GTA01V3/V4 LCD framebuffer problem

[Arnaud Patard (Rtp)]

Nils Faerber <nils.faerber at kernelconcepts.de> writes:

> Hi!

Hi !

> I am currently trying to get suspend/resume working with the current NEO
> GTA01 devices and encountered a strange issue...

Bv3 or Bv4 ? I may be wrong but iirc there was some suspend/resume
troubles on Bv3 hardware.

>
> With the latest kernel 2.6.21.5 I can successfully suspend and resume
> the machine. Most devices seem to be fine after resume, including USB,
> audio, touchscreen, modem, etc.
> But the LCD just displays a blank white screen with backlight switched
> on. And this does not change anymore until reboot.
>
> I looked through the s3c2410fb.c code and found it to be appearingly
> good, concerning suspend/resume. So in theory it should work just that
> it does not.

well, the s3c2410fb driver suspend/resume is working fine on my h1940
and I didn't get any report that it was not working.

>
> I also found by some additional debugging output that the LCD controller
> registers are still properly setup but in control register 5 I see that
> the status bits for the vertical and horizontal back porch are always set.
>
> Does that mean anything? Is it hanging at the back porch?
> Could it be possible that, though the LCD clock is reenabled on resume,
> the clock does not actually start to tick (for some reason)?
> Another suspicion I have is that probably the LCD_RESET is deasserted
> upon resume - I have not checked its state after resume (yet). This
> could probably also cause the TFT itself to become deaf.
>
> It's quite strange and also a pity - so close to be mostly usable ;)
>
> So any hint where and what for to look would be greatly appreciated! I
> am willing to do the work, I am just new to this platform and need some
> hints where to look for such kind of glitches...

If you're not in a big hurry for this issue to be solved, I'll try to
find some time this week-end to try to reproduce and fix this bug on my
neos. 

Regards,
Arnaud