<div dir="ltr"><div dir="ltr"><div>Hi Pekka, Alexandros,</div><div><br></div>Here's a patch containing all I had to do in order to test the explicit sync support Alexandros has implemented in Weston.<div><br></div><div>Thanks,</div><div>Tomek</div><div><br></div></div><br><div class="gmail_quote"><div dir="ltr">On Wed, 28 Nov 2018 at 14:35, Tomek Bury <<a href="mailto:tomek.bury@gmail.com">tomek.bury@gmail.com</a>> wrote:<br></div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex"><div dir="ltr"><div>Hi Pekka,</div><div><br></div>> I suppose the compositor-side copy of buffers you mentioned is for the<br>> lack of release fences, not acquire fences?<div>Yes, the lack of release fences is the very reason for the copy. I could cook something up for the acquire fence, but that wouldn't synchronise the buffer access anyway. The only way I can be sure the client doesn't overwrite a buffer still in use by the GPU was to texture from a copy and let the compositor release the original without waiting for the GPU and without a fence.</div><div><br></div><div>Cheers,</div><div>Tomek</div><div><br></div></div>
</blockquote></div></div>