[Xorg] Re: Damage/Composite + direct rendering clients

[Andy Ritger]

On Mon, 24 May 2004, Ian Romanick wrote:

> Andy Ritger wrote:
> 
> > The other concern (how to make sure direct rendering has completed
> > by the time the drawable is used as a source in a composite
> > operation) conceptually would be solved as you describe, but I
> > expect the implementation would be buried deeper
> 
> I guess I don't see what the problem is.  The graphics device processes 
> commands as a FIFO.  Groups of direct rendering commands are surrounded 
> by a lock.  If the compositor uses the same lock, synchronization and 
> proper ordering are guaranteed.

How does the composite manager use the same lock?  If it's simply
using Render requests to perform the compositing, does the X server
need to lock whenever it processes a Render request.  Does it do
so today?  What happens if the X server performs the Render request
in software?

You're also assuming you have one global fifo, whereas some hardware
has a separate fifo per hw graphics client and the hw context
switches between fifos.

> Determining if direct rendering has damaged a region is another kettle 
> of fish, though...

Yes, but I actually think that is going to be the easy part :)

Thanks,
- Andy