[Intel-gfx] [RFC] [PATCH] i965: better ClientWaitSync

[Daniel Vetter]

On Thu, Jun 07, 2012 at 09:12:42AM -0700, Ben Widawsky wrote:
> Use the new libdrm functionality to actually do timed waits on the sync
> object.
> 
> This patch is missing the configure.ac update to check for the correct
> libdrm supporting this function. As of now, libdrm has not yet received
> the version bump. That's mostly why this patch is "RFC"
> 
> Since intel_client_wait_sync function previously had no way of reporting
> back the errors/timeouts, and I do not have a good test case anyway,
> I've decided to not worry about this also.
> 
> Signed-off-by: Ben Widawsky <ben at bwidawsk.net>
> ---
>  src/mesa/drivers/dri/intel/intel_syncobj.c |    2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/src/mesa/drivers/dri/intel/intel_syncobj.c b/src/mesa/drivers/dri/intel/intel_syncobj.c
> index b303ea8..4dd8b7b 100644
> --- a/src/mesa/drivers/dri/intel/intel_syncobj.c
> +++ b/src/mesa/drivers/dri/intel/intel_syncobj.c
> @@ -93,7 +93,7 @@ static void intel_client_wait_sync(struct gl_context *ctx, struct gl_sync_object
>     struct intel_sync_object *sync = (struct intel_sync_object *)s;
>  
>     if (sync->bo) {
> -      drm_intel_bo_wait_rendering(sync->bo);
> +      drm_intel_gem_bo_wait(sync->bo, timeout);

If I've read the core mesa code correctly, StatusFlag = 1 means that the
fence has actually signalled. I guess you need to add a check for the
return value and do the below 3 lines only when the wait hasn't timed out
(or failed for another stupid reason).
-Daniel

>        s->StatusFlag = 1;
>        drm_intel_bo_unreference(sync->bo);
>        sync->bo = NULL;
> -- 
> 1.7.10.3
> 
> _______________________________________________
> Intel-gfx mailing list
> Intel-gfx at lists.freedesktop.org
> http://lists.freedesktop.org/mailman/listinfo/intel-gfx

-- 
Daniel Vetter
Mail: daniel at ffwll.ch
Mobile: +41 (0)79 365 57 48