[Intel-gfx] [RFC] drm/i915/display: Move cdclk checks to atomic check

Srivatsa, Anusha anusha.srivatsa at intel.com
Thu Dec 16 11:13:54 UTC 2021



> -----Original Message-----
> From: Jani Nikula <jani.nikula at linux.intel.com>
> Sent: Friday, December 10, 2021 4:17 PM
> To: Srivatsa, Anusha <anusha.srivatsa at intel.com>; intel-
> gfx at lists.freedesktop.org
> Subject: Re: [Intel-gfx] [RFC] drm/i915/display: Move cdclk checks to atomic
> check
> 
> On Thu, 09 Dec 2021, Anusha Srivatsa <anusha.srivatsa at intel.com> wrote:
> > i915 has squashing for DG2 and crawling for ADLP.
> > Moving the checks to atomic check phase so at a later phase we know
> > how the cdclk changes.
> 
> Just some high level comments:
> 
> - Functions named intel_cdclk_can_foo() must *not* change the state,
>   that's unexpected and surprising.
> 
> - There's a bunch of state handling already for cdclk, please don't just
>   dump new state in drm_i915_private, outside of the existing states. In
>   particular, storing yet another copy of cdclk is suspicious.
> 
> - Please don't add short stuff like CRAWL and SQUASH to our module
>   namespace.
> 
Will keep in mind for hopefully a better non-RFC version. Thanks for the feedback.

Anusha
> BR,
> Jani.
> 
> 
> >
> > Signed-off-by: Anusha Srivatsa <anusha.srivatsa at intel.com>
> > ---
> >  drivers/gpu/drm/i915/display/intel_cdclk.c | 49 +++++++++++++---------
> >  drivers/gpu/drm/i915/i915_drv.h            | 11 +++++
> >  2 files changed, 41 insertions(+), 19 deletions(-)
> >
> > diff --git a/drivers/gpu/drm/i915/display/intel_cdclk.c
> > b/drivers/gpu/drm/i915/display/intel_cdclk.c
> > index 639a64733f61..9382dd24d889 100644
> > --- a/drivers/gpu/drm/i915/display/intel_cdclk.c
> > +++ b/drivers/gpu/drm/i915/display/intel_cdclk.c
> > @@ -1707,9 +1707,27 @@ static void bxt_set_cdclk(struct
> drm_i915_private *dev_priv,
> >  		return;
> >  	}
> >
> > -	if (HAS_CDCLK_CRAWL(dev_priv) && dev_priv->cdclk.hw.vco > 0 &&
> vco > 0) {
> > -		if (dev_priv->cdclk.hw.vco != vco)
> > -			adlp_cdclk_pll_crawl(dev_priv, vco);
> > +	if (DISPLAY_VER(dev_priv) >= 12) {
> > +		int i = 0;
> > +		u32 squash_ctl = 0;
> > +		struct cdclk_steps *cdclk_steps = dev_priv->cdclk.steps;
> > +
> > +		for (i = 0; i < CDCLK_ACTIONS; i++) {
> > +			switch (cdclk_steps[i].action) {
> > +			case CRAWL:
> > +				adlp_cdclk_pll_crawl(dev_priv, vco);
> > +				break;
> > +			case SQUASH:
> > +				waveform =
> cdclk_squash_waveform(dev_priv, cdclk_steps[i].cdclk);
> > +				clock = vco / 2;
> > +				squash_ctl = CDCLK_SQUASH_ENABLE |
> > +				CDCLK_SQUASH_WINDOW_SIZE(0xf) |
> waveform;
> > +				intel_de_write(dev_priv,
> CDCLK_SQUASH_CTL, squash_ctl);
> > +				break;
> > +			default:
> > +				break;
> > +			}
> > +		}
> >  	} else if (DISPLAY_VER(dev_priv) >= 11) {
> >  		if (dev_priv->cdclk.hw.vco != 0 &&
> >  		    dev_priv->cdclk.hw.vco != vco)
> > @@ -1726,22 +1744,7 @@ static void bxt_set_cdclk(struct
> drm_i915_private *dev_priv,
> >  			bxt_de_pll_enable(dev_priv, vco);
> >  	}
> >
> > -	waveform = cdclk_squash_waveform(dev_priv, cdclk);
> > -
> > -	if (waveform)
> > -		clock = vco / 2;
> > -	else
> > -		clock = cdclk;
> > -
> > -	if (has_cdclk_squasher(dev_priv)) {
> > -		u32 squash_ctl = 0;
> > -
> > -		if (waveform)
> > -			squash_ctl = CDCLK_SQUASH_ENABLE |
> > -				CDCLK_SQUASH_WINDOW_SIZE(0xf) |
> waveform;
> > -
> > -		intel_de_write(dev_priv, CDCLK_SQUASH_CTL, squash_ctl);
> > -	}
> > +	clock = cdclk;
> >
> >  	val = bxt_cdclk_cd2x_div_sel(dev_priv, clock, vco) |
> >  		bxt_cdclk_cd2x_pipe(dev_priv, pipe) | @@ -1934,6 +1937,7
> @@ static
> > bool intel_cdclk_can_crawl(struct drm_i915_private *dev_priv,
> >  				  const struct intel_cdclk_config *a,
> >  				  const struct intel_cdclk_config *b)  {
> > +	struct cdclk_steps *cdclk_transition = dev_priv->cdclk.steps;
> >  	int a_div, b_div;
> >
> >  	if (!HAS_CDCLK_CRAWL(dev_priv))
> > @@ -1946,6 +1950,9 @@ static bool intel_cdclk_can_crawl(struct
> drm_i915_private *dev_priv,
> >  	a_div = DIV_ROUND_CLOSEST(a->vco, a->cdclk);
> >  	b_div = DIV_ROUND_CLOSEST(b->vco, b->cdclk);
> >
> > +	cdclk_transition[0].action = CRAWL;
> > +	cdclk_transition[0].cdclk = b->cdclk;
> > +
> >  	return a->vco != 0 && b->vco != 0 &&
> >  		a->vco != b->vco &&
> >  		a_div == b_div &&
> > @@ -1956,6 +1963,7 @@ static bool intel_cdclk_can_squash(struct
> drm_i915_private *dev_priv,
> >  				   const struct intel_cdclk_config *a,
> >  				   const struct intel_cdclk_config *b)  {
> > +	struct cdclk_steps *cdclk_transition = dev_priv->cdclk.steps;
> >  	/*
> >  	 * FIXME should store a bit more state in intel_cdclk_config
> >  	 * to differentiate squasher vs. cd2x divider properly. For @@
> > -1965,6 +1973,9 @@ static bool intel_cdclk_can_squash(struct
> drm_i915_private *dev_priv,
> >  	if (!has_cdclk_squasher(dev_priv))
> >  		return false;
> >
> > +	cdclk_transition[0].action = SQUASH;
> > +	cdclk_transition[0].cdclk = b->cdclk;
> > +
> >  	return a->cdclk != b->cdclk &&
> >  		a->vco != 0 &&
> >  		a->vco == b->vco &&
> > diff --git a/drivers/gpu/drm/i915/i915_drv.h
> > b/drivers/gpu/drm/i915/i915_drv.h index ae7dc7862b5d..c03299253b81
> > 100644
> > --- a/drivers/gpu/drm/i915/i915_drv.h
> > +++ b/drivers/gpu/drm/i915/i915_drv.h
> > @@ -117,6 +117,12 @@
> >
> >  struct drm_i915_gem_object;
> >
> > +enum cdclk_actions {
> > +	SQUASH = 0,
> > +	CRAWL,
> > +	CDCLK_ACTIONS
> > +};
> > +
> >  /* Threshold == 5 for long IRQs, 50 for short */  #define
> > HPD_STORM_DEFAULT_THRESHOLD 50
> >
> > @@ -782,6 +788,11 @@ struct drm_i915_private {
> >  		const struct intel_cdclk_vals *table;
> >
> >  		struct intel_global_obj obj;
> > +
> > +		struct cdclk_steps {
> > +			enum cdclk_actions action;
> > +			u32 cdclk;
> > +		} steps[CDCLK_ACTIONS];
> >  	} cdclk;
> >
> >  	struct {
> 
> --
> Jani Nikula, Intel Open Source Graphics Center


More information about the Intel-gfx mailing list