Don't do a pointless << 0.

src/opts/SkColor_opts_SSE2.h

Index: src/opts/SkColor_opts_SSE2.h
diff --git a/src/opts/SkColor_opts_SSE2.h b/src/opts/SkColor_opts_SSE2.h
index 95fb69cceb7355f0c4405fd55087239264e5b7ea..970abb859b980dff84dd4f1085ddd967ac2b7d3e 100644
--- a/src/opts/SkColor_opts_SSE2.h
+++ b/src/opts/SkColor_opts_SSE2.h
@@ -81,8 +81,12 @@ static inline __m128i SkAlphaMulQ_SSE2(const __m128i& c, const unsigned scale) {
 }
 
 static inline __m128i SkGetPackedA32_SSE2(const __m128i& src) {
+#if SK_A32_SHIFT == 24                // It's very common (universal?) that alpha is the top byte.
+    return _mm_srli_epi32(src, 24);   // You'd hope the compiler would remove the left shift then,
+#else                                 // but I've seen Clang just do a dumb left shift of zero. :(
     __m128i a = _mm_slli_epi32(src, (24 - SK_A32_SHIFT));
     return _mm_srli_epi32(a, 24);
+#endif
 }
 
 static inline __m128i SkGetPackedR32_SSE2(const __m128i& src) {