Today the process for applying an svg mask looks roughly like:

• CreateClippedDrawTarget(mask_size, RGBA32);
• Draw mask content
• call IntoLuminanceSource() on mask DrawTarget
• call PushLayer() with the result of IntoLuminanceSource()

If we instead add an API like PushLayerWithLuminance() we'll be convey the luminance conversion directly to Skia and have Skia do it in a single pass along with masking operation.

For the record here's what Chrome does.

• beginLayer (content layer, srcOver mode)
  • draw content
  • beginLayer (mask layer, dstIn mode + luminance-to-alpha filter)
    • draw mask
  • endLayer: dstIn(luma(mask layer), content layer)
• endLayer: srcOver(content layer, backdrop)

We draw things in a different order so it might look like something more like:

• PushLayerWithLuminance();
  • SkCanvas::saveLayer();
• draw content
• PopLayer()
  • draw mask with dstIn into layer
  • SkCanvas::restore() to do final composite