On Fri, Feb 17, 2012 at 11:57:58AM +0800, zhiwen...@linux.intel.com wrote:
> From: Alex Wu <zhiwen...@intel.com>
>
> If the scanout client was interrupted (e.g. by Ctrl-C),
> compositor-drm.c::page_flip_handler()
> was not guaranteed to be invoked for the last scanout
> drmModePageFlip(), so that pending_fs_surf_fb_id
> has no chance to reset to 0, and the output will still scanout the
> old framebuffer with pending_fs_surf_fb_id.
Ah yup, I see what could go wrong here. I think a more logic fix is to
just change:
if (output->pending_fs_surf_fb_id != 0) {
in drm_output_repaint() to
if (output->pending_scanout_surface) {
since that's really what indicates whether the other fields are valid
or not. Does that work for your case?
Kristian
> ---
> src/compositor-drm.c | 5 ++++-
> 1 files changed, 4 insertions(+), 1 deletions(-)
>
> diff --git a/src/compositor-drm.c b/src/compositor-drm.c
> index de6feb4..d5a8d28 100644
> --- a/src/compositor-drm.c
> +++ b/src/compositor-drm.c
> @@ -159,7 +159,10 @@ drm_output_repaint(struct weston_output *output_base)
> if (glCheckFramebufferStatus(GL_FRAMEBUFFER) != GL_FRAMEBUFFER_COMPLETE)
> return;
>
> - drm_output_prepare_scanout_surface(output);
> + if (drm_output_prepare_scanout_surface(output)) {
> + if (output->pending_fs_surf_fb_id != 0)
> + output->pending_fs_surf_fb_id = 0;
> + }
>
> wl_list_for_each_reverse(surface, &compositor->base.surface_list, link)
> weston_surface_draw(surface, &output->base);
> --
> 1.7.5.4
>
> _______________________________________________
> wayland-devel mailing list
> wayland-devel@lists.freedesktop.org
> http://lists.freedesktop.org/mailman/listinfo/wayland-devel
_______________________________________________
wayland-devel mailing list
wayland-devel@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/wayland-devel
