Re: [PATCH 17/21] drm/i915/display: Move vrr.guardband/pipeline_full out of !fastset block

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

On Sun, Mar 09, 2025 at 10:05:48PM +0530, Nautiyal, Ankit K wrote:
> 
> On 3/7/2025 7:56 PM, Ville Syrjälä wrote:
> > On Thu, Mar 06, 2025 at 06:40:56PM +0530, Ankit Nautiyal wrote:
> >> The vrr.guardband/pipeline_full depend on the vrr.vmin. Since we have
> >> set vrr.vmin to adjusted_mode->crtc_vtotal, this shouldn't change on the
> >> fly. With this we can move vrr.guardband/pipeline_full out from !fastset
> >> block.
> >>
> >> Signed-off-by: Ankit Nautiyal <ankit.k.nautiyal@intel.com>
> >> ---
> >>   drivers/gpu/drm/i915/display/intel_display.c | 5 +++--
> >>   1 file changed, 3 insertions(+), 2 deletions(-)
> >>
> >> diff --git a/drivers/gpu/drm/i915/display/intel_display.c b/drivers/gpu/drm/i915/display/intel_display.c
> >> index 322a05648f58..a642496e366c 100644
> >> --- a/drivers/gpu/drm/i915/display/intel_display.c
> >> +++ b/drivers/gpu/drm/i915/display/intel_display.c
> >> @@ -5393,8 +5393,6 @@ intel_pipe_config_compare(const struct intel_crtc_state *current_config,
> >>   		PIPE_CONF_CHECK_I(vrr.vmin);
> >>   		PIPE_CONF_CHECK_I(vrr.vmax);
> >>   		PIPE_CONF_CHECK_I(vrr.flipline);
> >> -		PIPE_CONF_CHECK_I(vrr.pipeline_full);
> >> -		PIPE_CONF_CHECK_I(vrr.guardband);
> >>   		PIPE_CONF_CHECK_I(vrr.vsync_start);
> >>   		PIPE_CONF_CHECK_I(vrr.vsync_end);
> >>   		PIPE_CONF_CHECK_LLI(cmrr.cmrr_m);
> >> @@ -5402,6 +5400,9 @@ intel_pipe_config_compare(const struct intel_crtc_state *current_config,
> >>   		PIPE_CONF_CHECK_BOOL(cmrr.enable);
> >>   	}
> >>   
> >> +	PIPE_CONF_CHECK_I(vrr.pipeline_full);
> >> +	PIPE_CONF_CHECK_I(vrr.guardband);
> >
> > Assuming we can't reprogram the guardband safely live
> > I think this would have to become
> >
> > if (!fastset || always_use_vrr_tg()) {
> >          ...
> > }
> >
> > which avoids breaking the LRR fastset on older platforms.
> >
> > Whether we can still risk the fastboot exception for the new
> > platforms I don't know. I guess leave it out for now and
> > ponder it further later.
> >
> > In the future I guess one option to resurrecting the fastsets
> > would be to somehow use a more fixed size gurdband instead of
> > the full vblank length, but that would need a lot of thought
> > and work, so definitely not something we can just do right
> > now.
> 
> 
> Yes I agree.
> 
> Will have the guardband under the new check.
> 
> Do we need to let pipeline_full as it is in !fastset since 
> always_use_vrr_tg() is anyways false for older platforms? Or we keep it 
> along with guardband?

I think we want to keep the two together. That way we can also
easily test the always_use_vrr_tg() codepaths on older platforms.

-- 
Ville Syrjälä
Intel