Hi folks, hi Daniel,
there is still an issue with flicker on panning with the 835GM chipset.
As already explained, the flicker only appears if the panning position
satisfies certain alignment constraints, in specific. As long as the
plane pointer is aligned to 64 byte boundaries, everything works correctly.
I tried now to add a parameter to the kernel to adjust the watermark
level of the chipset and see whether this makes any difference with
regards to the flicker. However, it does not. It seems that the trouble
is not there.
However, I noticed something interesting: If I use a non-native mode of
the display, i.e. enable the scaling feature of the DVO, everything
works, and no flicker ever appears. However, in these situations the
resolution is also lower than that of the native display, and hence the
issue may also be related to the PLL clock.
I also noticed that if I invert the display, i.e. use the rotation
feature, even in the native resolution, nothing flickers, though panning
is rather slow.
Does this somehow ring a bell? Can I somehow force the driver to use the
same logic for the regular display as for the inverted display to avoid
the flicker?
Any help would be appreciated.
Greetings,
Thomas
Hm, that sounds a bit more like the ddx is having fun with rendering.
Have you tried switching the backed from to either SNA or UXA? Also
adding relevant mailing lists ...
-Daniel
On Sun, Sep 22, 2013 at 7:06 PM, Thomas Richter <[email protected]> wrote:
> Hi folks, hi Daniel,
>
> there is still an issue with flicker on panning with the 835GM chipset. As
> already explained, the flicker only appears if the panning position
> satisfies certain alignment constraints, in specific. As long as the plane
> pointer is aligned to 64 byte boundaries, everything works correctly.
>
> I tried now to add a parameter to the kernel to adjust the watermark level
> of the chipset and see whether this makes any difference with regards to the
> flicker. However, it does not. It seems that the trouble is not there.
>
> However, I noticed something interesting: If I use a non-native mode of the
> display, i.e. enable the scaling feature of the DVO, everything works, and
> no flicker ever appears. However, in these situations the resolution is also
> lower than that of the native display, and hence the issue may also be
> related to the PLL clock.
>
> I also noticed that if I invert the display, i.e. use the rotation feature,
> even in the native resolution, nothing flickers, though panning is rather
> slow.
>
> Does this somehow ring a bell? Can I somehow force the driver to use the
> same logic for the regular display as for the inverted display to avoid the
> flicker?
>
> Any help would be appreciated.
>
> Greetings,
> Thomas
>
--
Daniel Vetter
Software Engineer, Intel Corporation
+41 (0) 79 365 57 48 - http://blog.ffwll.ch
On Sun, Sep 22, 2013 at 10:03:21PM +0200, Daniel Vetter wrote:
> Hm, that sounds a bit more like the ddx is having fun with rendering.
> Have you tried switching the backed from to either SNA or UXA? Also
> adding relevant mailing lists ...
If it was just the ddx, you would observe the same tearing with the
rotated setup as well. The question here would be whether it is tearing
or flicker? Tearing would just be the usual frontbuffer rendering
artifact. However, flicker could be misrendering of a DRI client (seeing
the backbuffer etc) or a clock timing issue (pipe underruns and whatnot).
Using a rotated framebuffer would disable pageflipping, and prevent the
historically common glitches when flipping. (Historically? That's
hysterical!)
-Chris
--
Chris Wilson, Intel Open Source Technology Centre
On 22.09.2013 22:03, Daniel Vetter wrote:
> Hm, that sounds a bit more like the ddx is having fun with rendering.
> Have you tried switching the backed from to either SNA or UXA? Also
> adding relevant mailing lists ...
>
No, whether I use uxa or sna makes no difference, same problem. I don't
think it's rendering related. The rendering is ok (except for some
defects in tuxracer if sna is used instead of uxa).
Greetings,
Thomas
On Mon, Sep 23, 2013 at 12:33:04AM +0200, Thomas Richter wrote:
> On 22.09.2013 22:03, Daniel Vetter wrote:
> >Hm, that sounds a bit more like the ddx is having fun with rendering.
> >Have you tried switching the backed from to either SNA or UXA? Also
> >adding relevant mailing lists ...
> >
> No, whether I use uxa or sna makes no difference, same problem. I
> don't think it's rendering related. The rendering is ok (except for
> some defects in tuxracer if sna is used instead of uxa).
Can you please test with
Option "LinearFramebuffer" "true"?
Thanks, Daniel
--
Daniel Vetter
Software Engineer, Intel Corporation
+41 (0) 79 365 57 48 - http://blog.ffwll.ch
Hi Daniel, hi others,
>> On 22.09.2013 22:03, Daniel Vetter wrote:
>>> Hm, that sounds a bit more like the ddx is having fun with rendering.
>>> Have you tried switching the backed from to either SNA or UXA? Also
>>> adding relevant mailing lists ...
>>>
>> No, whether I use uxa or sna makes no difference, same problem. I
>> don't think it's rendering related. The rendering is ok (except for
>> some defects in tuxracer if sna is used instead of uxa).
>
> Can you please test with
>
> Option "LinearFramebuffer" "true"?
Did. Still flickers. I'm not quite sure, but it seems to me that there
is at least one change (but allow me to test): If I'm now panning from
left to right (move the mouse to the right), then the screen is stable.
If I move it to the left, I get the flicker. Which is pretty strange.
Unless I'm completely mistaken, the tiled mode created flicker no matter
whether I scrolled left to right or right to left, it just depended on
the position. Wierd.
Greetings,
Thomas
Hi folks,
> Can you please test with
>
> Option "LinearFramebuffer" "true"?
>
> Thanks, Daniel
Tested again. Yes, indeed, LinearFramebuffer does make a difference.
Without it, panning to the right causes flicker, with it, moving the
mouse to the right - panning right (thus scrolling the screen to the
left) causes no flicker. Scrolling into the other direction always
causes flicker, no matter what.
Greetings,
Thomas
On Mon, Sep 23, 2013 at 12:21:30PM +0200, Thomas Richter wrote:
> Hi folks,
>
> >Can you please test with
> >
> > Option "LinearFramebuffer" "true"?
> >
> >Thanks, Daniel
>
> Tested again. Yes, indeed, LinearFramebuffer does make a difference.
> Without it, panning to the right causes flicker, with it, moving the
> mouse to the right - panning right (thus scrolling the screen to the
> left) causes no flicker. Scrolling into the other direction always
> causes flicker, no matter what.
Hm, this is something I've completely missed thus far: Does the flicker
only happen when you pan, or also when the image is at a stable postion
(but not aligned to one of the safe boundaries you've identified)?
Thanks, Daniel
--
Daniel Vetter
Software Engineer, Intel Corporation
+41 (0) 79 365 57 48 - http://blog.ffwll.ch