[Intel-gfx] [PATCH] drm/i915/cnl: WaFbcSkipSegments

Paulo Zanoni paulo.r.zanoni at intel.com
Fri Sep 1 21:04:43 UTC 2017


Em Ter, 2017-08-29 às 16:08 -0700, Rodrigo Vivi escreveu:
> Skip compressing 1 segment at the end of the frame,
> avoid a pixel count mismatch nuke event when last active
> pixel and dummy pixel has same color for Odd Plane
> Width / Height.
> 
> Cc: Paulo Zanoni <paulo.r.zanoni at intel.com>
> Signed-off-by: Rodrigo Vivi <rodrigo.vivi at intel.com>
> ---
>  drivers/gpu/drm/i915/i915_reg.h | 2 ++
>  drivers/gpu/drm/i915/intel_pm.c | 4 ++++
>  2 files changed, 6 insertions(+)
> 
> diff --git a/drivers/gpu/drm/i915/i915_reg.h
> b/drivers/gpu/drm/i915/i915_reg.h
> index e2908ae34004..0072ef79bf34 100644
> --- a/drivers/gpu/drm/i915/i915_reg.h
> +++ b/drivers/gpu/drm/i915/i915_reg.h
> @@ -2939,6 +2939,8 @@ enum i915_power_well_id {
>  #define ILK_DPFC_CHICKEN	_MMIO(0x43224)
>  #define   ILK_DPFC_DISABLE_DUMMY0 (1<<8)
>  #define   ILK_DPFC_NUKE_ON_ANY_MODIFICATION	(1<<23)
> +#define   CNL_SKIP_SEG_EN	(1<<12)
> +#define   CNL_SKIP_SEG_COUNT	(1<<10)

#define CNL_SKIP_SEG_COUNT_MASK (3 << 10)
#define CNL_SKIP_SEG_COUNT(x) ((x) << 10)

Also, since you're going to submit v2 anyway, I wouldn't complain if
you drive-by fixed the definitions of ILK_DPFC_CHICKEN bits to use
spaces as documented in the beginning of the file. If we fix every time
we touch a reg, we may at some point be consistent :).


>  #define ILK_FBC_RT_BASE		_MMIO(0x2128)
>  #define   ILK_FBC_RT_VALID	(1<<0)
>  #define   SNB_FBC_FRONT_BUFFER	(1<<1)
> diff --git a/drivers/gpu/drm/i915/intel_pm.c
> b/drivers/gpu/drm/i915/intel_pm.c
> index d5ff0b9f999f..acf793256507 100644
> --- a/drivers/gpu/drm/i915/intel_pm.c
> +++ b/drivers/gpu/drm/i915/intel_pm.c
> @@ -8283,6 +8283,10 @@ static void
> cannonlake_init_clock_gating(struct drm_i915_private *dev_priv)
>  		I915_WRITE(SLICE_UNIT_LEVEL_CLKGATE,
>  			   I915_READ(SLICE_UNIT_LEVEL_CLKGATE) |
>  			   SARBUNIT_CLKGATE_DIS);
> +
> +	/* WaFbcSkipSegments:cnl */

Since this is also documented in BSpec but it doesn't include the WA
name, can you please also document that this is WA #1133 in this
comment? I'd do something like:

/* Display WA #1133: WaFbcSkipSegments:cnl,glk. */

Please notice that we also need to apply this WA for GLK.


> +	I915_WRITE(ILK_DPFC_CHICKEN, I915_READ(ILK_DPFC_CHICKEN) |
> +		   CNL_SKIP_SEG_EN | CNL_SKIP_SEG_COUNT);

Need to clear bit 11 too.

val = READ();
val &= ~CNL_SKIP_SEG_COUNT_MASK;
val |= CNL_SKIP_SEG_EN | CNL_SKIP_SEG_COUNT(1);

>  }
>  
>  static void kabylake_init_clock_gating(struct drm_i915_private
> *dev_priv)


More information about the Intel-gfx mailing list