[bug report] drm: zte: add overlay plane support

[bug report] drm: zte: add overlay plane support

[Dan Carpenter]

Hello Shawn Guo,

The patch 4e986d3705df: "drm: zte: add overlay plane support" from
Nov 16, 2016, leads to the following static checker warning:

	drivers/gpu/drm/zte/zx_plane.c:170 zx_vl_rsz_setup()
	warn: always true condition '(fmt >= 0) => (0-u32max >= 0)'

drivers/gpu/drm/zte/zx_plane.c
   156  static void zx_vl_rsz_setup(struct zx_plane *zplane, uint32_t format,
   157                              u32 src_w, u32 src_h, u32 dst_w, u32 dst_h)
   158  {
   159          void __iomem *rsz = zplane->rsz;
   160          u32 src_chroma_w = src_w;
   161          u32 src_chroma_h = src_h;
   162          u32 fmt;
                ^^^^^^^
   163  
   164          /* Set up source and destination resolution */
   165          zx_writel(rsz + RSZ_SRC_CFG, RSZ_VER(src_h - 1) | RSZ_HOR(src_w - 1));
   166          zx_writel(rsz + RSZ_DEST_CFG, RSZ_VER(dst_h - 1) | RSZ_HOR(dst_w - 1));
   167  
   168          /* Configure data format for VL RSZ */
   169          fmt = zx_vl_rsz_get_fmt(format);
                ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
If fmt is -EINVAL then don't we just want to return anyway?

   170          if (fmt >= 0)
   171                  zx_writel_mask(rsz + RSZ_VL_CTRL_CFG, RSZ_VL_FMT_MASK, fmt);
   172  
   173          /* Calculate Chroma height and width */
   174          if (fmt == RSZ_VL_FMT_YCBCR420) {
   175                  src_chroma_w = src_w >> 1;
   176                  src_chroma_h = src_h >> 1;
   177          } else if (fmt == RSZ_VL_FMT_YCBCR422) {
   178                  src_chroma_w = src_w >> 1;
   179          }
   180  
   181          /* Set up Luma and Chroma step registers */
   182          zx_writel(rsz + RSZ_VL_LUMA_HOR, rsz_step_value(src_w, dst_w));
   183          zx_writel(rsz + RSZ_VL_LUMA_VER, rsz_step_value(src_h, dst_h));
   184          zx_writel(rsz + RSZ_VL_CHROMA_HOR, rsz_step_value(src_chroma_w, dst_w));
   185          zx_writel(rsz + RSZ_VL_CHROMA_VER, rsz_step_value(src_chroma_h, dst_h));
   186  
   187          zx_vl_rsz_set_update(zplane);
   188  }

regards,
dan carpenter
_______________________________________________
dri-devel mailing list
dri-devel@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/dri-devel

Re: [bug report] drm: zte: add overlay plane support

[Dan Carpenter]

Oh.  Also this one:

    drivers/gpu/drm/zte/zx_plane.c:253 zx_vl_plane_atomic_update()
    warn: always true condition '(fmt >= 0) => (0-u32max >= 0)'

regards,
dan carpenter



_______________________________________________
dri-devel mailing list
dri-devel@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/dri-devel

Re: [bug report] drm: zte: add overlay plane support

[Shawn Guo]

Hi Dan,

Thanks for the report.

On Wed, Feb 08, 2017 at 09:39:51AM +0300, Dan Carpenter wrote:
> Hello Shawn Guo,
> 
> The patch 4e986d3705df: "drm: zte: add overlay plane support" from
> Nov 16, 2016, leads to the following static checker warning:
> 
> 	drivers/gpu/drm/zte/zx_plane.c:170 zx_vl_rsz_setup()
> 	warn: always true condition '(fmt >= 0) => (0-u32max >= 0)'

I will send a follow-up patch to fix it by changing 'fmt' to type 'int'.

> 
> drivers/gpu/drm/zte/zx_plane.c
>    156  static void zx_vl_rsz_setup(struct zx_plane *zplane, uint32_t format,
>    157                              u32 src_w, u32 src_h, u32 dst_w, u32 dst_h)
>    158  {
>    159          void __iomem *rsz = zplane->rsz;
>    160          u32 src_chroma_w = src_w;
>    161          u32 src_chroma_h = src_h;
>    162          u32 fmt;
>                 ^^^^^^^
>    163  
>    164          /* Set up source and destination resolution */
>    165          zx_writel(rsz + RSZ_SRC_CFG, RSZ_VER(src_h - 1) | RSZ_HOR(src_w - 1));
>    166          zx_writel(rsz + RSZ_DEST_CFG, RSZ_VER(dst_h - 1) | RSZ_HOR(dst_w - 1));
>    167  
>    168          /* Configure data format for VL RSZ */
>    169          fmt = zx_vl_rsz_get_fmt(format);
>                 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
> If fmt is -EINVAL then don't we just want to return anyway?

The zx_vl_rsz_setup() is being called by atomic update function, which
can not fail anyway.  So we give it a best effort by going through the
hardware update procedure but skip invalid format.

Shawn

> 
>    170          if (fmt >= 0)
>    171                  zx_writel_mask(rsz + RSZ_VL_CTRL_CFG, RSZ_VL_FMT_MASK, fmt);
>    172  
>    173          /* Calculate Chroma height and width */
>    174          if (fmt == RSZ_VL_FMT_YCBCR420) {
>    175                  src_chroma_w = src_w >> 1;
>    176                  src_chroma_h = src_h >> 1;
>    177          } else if (fmt == RSZ_VL_FMT_YCBCR422) {
>    178                  src_chroma_w = src_w >> 1;
>    179          }
>    180  
>    181          /* Set up Luma and Chroma step registers */
>    182          zx_writel(rsz + RSZ_VL_LUMA_HOR, rsz_step_value(src_w, dst_w));
>    183          zx_writel(rsz + RSZ_VL_LUMA_VER, rsz_step_value(src_h, dst_h));
>    184          zx_writel(rsz + RSZ_VL_CHROMA_HOR, rsz_step_value(src_chroma_w, dst_w));
>    185          zx_writel(rsz + RSZ_VL_CHROMA_VER, rsz_step_value(src_chroma_h, dst_h));
>    186  
>    187          zx_vl_rsz_set_update(zplane);
>    188  }
_______________________________________________
dri-devel mailing list
dri-devel@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/dri-devel