On Die, 2010-04-27 at 10:31 +0200, Michel D?nzer wrote: > On Die, 2010-04-27 at 12:34 +1000, Dave Airlie wrote: > > From: Dave Airlie <airlied at redhat.com> > > > > On constrained r100 systems compiz would fail to start due to a lack > > of memory, we can just fallback place the objects rather than completely > > failing it works a lot better. > > > > v2: > > fixes issue identified by Michel when pinning could happen in a busy > > placement domain. > > [...] > > How about something like the below (completely untested) instead? This > should try to respect the userspace request first and only allow falling > back from VRAM to GTT if that failed. (Maybe the new bool parameter > shouldn't be called 'pinned' then but something like > 'allow_vram_fallback' and its sense inverted)
The alternative patch below is tested to not regress for me, but of course I'm not affected by the failure to put things in VRAM. Uwe / 'Chicken Shack', does this fix your problem with compiz? This doesn't cover radeon_bo_fault_reserve_notify() because my tree doesn't have the invisible VRAM stuff yet, but it'll be easy to extend this approach to there as well, and it shouldn't affect the immediate problem we're trying to solve.
