Re: [PATCH 15/16] drm/i915/display: Separate out functions to get/set VTOTAL register

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

On Tue, Mar 25, 2025 at 11:52:24AM +0530, Nautiyal, Ankit K wrote:
> 
> On 3/24/2025 11:32 PM, Ville Syrjälä wrote:
> > On Mon, Mar 24, 2025 at 07:02:47PM +0530, Ankit Nautiyal wrote:
> >> Introduce helpers to get and set TRANS_VTOTAL registers.
> >> This will pave way to avoid reading/writing VTOTAL.Vtotal bits for
> >> platforms that always use VRR timing generator.
> >>
> >> Signed-off-by: Ankit Nautiyal <ankit.k.nautiyal@intel.com>
> >> ---
> >>   drivers/gpu/drm/i915/display/intel_display.c | 41 +++++++++++++-------
> >>   1 file changed, 27 insertions(+), 14 deletions(-)
> >>
> >> diff --git a/drivers/gpu/drm/i915/display/intel_display.c b/drivers/gpu/drm/i915/display/intel_display.c
> >> index ae1dc32044fb..fa9c6793357e 100644
> >> --- a/drivers/gpu/drm/i915/display/intel_display.c
> >> +++ b/drivers/gpu/drm/i915/display/intel_display.c
> >> @@ -2638,6 +2638,15 @@ void intel_cpu_transcoder_set_m2_n2(struct intel_crtc *crtc,
> >>   		      PIPE_LINK_N2(display, transcoder));
> >>   }
> >>   
> >> +static void intel_crtc_set_vtotal(struct intel_display *display,
> >> +				  enum transcoder cpu_transcoder,
> >> +				  u32 crtc_vdisplay, u32 crtc_vtotal)
> >> +{
> >> +	intel_de_write(display, TRANS_VTOTAL(display, cpu_transcoder),
> >> +		       VACTIVE(crtc_vdisplay - 1) |
> >> +		       VTOTAL(crtc_vtotal - 1));
> >> +}
> >> +
> >>   static void intel_set_transcoder_timings(const struct intel_crtc_state *crtc_state)
> >>   {
> >>   	struct intel_display *display = to_intel_display(crtc_state);
> >> @@ -2702,9 +2711,8 @@ static void intel_set_transcoder_timings(const struct intel_crtc_state *crtc_sta
> >>   		       HSYNC_START(adjusted_mode->crtc_hsync_start - 1) |
> >>   		       HSYNC_END(adjusted_mode->crtc_hsync_end - 1));
> >>   
> >> -	intel_de_write(display, TRANS_VTOTAL(display, cpu_transcoder),
> >> -		       VACTIVE(crtc_vdisplay - 1) |
> >> -		       VTOTAL(crtc_vtotal - 1));
> >> +	intel_crtc_set_vtotal(display, cpu_transcoder, crtc_vdisplay, crtc_vtotal);
> >> +
> >>   	intel_de_write(display, TRANS_VBLANK(display, cpu_transcoder),
> >>   		       VBLANK_START(crtc_vblank_start - 1) |
> >>   		       VBLANK_END(crtc_vblank_end - 1));
> >> @@ -2718,9 +2726,8 @@ static void intel_set_transcoder_timings(const struct intel_crtc_state *crtc_sta
> >>   	 * bits. */
> >>   	if (display->platform.haswell && cpu_transcoder == TRANSCODER_EDP &&
> >>   	    (pipe == PIPE_B || pipe == PIPE_C))
> >> -		intel_de_write(display, TRANS_VTOTAL(display, pipe),
> >> -			       VACTIVE(crtc_vdisplay - 1) |
> >> -			       VTOTAL(crtc_vtotal - 1));
> >> +		intel_crtc_set_vtotal(display, (enum transcoder)pipe,
> >> +				      crtc_vdisplay, crtc_vtotal);
> >>   }
> >>   
> >>   static void intel_set_transcoder_timings_lrr(const struct intel_crtc_state *crtc_state)
> >> @@ -2766,9 +2773,7 @@ static void intel_set_transcoder_timings_lrr(const struct intel_crtc_state *crtc
> >>   	 * The double buffer latch point for TRANS_VTOTAL
> >>   	 * is the transcoder's undelayed vblank.
> >>   	 */
> >> -	intel_de_write(display, TRANS_VTOTAL(display, cpu_transcoder),
> >> -		       VACTIVE(crtc_vdisplay - 1) |
> >> -		       VTOTAL(crtc_vtotal - 1));
> >> +	intel_crtc_set_vtotal(display, cpu_transcoder, crtc_vdisplay, crtc_vtotal);
> > Not really a fan of special casing this that much. I think we should
> > probably handle it the same way we deal with the VBLANK_START vs.
> > TRANS_SET_CONTEXT_LATENCY.
> Hmm I can do away with the helper, and avoid having VTOTAL.Vtotal bits 
> set for specific cases in intel_vrr_set_transcoder_{timings, timings_lrr}.
> >
> > The readount should perhaps just be handled in intel_vrr_get_config().
> 
> I can try this out.
> 
> 
> > But I think we'll need somehting like transcoder_has_vrr() to exclude
> > the DSI transcoders in a consistent way.
> 
> If I understand correctly you mean that wherever we are avoiding VRR 
> related register read/write for DSI, use trans_has_vrr() instead of 
> !transcoder_is_dsi(),
> 
> with trans_has_vrr having call to transcoder_is_dsi()?

Yeah something like that.

> 
> Will perhaps add this as a separate patch.

These two last patches aren't needed to get this stuff actually
working right? Or is the GOP leaving TRANS_VOTAL.vtotal unset?
So yeah, if not needed right now probably best to leave this
stuff for a seaprate series.

-- 
Ville Syrjälä
Intel