1200899 - Take rotation into consideration when setting up maskQuadTransform for mask layer

Status: RESOLVED WONTFIX

(Core :: Graphics: Layers, defect)

Description

[Jeremy Chen ([e-mail])]

It can be seen from [1], we don't take rotation into consideration when setting up maskQuadTransform for mask layer. So whenever rotation happens, we use intermediate surface instead (see Bug 1149923). If we can find a way to pass rotation info to maskQuadTransform, we could no longer need intermediate surface for mask layer. This may have some performance improvement especially for FxOS, since desktop seems powerful enough to use intermediate surface all the time (see [2]).

[1] https://dxr.mozilla.org/mozilla-central/source/gfx/layers/opengl/CompositorOGL.cpp?from=compositorogl.cpp#1040-1043
[2] https://dxr.mozilla.org/mozilla-central/source/gfx/layers/Layers.cpp?from=layers.cpp#1228-1232

Comment 1

[Jeremy Chen ([e-mail])]

(In reply to Jeremy Chen [:jeremychen] UTC+8 from comment #0)
> It can be seen from [1], we don't take rotation into consideration when
> setting up maskQuadTransform for mask layer. So whenever rotation happens,
> we use intermediate surface instead (see Bug 1149923). If we can find a way
> to pass rotation info to maskQuadTransform, we could no longer need
> intermediate surface for mask layer. This may have some performance
> improvement especially for FxOS, since desktop seems powerful enough to use
> intermediate surface all the time (see [2]).
> 
> [1]
> https://dxr.mozilla.org/mozilla-central/source/gfx/layers/opengl/
> CompositorOGL.cpp?from=compositorogl.cpp#1040-1043
> [2]
> https://dxr.mozilla.org/mozilla-central/source/gfx/layers/Layers.
> cpp?from=layers.cpp#1228-1232

Update [1] to permalink, https://dxr.mozilla.org/mozilla-central/rev/66a77b9bfe5dcacd50eccf85de7c0e7e15ce0ffd/gfx/layers/opengl/CompositorOGL.cpp#1062-1065

Since we're removing all B2G-related code from mozilla-central (https://wiki.mozilla.org/B2G/Meeting/2016-09-27), I'm not sure if this is something we would fix in a foreseeable future. Close this bug for now. If someone think this could gain us any performance improvement on other mobile platform (ex. Fennec), he/she could reopen this bug.