[Intel-gfx] [RFC 1/2] drm/i915: Add RPL-U CDCLK table

Borah, Chaitanya Kumar chaitanya.kumar.borah at intel.com
Mon Jan 2 06:28:56 UTC 2023


Hello Jani,

> -----Original Message-----
> From: Jani Nikula <jani.nikula at linux.intel.com>
> Sent: Wednesday, November 30, 2022 1:58 PM
> To: Borah, Chaitanya Kumar <chaitanya.kumar.borah at intel.com>; intel-
> gfx at lists.freedesktop.org
> Cc: Syrjala, Ville <ville.syrjala at intel.com>
> Subject: Re: [Intel-gfx] [RFC 1/2] drm/i915: Add RPL-U CDCLK table
> 
> On Wed, 30 Nov 2022, Chaitanya Kumar Borah
> <chaitanya.kumar.borah at intel.com> wrote:
> > A new step of 480MHz has been added on SKUs that have a RPL-U device
> > id. Add a new table which include this new CDCLK step.
> >
> > BSpec: 55409
> >
> > Signed-off-by: Chaitanya Kumar Borah
> <chaitanya.kumar.borah at intel.com>
> > ---
> >  drivers/gpu/drm/i915/display/intel_cdclk.c | 39
> > ++++++++++++++++++++++
> >  1 file changed, 39 insertions(+)
> >
> > diff --git a/drivers/gpu/drm/i915/display/intel_cdclk.c
> > b/drivers/gpu/drm/i915/display/intel_cdclk.c
> > index 0c107a38f9d0..9bfeb1abba47 100644
> > --- a/drivers/gpu/drm/i915/display/intel_cdclk.c
> > +++ b/drivers/gpu/drm/i915/display/intel_cdclk.c
> > @@ -180,6 +180,18 @@ static void i85x_get_cdclk(struct drm_i915_private
> *dev_priv,
> >  	}
> >  }
> >
> > +static bool is_rplu(struct drm_i915_private *dev_priv) {
> > +	switch (INTEL_DEVID(dev_priv)) {
> > +	case 0xA7A1:
> > +	case 0xA721:
> > +	case 0xA7A9:
> 
> No. Under no circumstances are you to sprinkle PCI ID checks directly in code
> like this.
> 
> The only place where PCI IDs go are the macros in include/drm/i915_pciids.h,
> and each PCI ID should be present there only once.
> 
> Those macros should only be used in i915_pci.c and intel_device_info.c.
> 
> If you need a distinction between RPL-P and RPL-U like this (and I didn't check
> this in the spec, just commenting on the implementation) you need to split
> RPL-P and RPL-U definitions and add them as subplatforms in
> intel_device_info.c.
> 
> As a general tip, when you consider using a function or a macro, look up
> where it's used and how. INTEL_DEVID() isn't used like this anywhere.

Thank you for your feedback. I send out a new patch set which implements this change using quirks to identify
the PCI ids. Adding a separate sub platform just for this change seems to be a bit of an overkill. Let us know your thoughts.

Regards,

Chaitanya

> 
> 
> BR,
> Jani.
> 
> 
> > +		return true;
> > +	default:
> > +		return false;
> > +	}
> > +}
> > +
> >  static void i915gm_get_cdclk(struct drm_i915_private *dev_priv,
> >  			     struct intel_cdclk_config *cdclk_config)  { @@ -
> 1329,6
> > +1341,27 @@ static const struct intel_cdclk_vals adlp_cdclk_table[] = {
> >  	{}
> >  };
> >
> > +static const struct intel_cdclk_vals rplu_cdclk_table[] = {
> > +	{ .refclk = 19200, .cdclk = 172800, .divider = 3, .ratio = 27 },
> > +	{ .refclk = 19200, .cdclk = 192000, .divider = 2, .ratio = 20 },
> > +	{ .refclk = 19200, .cdclk = 480000, .divider = 2, .ratio = 50 },
> > +	{ .refclk = 19200, .cdclk = 556800, .divider = 2, .ratio = 58 },
> > +	{ .refclk = 19200, .cdclk = 652800, .divider = 2, .ratio = 68 },
> > +
> > +	{ .refclk = 24000, .cdclk = 176000, .divider = 3, .ratio = 22 },
> > +	{ .refclk = 24000, .cdclk = 192000, .divider = 2, .ratio = 16 },
> > +	{ .refclk = 24000, .cdclk = 480000, .divider = 2, .ratio = 40 },
> > +	{ .refclk = 24000, .cdclk = 552000, .divider = 2, .ratio = 46 },
> > +	{ .refclk = 24400, .cdclk = 648000, .divider = 2, .ratio = 54 },
> > +
> > +	{ .refclk = 38400, .cdclk = 179200, .divider = 3, .ratio = 14 },
> > +	{ .refclk = 38400, .cdclk = 192000, .divider = 2, .ratio = 10 },
> > +	{ .refclk = 38400, .cdclk = 480000, .divider = 2, .ratio = 25 },
> > +	{ .refclk = 38400, .cdclk = 556800, .divider = 2, .ratio = 29 },
> > +	{ .refclk = 38400, .cdclk = 652800, .divider = 2, .ratio = 34 },
> > +	{}
> > +};
> > +
> >  static const struct intel_cdclk_vals dg2_cdclk_table[] = {
> >  	{ .refclk = 38400, .cdclk = 163200, .divider = 2, .ratio = 34, .waveform
> = 0x8888 },
> >  	{ .refclk = 38400, .cdclk = 204000, .divider = 2, .ratio = 34,
> > .waveform = 0x9248 }, @@ -3353,6 +3386,12 @@ void
> intel_init_cdclk_hooks(struct drm_i915_private *dev_priv)
> >  		/* Wa_22011320316:adl-p[a0] */
> >  		if (IS_ADLP_DISPLAY_STEP(dev_priv, STEP_A0, STEP_B0))
> >  			dev_priv->display.cdclk.table =
> adlp_a_step_cdclk_table;
> > +		/*
> > +		 * BSpec: 55409
> > +		 * 480 MHz supported on SKUs that have a RPL-U Device ID
> > +		 */
> > +		else if (is_rplu(dev_priv))
> > +			dev_priv->cdclk.table = rplu_cdclk_table;
> >  		else
> >  			dev_priv->display.cdclk.table = adlp_cdclk_table;
> >  	} else if (IS_ROCKETLAKE(dev_priv)) {
> 
> --
> Jani Nikula, Intel Open Source Graphics Center


More information about the Intel-gfx mailing list