Re-proc SkBlitRow::Color32 for ARM.

src/core/SkBlitRow_D32.cpp

 /*
  * Copyright 2011 Google Inc.
  *
  * Use of this source code is governed by a BSD-style license that can be
  * found in the LICENSE file.
  */
 
 #include "SkBlitRow.h"
 #include "SkBlitMask.h"
 #include "SkColorPriv.h"
@@ skipping 126 matching lines @@
 // It's not quite perfect, but it's never wrong in the interesting edge cases,
 // and it's quite a bit faster than blend_perfect.
 //
 // blend_256_round_alt is our currently blessed algorithm.  Please use it or an analogous one.
 void SkBlitRow::Color32(SkPMColor dst[], const SkPMColor src[], int count, SkPMColor color) {
     switch (SkGetPackedA32(color)) {
         case   0: memmove(dst, src, count * sizeof(SkPMColor)); return;
         case 255: sk_memset32(dst, color, count);               return;
     }
 
+    // This Sk4px impl works great on other platforms or when we have NEON.
+#if defined(SK_CPU_ARM32) && !defined(SK_ARM_HAS_NEON)
+    if (auto proc = PlatformColor32Proc()) { return proc(dst, src, count, color); }
+#endif
+
     unsigned invA = 255 - SkGetPackedA32(color);
     invA += invA >> 7;
     SkASSERT(invA < 256);  // We've already handled alpha == 0 above.
 
     Sk16h colorHighAndRound = Sk4px(color).widenHi() + Sk16h(128);
     Sk16b invA_16x(invA);
 
     Sk4px::MapSrc(count, dst, src, [&](const Sk4px& src4) -> Sk4px {
         return src4.mulWiden(invA_16x).addNarrowHi(colorHighAndRound);
     });
 }