On Thu, Sep 05, 2019 at 02:38:24PM -0700, Matt Roper wrote:
> When calculating the PLL ratio we were still assuming that the CD2X
> divider is always /1. For the new frequencies that use a /2 divider
> that needs to be accounted for, otherwise our VCO result will be twice
> as large as it should be.
>
> Fixes: 3d1da92baffe ("drm/i915: Add 324mhz and 326.4mhz cdclks for gen11+")
> Cc: José Roberto de Souza <jose.souza@xxxxxxxxx>
> Cc: Lucas De Marchi <lucas.demarchi@xxxxxxxxx>
> Signed-off-by: Matt Roper <matthew.d.roper@xxxxxxxxx>
> ---
> drivers/gpu/drm/i915/display/intel_cdclk.c | 17 ++++++++++++++---
> 1 file changed, 14 insertions(+), 3 deletions(-)
>
> diff --git a/drivers/gpu/drm/i915/display/intel_cdclk.c b/drivers/gpu/drm/i915/display/intel_cdclk.c
> index 1329d3e60e26..55801aeefd1c 100644
> --- a/drivers/gpu/drm/i915/display/intel_cdclk.c
> +++ b/drivers/gpu/drm/i915/display/intel_cdclk.c
> @@ -1794,6 +1794,7 @@ static int icl_calc_cdclk(int min_cdclk, unsigned int ref)
>
> static int icl_calc_cdclk_pll_vco(struct drm_i915_private *dev_priv, int cdclk)
> {
> + int cd2xdiv;
> int ratio;
>
> if (cdclk == dev_priv->cdclk.hw.bypass)
> @@ -1805,27 +1806,37 @@ static int icl_calc_cdclk_pll_vco(struct drm_i915_private *dev_priv, int cdclk)
> /* fall through */
> case 172800:
> case 307200:
> - case 326400:
> case 556800:
> case 652800:
> + cd2xdiv = 1;
> WARN_ON(dev_priv->cdclk.hw.ref != 19200 &&
> dev_priv->cdclk.hw.ref != 38400);
> break;
> case 180000:
> case 312000:
> - case 324000:
> case 552000:
> case 648000:
> + cd2xdiv = 1;
> WARN_ON(dev_priv->cdclk.hw.ref != 24000);
> break;
> case 192000:
> + cd2xdiv = 1;
> WARN_ON(dev_priv->cdclk.hw.ref != 19200 &&
> dev_priv->cdclk.hw.ref != 38400 &&
> dev_priv->cdclk.hw.ref != 24000);
> break;
> + case 326400:
> + cd2xdiv = 2;
> + WARN_ON(dev_priv->cdclk.hw.ref != 19200 &&
> + dev_priv->cdclk.hw.ref != 38400);
> + break;
> + case 324000:
> + cd2xdiv = 2;
> + WARN_ON(dev_priv->cdclk.hw.ref != 24000);
> + break;
> }
>
> - ratio = cdclk / (dev_priv->cdclk.hw.ref / 2);
> + ratio = cdclk / (dev_priv->cdclk.hw.ref / cd2xdiv / 2);
>
> return dev_priv->cdclk.hw.ref * ratio;
Isn't that just cdclk*cd2xdiv*2 ?
I do kinda like the idea of not hardcoding the vco ratio here.
We should probably do the same for the older platforms as well...
> }
> --
> 2.20.1
>
> _______________________________________________
> Intel-gfx mailing list
> Intel-gfx@xxxxxxxxxxxxxxxxxxxxx
> https://lists.freedesktop.org/mailman/listinfo/intel-gfx
--
Ville Syrjälä
Intel
_______________________________________________
Intel-gfx mailing list
Intel-gfx@xxxxxxxxxxxxxxxxxxxxx
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
