1036785 - [OMTC] Use DIB surfaces when not accelerating content drawing

Status: RESOLVED FIXED

(Core :: Graphics: Layers, defect)

Description

[Bas Schouten (:bas.schouten)]

Currently we're using Cairo ImageSurfaces when not using accelerated content drawing (except with a D3D9 compositor), this is bad since it means we don't get subpixel AA, even when a surface is opaque. We should fix this.

Comment 1

[Bas Schouten (:bas.schouten)]

Attachment: Use DIBs for opaque surfaces

This patch makes us use DIBs for opaque surfaces, and makes all that code shared between all the different compositors. We might want to consider DIBs even for transparent surfaces on windows, but let's hold off on that for now.

Comment 2

[Bas Schouten (:bas.schouten)]

Attachment: Use DIBs for opaque surfaces v2

This time including the new files :-)

Comment 3

[Nicolas Silva [:nical]]

Comment on attachment 8453558 [details] [diff] [review]
Use DIBs for opaque surfaces v2

Review of attachment 8453558 [details] [diff] [review]:
-----------------------------------------------------------------

Nice

::: gfx/layers/TextureDIB.h
@@ +28,5 @@
> +  virtual ~DIBTextureClient();
> +
> +  // TextureClient
> +
> +  virtual bool IsAllocated() const MOZ_OVERRIDE{ return !!mSurface; }

nit: missing space between MOZ_OVERRIDE and {
same thing with IsLocked, CanExposeDrawTarget and HasInternalBuffer

@@ +38,5 @@
> +  virtual bool IsLocked() const MOZ_OVERRIDE{ return mIsLocked; }
> +
> +  virtual bool ToSurfaceDescriptor(SurfaceDescriptor& aOutDescriptor) MOZ_OVERRIDE;
> +
> +  virtual gfx::IntSize GetSize() const { return mSize; }

nit: MOZ_OVERRIDE

@@ +40,5 @@
> +  virtual bool ToSurfaceDescriptor(SurfaceDescriptor& aOutDescriptor) MOZ_OVERRIDE;
> +
> +  virtual gfx::IntSize GetSize() const { return mSize; }
> +
> +  virtual gfx::SurfaceFormat GetFormat() const { return mFormat; }

nit: MOZ_OVERRIDE

Comment 4

[Bas Schouten (:bas.schouten)]

Attachment: Fuzz some tests to mask subpixel-AA differences.

This patch masks some subpixel AA differences I can't quite explain, but seem to be there on the try-servers.

Comment 5

[Bas Schouten (:bas.schouten)]

https://hg.mozilla.org/integration/mozilla-inbound/rev/c98c4be7d7bf
https://hg.mozilla.org/integration/mozilla-inbound/rev/65a89b89cb1e

Comment 6

[Carsten Book [:Tomcat]]

https://hg.mozilla.org/mozilla-central/rev/c98c4be7d7bf
https://hg.mozilla.org/mozilla-central/rev/65a89b89cb1e