rounding for arm filter

source/scale_neon64.cc

 /*
  *  Copyright 2014 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 608 matching lines @@
     "mov       v7.16b, v2.16b                  \n"
     "uzp1      v6.8h, v6.8h, v7.8h             \n"
     "ushll     v4.8h, v4.8b, #0                \n"
     "ushll     v5.8h, v5.8b, #0                \n"
     "ssubl     v16.4s, v5.4h, v4.4h            \n"
     "ssubl2    v17.4s, v5.8h, v4.8h            \n"
     "ushll     v7.4s, v6.4h, #0                \n"
     "ushll2    v6.4s, v6.8h, #0                \n"
     "mul       v16.4s, v16.4s, v7.4s           \n"
     "mul       v17.4s, v17.4s, v6.4s           \n"
-    "shrn      v6.4h, v16.4s, #16              \n"
-    "shrn2     v6.8h, v17.4s, #16              \n"
+    "rshrn      v6.4h, v16.4s, #16             \n"
+    "rshrn2     v6.8h, v17.4s, #16             \n"
     "add       v4.8h, v4.8h, v6.8h             \n"
     "xtn       v4.8b, v4.8h                    \n"
 
     MEMACCESS(0)
     "st1       {v4.8b}, [%0], #8               \n"  // store pixels
     "add       v1.4s, v1.4s, v0.4s             \n"
     "add       v2.4s, v2.4s, v0.4s             \n"
     "subs      %w2, %w2, #8                    \n"  // 8 processed per loop
     "b.gt      1b                              \n"
   : "+r"(dst_ptr),          // %0
@@ skipping 392 matching lines @@
 }
 
 #undef LOAD2_DATA32_LANE
 
 #endif  // !defined(LIBYUV_DISABLE_NEON) && defined(__aarch64__)
 
 #ifdef __cplusplus
 }  // extern "C"
 }  // namespace libyuv
 #endif