Re: [PATCH 5/5] drm/stm: ltdc: add support of ycbcr pixel formats

From: Nathan Chancellor
Date: Wed Feb 02 2022 - 11:55:18 EST


Hi Yannick,

On Wed, Dec 15, 2021 at 10:48:43PM +0100, Yannick Fertre wrote:
> This patch adds the following YCbCr input pixel formats on the latest
> LTDC hardware version:
>
> 1 plane (co-planar) : YUYV, YVYU, UYVY, VYUY
> 2 planes (semi-planar): NV12, NV21
> 3 planes (full-planar): YU12=I420=DRM YUV420, YV12=DRM YVU420
>
> Signed-off-by: Yannick Fertre <yannick.fertre@xxxxxxxxxxx>

<snip>

> +static inline void ltdc_set_ycbcr_config(struct drm_plane *plane, u32 drm_pix_fmt)
> +{
> + struct ltdc_device *ldev = plane_to_ltdc(plane);
> + struct drm_plane_state *state = plane->state;
> + u32 lofs = plane->index * LAY_OFS;
> + u32 val;
> +
> + switch (drm_pix_fmt) {
> + case DRM_FORMAT_YUYV:
> + val = (YCM_I << 4) | LxPCR_YF | LxPCR_CBF;
> + break;
> + case DRM_FORMAT_YVYU:
> + val = (YCM_I << 4) | LxPCR_YF;
> + break;
> + case DRM_FORMAT_UYVY:
> + val = (YCM_I << 4) | LxPCR_CBF;
> + break;
> + case DRM_FORMAT_VYUY:
> + val = (YCM_I << 4);
> + break;
> + case DRM_FORMAT_NV12:
> + val = (YCM_SP << 4) | LxPCR_CBF;
> + break;
> + case DRM_FORMAT_NV21:
> + val = (YCM_SP << 4);
> + break;
> + case DRM_FORMAT_YUV420:
> + case DRM_FORMAT_YVU420:
> + val = (YCM_FP << 4);
> + break;
> + default:
> + /* RGB or not a YCbCr supported format */
> + break;
> + }
> +
> + /* Enable limited range */
> + if (state->color_range == DRM_COLOR_YCBCR_LIMITED_RANGE)
> + val |= LxPCR_YREN;
> +
> + /* enable ycbcr conversion */
> + val |= LxPCR_YCEN;
> +
> + regmap_write(ldev->regmap, LTDC_L1PCR + lofs, val);
> +}

This patch as commit 484e72d3146b ("drm/stm: ltdc: add support of ycbcr
pixel formats") in -next introduced the following clang warning:

drivers/gpu/drm/stm/ltdc.c:625:2: warning: variable 'val' is used uninitialized whenever switch default is taken [-Wsometimes-uninitialized]
default:
^~~~~~~
drivers/gpu/drm/stm/ltdc.c:635:2: note: uninitialized use occurs here
val |= LxPCR_YCEN;
^~~
drivers/gpu/drm/stm/ltdc.c:600:9: note: initialize the variable 'val' to silence this warning
u32 val;
^
= 0
1 warning generated.

Would it be okay to just return in the default case (maybe with a
message about an unsupported format?) or should there be another fix?

Cheers,