On Mon, Mar 26, 2012 at 09:48:01AM +0100, Chris Wilson wrote: > On Mon, 26 Mar 2012 09:45:40 +0200, Daniel Vetter <daniel.vetter at ffwll.ch> wrote: > > int > > i915_gem_init_ioctl(struct drm_device *dev, void *data, > > struct drm_file *file) > > @@ -155,7 +136,8 @@ i915_gem_init_ioctl(struct drm_device *dev, void *data, > > return -EINVAL; > > > > mutex_lock(&dev->struct_mutex); > > - i915_gem_do_init(dev, args->gtt_start, args->gtt_end, args->gtt_end); > > + i915_gem_init_global_gtt(dev, args->gtt_start, > > + args->gtt_end, args->gtt_end); > > mutex_unlock(&dev->struct_mutex); > > > > return 0; > > This raises the interesting question of ppgtt interacting with DRI1, a > never supported combination. Do we start making that expressly clear > with a few if (HAS_ALIASING_PPGTT()) return -ENODEV; or even if (gen >= > 6) return -ENODEV; ? Atm ppgtt setup is only done for the modeset case, so even if someone is nutty enough to backport gen6 kms to ums it'll just work. But I'll plan to rework intel-gtt some more and bail out with -EINVAL or whatever for the ums-only ioctls on gen6+. -Daniel -- Daniel Vetter Mail: daniel at ffwll.ch Mobile: +41 (0)79 365 57 48
