Re: [Intel-gfx] [PATCH 2/4] drm/i915: Add belts and suspenders locking for seamless M/N changes

[Jani Nikula <jani.nikula@linux.intel.com>]

On Mon, 06 Mar 2023, Ville Syrjala <ville.syrjala@linux.intel.com> wrote:
> From: Ville Syrjälä <ville.syrjala@linux.intel.com>
>
> Add some (probably overkill) locking to protect the vblank
> timestamping constants updates during seamless M/N fastsets.
>
> As everything should be naturally aligned I think the individual
> pieces should probably end up updating atomically enough. So this
> is only really meant to guarantee everyone sees a consistent whole.
>
> All the drm_vblank.c usage is covered by vblank_time_lock,
> and uncore.lock will take care of __intel_get_crtc_scanline()
> that can also be called from outside the core vblank functionality.

The patch seems to do what it says on the box, but I increasingly
dislike the use of uncore.lock for anything other than the nuts and
bolts of uncore.

BR,
Jani.

>
> Currently only crtc_clock and framedur_ns can change, but in
> the future might fastset also across eg. vtotal/vblank_end
> changes, so let's just grab the locks across the whole thing.
>
> Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
> ---
>  drivers/gpu/drm/i915/display/intel_display.c | 24 +++++++++++++++++++-
>  1 file changed, 23 insertions(+), 1 deletion(-)
>
> diff --git a/drivers/gpu/drm/i915/display/intel_display.c b/drivers/gpu/drm/i915/display/intel_display.c
> index a1fbdf32bd21..020320468967 100644
> --- a/drivers/gpu/drm/i915/display/intel_display.c
> +++ b/drivers/gpu/drm/i915/display/intel_display.c
> @@ -5908,6 +5908,8 @@ void intel_crtc_update_active_timings(const struct intel_crtc_state *crtc_state)
>  	struct intel_crtc *crtc = to_intel_crtc(crtc_state->uapi.crtc);
>  	struct drm_i915_private *dev_priv = to_i915(crtc->base.dev);
>  	struct drm_display_mode adjusted_mode;
> +	int vmax_vblank_start = 0;
> +	unsigned long irqflags;
>  
>  	drm_mode_init(&adjusted_mode, &crtc_state->hw.adjusted_mode);
>  
> @@ -5915,11 +5917,28 @@ void intel_crtc_update_active_timings(const struct intel_crtc_state *crtc_state)
>  		adjusted_mode.crtc_vtotal = crtc_state->vrr.vmax;
>  		adjusted_mode.crtc_vblank_end = crtc_state->vrr.vmax;
>  		adjusted_mode.crtc_vblank_start = intel_vrr_vmin_vblank_start(crtc_state);
> -		crtc->vmax_vblank_start = intel_vrr_vmax_vblank_start(crtc_state);
> +		vmax_vblank_start = intel_vrr_vmax_vblank_start(crtc_state);
>  	}
>  
> +	/*
> +	 * Belts and suspenders locking to guarantee everyone sees 100%
> +	 * consistent state during fastset seamless refresh rate changes.
> +	 *
> +	 * vblank_time_lock takes care of all drm_vblank.c stuff, and
> +	 * uncore.lock takes care of __intel_get_crtc_scanline() which
> +	 * may get called elsewhere as well.
> +	 *
> +	 * TODO maybe just protect everything (including
> +	 * __intel_get_crtc_scanline()) with vblank_time_lock?
> +	 * Need to audit everything to make sure it's safe.
> +	 */
> +	spin_lock_irqsave(&dev_priv->drm.vblank_time_lock, irqflags);
> +	spin_lock(&dev_priv->uncore.lock);
> +
>  	drm_calc_timestamping_constants(&crtc->base, &adjusted_mode);
>  
> +	crtc->vmax_vblank_start = vmax_vblank_start;
> +
>  	crtc->mode_flags = crtc_state->mode_flags;
>  
>  	/*
> @@ -5963,6 +5982,9 @@ void intel_crtc_update_active_timings(const struct intel_crtc_state *crtc_state)
>  	} else {
>  		crtc->scanline_offset = 1;
>  	}
> +
> +	spin_unlock(&dev_priv->uncore.lock);
> +	spin_unlock_irqrestore(&dev_priv->drm.vblank_time_lock, irqflags);
>  }
>  
>  /*

-- 
Jani Nikula, Intel Open Source Graphics Center