[Intel-gfx] [PATCH] drm/i915/skl: Correct other-pipe watermark update condition check (v2)

Kumar, Mahesh mahesh1.kumar at intel.com
Mon Oct 26 23:16:25 PDT 2015

Yes, it doesn't solve all warnings.
Other than fbdev restore, there is one warning which I faced during 
DPMS/Hot-plug, which is due to calculation of watermarks twice. Once for 
first CRTC "Pipe-A" (during which it consider/calculates watermark for 
other pipe "Pipe-B" as well & stores in structure 
"intel_crtc->wm.active.skl") & Now in same atomic_commit, when it 
calculate watermark for second CRTC "Pipe-B" , that time watermark for 
other-pipe (in this case pipe-A) already have calculated values so it 
gives !wm_changed warning.

Hoping 2-stage watermark programming will take care of this, as in that 
case, If I'm not misunderstood  final results should be stored after 
calculation for all pipes.


On 10/23/2015 10:11 PM, Matt Roper wrote:
> From: "Kumar, Mahesh" <mahesh1.kumar at intel.com>
> If ddb allocation for planes in current CRTC is changed, that doesn't
> lead to ddb allocation change for other CRTCs, because our DDB allocation
> is not dynamic according to plane parameters, ddb is allocated according
> to number of CRTC enabled, & divided equally among CTRC's.
> In current condition check during Watermark calculation, if number of
> plane/ddb allocation changes for current CRTC, Watermark for other pipes
> are recalculated. But there is no change in DDB allocation of other pipe
> so watermark is also not changed, This leads to warning messages.
> WARN_ON(!wm_changed)
> This patch corrects this and check if DDB allocation for pipes is changed,
> then only recalculate watermarks.
> v2 (by Matt): Rebased to latest -nightly and fixed a typo
> Signed-off-by: Kumar, Mahesh <mahesh1.kumar at intel.com>
> Reviewed-by(v1): Ville Syrjälä <ville.syrjala at linux.intel.com>
> Signed-off-by: Matt Roper <matthew.d.roper at intel.com>
> ---
> I don't think this solves all the !wm_changed warnings (I still see one during
> fbdev restore following various igt tests), but it seems like a move in the
> right direction so I figured I'd go ahead and rebase Mahesh' patch so it can
> get merged.
>   drivers/gpu/drm/i915/intel_pm.c | 12 +++++-------
>   1 file changed, 5 insertions(+), 7 deletions(-)
> diff --git a/drivers/gpu/drm/i915/intel_pm.c b/drivers/gpu/drm/i915/intel_pm.c
> index 0fb0459..0467e34 100644
> --- a/drivers/gpu/drm/i915/intel_pm.c
> +++ b/drivers/gpu/drm/i915/intel_pm.c
> @@ -3064,14 +3064,12 @@ static bool skl_ddb_allocation_changed(const struct skl_ddb_allocation *new_ddb,
>   	struct drm_device *dev = intel_crtc->base.dev;
>   	struct drm_i915_private *dev_priv = dev->dev_private;
>   	const struct skl_ddb_allocation *cur_ddb = &dev_priv->wm.skl_hw.ddb;
> -	enum pipe pipe = intel_crtc->pipe;
> -
> -	if (memcmp(new_ddb->plane[pipe], cur_ddb->plane[pipe],
> -		   sizeof(new_ddb->plane[pipe])))
> -		return true;
> -	if (memcmp(&new_ddb->plane[pipe][PLANE_CURSOR], &cur_ddb->plane[pipe][PLANE_CURSOR],
> -		    sizeof(new_ddb->plane[pipe][PLANE_CURSOR])))
> +	/*
> +	 * If ddb allocation of pipes changed, it may require recalculation of
> +	 * watermarks
> +	 */
> +	if (memcmp(new_ddb->pipe, cur_ddb->pipe, sizeof(new_ddb->pipe)))
>   		return true;
>   	return false;

More information about the Intel-gfx mailing list