From: Jesse B. <jb...@vi...> - 2008-11-25 21:18:16
|
On Tuesday, November 25, 2008 12:18 pm Jesse Barnes wrote: > On Tuesday, November 25, 2008 11:44 am Eric Anholt wrote: > > I don't like this plan. If my app has a new frame all ready to go, I > > want it to get displayed the next time the compositor wakes up to do a > > vblank-synced composite, not wait for the compositor to grab the stale > > frame before continuing. > > I don't disagree, that's what I was trying to say. Obviously I didn't say > it very well. I should expand here a little. There's a difference between glXSwapBuffers for redirected clients vs. the compositing manager. The former will be waiting for the compositing manager if they want to make sure their frames get displayed before overwriting their last frame's contents. The compositing manager will be the one actually waiting on vertical blank events using the new "queue vblank sync'd blit" ioctl. If an app doesn't want to wait, that's fine, it can do glXSwapBuffers with the swap interval set to 0. But apps that want synchronization will have to wait on the compositing manager, so we'll need some way of knowing when it has displayed a frame. -- Jesse Barnes, Intel Open Source Technology Center |