Use none SSE functions when data is not 16 byte aligned

core/src/fxcodec/fx_libopenjpeg/libopenjpeg20/dwt.c

Index: core/src/fxcodec/fx_libopenjpeg/libopenjpeg20/dwt.c
diff --git a/core/src/fxcodec/fx_libopenjpeg/libopenjpeg20/dwt.c b/core/src/fxcodec/fx_libopenjpeg/libopenjpeg20/dwt.c
index e1f8a337d414b70d7e5e8ff6d33dafcf19e52ae7..8c1fe695e7aa8fceaed65d8daa9c7ef4947e4625 100644
--- a/core/src/fxcodec/fx_libopenjpeg/libopenjpeg20/dwt.c
+++ b/core/src/fxcodec/fx_libopenjpeg/libopenjpeg20/dwt.c
@@ -140,7 +140,8 @@ static void opj_v4dwt_interleave_h(opj_v4dwt_t* restrict w, OPJ_FLOAT32* restric
 
 static void opj_v4dwt_interleave_v(opj_v4dwt_t* restrict v , OPJ_FLOAT32* restrict a , OPJ_INT32 x, OPJ_INT32 nb_elts_read);
 
-#ifdef __SSE__
+//#ifdef __SSE__
+#if 0

[jabdelmalek, 2014/06/06 17:25:31]

here and below where you comment this out, please add "http://crbug.com/373619"
in a comment so that someone looking at this knows why it's commented out

 static void opj_v4dwt_decode_step1_sse(opj_v4_t* w, OPJ_INT32 count, const __m128 c);
 
 static void opj_v4dwt_decode_step2_sse(opj_v4_t* l, opj_v4_t* w, OPJ_INT32 k, OPJ_INT32 m, __m128 c);
@@ -671,8 +672,8 @@ void opj_v4dwt_interleave_v(opj_v4dwt_t* restrict v , OPJ_FLOAT32* restrict a ,
 	}
 }
 
-#ifdef __SSE__
-
+//#ifdef __SSE__
+#if 0
 void opj_v4dwt_decode_step1_sse(opj_v4_t* w, OPJ_INT32 count, const __m128 c){
 	__m128* restrict vw = (__m128*) w;
 	OPJ_INT32 i;
@@ -808,7 +809,8 @@ void opj_v4dwt_decode(opj_v4dwt_t* restrict dwt)
 		a = 1;
 		b = 0;
 	}
-#ifdef __SSE__
+#if 0
+//#ifdef __SSE__
 	opj_v4dwt_decode_step1_sse(dwt->wavelet+a, dwt->sn, _mm_set1_ps(opj_K));
 	opj_v4dwt_decode_step1_sse(dwt->wavelet+b, dwt->dn, _mm_set1_ps(opj_c13318));
 	opj_v4dwt_decode_step2_sse(dwt->wavelet+b, dwt->wavelet+a+1, dwt->sn, opj_int_min(dwt->sn, dwt->dn-a), _mm_set1_ps(opj_dwt_delta));