Sync to upstream libopus

Sync to commit bb4b6885a139644cf3ac14e7deda9f633ec2d93c

This brings in a bunch of optimizations to decode speed
and memory usage. Allocations are switched from using
the pseudostack to using the real stack. Enabled hacks
to reduce stack usage.

This should fix crashes on sansa clip, although some
files will not play due to failing allocations in the
codec buffer.

Speeds up decoding of the following test files:

                 H300 (cf)   C200 (arm7tdmi)  ipod classic (arm9e)
16 kbps (silk)   14.28 MHz   4.00 MHz         2.61 MHz
64 kbps (celt)   4.09 MHz    8.08 MHz         6.24 MHz
128 kbps (celt)  1.93 MHz    8.83 MHz         6.53 MHz

Change-Id: I851733a8a5824b61feb363a173091bc7e6629b58

2 files changed, 39 insertions, 0 deletions

diff --git a/lib/rbcodec/codecs/libopus/celt/arm/fixed_armv4.h b/lib/rbcodec/codecs/libopus/celt/arm/fixed_armv4.h
index b690bc8cea..efb3b1896a 100644
--- a/lib/rbcodec/codecs/libopus/celt/arm/fixed_armv4.h
+++ b/lib/rbcodec/codecs/libopus/celt/arm/fixed_armv4.h
@@ -68,6 +68,10 @@ static OPUS_INLINE opus_val32 MULT16_32_Q15_armv4(opus_val16 a, opus_val32 b)
 #undef MAC16_32_Q15
 #define MAC16_32_Q15(c, a, b) ADD32(c, MULT16_32_Q15(a, b))
 
+/** 16x32 multiply, followed by a 16-bit shift right and 32-bit add.
+    Result fits in 32 bits. */
+#undef MAC16_32_Q16
+#define MAC16_32_Q16(c, a, b) ADD32(c, MULT16_32_Q16(a, b))
 
 /** 32x32 multiplication, followed by a 31-bit shift right. Results fits in 32 bits */
 #undef MULT32_32_Q31
diff --git a/lib/rbcodec/codecs/libopus/celt/arm/fixed_armv5e.h b/lib/rbcodec/codecs/libopus/celt/arm/fixed_armv5e.h
index 1194a7d3ec..36a6321101 100644
--- a/lib/rbcodec/codecs/libopus/celt/arm/fixed_armv5e.h
+++ b/lib/rbcodec/codecs/libopus/celt/arm/fixed_armv5e.h
@@ -82,6 +82,23 @@ static OPUS_INLINE opus_val32 MAC16_32_Q15_armv5e(opus_val32 c, opus_val16 a,
 }
 #define MAC16_32_Q15(c, a, b) (MAC16_32_Q15_armv5e(c, a, b))
 
+/** 16x32 multiply, followed by a 16-bit shift right and 32-bit add.
+    Result fits in 32 bits. */
+#undef MAC16_32_Q16
+static OPUS_INLINE opus_val32 MAC16_32_Q16_armv5e(opus_val32 c, opus_val16 a,
+ opus_val32 b)
+{
+  int res;
+  __asm__(
+      "#MAC16_32_Q16\n\t"
+      "smlawb %0, %1, %2, %3;\n"
+      : "=r"(res)
+      : "r"(b), "r"(a), "r"(c)
+  );
+  return res;
+}
+#define MAC16_32_Q16(c, a, b) (MAC16_32_Q16_armv5e(c, a, b))
+
 /** 16x16 multiply-add where the result fits in 32 bits */
 #undef MAC16_16
 static OPUS_INLINE opus_val32 MAC16_16_armv5e(opus_val32 c, opus_val16 a,
@@ -113,4 +130,22 @@ static OPUS_INLINE opus_val32 MULT16_16_armv5e(opus_val16 a, opus_val16 b)
 }
 #define MULT16_16(a, b) (MULT16_16_armv5e(a, b))
 
+#ifdef OPUS_ARM_INLINE_MEDIA
+
+#undef SIG2WORD16
+static OPUS_INLINE opus_val16 SIG2WORD16_armv6(opus_val32 x)
+{
+   celt_sig res;
+   __asm__(
+       "#SIG2WORD16\n\t"
+       "ssat %0, #16, %1, ASR #12\n\t"
+       : "=r"(res)
+       : "r"(x+2048)
+   );
+   return EXTRACT16(res);
+}
+#define SIG2WORD16(x) (SIG2WORD16_armv6(x))
+
+#endif /* OPUS_ARM_INLINE_MEDIA */
+
 #endif