[PATCH 5/5] drm/imx: add exclusive fence to plane state

[Daniel Vetter]

On Thu, Aug 11, 2016 at 11:18 AM, Lucas Stach <l.stach at pengutronix.de> wrote:
> This allows the atomic helper to wait on them, instead of open-coding
> the same in the imx-drm driver.
>
> Signed-off-by: Lucas Stach <l.stach at pengutronix.de>
> ---
>  drivers/gpu/drm/imx/imx-drm-core.c | 63 +++++++++++++++++---------------------
>  1 file changed, 28 insertions(+), 35 deletions(-)
>
> diff --git a/drivers/gpu/drm/imx/imx-drm-core.c b/drivers/gpu/drm/imx/imx-drm-core.c
> index 66b3556f7b79..2d3f32f1b13b 100644
> --- a/drivers/gpu/drm/imx/imx-drm-core.c
> +++ b/drivers/gpu/drm/imx/imx-drm-core.c
> @@ -152,50 +152,43 @@ static void imx_drm_output_poll_changed(struct drm_device *drm)
>         drm_fbdev_cma_hotplug_event(imxdrm->fbhelper);
>  }
>
> +static int imx_drm_atomic_commit(struct drm_device *dev,
> +                                struct drm_atomic_state *state,
> +                                bool nonblock)
> +{
> +       struct drm_plane_state *plane_state;
> +       struct drm_plane *plane;
> +       struct dma_buf *dma_buf;
> +       int i;
> +
> +       /*
> +        * If the plane fb has an dma-buf attached, fish out the exclusive
> +        * fence for the atomic helper to wait on.
> +        */
> +       for_each_plane_in_state(state, plane, plane_state, i) {
> +               if ((plane->state->fb != plane_state->fb) && plane_state->fb) {
> +                       dma_buf = drm_fb_cma_get_gem_obj(plane_state->fb,
> +                                                        0)->base.dma_buf;
> +                       if (!dma_buf)
> +                               continue;
> +                       plane_state->fence =
> +                               reservation_object_get_excl_rcu(dma_buf->resv);
> +               }
> +       }

Just an fyi, but the idea was that you could do this in youre
prepare_plane hook, assuming that only when userspace flips is it
interested in syncing (for backwards compat reasons). But this gets
the job done too.

The upshot of doing this consistently in a prepare_plane hook is that
it would then allow us to have a shared implementation for all cma
based kms drivers. If you feel bored, I'd be happy to take a look at a
refactor patch which would extract that helper (and wire it up for
imx).

Just a drive-by comment since I stumbled over this as a conflict
between -next and -fixes when rebuilding intel trees.
-Daniel

> +
> +       return drm_atomic_helper_commit(dev, state, nonblock);
> +}
> +
>  static const struct drm_mode_config_funcs imx_drm_mode_config_funcs = {
>         .fb_create = drm_fb_cma_create,
>         .output_poll_changed = imx_drm_output_poll_changed,
>         .atomic_check = drm_atomic_helper_check,
> -       .atomic_commit = drm_atomic_helper_commit,
> +       .atomic_commit = imx_drm_atomic_commit,
>  };
>
>  static void imx_drm_atomic_commit_tail(struct drm_atomic_state *state)
>  {
>         struct drm_device *dev = state->dev;
> -       struct drm_crtc *crtc;
> -       struct drm_crtc_state *crtc_state;
> -       struct drm_plane_state *plane_state;
> -       struct drm_gem_cma_object *cma_obj;
> -       struct fence *excl;
> -       unsigned shared_count;
> -       struct fence **shared;
> -       unsigned int i, j;
> -       int ret;
> -
> -       /* Wait for fences. */
> -       for_each_crtc_in_state(state, crtc, crtc_state, i) {
> -               plane_state = crtc->primary->state;
> -               if (plane_state->fb) {
> -                       cma_obj = drm_fb_cma_get_gem_obj(plane_state->fb, 0);
> -                       if (cma_obj->base.dma_buf) {
> -                               ret = reservation_object_get_fences_rcu(
> -                                       cma_obj->base.dma_buf->resv, &excl,
> -                                       &shared_count, &shared);
> -                               if (unlikely(ret))
> -                                       DRM_ERROR("failed to get fences "
> -                                                 "for buffer\n");
> -
> -                               if (excl) {
> -                                       fence_wait(excl, false);
> -                                       fence_put(excl);
> -                               }
> -                               for (j = 0; j < shared_count; i++) {
> -                                       fence_wait(shared[j], false);
> -                                       fence_put(shared[j]);
> -                               }
> -                       }
> -               }
> -       }
>
>         drm_atomic_helper_commit_modeset_disables(dev, state);
>
> --
> 2.8.1
>
> _______________________________________________
> dri-devel mailing list
> dri-devel at lists.freedesktop.org
> https://lists.freedesktop.org/mailman/listinfo/dri-devel



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