[Mesa-dev] [PATCH 01/10] gallium/radeon: flush the context after in-place texture realloc before export

[Nicolai Hähnle]

On 08.07.2017 02:42, Marek Olšák wrote:
> From: Marek Olšák <marek.olsak at amd.com>
> 
> ---
>   src/gallium/drivers/radeon/r600_texture.c | 1 +
>   1 file changed, 1 insertion(+)
> 
> diff --git a/src/gallium/drivers/radeon/r600_texture.c b/src/gallium/drivers/radeon/r600_texture.c
> index 2deb56a..f5f7d10 100644
> --- a/src/gallium/drivers/radeon/r600_texture.c
> +++ b/src/gallium/drivers/radeon/r600_texture.c
> @@ -563,20 +563,21 @@ static boolean r600_texture_get_handle(struct pipe_screen* screen,
>   		 * interop in the future.
>   		 */
>   		if (resource->nr_samples > 1 || rtex->is_depth)
>   			return false;
>   
>   		/* Move a suballocated texture into a non-suballocated allocation. */
>   		if (rscreen->ws->buffer_is_suballocated(res->buf)) {
>   			assert(!res->b.is_shared);
>   			r600_reallocate_texture_inplace(rctx, rtex,
>   							PIPE_BIND_SHARED, false);
> +			rctx->b.flush(&rctx->b, NULL, 0);

Hmm. The caller anyway is responsible for flushing the context before 
the texture is used externally. So this is redundant for callers that 
flush after calling get_handle, but it fixes callers that flush before 
calling get_handle. Seems fine, maybe you could add that to the 
explanation. Either way,

Reviewed-by: Nicolai Hähnle <nicolai.haehnle at amd.com>


>   			assert(res->b.b.bind & PIPE_BIND_SHARED);
>   			assert(res->flags & RADEON_FLAG_NO_SUBALLOC);
>   		}
>   
>   		/* Since shader image stores don't support DCC on VI,
>   		 * disable it for external clients that want write
>   		 * access.
>   		 */
>   		if (usage & PIPE_HANDLE_USAGE_WRITE && rtex->dcc_offset) {
>   			if (r600_texture_disable_dcc(rctx, rtex))
> 


-- 
Lerne, wie die Welt wirklich ist,
Aber vergiss niemals, wie sie sein sollte.