[Intel-gfx] [PATCH v4] drm/i915: Fix FBC cfb stride programming for non X-tiled FB

Paulo Zanoni paulo.r.zanoni at intel.com
Fri Jul 14 19:34:51 UTC 2017


Em Sex, 2017-03-24 às 17:12 +0530, Praveen Paneri escreveu:
> When FBC is enabled for linear, legacy Y-tiled and Yf-tiled
> surfaces on gen9, the cfb stride must be programmed by SW as
> 
> cfb_stride = ceiling[(at least plane width in pixels)/
> 		     (32 * compression limit factor)] * 8
> 
> v2: Minor fix for a build error
> 
> v3: Fixed subject, register name and platform check (Ville)
> 
> v4: Added WA details in comment (Paulo)
> 
> Cc: Paulo Zanoni <paulo.r.zanoni at intel.com>
> Cc: Ville Syrjälä <ville.syrjala at linux.intel.com>
> Signed-off-by: Praveen Paneri <praveen.paneri at intel.com>
> ---
>  drivers/gpu/drm/i915/i915_reg.h  | 3 +++
>  drivers/gpu/drm/i915/intel_fbc.c | 9 +++++++++
>  2 files changed, 12 insertions(+)
> 
> diff --git a/drivers/gpu/drm/i915/i915_reg.h
> b/drivers/gpu/drm/i915/i915_reg.h
> index 04c8f69..4adf2e6 100644
> --- a/drivers/gpu/drm/i915/i915_reg.h
> +++ b/drivers/gpu/drm/i915/i915_reg.h
> @@ -6504,6 +6504,9 @@ enum {
>  #define  GLK_CL1_PWR_DOWN	(1 << 11)
>  #define  GLK_CL2_PWR_DOWN	(1 << 12)
>  
> +#define CHICKEN_MISC_4		_MMIO(0x4208c)
> +#define   FBC_STRIDE_OVERRIDE	(1<<13)
> +
>  #define _CHICKEN_PIPESL_1_A	0x420b0
>  #define _CHICKEN_PIPESL_1_B	0x420b4
>  #define  HSW_FBCQ_DIS			(1 << 22)
> diff --git a/drivers/gpu/drm/i915/intel_fbc.c
> b/drivers/gpu/drm/i915/intel_fbc.c
> index ded2add..1183633 100644
> --- a/drivers/gpu/drm/i915/intel_fbc.c
> +++ b/drivers/gpu/drm/i915/intel_fbc.c
> @@ -298,9 +298,18 @@ static bool ilk_fbc_is_active(struct
> drm_i915_private *dev_priv)
>  static void gen7_fbc_activate(struct drm_i915_private *dev_priv)
>  {
>  	struct intel_fbc_reg_params *params = &dev_priv->fbc.params;
> +	struct intel_fbc_state_cache *cache = &dev_priv-
> >fbc.state_cache;

Please take a look at b183b3f14395. We need to store the calculated
stride value in reg_params so the stride value can actually be taken
into consideration when comparing different FBC configurations.


>  	u32 dpfc_ctl;
>  	int threshold = dev_priv->fbc.threshold;
>  
> +	/* Display WA #0529: skl, kbl, bxt, glk */
> +	if (IS_GEN9(dev_priv) &&

AFAIU, this WA shouldn't be applied for GLK.


> +	    i915_gem_object_get_tiling(cache->vma->obj) !=
> I915_TILING_X) {
> +		int cfb_stride = DIV_ROUND_UP(cache->plane.src_w,
> +					      (32 * threshold)) * 8;
> +		I915_WRITE(CHICKEN_MISC_4, FBC_STRIDE_OVERRIDE |
> cfb_stride);

We need to zero those bits when we're not using Y tiling. I mean, at
least bit 13.

Also, CHICKEN_MISC_4 contains other fields not touched here, so we
should probably do a read-modify-write on it, preserving everything
we're not touching.

Thanks,
Paulo


> +	}
> +
>  	dpfc_ctl = 0;
>  	if (IS_IVYBRIDGE(dev_priv))
>  		dpfc_ctl |= IVB_DPFC_CTL_PLANE(params->crtc.plane);


More information about the Intel-gfx mailing list