[MPlayer-dev-eng] [PATCH][RESEND] XV page flipping sync

[Ivan Kalvachev]

Joey Parrish said:
> On Tue, Jun 22, 2004 at 10:42:09AM +0200, Diego Biurrun wrote:
>> Reimar Döffinger writes:
>> > > 01:44 <@KotH> DonDiego: whether it catches all XFlush calls 01:44
<@KotH> DonDiego: if he is right, then there should be no XFlush
>> in the x11 code
>> >
>> > My theory was always that the XFlush is rather useless. Am I correct?
Should they be removed as well? (The problem described in the original
mail can't happen here, as XSync is called as well).
>> Been talking to Attila on IRC again:
>> 02:50 <@KotH> DonDiego: imho, the patch should be extended to
>>               xv/x11/x11_common, then it's ok to apply
>> 02:51 <@KotH> DonDiego: and to answer reimars question: according to the
>>               manpages, doing xflush+xsync is useless, xsync alone is
enough
>> So could somebody extend this patch?
>
> Okay, the first patch removes XFlush whenever it's redundant with XSync
calls.  The second patch removes all XFlush everywhere.  I know little of X,
but I feel like the first one is probably safe and the second one is
probably wrong.  I leave it to somebody else to apply.
>
> Patches actually attached this time.  :)
>
> --Joey

Somehow I don't like both patches.
I agree it is pointless to have BOTH XFlush and XSync, but I'm not sure that
XFlush is always the one that should be removed.

I particularry don't like this:
@@ -595,12 +593,9 @@
                    (vo_fs ? vo_dheight - 1 : vo_dheight) + vo_panscan_y);
     }
     if (num_buffers > 1)
-    {
         current_buf =
             vo_directrendering ? 0 : ((current_buf + 1) % num_buffers);
-        XFlush(mDisplay);
-    } else
-        XSync(mDisplay, False);
+    XSync(mDisplay, False);
     return;
 }
Are you sure this is the right solution for all cases?
Anyway are you sure that this syntax doesn't have some purpose?

Wish You Best
   Ivan Kalvachev
  iive