[Mesa-dev] [PATCH kmscube] Fix GBM image formats

[Emil Velikov]

On 4 April 2017 at 13:49, Thomas Hellstrom <thellstrom at vmware.com> wrote:
> When the GBM BOs are created, they are created using a bo with a correct
> memory requirement but with an incorrect channel layout. This makes
> drivers that are picky about matching formats (svga/vmwgfx) complain.
>
> Use the correct GBM formats and require GBM >= 13.0 to make sure they
> are available.
>
> Signed-off-by: Thomas Hellstrom <thellstrom at vmware.com>
> ---
>  configure.ac |  2 +-
>  cube-tex.c   | 10 +++++-----
>  2 files changed, 6 insertions(+), 6 deletions(-)
>
> diff --git a/configure.ac b/configure.ac
> index 242164f..aa0ef36 100644
> --- a/configure.ac
> +++ b/configure.ac
> @@ -36,7 +36,7 @@ m4_ifdef([AM_SILENT_RULES], [AM_SILENT_RULES([yes])])
>
>  # Obtain compiler/linker options for depedencies
>  PKG_CHECK_MODULES(DRM, libdrm)
> -PKG_CHECK_MODULES(GBM, gbm)
> +PKG_CHECK_MODULES(GBM, gbm >= 13.0)
>  PKG_CHECK_MODULES(EGL, egl)
>  PKG_CHECK_MODULES(GLES2, glesv2)
>
> diff --git a/cube-tex.c b/cube-tex.c
> index b11023f..d97ab77 100644
> --- a/cube-tex.c
> +++ b/cube-tex.c
> @@ -223,7 +223,7 @@ static int get_fd_rgba(uint32_t *pstride)
>         int fd;
>
>         /* NOTE: do not actually use GBM_BO_USE_WRITE since that gets us a dumb buffer: */
> -       bo = gbm_bo_create(gl.gbm->dev, texw, texh, GBM_FORMAT_ARGB8888, GBM_BO_USE_LINEAR);
> +       bo = gbm_bo_create(gl.gbm->dev, texw, texh, GBM_FORMAT_ABGR8888, GBM_BO_USE_LINEAR);
>
>         map = gbm_bo_map(bo, 0, 0, texw, texh, GBM_BO_TRANSFER_WRITE, &stride, &map_data);
>
> @@ -254,9 +254,9 @@ static int get_fd_y(uint32_t *pstride)
>
>         /* NOTE: do not actually use GBM_BO_USE_WRITE since that gets us a dumb buffer: */
>         /* hmm, no R8/R8G8 gbm formats?? */
This ...

> -       bo = gbm_bo_create(gl.gbm->dev, texw/4, texh, GBM_FORMAT_ARGB8888, GBM_BO_USE_LINEAR);
> +       bo = gbm_bo_create(gl.gbm->dev, texw, texh, GBM_FORMAT_R8, GBM_BO_USE_LINEAR);
>
> -       map = gbm_bo_map(bo, 0, 0, texw/4, texh, GBM_BO_TRANSFER_WRITE, &stride, &map_data);
> +       map = gbm_bo_map(bo, 0, 0, texw, texh, GBM_BO_TRANSFER_WRITE, &stride, &map_data);
>
>         for (uint32_t i = 0; i < texh; i++) {
>                 memcpy(&map[stride * i], &src[texw * i], texw);
> @@ -285,9 +285,9 @@ static int get_fd_uv(uint32_t *pstride)
>
>         /* NOTE: do not actually use GBM_BO_USE_WRITE since that gets us a dumb buffer: */
>         /* hmm, no R8/R8G8 gbm formats?? */
... and this line should be removed, right ?

With that
Reviewed-by: Emil Velikov <emil.velikov at collabora.com>

-Emil