Re: [Intel-gfx] [PATCH v2 06/22] drm/i915: Add helper to modeset a set of pipes

[Imre Deak <imre.deak@intel.com>]

On Fri, Sep 08, 2023 at 10:25:59PM +0300, Ville Syrjälä wrote:
> On Thu, Aug 24, 2023 at 11:05:01AM +0300, Imre Deak wrote:
> > Add intel_modeset_pipes_in_mask() to modeset a provided set of pipes,
> > used in a follow-up patch.
> > 
> > While at it add DocBook descriptions for the exported functions.
> > 
> > v2:
> > - Add a flag controlling if active planes are force updated as well.
> > - Add DockBook descriptions.
> > 
> > Signed-off-by: Imre Deak <imre.deak@intel.com>
> > ---
> >  drivers/gpu/drm/i915/display/intel_display.c | 43 ++++++++++++++++++--
> >  drivers/gpu/drm/i915/display/intel_display.h |  3 ++
> >  2 files changed, 42 insertions(+), 4 deletions(-)
> > 
> > diff --git a/drivers/gpu/drm/i915/display/intel_display.c b/drivers/gpu/drm/i915/display/intel_display.c
> > index db3c26e013e3b..a1956b89fd75d 100644
> > --- a/drivers/gpu/drm/i915/display/intel_display.c
> > +++ b/drivers/gpu/drm/i915/display/intel_display.c
> > @@ -5420,8 +5420,24 @@ intel_verify_planes(struct intel_atomic_state *state)
> >  			     plane_state->uapi.visible);
> >  }
> >  
> > -int intel_modeset_all_pipes(struct intel_atomic_state *state,
> > -			    const char *reason)
> > +/**
> > + * intel_modeset_pipes_in_mask - force a full modeset on a set of pipes
> > + * @state: intel atomic state
> > + * @reason: the reason for the full modeset
> > + * @mask: mask of pipes to modeset
> > + * @update_active_planes: force updating all active planes
> > + *
> > + * Force a full modeset on CRTCs in @mask due to the description in @reason.
> > + * Also force updating all active planes in each modeset CRTC if
> > + * @update_active_planes is %true. This flag must be set to %true if the
> > + * function is called after new plane states are computed already and
> > + * set to %false otherwise.
> > + *
> > + * Returns 0 in case of success, negative error code otherwise.
> > + */
> > +int intel_modeset_pipes_in_mask(struct intel_atomic_state *state,
> > +				const char *reason, u8 mask,
> > +				bool update_active_planes)
> 
> Not really a fan of this parametrized behaviour. Also pretty sure we
> have several other places that trigger modesets early in the atomic
> check and all those just hand roll currently. So if we want a helper
> then it might make more sense to try to combine all those early cases
> into a new function. We could rename the current thing _late() or
> something to make it clearer when it should be used.

Ok, so the exported functions would be
intel_modeset_pipes_in_mask_early() / intel_modeset_all_pipes_late().

Yes, noticed the other early users as well, I suppose it's ok to
refactor those as a follow-up (iiuc I have even some patches for that).

> 
> >  {
> >  	struct drm_i915_private *dev_priv = to_i915(state->base.dev);
> >  	struct intel_crtc *crtc;
> > @@ -5430,7 +5446,7 @@ int intel_modeset_all_pipes(struct intel_atomic_state *state,
> >  	 * Add all pipes to the state, and force
> >  	 * a modeset on all the active ones.
> >  	 */
> > -	for_each_intel_crtc(&dev_priv->drm, crtc) {
> > +	for_each_intel_crtc_in_pipe_mask(&dev_priv->drm, crtc, mask) {
> >  		struct intel_crtc_state *crtc_state;
> >  		int ret;
> >  
> > @@ -5461,7 +5477,9 @@ int intel_modeset_all_pipes(struct intel_atomic_state *state,
> >  		if (ret)
> >  			return ret;
> >  
> > -		crtc_state->update_planes |= crtc_state->active_planes;
> > +		if (update_active_planes)
> > +			crtc_state->update_planes |= crtc_state->active_planes;
> > +
> >  		crtc_state->async_flip_planes = 0;
> >  		crtc_state->do_async_flip = false;
> >  	}
> > @@ -5469,6 +5487,23 @@ int intel_modeset_all_pipes(struct intel_atomic_state *state,
> >  	return 0;
> >  }
> >  
> > +/**
> > + * intel_modeset_all_pipes - force a full modeset on all pipes
> > + * @state: intel atomic state
> > + * @reason: the reason for the full modeset
> > + *
> > + * Force a full modeset on all CRTCs due to the description in @reason.
> > + * This function can be called only after new plane states are computed
> > + * already.
> > + *
> > + * Returns 0 in case of success, negative error code otherwise.
> > + */
> > +int intel_modeset_all_pipes(struct intel_atomic_state *state,
> > +			    const char *reason)
> > +{
> > +	return intel_modeset_pipes_in_mask(state, reason, -1, true);
> > +}
> > +
> >  /*
> >   * This implements the workaround described in the "notes" section of the mode
> >   * set sequence documentation. When going from no pipes or single pipe to
> > diff --git a/drivers/gpu/drm/i915/display/intel_display.h b/drivers/gpu/drm/i915/display/intel_display.h
> > index 49ac8473b988b..d9a54610d9d5e 100644
> > --- a/drivers/gpu/drm/i915/display/intel_display.h
> > +++ b/drivers/gpu/drm/i915/display/intel_display.h
> > @@ -515,6 +515,9 @@ void intel_update_watermarks(struct drm_i915_private *i915);
> >  /* modesetting */
> >  int intel_modeset_all_pipes(struct intel_atomic_state *state,
> >  			    const char *reason);
> > +int intel_modeset_pipes_in_mask(struct intel_atomic_state *state,
> > +				const char *reason, u8 pipe_mask,
> > +				bool update_active_planes);
> >  void intel_modeset_get_crtc_power_domains(struct intel_crtc_state *crtc_state,
> >  					  struct intel_power_domain_mask *old_domains);
> >  void intel_modeset_put_crtc_power_domains(struct intel_crtc *crtc,
> > -- 
> > 2.37.2
> 
> -- 
> Ville Syrjälä
> Intel