Add MIPS SIMD Arch (MSA) optimized MirrorRow function

include/libyuv/row.h

Index: include/libyuv/row.h
diff --git a/include/libyuv/row.h b/include/libyuv/row.h
index 013a7e53e326ecabb13b365d1dca8430b8952976..06981cc266ddbdee4c76b4322a15a17dcd31e855 100644
--- a/include/libyuv/row.h
+++ b/include/libyuv/row.h
@@ -362,8 +362,8 @@ extern "C" {
 // The following are available on Mips platforms:
 #if !defined(LIBYUV_DISABLE_MIPS) && defined(__mips__) && \
     (_MIPS_SIM == _MIPS_SIM_ABI32) && (__mips_isa_rev < 6)
-#define HAS_COPYROW_MIPS
 #if defined(__mips_dsp) && (__mips_dsp_rev >= 2)
+#define HAS_COPYROW_MIPS
 #define HAS_I422TOARGBROW_DSPR2
 #define HAS_INTERPOLATEROW_DSPR2
 #define HAS_MIRRORROW_DSPR2
@@ -372,6 +372,11 @@ extern "C" {
 #endif
 #endif
 
+#if !defined(LIBYUV_DISABLE_MSA) && defined(__mips__) && \
+	 defined(__mips_msa)

[fbarchard1, 2016/09/14 01:48:04]

indent should be 4 for line continuations.
you could probably remove the  && defined(__mips__) since it is implied by 
defined(__mips_msa)

[manojkumar.bhosale, 2016/09/14 12:45:29]

Done.

+#define HAS_MIRRORROW_MSA
+#endif
+
 #if defined(_MSC_VER) && !defined(__CLR_VER) && !defined(__clang__)
 #if defined(VISUALC_HAS_AVX2)
 #define SIMD_ALIGNED(var) __declspec(align(32)) var
@@ -809,11 +814,13 @@ void MirrorRow_AVX2(const uint8* src, uint8* dst, int width);
 void MirrorRow_SSSE3(const uint8* src, uint8* dst, int width);
 void MirrorRow_NEON(const uint8* src, uint8* dst, int width);
 void MirrorRow_DSPR2(const uint8* src, uint8* dst, int width);
+void MirrorRow_MSA(const uint8* src, uint8* dst, int width);
 void MirrorRow_C(const uint8* src, uint8* dst, int width);
 void MirrorRow_Any_AVX2(const uint8* src, uint8* dst, int width);
 void MirrorRow_Any_SSSE3(const uint8* src, uint8* dst, int width);
 void MirrorRow_Any_SSE2(const uint8* src, uint8* dst, int width);
 void MirrorRow_Any_NEON(const uint8* src, uint8* dst, int width);
+void MirrorRow_Any_MSA(const uint8* src, uint8* dst, int width);
 
 void MirrorUVRow_SSSE3(const uint8* src_uv, uint8* dst_u, uint8* dst_v,
                        int width);