Clean up SkXfermode_opts.h

It seems that MSVC + __vectorcall don't play well together,
so back ourselves out into a situation where we don't need it.

   - Inline transfermode functions.  This removes the need for SK_VECTORCALL.
   - Remove 565 destination specializations.
     Blending into 565 is not speed-critical enough to merit the code bloat.
   - Removing 565 specializations means a bunch of Sk4px code is now dead.

8888 xfermodes generally speed up a bit from inlining, smoothly ranging from no change down to 0.65x for the fastest functions like Plus or Modulate.

565 xfermodes generally slow down because we're doing 565 -> 8888 and 8888->565 conversion serially[1] and using the stack, smoothly ranging from no change up to 2x slower for the fastest functions like Plus and Modulate.

[1] the 565->8888 conversion is actually being autovectorized

BUG=skia:4765,skia:4776
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1565223002
CQ_EXTRA_TRYBOTS=client.skia:Test-Ubuntu-GCC-GCE-CPU-AVX2-x86_64-Release-SKNX_NO_SIMD-Trybot

No public API changes.
TBR=reed@google.com

Review URL: https://codereview.chromium.org/1565223002

1 files changed, 8 insertions, 16 deletions

diff --git a/src/core/Sk4px.h b/src/core/Sk4px.h
index 3755488a4a..05378a8bc8 100644
--- a/src/core/Sk4px.h
+++ b/src/core/Sk4px.h
@@ -48,14 +48,6 @@ public:
     void store2(SkPMColor[2]) const;
     void store1(SkPMColor[1]) const;
 
-    // Same as above for 565.
-    static Sk4px Load4(const SkPMColor16 src[4]);
-    static Sk4px Load2(const SkPMColor16 src[2]);
-    static Sk4px Load1(const SkPMColor16 src[1]);
-    void store4(SkPMColor16 dst[4]) const;
-    void store2(SkPMColor16 dst[2]) const;
-    void store1(SkPMColor16 dst[1]) const;
-
     // 1, 2, or 4 SkPMColors with 16-bit components.
     // This is most useful as the result of a multiply, e.g. from mulWiden().
     class Wide : public Sk16h {
@@ -105,8 +97,8 @@ public:
 
     // A generic driver that maps fn over a src array into a dst array.
     // fn should take an Sk4px (4 src pixels) and return an Sk4px (4 dst pixels).
-    template <typename Fn, typename Dst>
-    static void MapSrc(int n, Dst* dst, const SkPMColor* src, const Fn& fn) {
+    template <typename Fn>
+    static void MapSrc(int n, SkPMColor* dst, const SkPMColor* src, const Fn& fn) {
         SkASSERT(dst);
         SkASSERT(src);
         // This looks a bit odd, but it helps loop-invariant hoisting across different calls to fn.
@@ -137,8 +129,8 @@ public:
     }
 
     // As above, but with dst4' = fn(dst4, src4).
-    template <typename Fn, typename Dst>
-    static void MapDstSrc(int n, Dst* dst, const SkPMColor* src, const Fn& fn) {
+    template <typename Fn>
+    static void MapDstSrc(int n, SkPMColor* dst, const SkPMColor* src, const Fn& fn) {
         SkASSERT(dst);
         SkASSERT(src);
         while (n > 0) {
@@ -167,8 +159,8 @@ public:
     }
 
     // As above, but with dst4' = fn(dst4, alpha4).
-    template <typename Fn, typename Dst>
-    static void MapDstAlpha(int n, Dst* dst, const SkAlpha* a, const Fn& fn) {
+    template <typename Fn>
+    static void MapDstAlpha(int n, SkPMColor* dst, const SkAlpha* a, const Fn& fn) {
         SkASSERT(dst);
         SkASSERT(a);
         while (n > 0) {
@@ -197,8 +189,8 @@ public:
     }
 
     // As above, but with dst4' = fn(dst4, src4, alpha4).
-    template <typename Fn, typename Dst>
-    static void MapDstSrcAlpha(int n, Dst* dst, const SkPMColor* src, const SkAlpha* a,
+    template <typename Fn>
+    static void MapDstSrcAlpha(int n, SkPMColor* dst, const SkPMColor* src, const SkAlpha* a,
                                const Fn& fn) {
         SkASSERT(dst);
         SkASSERT(src);