On Wed, Nov 26, 2014 at 11:39 AM, Chris Wilson <chris@xxxxxxxxxxxxxxxxxx> wrote: > On Wed, Nov 26, 2014 at 10:28:13AM +0100, Daniel Vetter wrote: >> With refcounting it looks like you can just drop that refcount, but >> that's not really the case. So make sure no one forgets. >> >> Motivated by the unlocked call in the mmio flip code. > > I had an unlocked variant for exactly this purpose (and a few others > where we do not want to take the lock again) and so also had the WARN > inside i915_request_free. > > Drop the _gem_, the requests are lower level than GEM itself. Yeah there's definitely more to do. The one I actually expect you to scream about is that it keeps an explicit refcount for all the obj->las_* fences, instead of relying upon the implicit fence that a correct retire sequence allows us. This here is just about enforcing correctness. -Daniel -- Daniel Vetter Software Engineer, Intel Corporation +41 (0) 79 365 57 48 - http://blog.ffwll.ch _______________________________________________ Intel-gfx mailing list Intel-gfx@xxxxxxxxxxxxxxxxxxxxx http://lists.freedesktop.org/mailman/listinfo/intel-gfx
