     asm volatile ("ldmia %[d], {r0, r1, r2, r3};"
diff --git a/apps/codecs/lib/asm_mcf5249.h b/apps/codecs/lib/asm_mcf5249.h
index 88d4396..841c413 100644
--- a/apps/codecs/lib/asm_mcf5249.h
+++ b/apps/codecs/lib/asm_mcf5249.h
@@ -44,17 +44,17 @@ static inline int32_t MULT31(int32_t x, int32_t y) {
 }
 
 #define INCL_OPTIMIZED_MULT31_SHIFT15
+/* NOTE: this requires that the emac is *NOT* rounding */
 static inline int32_t MULT31_SHIFT15(int32_t x, int32_t y) {
   int32_t r;
 
   asm volatile ("mac.l %[x], %[y], %%acc0;"  /* multiply */
                 "mulu.l %[y], %[x];"         /* get lower half, avoid emac stall */
                 "movclr.l %%acc0, %[r];"     /* get higher half */
-                "asl.l #8, %[r];"            /* hi<<16, plus one free */
-                "asl.l #8, %[r];"
+                "swap %[r];"                 /* hi<<16, plus one free */
                 "lsr.l #8, %[x];"            /* (unsigned)lo >> 15 */
                 "lsr.l #7, %[x];"
-                "or.l %[x], %[r];"           /* logical-or results */
+                "move.w %[x], %[r];"         /* logical-or results */
                 : [r] "=&d" (r), [x] "+d" (x)
                 : [y] "d" (y)
                 : "cc");
@@ -202,7 +202,7 @@ void vect_add(int32_t *x, const int32_t *y, int n)
 }
 
 static inline
-void vect_copy(int32_t *x, int32_t *y, int n)
+void vect_copy(int32_t *x, const int32_t *y, int n)
 {
   /* align to 16 bytes */
   while(n>0 && (int)x&15) {
@@ -228,7 +228,7 @@ void vect_copy(int32_t *x, int32_t *y, int n)
 }
 
 static inline
-void vect_mult_fw(int32_t *data, int32_t *window, int n)
+void vect_mult_fw(int32_t *data, const int32_t *window, int n)
 {
   /* ensure data is aligned to 16-bytes */
   while(n>0 && (int)data&15) {
@@ -282,7 +282,7 @@ void vect_mult_fw(int32_t *data, int32_t *window, int n)
 }
 
 static inline
-void vect_mult_bw(int32_t *data, int32_t *window, int n)
+void vect_mult_bw(int32_t *data, const int32_t *window, int n)
 {
   /* ensure at least data is aligned to 16-bytes */
   while(n>0 && (int)data&15) {
diff --git a/apps/codecs/lib/codeclib_misc.h b/apps/codecs/lib/codeclib_misc.h
index 08be937..8ebe22e 100644
--- a/apps/codecs/lib/codeclib_misc.h
+++ b/apps/codecs/lib/codeclib_misc.h
@@ -187,7 +187,7 @@ static inline void XNPROD31(int32_t a, int32_t b,
 #define _V_VECT_OPS
 
 static inline
-void vect_add(int32_t *x, int32_t *y, int n)
+void vect_add(int32_t *x, const int32_t *y, int n)
 {
   while (n>0) {
     *x++ += *y++;
@@ -196,7 +196,7 @@ void vect_add(int32_t *x, int32_t *y, int n)
 }
 
 static inline
-void vect_copy(int32_t *x, int32_t *y, int n)
+void vect_copy(int32_t *x, const int32_t *y, int n)
 {
   while (n>0) {
     *x++ = *y++;
@@ -205,7 +205,7 @@ void vect_copy(int32_t *x, int32_t *y, int n)
 }
 
 static inline
-void vect_mult_fw(int32_t *data, int32_t *window, int n)
+void vect_mult_fw(int32_t *data, const int32_t *window, int n)
 {
   while(n>0) {
     *data = MULT31(*data, *window);
@@ -216,7 +216,7 @@ void vect_mult_fw(int32_t *data, int32_t *window, int n)
 }
 
 static inline
-void vect_mult_bw(int32_t *data, int32_t *window, int n)
+void vect_mult_bw(int32_t *data, const int32_t *window, int n)
 {
   while(n>0) {
     *data = MULT31(*data, *window);