Quoting Joonas Lahtinen (2017-10-09 10:19:40)
> On Sat, 2017-10-07 at 16:28 +0100, Chris Wilson wrote:
> > Quoting Michal Wajdeczko (2017-10-07 12:14:30)
> > > On Sat, 07 Oct 2017 10:56:58 +0200, Chris Wilson
> > > > +void intel_uncore_init_mmio(struct drm_i915_private *dev_priv)
> > > > +{
> > > > + i915_check_vgpu(dev_priv);
> > > > +
> > > > + intel_uncore_edram_detect(dev_priv);
> > > > + intel_uncore_fw_domains_init(dev_priv);
> > >
> > > Do you plan to split fw_domains_init into early/mmio parts?
> > > Without such split use of intel_uncore_forcewake_for_reg
> > > from uc_init_early will still be not possible.
> >
> > Ah, I was mistaken in thinking the vfunc tables were enough. We still
> > have some mmio access in fw_domains_init that we need to split into
> > _early and _mmio parts. On the whole, what do we think, is the motion
> > still justified? I think it still is -- we have a lot of static
> > information here that is useful during early setup.
>
> Not sure if I fully understood the quesiton but the init code split
> makes it more coherent across driver.
Yes, I was asking if the proposed fixed patch still made sense when the
immediate use case (uc_init) had already been fixed.
-Chris
_______________________________________________
Intel-gfx mailing list
Intel-gfx@xxxxxxxxxxxxxxxxxxxxx
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
