This avoids excessive amounts of test data accumulating which can slow tests down.
As part of this change, the pref for enabling the logging was made 'Live'.
For frame statistics to work properly, we have to notify an ImageContainer
when it has been composited. This requires a few changes, which have
been lumped together in this patch:
-- Create PImageContainer and ImageContainerParent/ImageContainerChild.
-- Add mFrameID and mProducerID everywhere we're passing around images.
-- Route composition notifications from the compositor back to
ImageContainerChild.
The ScrollingBehavior passed around extensively in TabParent/TabChild is
entirely redundant now that the layers.async-pan-zoom.enabled pref definitively
controls whether or not APZ is enabled. All the code related to ScrollingBehavior
can be removed.
This patch adds a cross platform 'sync object' that is used to synchronize the drawing of individual textures. For the D3D11 implementation all textures that are written to will have one pixel copied into the D3D11 sync texture while holding its lock. The compositor will then, before composition acquire and release sync once, this should ensure all drawing on the content side has completed.
This patch adds a cross platform 'sync object' that is used to synchronize the drawing of individual textures. For the D3D11 implementation all textures that are written to will have one pixel copied into the D3D11 sync texture while holding its lock. The compositor will then, before composition acquire and release sync once, this should ensure all drawing on the content side has completed.
Calling AcquireSync on textures created with
D3D11_RESOURCE_MISC_SHARED_KEYEDMUTEX appears to be really slow, this really
bites us with the number of textures involved when tiling.
I've also tested using ID3D11Query to check when work is completed, this also
appears to be pretty slow.
This instead just uses a single texture with a keyedmutex, and makes sure we
draw to it last and lock it first. It's pretty hacky, but seems to work really
well so far.
Currently we use a world transform on the LayerManager. This gets passed to the
screen render target which adjusts the gl transform to accomplish the rotation.
This causes a lot of the system to have to know about the transform.
Instead we can just bake the transform into the root layer's shadow transform.
Everything now mostly just works.
