Re: [PATCH v3 04/11] drm/sun4i: abstract the layer type

From: Maxime Ripard <maxime.ripard@free-electrons.com>
To: Icenowy Zheng <icenowy@aosc.io>
Cc: Rob Herring <robh+dt@kernel.org>, Chen-Yu Tsai <wens@csie.org>,
	Jernej Skrabec <jernej.skrabec@siol.net>,
	linux-clk@vger.kernel.org, devicetree@vger.kernel.org,
	linux-arm-kernel@lists.infradead.org,
	linux-kernel@vger.kernel.org, dri-devel@lists.freedesktop.org,
	linux-sunxi@googlegroups.com
Subject: Re: [PATCH v3 04/11] drm/sun4i: abstract the layer type
Date: Mon, 3 Apr 2017 10:14:23 +0200	[thread overview]
Message-ID: <20170403081423.p6x4ytfoga2krygv@lukather> (raw)
In-Reply-To: <20170329194613.55548-5-icenowy@aosc.io>

[-- Attachment #1: Type: text/plain, Size: 3582 bytes --]

Hi,

On Thu, Mar 30, 2017 at 03:46:06AM +0800, Icenowy Zheng wrote:
> As we are going to add support for the Allwinner DE2 Mixer in sun4i-drm
> driver, we will finally have two types of layer.
> 
> Abstract the layer type to void * and a ops struct, which contains the
> only function used by crtc -- get the drm_plane struct of the layer.
> 
> Signed-off-by: Icenowy Zheng <icenowy@aosc.io>
> ---
> Refactored patch in v3.
> 
>  drivers/gpu/drm/sun4i/sun4i_crtc.c  | 19 +++++++++++--------
>  drivers/gpu/drm/sun4i/sun4i_crtc.h  |  3 ++-
>  drivers/gpu/drm/sun4i/sun4i_layer.c | 19 ++++++++++++++++++-
>  drivers/gpu/drm/sun4i/sun4i_layer.h |  2 +-
>  drivers/gpu/drm/sun4i/sunxi_layer.h | 17 +++++++++++++++++
>  5 files changed, 49 insertions(+), 11 deletions(-)
>  create mode 100644 drivers/gpu/drm/sun4i/sunxi_layer.h
> 
> diff --git a/drivers/gpu/drm/sun4i/sun4i_crtc.c b/drivers/gpu/drm/sun4i/sun4i_crtc.c
> index 3c876c3a356a..33854ee7f636 100644
> --- a/drivers/gpu/drm/sun4i/sun4i_crtc.c
> +++ b/drivers/gpu/drm/sun4i/sun4i_crtc.c
> @@ -29,6 +29,7 @@
>  #include "sun4i_crtc.h"
>  #include "sun4i_drv.h"
>  #include "sun4i_layer.h"
> +#include "sunxi_layer.h"
>  #include "sun4i_tcon.h"
>  
>  static void sun4i_crtc_atomic_begin(struct drm_crtc *crtc,
> @@ -149,7 +150,7 @@ struct sun4i_crtc *sun4i_crtc_init(struct drm_device *drm,
>  	scrtc->tcon = tcon;
>  
>  	/* Create our layers */
> -	scrtc->layers = sun4i_layers_init(drm, scrtc->backend);
> +	scrtc->layers = (void **)sun4i_layers_init(drm, scrtc);
>  	if (IS_ERR(scrtc->layers)) {
>  		dev_err(drm->dev, "Couldn't create the planes\n");
>  		return NULL;
> @@ -157,14 +158,15 @@ struct sun4i_crtc *sun4i_crtc_init(struct drm_device *drm,
>  
>  	/* find primary and cursor planes for drm_crtc_init_with_planes */
>  	for (i = 0; scrtc->layers[i]; i++) {
> -		struct sun4i_layer *layer = scrtc->layers[i];
> +		void *layer = scrtc->layers[i];
> +		struct drm_plane *plane = scrtc->layer_ops->get_plane(layer);
>  
> -		switch (layer->plane.type) {
> +		switch (plane->type) {
>  		case DRM_PLANE_TYPE_PRIMARY:
> -			primary = &layer->plane;
> +			primary = plane;
>  			break;
>  		case DRM_PLANE_TYPE_CURSOR:
> -			cursor = &layer->plane;
> +			cursor = plane;
>  			break;
>  		default:
>  			break;
> @@ -190,10 +192,11 @@ struct sun4i_crtc *sun4i_crtc_init(struct drm_device *drm,
>  	/* Set possible_crtcs to this crtc for overlay planes */
>  	for (i = 0; scrtc->layers[i]; i++) {
>  		uint32_t possible_crtcs = BIT(drm_crtc_index(&scrtc->crtc));
> -		struct sun4i_layer *layer = scrtc->layers[i];
> +		void *layer = scrtc->layers[i];
> +		struct drm_plane *plane = scrtc->layer_ops->get_plane(layer);
>  
> -		if (layer->plane.type == DRM_PLANE_TYPE_OVERLAY)
> -			layer->plane.possible_crtcs = possible_crtcs;
> +		if (plane->type == DRM_PLANE_TYPE_OVERLAY)
> +			plane->possible_crtcs = possible_crtcs;

I think the logic should be reversed here, the CRTC shouldn't care
(much) about the layers at all.

We should modify sun4i_crtc_init to get the argument it needs (primary
and cursor planes for example) through its parameters, and have the
caller (which iirc is sun4i_drv) call it with the right parameters
depending on whether you're using DE or DE2.

If we're doing that, I don't think we even need the pointer to the
array of layers in struct sun4i_crtc, which will make it easier to
deal with.

Maxime

-- 
Maxime Ripard, Free Electrons
Embedded Linux and Kernel engineering
http://free-electrons.com

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 801 bytes --]