[Intel-gfx] [PATCH v2 13/19] drm/i915/dsb: Use non-posted register writes for legacy LUT

[Shankar, Uma]

> -----Original Message-----
> From: Intel-gfx <intel-gfx-bounces at lists.freedesktop.org> On Behalf Of Ville
> Syrjala
> Sent: Wednesday, June 7, 2023 12:45 AM
> To: intel-gfx at lists.freedesktop.org
> Subject: [Intel-gfx] [PATCH v2 13/19] drm/i915/dsb: Use non-posted register
> writes for legacy LUT
> 
> From: Ville Syrjälä <ville.syrjala at linux.intel.com>
> 
> The DSB has problems writing the legacy LUT. The two workarounds I've
> discoverted are:
> - write each entry twice back to back
> - use non-posted writes
> 
> Let's use non-posted writes as that seems a bit more standard.

Change looks good but I feel it will be good to get this documented in spec.
Not able to locate any reference.

Anyways, with empirical data based on your findings no concern as such.
Reviewed-by: Uma Shankar <uma.shankar at intel.com>

> TODO: measure which is faster
> 
> Signed-off-by: Ville Syrjälä <ville.syrjala at linux.intel.com>
> ---
>  drivers/gpu/drm/i915/display/intel_color.c | 11 +++++++++++
>  1 file changed, 11 insertions(+)
> 
> diff --git a/drivers/gpu/drm/i915/display/intel_color.c
> b/drivers/gpu/drm/i915/display/intel_color.c
> index 077e45372dab..b3dd4013d058 100644
> --- a/drivers/gpu/drm/i915/display/intel_color.c
> +++ b/drivers/gpu/drm/i915/display/intel_color.c
> @@ -1265,9 +1265,20 @@ static void ilk_load_lut_8(const struct intel_crtc_state
> *crtc_state,
> 
>  	lut = blob->data;
> 
> +	/*
> +	 * DSB fails to correctly load the legacy LUT
> +	 * unless we either write each entry twice,
> +	 * or use non-posted writes
> +	 */
> +	if (crtc_state->dsb)
> +		intel_dsb_nonpost_start(crtc_state->dsb);
> +
>  	for (i = 0; i < 256; i++)
>  		ilk_lut_write(crtc_state, LGC_PALETTE(pipe, i),
>  			      i9xx_lut_8(&lut[i]));
> +
> +	if (crtc_state->dsb)
> +		intel_dsb_nonpost_end(crtc_state->dsb);
>  }
> 
>  static void ilk_load_lut_10(const struct intel_crtc_state *crtc_state,
> --
> 2.39.3