[Mesa-stable] [PATCH] st/va: use provided sizes and coords for vlVaGetImage

Ilia Mirkin imirkin at alum.mit.edu
Wed Oct 10 03:32:30 UTC 2018


On Tue, Oct 9, 2018 at 4:32 PM <boyuan.zhang at amd.com> wrote:
>
> From: Boyuan Zhang <boyuan.zhang at amd.com>
>
> vlVaGetImage should respect the width, height, and coordinates x and y that
> passed in. Therefore, pipe_box should be created with the passed in values
> instead of surface width/height.
>
> v2: add input size check, return error when size out of bounds
> v3: fix the size check for vaimage
>
> Signed-off-by: Boyuan Zhang <boyuan.zhang at amd.com>
> Cc: "18.2" <mesa-stable at lists.freedesktop.org>
> Reviewed-by: Leo Liu <leo.liu at amd.com>
> ---
>  src/gallium/state_trackers/va/image.c | 26 +++++++++++++++++++++++---
>  1 file changed, 23 insertions(+), 3 deletions(-)
>
> diff --git a/src/gallium/state_trackers/va/image.c b/src/gallium/state_trackers/va/image.c
> index 3f892c9..71d2f49 100644
> --- a/src/gallium/state_trackers/va/image.c
> +++ b/src/gallium/state_trackers/va/image.c
> @@ -353,6 +353,23 @@ vlVaGetImage(VADriverContextP ctx, VASurfaceID surface, int x, int y,
>        return VA_STATUS_ERROR_INVALID_IMAGE;
>     }
>
> +   if (x < 0 || y < 0) {
> +      mtx_unlock(&drv->mutex);
> +      return VA_STATUS_ERROR_INVALID_PARAMETER;
> +   }
> +
> +   if (x + width > surf->templat.width ||
> +       y + height > surf->templat.height) {
> +      mtx_unlock(&drv->mutex);
> +      return VA_STATUS_ERROR_INVALID_PARAMETER;
> +   }
> +
> +   if (width > vaimage->width ||
> +       height > vaimage->height) {
> +      mtx_unlock(&drv->mutex);
> +      return VA_STATUS_ERROR_INVALID_PARAMETER;
> +   }
> +
>     img_buf = handle_table_get(drv->htab, vaimage->buf);
>     if (!img_buf) {
>        mtx_unlock(&drv->mutex);
> @@ -400,11 +417,14 @@ vlVaGetImage(VADriverContextP ctx, VASurfaceID surface, int x, int y,
>     }
>
>     for (i = 0; i < vaimage->num_planes; i++) {
> -      unsigned width, height;
> +      unsigned w = align(width, 2);
> +      unsigned h = align(height, 2);
>        if (!views[i]) continue;
> -      vlVaVideoSurfaceSize(surf, i, &width, &height);
> +      vl_video_buffer_adjust_size(&w, &h, i,
> +                                  surf->templat.chroma_format,
> +                                  surf->templat.interlaced);

You should adjust the x/y coords too (and probably round them down to
2, i.e. & ~1). Same transformation as for the w/h ought to work.

All this round-by-2 stuff seems sketchy though. It's clearly bogus for
the RGB formats, but it does seem prevalent elsewhere. I don't think
it's necessary to fix here.

Cheers,

  -ilia

>        for (j = 0; j < views[i]->texture->array_size; ++j) {
> -         struct pipe_box box = {0, 0, j, width, height, 1};
> +         struct pipe_box box = {x, y, j, w, h, 1};
>           struct pipe_transfer *transfer;
>           uint8_t *map;
>           map = drv->pipe->transfer_map(drv->pipe, views[i]->texture, 0,
> --
> 2.7.4
>


More information about the mesa-stable mailing list