refactor ARGBToI422 using ARGBToI420 internally

source/row_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 1459 matching lines @@
 #endif  // HAS_ARGBTOUV444ROW_NEON
 
 #define RGBTOUV_SETUP_REG                                                      \
     "movi       v20.8h, #56, lsl #0  \n"  /* UB/VR coefficient (0.875) / 2 */  \
     "movi       v21.8h, #37, lsl #0  \n"  /* UG coefficient (-0.5781) / 2  */  \
     "movi       v22.8h, #19, lsl #0  \n"  /* UR coefficient (-0.2969) / 2  */  \
     "movi       v23.8h, #9,  lsl #0  \n"  /* VB coefficient (-0.1406) / 2  */  \
     "movi       v24.8h, #47, lsl #0  \n"  /* VG coefficient (-0.7344) / 2  */  \
     "movi       v25.16b, #0x80       \n"  /* 128.5 (0x8080 in 16-bit)      */
 
-// 16x1 pixels -> 8x1.  width is number of argb pixels. e.g. 16.
-#ifdef HAS_ARGBTOUV422ROW_NEON
-void ARGBToUV422Row_NEON(const uint8* src_argb, uint8* dst_u, uint8* dst_v,
-                         int width) {
-  asm volatile (
-    RGBTOUV_SETUP_REG
-  "1:                                          \n"
-    MEMACCESS(0)
-    "ld4        {v0.16b,v1.16b,v2.16b,v3.16b}, [%0], #64 \n"  // load 16 pixels.
-
-    "uaddlp     v0.8h, v0.16b                  \n"  // B 16 bytes -> 8 shorts.
-    "uaddlp     v1.8h, v1.16b                  \n"  // G 16 bytes -> 8 shorts.
-    "uaddlp     v2.8h, v2.16b                  \n"  // R 16 bytes -> 8 shorts.
-
-    "subs       %w3, %w3, #16                  \n"  // 16 processed per loop.
-    "mul        v3.8h, v0.8h, v20.8h           \n"  // B
-    "mls        v3.8h, v1.8h, v21.8h           \n"  // G
-    "mls        v3.8h, v2.8h, v22.8h           \n"  // R
-    "add        v3.8h, v3.8h, v25.8h           \n"  // +128 -> unsigned
-
-    "mul        v4.8h, v2.8h, v20.8h           \n"  // R
-    "mls        v4.8h, v1.8h, v24.8h           \n"  // G
-    "mls        v4.8h, v0.8h, v23.8h           \n"  // B
-    "add        v4.8h, v4.8h, v25.8h           \n"  // +128 -> unsigned
-
-    "uqshrn     v0.8b, v3.8h, #8               \n"  // 16 bit to 8 bit U
-    "uqshrn     v1.8b, v4.8h, #8               \n"  // 16 bit to 8 bit V
-
-    MEMACCESS(1)
-    "st1        {v0.8b}, [%1], #8              \n"  // store 8 pixels U.
-    MEMACCESS(2)
-    "st1        {v1.8b}, [%2], #8              \n"  // store 8 pixels V.
-    "b.gt       1b                             \n"
-  : "+r"(src_argb),  // %0
-    "+r"(dst_u),     // %1
-    "+r"(dst_v),     // %2
-    "+r"(width)        // %3
-  :
-  : "cc", "memory", "v0", "v1", "v2", "v3", "v4", "v5", "v6", "v7",
-    "v20", "v21", "v22", "v23", "v24", "v25"
-  );
-}
-#endif  // HAS_ARGBTOUV422ROW_NEON
-
 // 32x1 pixels -> 8x1.  width is number of argb pixels. e.g. 32.
 #ifdef HAS_ARGBTOUV411ROW_NEON
 void ARGBToUV411Row_NEON(const uint8* src_argb, uint8* dst_u, uint8* dst_v,
                          int width) {
   asm volatile (
     RGBTOUV_SETUP_REG
   "1:                                          \n"
     MEMACCESS(0)
     "ld4        {v0.16b,v1.16b,v2.16b,v3.16b}, [%0], #64 \n"  // load 16 pixels.
     "uaddlp     v0.8h, v0.16b                  \n"  // B 16 bytes -> 8 shorts.
@@ skipping 1462 matching lines @@
   : "cc", "memory", "v0", "v1", "v2", "v3"  // Clobber List
   );
 }
 #endif  // HAS_SOBELYROW_NEON
 #endif  // !defined(LIBYUV_DISABLE_NEON) && defined(__aarch64__)
 
 #ifdef __cplusplus
 }  // extern "C"
 }  // namespace libyuv
 #endif