rounding for arm filter

source/scale_common.cc

 /*
  *  Copyright 2013 The LibYuv Project Authors. All rights reserved.
  *
  *  Use of this source code is governed by a BSD-style license
  *  that can be found in the LICENSE file in the root of the source
  *  tree. An additional intellectual property rights grant can be found
  *  in the file PATENTS. All contributing project authors may
  *  be found in the AUTHORS file in the root of the source tree.
  */
 
@@ skipping 399 matching lines @@
     dst_ptr[1] = dst_ptr[0] = src_ptr[0];
     src_ptr += 1;
     dst_ptr += 2;
   }
   if (dst_width & 1) {
     dst_ptr[0] = src_ptr[0];
   }
 }
 
 // (1-f)a + fb can be replaced with a + f(b-a)
-#if defined(__arm__)
-// arm uses 16 bit math with truncation.
-// TODO(fbarchard): add rounding.
+#if defined(__arm__) || defined(__aarch64__)
 #define BLENDER(a, b, f) (uint8)((int)(a) + \
-    (((int)((f)) * ((int)(b) - (int)(a))) >> 16))
+    ((((int)((f)) * ((int)(b) - (int)(a))) + 0x8000) >> 16))
 #else
 // inteluses 7 bit math with rounding.
 #define BLENDER(a, b, f) (uint8)((int)(a) + \
     (((int)((f) >> 9) * ((int)(b) - (int)(a)) + 0x40) >> 7))
 #endif
 
 void ScaleFilterCols_C(uint8* dst_ptr, const uint8* src_ptr,
                        int dst_width, int x, int dx) {
   int j;
   for (j = 0; j < dst_width - 1; j += 2) {
@@ skipping 38 matching lines @@
     int64 xi = x >> 16;
     int a = src_ptr[xi];
     int b = src_ptr[xi + 1];
     dst_ptr[0] = BLENDER(a, b, x & 0xffff);
   }
 }
 #undef BLENDER
 
 // Same as 8 bit arm blender but return is cast to uint16
 #define BLENDER(a, b, f) (uint16)((int)(a) + \
-    (((int)((f)) * ((int)(b) - (int)(a))) >> 16))
+    ((((int)((f)) * ((int)(b) - (int)(a))) + 0x8000) >> 16))
 
 void ScaleFilterCols_16_C(uint16* dst_ptr, const uint16* src_ptr,
                        int dst_width, int x, int dx) {
   int j;
   for (j = 0; j < dst_width - 1; j += 2) {
     int xi = x >> 16;
     int a = src_ptr[xi];
     int b = src_ptr[xi + 1];
     dst_ptr[0] = BLENDER(a, b, x & 0xffff);
     x += dx;
@@ skipping 317 matching lines @@
   for (j = 0; j < dst_width - 1; j += 2) {
     dst[1] = dst[0] = src[0];
     src += 1;
     dst += 2;
   }
   if (dst_width & 1) {
     dst[0] = src[0];
   }
 }
 
-// TODO(fbarchard): Replace 0x7f ^ f with 128-f.  bug=605.
+// TODO(fbarchard): Replace 0x7f ^ f with 128-f.  bug=607.
 // Mimics SSSE3 blender
 #define BLENDER1(a, b, f) ((a) * (0x7f ^ f) + (b) * f) >> 7
 #define BLENDERC(a, b, f, s) (uint32)( \
     BLENDER1(((a) >> s) & 255, ((b) >> s) & 255, f) << s)
 #define BLENDER(a, b, f) \
     BLENDERC(a, b, f, 24) | BLENDERC(a, b, f, 16) | \
     BLENDERC(a, b, f, 8) | BLENDERC(a, b, f, 0)
 
 void ScaleARGBFilterCols_C(uint8* dst_argb, const uint8* src_argb,
                            int dst_width, int x, int dx) {
@@ skipping 320 matching lines @@
     *dx = -*dx;
     // src_width = -src_width;   // Caller must do this.
   }
 }
 #undef CENTERSTART
 
 #ifdef __cplusplus
 }  // extern "C"
 }  // namespace libyuv
 #endif