gma500: color distortion on framebuffer console unbind

gma500: color distortion on framebuffer console unbind

[Baruch Siach]

Hi Patrik,

I am using the gma500 driver of the latest stable kernel, 4.8.1. My hardware
is Atom N2600 processor integrated graphics controller, PCI PID: 0x0be1. 
Output is LVDS. The system starts with framebuffer console enabled. Later in 
the boot process I unbind the console with the following command:

  echo 0 > /sys/class/vtconsole/vtcon1/bind

Immediately as the console is unbound some colors on the screen appear
distorted. The links below show the output of 'fb-test'
(https://github.com/prpplague/fb-test-app) before and after unbind.

  http://filebin.ca/2xxwlwtBBzBN/display-good.jpg
  http://filebin.ca/2xxxTyIboJnR/display-bad.jpg

Suggestions are welcome.

Thanks,
baruch

-- 
     http://baruch.siach.name/blog/                  ~. .~   Tk Open Systems
=}------------------------------------------------ooO--U--Ooo------------{=
   - baruch@tkos.co.il - tel: +972.2.679.5364, http://www.tkos.co.il -
_______________________________________________
dri-devel mailing list
dri-devel@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/dri-devel

Re: gma500: color distortion on framebuffer console unbind

[Patrik Jakobsson]

Hi,

Sorry for late reply. Could be that we're not restoring the state
properly. Not sure though that we guarantee that the framebuffer
contents is valid after unbind. Perhaps take a look at what other
drivers do. I see the same issue on my PSB systems which run SDVO so
it is at least not LVDS related. Unfortunately I wont have time to
look at anything gma500 related at the moment. Feel free to dig into
this and I'll also put it on my todo-list.

Cheers
Patrik


On Mon, Oct 10, 2016 at 10:01 AM, Baruch Siach <baruch@tkos.co.il> wrote:
> Hi Patrik,
>
> I am using the gma500 driver of the latest stable kernel, 4.8.1. My hardware
> is Atom N2600 processor integrated graphics controller, PCI PID: 0x0be1.
> Output is LVDS. The system starts with framebuffer console enabled. Later in
> the boot process I unbind the console with the following command:
>
>   echo 0 > /sys/class/vtconsole/vtcon1/bind
>
> Immediately as the console is unbound some colors on the screen appear
> distorted. The links below show the output of 'fb-test'
> (https://github.com/prpplague/fb-test-app) before and after unbind.
>
>   http://filebin.ca/2xxwlwtBBzBN/display-good.jpg
>   http://filebin.ca/2xxxTyIboJnR/display-bad.jpg
>
> Suggestions are welcome.
>
> Thanks,
> baruch
>
> --
>      http://baruch.siach.name/blog/                  ~. .~   Tk Open Systems
> =}------------------------------------------------ooO--U--Ooo------------{=
>    - baruch@tkos.co.il - tel: +972.2.679.5364, http://www.tkos.co.il -
_______________________________________________
dri-devel mailing list
dri-devel@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/dri-devel

Re: gma500: color distortion on framebuffer console unbind

[Baruch Siach]

Hi Patrik,

On Thu, Oct 20, 2016 at 12:39:35AM +0200, Patrik Jakobsson wrote:
> Sorry for late reply. Could be that we're not restoring the state
> properly. Not sure though that we guarantee that the framebuffer
> contents is valid after unbind. Perhaps take a look at what other
> drivers do. I see the same issue on my PSB systems which run SDVO so
> it is at least not LVDS related. Unfortunately I wont have time to
> look at anything gma500 related at the moment. Feel free to dig into
> this and I'll also put it on my todo-list.

Thanks for taking the time to reproduce this issue. This rules out any local 
board layout issue on my side. For the record, my hardware is the Advantech 
SIMB-M02 Mini-ITX board.

I should have mentioned that the issue persists after unbind. That is, when I 
run fb-test only after unbind, I see the same corrupted display. So 
framebuffer content invalidation on console unbind is unlikely to be the issue 
here.

One more point that we see a similar issue with our Qt based application that 
uses the "linuxfb" plugin. The interesting twist here is that SOMETIMES the 
display corruption issue just disappears after a few minutes. The application 
is idle during that time. Not sure what to make of that.

Thanks,
baruch

> On Mon, Oct 10, 2016 at 10:01 AM, Baruch Siach <baruch@tkos.co.il> wrote:
> > I am using the gma500 driver of the latest stable kernel, 4.8.1. My 
> > hardware is Atom N2600 processor integrated graphics controller, PCI PID: 
> > 0x0be1. Output is LVDS. The system starts with framebuffer console 
> > enabled. Later in the boot process I unbind the console with the following 
> > command:
> >
> >   echo 0 > /sys/class/vtconsole/vtcon1/bind
> >
> > Immediately as the console is unbound some colors on the screen appear
> > distorted. The links below show the output of 'fb-test'
> > (https://github.com/prpplague/fb-test-app) before and after unbind.
> >
> >   http://filebin.ca/2xxwlwtBBzBN/display-good.jpg
> >   http://filebin.ca/2xxxTyIboJnR/display-bad.jpg
> >
> > Suggestions are welcome.

-- 
     http://baruch.siach.name/blog/                  ~. .~   Tk Open Systems
=}------------------------------------------------ooO--U--Ooo------------{=
   - baruch@tkos.co.il - tel: +972.52.368.4656, http://www.tkos.co.il -
_______________________________________________
dri-devel mailing list
dri-devel@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/dri-devel