On Mon, 20 Aug 2012 23:17:06 +0200, Daniel Vetter <daniel at ffwll.ch> wrote: > On Sat, Aug 11, 2012 at 03:41:26PM +0100, Chris Wilson wrote: > > Signed-off-by: Chris Wilson <chris at chris-wilson.co.uk> > > Since most of the overlay-supporting hw uses physical mem for the overlay > I think this isn't much worth it: The additional frobbery in > attach/detach_phys object is likely more work than we'll anything we'll > ever gain from using stolen mem here. Especially since we'll use stolen > mem already for the rings. In a straw poll of the machines on my desk, non-physical machines outnumber the physical overlay machines. :-p However, hooking up the physical to use stolen is also a good idea. Too bad, I haven't found a way to detect the base of stolen memory on gen2 devices without arch specific internals. It worked nicely right up until I tried to build i915.ko as a module. -Chris -- Chris Wilson, Intel Open Source Technology Centre
