v3.10-rc1: backlight brightness after boot on N900

v3.10-rc1: backlight brightness after boot on N900

[Aaro Koskinen]

Hi,

Is it expected that after boot you get 0 brightness i.e. a seemingly
blank display on N900 with 3.10-rc1?

First thought after seeing a blank display was that the probe
issues are still there, but everything was fine and after setting
/sys/class/backlight/acx565akm/brightness you get a picture.

Just wondering how the initial brightness is determined anyway?

A.

Re: v3.10-rc1: backlight brightness after boot on N900

[Tomi Valkeinen]

[-- Attachment #1: Type: text/plain, Size: 1409 bytes --]

On 15/05/13 20:18, Aaro Koskinen wrote:
> Hi,
> 
> Is it expected that after boot you get 0 brightness i.e. a seemingly
> blank display on N900 with 3.10-rc1?
> 
> First thought after seeing a blank display was that the probe
> issues are still there, but everything was fine and after setting
> /sys/class/backlight/acx565akm/brightness you get a picture.
> 
> Just wondering how the initial brightness is determined anyway?

There were some small changes merged in v3.10 related to acx's
backlight, but they were removing dead code so I don't see how those
could affect...

Looking at the driver, the bl is initialized:

	if (md->has_bc)
		brightness = acx565akm_get_actual_brightness(md);
	else
		brightness = 0;

And has_bc is set only if the model is detected as MIPID_VER_ACX565AKM
(that's the panel in final N900, isn't it?). So what comes to my mind is
that either the model is detected wrong, or the
acx565akm_get_actual_brightness() which reads the current BL from the
panel happens to return 0, either because of an error, or because the
bootloader has set the BL to 0.

But why either of those would've changed with v3.10... No idea...

Can you add debug prints to see which code paths are taken there
(drivers/video/omap2/displays/panel-acx565akm.c, around line 580). The
panel model should also be printed with "omapfb: %s rev %02x LCD detected".

 Tomi



[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 901 bytes --]

Re: v3.10-rc1: backlight brightness after boot on N900

[Aaro Koskinen]

On Thu, May 16, 2013 at 07:52:12AM +0300, Tomi Valkeinen wrote:
> On 15/05/13 20:18, Aaro Koskinen wrote:
> > Is it expected that after boot you get 0 brightness i.e. a seemingly
> > blank display on N900 with 3.10-rc1?
> 
> There were some small changes merged in v3.10 related to acx's
> backlight, but they were removing dead code so I don't see how those
> could affect...

[...]

> So what comes to my mind is that either the model is detected wrong,
> or the acx565akm_get_actual_brightness() which reads the current BL from
> the panel happens to return 0, either because of an error, or because
> the bootloader has set the BL to 0.
> 
> But why either of those would've changed with v3.10... No idea...

The panel is detected and probed correctly. So something else is wrong.
It's not a bootloader issue, because booting 3.9 right after 3.10-rc1
makes the display work again correctly.

So I bisected and it points to:

	commit a2797beadfcb5a0974a30929b613b4d1adb023d1
	Author: Jon Hunter <jon-hunter@ti.com>
	Date:   Thu Apr 4 15:16:15 2013 -0500

	    gpio/omap: force restore if context loss is not detectable

When I revert that commit from 3.10-rc1, the display works again as
expected.

A.

Re: v3.10-rc1: backlight brightness after boot on N900

[Tony Lindgren]

* Aaro Koskinen <aaro.koskinen@iki.fi> [130517 11:30]:
> On Thu, May 16, 2013 at 07:52:12AM +0300, Tomi Valkeinen wrote:
> > On 15/05/13 20:18, Aaro Koskinen wrote:
> > > Is it expected that after boot you get 0 brightness i.e. a seemingly
> > > blank display on N900 with 3.10-rc1?
> > 
> > There were some small changes merged in v3.10 related to acx's
> > backlight, but they were removing dead code so I don't see how those
> > could affect...
> 
> [...]
> 
> > So what comes to my mind is that either the model is detected wrong,
> > or the acx565akm_get_actual_brightness() which reads the current BL from
> > the panel happens to return 0, either because of an error, or because
> > the bootloader has set the BL to 0.
> > 
> > But why either of those would've changed with v3.10... No idea...
> 
> The panel is detected and probed correctly. So something else is wrong.
> It's not a bootloader issue, because booting 3.9 right after 3.10-rc1
> makes the display work again correctly.
> 
> So I bisected and it points to:
> 
> 	commit a2797beadfcb5a0974a30929b613b4d1adb023d1
> 	Author: Jon Hunter <jon-hunter@ti.com>
> 	Date:   Thu Apr 4 15:16:15 2013 -0500
> 
> 	    gpio/omap: force restore if context loss is not detectable
> 
> When I revert that commit from 3.10-rc1, the display works again as
> expected.

OK that should be fixed by ab6ec0fb (gpio/omap: ensure gpio context is initialised)
in linux-next. Can you check if that helps?

Looks like it has not yet been merged to mainline tree.
Linus W, any news on when you might be sending that fix upstream?

Regards,

Tony

Re: v3.10-rc1: backlight brightness after boot on N900

[Aaro Koskinen]

Hi,

On Fri, May 17, 2013 at 11:45:35AM -0700, Tony Lindgren wrote:
> * Aaro Koskinen <aaro.koskinen@iki.fi> [130517 11:30]:
> > So I bisected and it points to:
> > 
> > 	commit a2797beadfcb5a0974a30929b613b4d1adb023d1
> > 	Author: Jon Hunter <jon-hunter@ti.com>
> > 	Date:   Thu Apr 4 15:16:15 2013 -0500
> > 
> > 	    gpio/omap: force restore if context loss is not detectable
> > 
> > When I revert that commit from 3.10-rc1, the display works again as
> > expected.
> 
> OK that should be fixed by ab6ec0fb (gpio/omap: ensure gpio context is initialised)
> in linux-next. Can you check if that helps?

Yeah, that fixes it. Thanks,

A.

Re: v3.10-rc1: backlight brightness after boot on N900

[Linus Walleij]

On Fri, May 17, 2013 at 8:45 PM, Tony Lindgren <tony@atomide.com> wrote:

> OK that should be fixed by ab6ec0fb (gpio/omap: ensure gpio context is initialised)
> in linux-next. Can you check if that helps?
>
> Looks like it has not yet been merged to mainline tree.
> Linus W, any news on when you might be sending that fix upstream?

It fell out because of the other debacle during the merge window.

I'm more than overloaded (due to the ever increasing development pace
of the kernel I guess) so queueing fixes takes some time.

Yours,
Linus Walleij

Re: v3.10-rc1: backlight brightness after boot on N900

[Tony Lindgren]

* Linus Walleij <linus.walleij@linaro.org> [130517 14:56]:
> On Fri, May 17, 2013 at 8:45 PM, Tony Lindgren <tony@atomide.com> wrote:
> 
> > OK that should be fixed by ab6ec0fb (gpio/omap: ensure gpio context is initialised)
> > in linux-next. Can you check if that helps?
> >
> > Looks like it has not yet been merged to mainline tree.
> > Linus W, any news on when you might be sending that fix upstream?
> 
> It fell out because of the other debacle during the merge window.
> 
> I'm more than overloaded (due to the ever increasing development pace
> of the kernel I guess) so queueing fixes takes some time.

Yes I can totally understand that, thanks for the update!

Regards,

Tony