Hi Jonas,

On 13 January 2017 at 09:05, Jonas Ådahl <[email protected]> wrote:
> @@ -1876,9 +1878,7 @@ dri2_initialize_wayland_swrast(_EGLDriver *drv, 
> _EGLDisplay *disp)
>     if (dri2_dpy->own_device)
>        wl_display_dispatch_pending(dri2_dpy->wl_dpy);
>
> -   dri2_dpy->wl_registry = wl_display_get_registry(dri2_dpy->wl_dpy);
> -   wl_proxy_set_queue((struct wl_proxy *) dri2_dpy->wl_registry,
> -                      dri2_dpy->wl_queue);
> +   dri2_dpy->wl_registry = wl_display_get_registry(dri2_dpy->wl_dpy_wrapper);
>     wl_registry_add_listener(dri2_dpy->wl_registry,
>                              &registry_listener_swrast, dri2_dpy);

We never actually create the proxy wrapper in dri2_initialize_wayland_swrast. :(

Emil, if you pull the @@ -1159,12 +1156,17 @@ hunk to create the
wl_display proxy wrapper and set the queue into the same position
here, so the two initialisation functions are identical from top to
the first roundtrip() call, plus the equivalent cleanup, this patch
is:
Reviewed-by: Daniel Stone <[email protected]>

I went through and the rest seemed fine. I would've preferred the
wrapped display to be called wl_dpy and the previous wl_dpy to be
renamed to wl_dpy_unwrapped to make it as difficult as possible to get
this wrong, but having this patch miss the release for that reason
alone seems pretty petty, even for me.

Cheers,
Daniel
_______________________________________________
mesa-dev mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/mesa-dev

Reply via email to