Re: [PATCH 10/16] drm/i915: fix CFB size calculation

From: "Ville Syrjälä" <ville.syrjala@linux.intel.com>
To: Paulo Zanoni <paulo.r.zanoni@intel.com>
Cc: intel-gfx@lists.freedesktop.org
Subject: Re: [PATCH 10/16] drm/i915: fix CFB size calculation
Date: Fri, 28 Aug 2015 20:25:42 +0300	[thread overview]
Message-ID: <20150828172542.GD5176@intel.com> (raw)
In-Reply-To: <1439588061-18064-11-git-send-email-paulo.r.zanoni@intel.com>

On Fri, Aug 14, 2015 at 06:34:15PM -0300, Paulo Zanoni wrote:
> We were considering the whole framebuffer height, but the spec clearly
> says that we should only consider the active display height size.
> 
> On my current testing machine, this moves us from 124 successes and
> 502 skips to 209 successes and 417 skips on "kms_frontbuffer_tracking
> --fbc-only". The high amount of skips is due to the --fbc-only
> argument. We had those skips due to not enough stolen memory for the
> tests. We're now passing the maximum possible amount: 209.
> 
> Note: when this patch was written, the amount of tests we had for FBC
> was different than what we have now.
> 
> Testcase: igt/kms_frontbuffer_tracking/fbc-*
> Signed-off-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
> ---
>  drivers/gpu/drm/i915/intel_fbc.c | 13 +++++++++----
>  1 file changed, 9 insertions(+), 4 deletions(-)
> 
> diff --git a/drivers/gpu/drm/i915/intel_fbc.c b/drivers/gpu/drm/i915/intel_fbc.c
> index dc84e67..cfd4cba 100644
> --- a/drivers/gpu/drm/i915/intel_fbc.c
> +++ b/drivers/gpu/drm/i915/intel_fbc.c
> @@ -695,9 +695,15 @@ void intel_fbc_cleanup_cfb(struct drm_i915_private *dev_priv)
>  	mutex_unlock(&dev_priv->fbc.lock);
>  }
>  
> -static int intel_fbc_setup_cfb(struct drm_i915_private *dev_priv, int size,
> -			       int fb_cpp)
> +static int intel_fbc_setup_cfb(struct intel_crtc *crtc)
>  {
> +	struct drm_i915_private *dev_priv = crtc->base.dev->dev_private;
> +	struct drm_framebuffer *fb = crtc->base.primary->fb;
> +	int size, fb_cpp;
> +
> +	size = crtc->config->pipe_src_h * fb->pitches[0];
> +	fb_cpp = drm_format_plane_cpp(fb->pixel_format, 0);
> +

From the looks of it SKL ought to be able to use FBC even with a
non-fullscreen plane, so this should look up the size from the plane
state. What I don't know is whether FBC would still work when plane scaling
is enabled. But I suppose it if can, it would deal with the data before
scaling, so using the clipped src size would be the right thing here I think.

>  	if (size <= dev_priv->fbc.uncompressed_size)
>  		return 0;
>  
> @@ -844,8 +850,7 @@ static void __intel_fbc_update(struct drm_i915_private *dev_priv)
>  		goto out_disable;
>  	}
>  
> -	if (intel_fbc_setup_cfb(dev_priv, obj->base.size,
> -				drm_format_plane_cpp(fb->pixel_format, 0))) {
> +	if (intel_fbc_setup_cfb(intel_crtc)) {
>  		set_no_fbc_reason(dev_priv, FBC_STOLEN_TOO_SMALL);
>  		goto out_disable;
>  	}
> -- 
> 2.4.6
> 
> _______________________________________________
> Intel-gfx mailing list
> Intel-gfx@lists.freedesktop.org
> http://lists.freedesktop.org/mailman/listinfo/intel-gfx

-- 
Ville Syrjälä
Intel OTC
_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/intel-gfx