Re: [PATCH] drm/i915/pc8: Add parent interface for PC8 forcewake tricks

["Ville Syrjälä" <ville.syrjala@linux.intel.com>]

On Tue, Dec 09, 2025 at 02:13:47PM +0200, Jani Nikula wrote:
> On Tue, 09 Dec 2025, Ville Syrjala <ville.syrjala@linux.intel.com> wrote:
<snip>
> > diff --git a/drivers/gpu/drm/i915/display/intel_parent.h b/drivers/gpu/drm/i915/display/intel_parent.h
> > index 8f91a6f75c53..974a016ab3be 100644
> > --- a/drivers/gpu/drm/i915/display/intel_parent.h
> > +++ b/drivers/gpu/drm/i915/display/intel_parent.h
> > @@ -22,6 +22,9 @@ void intel_parent_hdcp_gsc_context_free(struct intel_display *display,
> >  bool intel_parent_irq_enabled(struct intel_display *display);
> >  void intel_parent_irq_synchronize(struct intel_display *display);
> >  
> > +void intel_parent_pc8_block(struct intel_display *display);
> > +void intel_parent_pc8_unblock(struct intel_display *display);
> > +
> >  bool intel_parent_rps_available(struct intel_display *display);
> >  void intel_parent_rps_boost_if_not_started(struct intel_display *display, struct dma_fence *fence);
> >  void intel_parent_rps_mark_interactive(struct intel_display *display, bool interactive);
> > diff --git a/drivers/gpu/drm/i915/i915_display_pc8.c b/drivers/gpu/drm/i915/i915_display_pc8.c
> > new file mode 100644
> > index 000000000000..443935d282e3
> > --- /dev/null
> > +++ b/drivers/gpu/drm/i915/i915_display_pc8.c
> > @@ -0,0 +1,30 @@
> > +// SPDX-License-Identifier: MIT
> > +/*
> > + * Copyright 2025, Intel Corporation.
> > + */
> > +
> > +#include <drm/drm_print.h>
> > +#include <drm/intel/display_parent_interface.h>
> > +
> > +#include "i915_drv.h"
> > +#include "intel_uncore.h"
> 
> For completeness, I think this should include i915_display_pc8.h. I'm a
> bit surprised the compilers only warn about non-static functions without
> declarations, not about non-static variables.

A bit of googling gave me, among other things, this:
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=110947

And indeed enabling -Wmissing-variable-declarations still
triggers on 'current_stack_pointer'. So I guess no progress
on that since 2023 :(

The other surprising thing here is that the extern decalaration
doesn't seem to need the type definition. Or at least HDRTEST
passses w/o including display_parent_interface.h in
i915_display_pc8.h.

-- 
Ville Syrjälä
Intel