git pull - ickle dri patch

[Eric Anholt]

Dave Airlie <airlied at gmail.com> writes:
> This patch has gone around long enough and seem to not make it.
>
> Dave.
>
> The following changes since commit 70e5766874a919039678bb2ed75f2ccea0cb4345:
>
>   xf86: fix multi-seat video device support. (v2) (2012-09-19 15:48:50 +1000)
>
> are available in the git repository at:
>
>   ssh://people.freedesktop.org/~airlied/xserver for-keithp
>
> for you to fetch changes up to e8b2000e5f6e19f1da23bfef8ed9e833c7643e54:
>
>   dri2: Pass AsyncSwap [swap_interval=0] requests to the drivers
> (2012-09-20 09:32:14 +1000)

I still don't understand why we need a new entrypoint for this.  Is it
just to avoid synced behavior if the existing entrypoint is not
expecting to see swap_interval=0 requests?  Or is it that you're trying
to avoid invalidates for the blit case?  That seems totally orthogonal
to vblank_mode=0 support.  There are plenty of swap-blit paths in the
vblank_mode!=0 code.

I actually tried this code out anyway (sna with --enable-async-swap),
and I've got to say I'm not impressed.  It did get me pageflipping for
vblank_mode=0, and it quad-buffers so you avoid stalling (because memory
is awesome), but it also manages to trivially hang the X Server: under
metacity, vblank_mode=0 glxgears -fullscreen, hit escape, boom.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 197 bytes
Desc: not available
URL: <http://lists.x.org/archives/xorg-devel/attachments/20120925/134b4e17/attachment.pgp>