[PATCH weston 2/2] compositor-drm: Support linux_dmabuf output for sprite planes without gbm
Pekka Paalanen
ppaalanen at gmail.com
Wed Jun 15 08:22:22 UTC 2016
On Fri, 27 May 2016 18:52:28 +0900
Tomohito Esaki <etom at igel.co.jp> wrote:
> Multiplanar formats are supported by using drmModeAddFB2 and bypassing
> gbm. If drmModeAddFB2 isn't available, the existing gbm bo import path
> is used and multiplanar formats are unsupported.
>
> Signed-off-by: Tomohito Esaki <etom at igel.co.jp>
> ---
> src/compositor-drm.c | 47 +++++++++++++++++++----------------------------
> 1 file changed, 19 insertions(+), 28 deletions(-)
Hi,
I'm not quite sure the testing for the AddFB2 working is totally
correct, as the previous patch is adding a new call to AddFB2 without
dealing with no_addfb2, but I do not think that is significant. I'm
fine leaving it the way you have it.
> diff --git a/src/compositor-drm.c b/src/compositor-drm.c
> index cd2fb71..d6a4a5a 100644
> --- a/src/compositor-drm.c
> +++ b/src/compositor-drm.c
> @@ -960,11 +960,9 @@ page_flip_handler(int fd, unsigned int frame,
>
> static uint32_t
> drm_output_check_sprite_format(struct drm_sprite *s,
> - struct weston_view *ev, struct gbm_bo *bo)
> + struct weston_view *ev, uint32_t format)
> {
> - uint32_t i, format;
> -
> - format = gbm_bo_get_format(bo);
> + uint32_t i;
>
> if (format == GBM_FORMAT_ARGB8888) {
> pixman_region32_t r;
> @@ -1005,15 +1003,12 @@ drm_output_prepare_overlay_view(struct drm_output *output,
> struct drm_sprite *s;
> struct linux_dmabuf_buffer *dmabuf;
> int found = 0;
> - struct gbm_bo *bo;
> + struct gbm_bo *bo = NULL;
> pixman_region32_t dest_rect, src_rect;
> pixman_box32_t *box, tbox;
> uint32_t format;
> wl_fixed_t sx1, sy1, sx2, sy2;
>
> - if (b->gbm == NULL)
> - return NULL;
> -
> if (viewport->buffer.transform != output->base.transform)
> return NULL;
>
> @@ -1053,15 +1048,9 @@ drm_output_prepare_overlay_view(struct drm_output *output,
> if (!found)
> return NULL;
>
> - if ((dmabuf = linux_dmabuf_buffer_get(buffer_resource))) {
> + if ((dmabuf = linux_dmabuf_buffer_get(buffer_resource)) &&
> + b->no_addfb2 && b->gbm) {
> #ifdef HAVE_GBM_FD_IMPORT
> - /* XXX: TODO:
> - *
> - * Use AddFB2 directly, do not go via GBM.
> - * Add support for multiplanar formats.
> - * Both require refactoring in the DRM-backend to
> - * support a mix of gbm_bos and drmfbs.
> - */
> struct gbm_import_fd_data gbm_dmabuf = {
> .fd = dmabuf->attributes.fd[0],
> .width = dmabuf->attributes.width,
> @@ -1078,22 +1067,24 @@ drm_output_prepare_overlay_view(struct drm_output *output,
> #else
> return NULL;
> #endif
> - } else {
> + } else if (b->gbm) {
> bo = gbm_bo_import(b->gbm, GBM_BO_IMPORT_WL_BUFFER,
> buffer_resource, GBM_BO_USE_SCANOUT);
> }
> - if (!bo)
> - return NULL;
> -
> - format = drm_output_check_sprite_format(s, ev, bo);
> - if (format == 0) {
> - gbm_bo_destroy(bo);
> - return NULL;
> - }
>
> - s->next = drm_fb_get_from_bo(bo, b, format);
> - if (!s->next) {
> - gbm_bo_destroy(bo);
> + if (bo) {
> + if (((format = drm_output_check_sprite_format(
> + s, ev, gbm_bo_get_format(bo))) == 0) ||
> + (!(s->next = drm_fb_get_from_bo(bo, b, format)))) {
> + gbm_bo_destroy(bo);
> + return NULL;
> + }
> + } else if (dmabuf) {
> + if (((format = drm_output_check_sprite_format(
> + s, ev, dmabuf->attributes.format)) == 0) ||
> + (!(s->next = drm_fb_create_dmabuf(dmabuf, b, format))))
> + return NULL;
The same comments here as for the previous patch for splitting
assignments into separate statements for readability.
> + } else {
> return NULL;
> }
>
Otherwise this patch looks good. Would you like to fix and re-post both?
Some ideas for further development if you are interested:
- Check that dmabuf_attributes::flags does not contain anything we
don't support in the direct scanout paths.
- Handle dmabuf_attributes::modifier[] when possible, or reject if
non-zero.
Thanks,
pq
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 811 bytes
Desc: OpenPGP digital signature
URL: <https://lists.freedesktop.org/archives/wayland-devel/attachments/20160615/e4585467/attachment-0001.sig>
More information about the wayland-devel
mailing list