Re: [PATCH v2 2/2] drm/msm/mdp4: Support NV12MT format in mdp4

[Daniel Vetter <daniel@ffwll.ch>]

On Fri, Mar 13, 2015 at 03:12:10PM -0400, Stephane Viau wrote:
> From: Beeresh Gopal <gbeeresh@codeaurora.org>
> 
> Using fb modifier flag, support NV12MT format in MDP4.
> 
> v2:
> - rework the modifier's description [Daniel Vetter's comment]
> - drop .set_mode_config() callback [Rob Clark's comment]
> 
> Signed-off-by: Beeresh Gopal <gbeeresh@codeaurora.org>
> Signed-off-by: Stephane Viau <sviau@codeaurora.org>
> ---
>  drivers/gpu/drm/msm/mdp/mdp4/mdp4_kms.c   |  2 ++
>  drivers/gpu/drm/msm/mdp/mdp4/mdp4_plane.c | 22 ++++++++++++++++++++++
>  include/uapi/drm/drm_fourcc.h             |  5 +++++
>  3 files changed, 29 insertions(+)
> 
> diff --git a/drivers/gpu/drm/msm/mdp/mdp4/mdp4_kms.c b/drivers/gpu/drm/msm/mdp/mdp4/mdp4_kms.c
> index d81e19d..6387881 100644
> --- a/drivers/gpu/drm/msm/mdp/mdp4/mdp4_kms.c
> +++ b/drivers/gpu/drm/msm/mdp/mdp4/mdp4_kms.c
> @@ -119,6 +119,8 @@ static int mdp4_hw_init(struct msm_kms *kms)
>  	if (mdp4_kms->rev > 1)
>  		mdp4_write(mdp4_kms, REG_MDP4_RESET_STATUS, 1);
>  
> +	dev->mode_config.allow_fb_modifiers = true;
> +
>  out:
>  	pm_runtime_put_sync(dev->dev);
>  
> diff --git a/drivers/gpu/drm/msm/mdp/mdp4/mdp4_plane.c b/drivers/gpu/drm/msm/mdp/mdp4/mdp4_plane.c
> index cde2500..2c2d6a5 100644
> --- a/drivers/gpu/drm/msm/mdp/mdp4/mdp4_plane.c
> +++ b/drivers/gpu/drm/msm/mdp/mdp4/mdp4_plane.c
> @@ -33,6 +33,21 @@ struct mdp4_plane {
>  };
>  #define to_mdp4_plane(x) container_of(x, struct mdp4_plane, base)
>  
> +/* MDP format helper functions */
> +static inline
> +enum mdp4_frame_format mdp4_get_frame_format(struct drm_framebuffer *fb)
> +{
> +	bool is_tile = false;
> +
> +	if (fb->modifier[1] == DRM_FORMAT_MOD_SAMSUNG_64_32_TILE)
> +		is_tile = true;
> +
> +	if (fb->pixel_format == DRM_FORMAT_NV12 && is_tile)
> +		return FRAME_TILE_YCBCR_420;
> +
> +	return FRAME_LINEAR;
> +}
> +
>  static void mdp4_plane_set_scanout(struct drm_plane *plane,
>  		struct drm_framebuffer *fb);
>  static int mdp4_plane_mode_set(struct drm_plane *plane,
> @@ -203,6 +218,7 @@ static int mdp4_plane_mode_set(struct drm_plane *plane,
>  	uint32_t op_mode = 0;
>  	uint32_t phasex_step = MDP4_VG_PHASE_STEP_DEFAULT;
>  	uint32_t phasey_step = MDP4_VG_PHASE_STEP_DEFAULT;
> +	enum mdp4_frame_format frame_type = mdp4_get_frame_format(fb);
>  
>  	if (!(crtc && fb)) {
>  		DBG("%s: disabled!", mdp4_plane->name);
> @@ -302,6 +318,7 @@ static int mdp4_plane_mode_set(struct drm_plane *plane,
>  			MDP4_PIPE_SRC_FORMAT_UNPACK_COUNT(format->unpack_count - 1) |
>  			MDP4_PIPE_SRC_FORMAT_FETCH_PLANES(format->fetch_type) |
>  			MDP4_PIPE_SRC_FORMAT_CHROMA_SAMP(format->chroma_sample) |
> +			MDP4_PIPE_SRC_FORMAT_FRAME_FORMAT(frame_type) |
>  			COND(format->unpack_tight, MDP4_PIPE_SRC_FORMAT_UNPACK_TIGHT));
>  
>  	mdp4_write(mdp4_kms, REG_MDP4_PIPE_SRC_UNPACK(pipe),
> @@ -322,6 +339,11 @@ static int mdp4_plane_mode_set(struct drm_plane *plane,
>  	mdp4_write(mdp4_kms, REG_MDP4_PIPE_PHASEX_STEP(pipe), phasex_step);
>  	mdp4_write(mdp4_kms, REG_MDP4_PIPE_PHASEY_STEP(pipe), phasey_step);
>  
> +	if (frame_type != FRAME_LINEAR)
> +		mdp4_write(mdp4_kms, REG_MDP4_PIPE_SSTILE_FRAME_SIZE(pipe),
> +				MDP4_PIPE_SSTILE_FRAME_SIZE_WIDTH(src_w) |
> +				MDP4_PIPE_SSTILE_FRAME_SIZE_HEIGHT(src_h));
> +
>  	return 0;
>  }
>  
> diff --git a/include/uapi/drm/drm_fourcc.h b/include/uapi/drm/drm_fourcc.h
> index 188e61f..2ff79cb 100644
> --- a/include/uapi/drm/drm_fourcc.h
> +++ b/include/uapi/drm/drm_fourcc.h
> @@ -161,4 +161,9 @@
>   * authoritative source for all of these.
>   */
>  
> +/* Samsung framebuffer modifiers */
> +
> +/* Tiled: 64x32 pixel macroblocks */

Since this seems shared by a lot of vendors (I still don't believe Samsung
invented this really ...) can you please describe this thing a bit in more
detail? Somewhat important how macroblocks are laid out and pixels within.
Also with a planar format like NV12 "pixel" is a bit unclear, e.g. what
happens if you throw 10bit plane formats at this? So maybe also add a note
that for now this is only used together with NV12T.

Then there's the question of validating the input - stride should probably
be a full multiple of the macroblock size. Since this is a shared format
imo this kind of checking should be done in drm core.
-Daniel


> +#define DRM_FORMAT_MOD_SAMSUNG_64_32_TILE fourcc_mod_code(SAMSUNG, 1)
> +
>  #endif /* DRM_FOURCC_H */
> -- 
> Qualcomm Innovation Center, Inc.
> 
> The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum, a Linux Foundation Collaborative Project
> 

-- 
Daniel Vetter
Software Engineer, Intel Corporation
+41 (0) 79 365 57 48 - http://blog.ffwll.ch