Planar ramp() for S32 linear gradients

src/effects/gradients/Sk4fGradientPriv.h

 /*
  * Copyright 2016 Google Inc.
  *
  * Use of this source code is governed by a BSD-style license that can be
  * found in the LICENSE file.
  */
 
 #ifndef Sk4fGradientPriv_DEFINED
 #define Sk4fGradientPriv_DEFINED
 
@@ skipping 90 matching lines @@
             store(c3, dst + 3);
         }
     }
 };
 
 template <ApplyPremul premul>
 struct DstTraits<DstType::S32, premul> {
     using PM   = PremulTraits<premul>;
     using Type = SkPMColor;
 
-    // TODO: prescale by something like (255^2, 255^2, 255^2, 255) and use
-    //       linear_to_srgb to save a mult in store?
     static Sk4f load(const SkPM4f& c) {
-        return c.to4f_pmorder();
+        // Prescaling by (255^2, 255^2, 255^2, 255) on load, to avoid a 255 multiply on
+        // each store (S32 conversion yields a uniform 255 factor).
+        return c.to4f_pmorder() * Sk4f(255 * 255, 255 * 255, 255 * 255, 255);
     }
 
     static void store(const Sk4f& c, Type* dst) {
         // FIXME: this assumes opaque colors.  Handle unpremultiplication.
-        *dst = Sk4f_toS32(PM::apply(c));
+        *dst = to_4b(linear_to_srgb(PM::apply(c)));
     }
 
     static void store(const Sk4f& c, Type* dst, int n) {
-        sk_memset32(dst, Sk4f_toS32(PM::apply(c)), n);
+        sk_memset32(dst, to_4b(linear_to_srgb(PM::apply(c))), n);
     }
 
     static void store4x(const Sk4f& c0, const Sk4f& c1,
                         const Sk4f& c2, const Sk4f& c3,
                         Type* dst) {
         store(c0, dst + 0);
         store(c1, dst + 1);
         store(c2, dst + 2);
         store(c3, dst + 3);
     }
@@ skipping 52 matching lines @@
         store(c0, dst + 0);
         store(c1, dst + 1);
         store(c2, dst + 2);
         store(c3, dst + 3);
     }
 };
 
 } // anonymous namespace
 
 #endif // Sk4fGradientPriv_DEFINED