[Mesa-dev] [PATCH 4/4] st/mesa: use asynchronous flushes for glFlush
Nicolai Hähnle
nhaehnle at gmail.com
Mon Nov 20 15:23:36 UTC 2017
On 09.11.2017 14:45, Nicolai Hähnle wrote:
> From: Nicolai Hähnle <nicolai.haehnle at amd.com>
>
> Having the gallium driver thread flush in the background should be
> sufficient for glFlush semantics.
>
> Various end-of-frame flushes (from st_context_flush and st/dri) still
> use a synchronous flush. We should eventually be able to transition
> those to asynchronous flushes as well by passing fences explicitly
> via the X protocol.
Thanks for the reviews, folks.
This last patch causes a non-deterministic regression in
dEQP-EGL.functional.image.render_multiple_contexts.gles2_*_read_pixels
I'm pretty sure that's a test bug, and I filed an issue on Khronos'
internal bug tracker
(https://gitlab.khronos.org/Tracker/vk-gl-cts/issues/857). Still, I'll
hold this particular patch for now, and only push the first three after
some more double-checking.
Cheers,
Nicolai
> ---
> src/mesa/state_tracker/st_cb_flush.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/src/mesa/state_tracker/st_cb_flush.c b/src/mesa/state_tracker/st_cb_flush.c
> index 14bfd5a4684..5f4e2ac3cc1 100644
> --- a/src/mesa/state_tracker/st_cb_flush.c
> +++ b/src/mesa/state_tracker/st_cb_flush.c
> @@ -81,21 +81,21 @@ void st_finish( struct st_context *st )
> */
> static void st_glFlush(struct gl_context *ctx)
> {
> struct st_context *st = st_context(ctx);
>
> /* Don't call st_finish() here. It is not the state tracker's
> * responsibilty to inject sleeps in the hope of avoiding buffer
> * synchronization issues. Calling finish() here will just hide
> * problems that need to be fixed elsewhere.
> */
> - st_flush(st, NULL, 0);
> + st_flush(st, NULL, PIPE_FLUSH_ASYNC);
>
> st_manager_flush_frontbuffer(st);
> }
>
>
> /**
> * Called via ctx->Driver.Finish()
> */
> static void st_glFinish(struct gl_context *ctx)
> {
>
--
Lerne, wie die Welt wirklich ist,
Aber vergiss niemals, wie sie sein sollte.
More information about the mesa-dev
mailing list