Hi Pekka,
On Tue, 30 Jan 2018 at 09:58, Pekka Paalanen <ppaala...@gmail.com> wrote:
> On Wed, 20 Dec 2017 12:26:58 +0000
> Daniel Stone <dani...@collabora.com> wrote:
> > + for (i = 0; i < plane->count_formats; i++) {
> > + if (plane->formats[i].format == output->gbm_format)
>
> Here it is output->gbm_format...
>
> > + break;
> > + }
> > +
> > + if (i == plane->count_formats) {
> > + /* XXX: better error message */
> > + weston_log("can't use format for output\n");
>
> XXX :-)
>
> Maybe I could add name strings to the pixel format list and a helper
> that would be safe for arbitrary and even invalid format codes for
> turning a format into pretty string.
>
> Should this fall back to the fallback format as well?
We shouldn't fall back to the fallback format. For XRGB, the format is
ARGB, which will enable alpha blending. Maybe we could try doing so in
a separate patch, if we know we always set solid alpha channels, but
...
> > - output->gbm_surface = gbm_surface_create(b->gbm,
> > - output->base.current_mode->width,
> > -
> > output->base.current_mode->height,
> > - format[0],
> > - GBM_BO_USE_SCANOUT |
> > - GBM_BO_USE_RENDERING);
> > if (!output->gbm_surface) {
> > weston_log("failed to create gbm surface\n");
> > return -1;
>
> This is not new in this patch, but the GBM surface is created with
> gbm_format, but the EGLConfig may be chosen by gbm_format or its
> fallback format. If it picks with the fallback format, do we have a
> problem?
It's not a problem that I know of, since the formats are chosen to be
completely interchangeable. Mesa at least will happily render to the
other format, and I believe this is also true of other
implementations.
Cheers,
Daniel
_______________________________________________
wayland-devel mailing list
wayland-devel@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/wayland-devel
