Some Wayland compositors, notably Exo, do not always release buffers
fast enough, and not in sync with their frame callbacks, to guarantee
that a free buffer is available the next time a client calls
`eglSwapBuffers()`.
This currently leads to a crash in `dri2_wl_swrast_get_backbuffer_data()`
with the swrast backend. To avoid this, simply block until the
compositor releases a buffer eventually.
While arguably compositors should release buffers they don't need any
more for the next frame, this can be quite complex depending on
the architecture - notably multi-process/IPC in case of Exo.
cc: mesa-stable
Signed-off-by: Robert Mader <robert.mader@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24091>
}
/* else choose any another free location */
- if (!dri2_surf->back) {
+ while (!dri2_surf->back) {
for (int i = 0; i < ARRAY_SIZE(dri2_surf->color_buffers); i++) {
if (!dri2_surf->color_buffers[i].locked) {
dri2_surf->back = &dri2_surf->color_buffers[i];
break;
}
}
- }
- if (!dri2_surf->back) {
- _eglError(EGL_BAD_ALLOC, "failed to find free buffer");
- return -1;
+ /* wait for the compositor to release a buffer */
+ if (!dri2_surf->back) {
+ if (wl_display_dispatch_queue(dri2_dpy->wl_dpy, dri2_surf->wl_queue) ==
+ -1) {
+ _eglError(EGL_BAD_ALLOC, "waiting for a free buffer failed");
+ return -1;
+ }
+ }
}
dri2_surf->back->locked = true;