[PATCH 1/2] drm/udl: make page flips asynchronous.

Chris Wilson chris at chris-wilson.co.uk
Fri Mar 6 00:14:15 PST 2015


On Thu, Mar 05, 2015 at 05:52:50PM -0800, Haixia Shi wrote:
> On Thu, Mar 5, 2015 at 2:33 AM, Chris Wilson <chris at chris-wilson.co.uk> wrote:
> > On Wed, Mar 04, 2015 at 06:26:23PM -0800, Haixia Shi wrote:
> >> This also limits the maximum frequency of page flips by the vrefresh rate.
> >>
> >> Signed-off-by: Haixia Shi <hshi at chromium.org>
> >> Reviewed-by: Stéphane Marchesin <marcheu at chromium.org>
> >> Tested-by: Haixia Shi <hshi at chromium.org>
> >
> > I think the better approach would have been to push the task down to
> > udl_handle_damage(). Almost all of the callers would prefer deferred
> > updates (with the notable exception of modesetting).
> > -Chris
> 
> I'm not sure this is correct. While there's reasonable expectation for
> page flips to be asynchronous (when flip completes it would send the
> vblank event), my understanding is that fb_fillrect, fb_copyarea and
> fb_imageblit are synchronous.

No, they are not synchronous with monitor refresh. Conceptually running
the usb xfer from a timer is the equivalent of the monitor reading from
the scanout at a distinct point in time from when we write into the
framebuffer.
-Chris

-- 
Chris Wilson, Intel Open Source Technology Centre


More information about the dri-devel mailing list