libopus: asm C_MUL for coldfire

The recent merge of upstream changed the fft to use C_MUL which
wasn't implemented in asm for coldfire.

Speeds up decoding 64 kbps test file by 2.68 MHz and 128 kbps
test file by 2.80 MHz on H300.

Change-Id: I8b61fc0f9568d6350431e311a12e44fe4f60f72e

lib/rbcodec/codecs/libopus/celt/cf/kiss_fft_cf.h

@@ -32,6 +32,27 @@
 
 #ifdef FIXED_POINT
 
+#undef C_MUL
+#define C_MUL(m,a,b) \
+    { \
+      asm volatile("move.l (%[bp]), %%d2;" \
+                   "clr.l %%d3;" \
+                   "move.w %%d2, %%d3;" \
+                   "swap %%d3;" \
+                   "clr.w %%d2;" \
+                   "movem.l (%[ap]), %%d0-%%d1;" \
+                   "mac.l %%d0, %%d2, %%acc0;" \
+                   "msac.l %%d1, %%d3, %%acc0;" \
+                   "mac.l %%d1, %%d2, %%acc1;" \
+                   "mac.l %%d0, %%d3, %%acc1;" \
+                   "movclr.l %%acc0, %[mr];" \
+                   "movclr.l %%acc1, %[mi];" \
+                   : [mr] "=r" ((m).r), [mi] "=r" ((m).i) \
+                   : [ap] "a" (&(a)), [bp] "a" (&(b)) \
+                   : "d0", "d1", "d2", "d3", "cc"); \
+    }
+
+
 #undef C_MULC
 #define C_MULC(m,a,b) \
     { \