Nokia N900 FB regression?

Nokia N900 FB regression?

[Aaro Koskinen]

Hi,

Somewhere between 3.16-rc2 and rc6 (I was on holidays...) I noticed
the framebuffer stopped working on N900 (nothing on screen).

I bisected this to 9188883fd66e9809e93e06d5bbd49cf99a6cdbee (ARM: dts:
Enable twl4030 off-idle configuration for selected omaps).
With this commit reverted, I can see the Tux again with 3.16-rc6.

Any ideas?

A.

Re: Nokia N900 FB regression?

[Nick Krause]

On Tue, Jul 22, 2014 at 5:23 PM, Aaro Koskinen <aaro.koskinen@iki.fi> wrote:
> Hi,
>
> Somewhere between 3.16-rc2 and rc6 (I was on holidays...) I noticed
> the framebuffer stopped working on N900 (nothing on screen).
>
> I bisected this to 9188883fd66e9809e93e06d5bbd49cf99a6cdbee (ARM: dts:
> Enable twl4030 off-idle configuration for selected omaps).
> With this commit reverted, I can see the Tux again with 3.16-rc6.
>
> Any ideas?
>
> A.
> --
> To unsubscribe from this list: send the line "unsubscribe linux-omap" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html


I am new to the kernel but I can take a look into this this commit.
Cheers Nick

Re: Nokia N900 FB regression?

[Nick Krause]

On Tue, Jul 22, 2014 at 6:02 PM, Nick Krause <xerofoify@gmail.com> wrote:
> On Tue, Jul 22, 2014 at 5:23 PM, Aaro Koskinen <aaro.koskinen@iki.fi> wrote:
>> Hi,
>>
>> Somewhere between 3.16-rc2 and rc6 (I was on holidays...) I noticed
>> the framebuffer stopped working on N900 (nothing on screen).
>>
>> I bisected this to 9188883fd66e9809e93e06d5bbd49cf99a6cdbee (ARM: dts:
>> Enable twl4030 off-idle configuration for selected omaps).
>> With this commit reverted, I can see the Tux again with 3.16-rc6.
>>
>> Any ideas?
>>
>> A.
>> --
>> To unsubscribe from this list: send the line "unsubscribe linux-omap" in
>> the body of a message to majordomo@vger.kernel.org
>> More majordomo info at  http://vger.kernel.org/majordomo-info.html
>
>
> I am new to the kernel but I can take a look into this this commit.
> Cheers Nick

+
+       twl_power: power {
+               compatible = "ti,twl4030-power-n900",
"ti,twl4030-power-idle-osc-off";
+               ti,use_poweroff;
+       };
 };
This seems to be the case of no boot. Perhaps compatible functions are
incorrect?
I am new so asking a maintainer may be better but seems the idle function set
here turns off your frame butter. I would also test this on the beagleboard xm
as this patch may affect it too.
Nick

Re: Nokia N900 FB regression?

[Tony Lindgren]

* Aaro Koskinen <aaro.koskinen@iki.fi> [140722 14:25]:
> Hi,
> 
> Somewhere between 3.16-rc2 and rc6 (I was on holidays...) I noticed
> the framebuffer stopped working on N900 (nothing on screen).
> 
> I bisected this to 9188883fd66e9809e93e06d5bbd49cf99a6cdbee (ARM: dts:
> Enable twl4030 off-idle configuration for selected omaps).
> With this commit reverted, I can see the Tux again with 3.16-rc6.
> 
> Any ideas?

Hmm, OK yeah that enables the deeper idle states. I have tested
this on n900, but only have remote access to it so could not see
the framebuffer.

Sounds like there's some twl regulator that we cannot idle
by default on n900. Here are the steps to narrow it down:

1. Try ti,twl4030-power-idle in omap3-n900.dts instead of
   ti,twl4030-power-idle-osc-off. This probably won't help
   unless you're enabling off-idle though, so if it does not
   help, just keep on using ti,twl4030-power-idle-osc-off.

2. Try commenting out one or more of the TWL_REMAP_SLEEP or
   TWL_REMAP_OFF lines in omap3_idle_rconfig in file
   drivers/mfd/twl4030-power.c. That's the recommended init
   sequence like the comments in the code mention. Chances
   are we're missing a regulator_get somewhere but finding
   out which one in omap3_idle_rconfig will help narrow it
   down :)

3. If it looks like a more complicated fix, we can just use
   compatible = "ti,twl4030-power-n900" to disable the twl
   features for v3.16 and re-enable it for v3.17.

Regards,

Tony

Re: Nokia N900 FB regression?

[Nick Krause]

On Wed, Jul 23, 2014 at 3:37 AM, Tony Lindgren <tony@atomide.com> wrote:
> * Aaro Koskinen <aaro.koskinen@iki.fi> [140722 14:25]:
>> Hi,
>>
>> Somewhere between 3.16-rc2 and rc6 (I was on holidays...) I noticed
>> the framebuffer stopped working on N900 (nothing on screen).
>>
>> I bisected this to 9188883fd66e9809e93e06d5bbd49cf99a6cdbee (ARM: dts:
>> Enable twl4030 off-idle configuration for selected omaps).
>> With this commit reverted, I can see the Tux again with 3.16-rc6.
>>
>> Any ideas?
>
> Hmm, OK yeah that enables the deeper idle states. I have tested
> this on n900, but only have remote access to it so could not see
> the framebuffer.
>
> Sounds like there's some twl regulator that we cannot idle
> by default on n900. Here are the steps to narrow it down:
>
> 1. Try ti,twl4030-power-idle in omap3-n900.dts instead of
>    ti,twl4030-power-idle-osc-off. This probably won't help
>    unless you're enabling off-idle though, so if it does not
>    help, just keep on using ti,twl4030-power-idle-osc-off.
>
> 2. Try commenting out one or more of the TWL_REMAP_SLEEP or
>    TWL_REMAP_OFF lines in omap3_idle_rconfig in file
>    drivers/mfd/twl4030-power.c. That's the recommended init
>    sequence like the comments in the code mention. Chances
>    are we're missing a regulator_get somewhere but finding
>    out which one in omap3_idle_rconfig will help narrow it
>    down :)
>
> 3. If it looks like a more complicated fix, we can just use
>    compatible = "ti,twl4030-power-n900" to disable the twl
>    features for v3.16 and re-enable it for v3.17.
>
> Regards,
>
> Tony
> --
> To unsubscribe from this list: send the line "unsubscribe linux-omap" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html


Thanks Tony for the help, seems that this will be fixed before the 3.17 and
needs no more assist from me.
Cheers Nick

Re: Nokia N900 FB regression?

[Aaro Koskinen]

Hi,

On Wed, Jul 23, 2014 at 12:37:06AM -0700, Tony Lindgren wrote:
> * Aaro Koskinen <aaro.koskinen@iki.fi> [140722 14:25]:
> > Hi,
> > 
> > Somewhere between 3.16-rc2 and rc6 (I was on holidays...) I noticed
> > the framebuffer stopped working on N900 (nothing on screen).
> > 
> > I bisected this to 9188883fd66e9809e93e06d5bbd49cf99a6cdbee (ARM: dts:
> > Enable twl4030 off-idle configuration for selected omaps).
> > With this commit reverted, I can see the Tux again with 3.16-rc6.
> > 
> > Any ideas?
> 
> Hmm, OK yeah that enables the deeper idle states. I have tested
> this on n900, but only have remote access to it so could not see
> the framebuffer.
> 
> Sounds like there's some twl regulator that we cannot idle
> by default on n900. Here are the steps to narrow it down:
> 
> 1. Try ti,twl4030-power-idle in omap3-n900.dts instead of
>    ti,twl4030-power-idle-osc-off. This probably won't help
>    unless you're enabling off-idle though, so if it does not
>    help, just keep on using ti,twl4030-power-idle-osc-off.
> 
> 2. Try commenting out one or more of the TWL_REMAP_SLEEP or
>    TWL_REMAP_OFF lines in omap3_idle_rconfig in file
>    drivers/mfd/twl4030-power.c. That's the recommended init
>    sequence like the comments in the code mention. Chances
>    are we're missing a regulator_get somewhere but finding
>    out which one in omap3_idle_rconfig will help narrow it
>    down :)

It's VAUX1 - commenting that out alone fixes the issue. But I think we
are defining that one properly in the DTS (vdds_sdi-supply = <&vaux1>),
and also I checked that sdi_init_regulator() succeeds...

BTW, this issue can be also detected without seeing the framebuffer,
from dmesg:

# dmesg|grep omapfb
[    2.785705] omapfb omapfb: no displays
[    2.786041] omapfb omapfb: failed to setup omapfb
[    2.786102] platform omapfb: Driver omapfb requests probe deferral
[    4.038055] acx565akm spi1.2: omapfb: acx565akm rev ab LCD detected
[    5.421325] omapfb omapfb: Failed to enable display 'lcd'
[    5.431335] omapfb omapfb: failed to initialize default display
[    5.450408] omapfb omapfb: failed to setup omapfb
[    5.455444] omapfb: probe of omapfb failed with error -1

A.

Re: Nokia N900 FB regression?

[Tony Lindgren]

* Aaro Koskinen <aaro.koskinen@iki.fi> [140724 13:03]:
> Hi,
> 
> On Wed, Jul 23, 2014 at 12:37:06AM -0700, Tony Lindgren wrote:
> > * Aaro Koskinen <aaro.koskinen@iki.fi> [140722 14:25]:
> > > Hi,
> > > 
> > > Somewhere between 3.16-rc2 and rc6 (I was on holidays...) I noticed
> > > the framebuffer stopped working on N900 (nothing on screen).
> > > 
> > > I bisected this to 9188883fd66e9809e93e06d5bbd49cf99a6cdbee (ARM: dts:
> > > Enable twl4030 off-idle configuration for selected omaps).
> > > With this commit reverted, I can see the Tux again with 3.16-rc6.
> > > 
> > > Any ideas?
> > 
> > Hmm, OK yeah that enables the deeper idle states. I have tested
> > this on n900, but only have remote access to it so could not see
> > the framebuffer.
> > 
> > Sounds like there's some twl regulator that we cannot idle
> > by default on n900. Here are the steps to narrow it down:
> > 
> > 1. Try ti,twl4030-power-idle in omap3-n900.dts instead of
> >    ti,twl4030-power-idle-osc-off. This probably won't help
> >    unless you're enabling off-idle though, so if it does not
> >    help, just keep on using ti,twl4030-power-idle-osc-off.
> > 
> > 2. Try commenting out one or more of the TWL_REMAP_SLEEP or
> >    TWL_REMAP_OFF lines in omap3_idle_rconfig in file
> >    drivers/mfd/twl4030-power.c. That's the recommended init
> >    sequence like the comments in the code mention. Chances
> >    are we're missing a regulator_get somewhere but finding
> >    out which one in omap3_idle_rconfig will help narrow it
> >    down :)
> 
> It's VAUX1 - commenting that out alone fixes the issue. But I think we
> are defining that one properly in the DTS (vdds_sdi-supply = <&vaux1>),
> and also I checked that sdi_init_regulator() succeeds...

OK, looks like the LCD panel is also using V28 from vaux1. It
goes to the "Main flex connector" pin 26.

It also seems that other devices are also using V28 like
accelerometer, main camera cover switch, usd card hot-swap
detection sensor, tvout analog switch, fm transmitter.. And
it's marked as regulator-always-on because of battery cover
switch.

So I'll just send a patch to disable the twl configuration
for n900 for now as anything else would be quite intrusive. I'll
look into it more later on once I have access to my n900.
 
> BTW, this issue can be also detected without seeing the framebuffer,
> from dmesg:
> 
> # dmesg|grep omapfb
> [    2.785705] omapfb omapfb: no displays
> [    2.786041] omapfb omapfb: failed to setup omapfb
> [    2.786102] platform omapfb: Driver omapfb requests probe deferral
> [    4.038055] acx565akm spi1.2: omapfb: acx565akm rev ab LCD detected
> [    5.421325] omapfb omapfb: Failed to enable display 'lcd'
> [    5.431335] omapfb omapfb: failed to initialize default display
> [    5.450408] omapfb omapfb: failed to setup omapfb
> [    5.455444] omapfb: probe of omapfb failed with error -1

OK thanks. I bet I was doing testing on omap2plus_defconfig
only that does not even have the LCD enabled.

Regards,

Tony