[Letux-kernel] Pyra display driver once again broken in v5.2-rc1

[H. Nikolaus Schaller]

> Am 29.05.2019 um 07:16 schrieb Andreas Kemnade <andreas at kemnade.info>:
> 
> On Wed, 29 May 2019 06:39:22 +0200
> "H. Nikolaus Schaller" <hns at goldelico.com> wrote:
> 
>>> Am 28.05.2019 um 23:40 schrieb Andreas Kemnade <andreas at kemnade.info>:
>>> 
>>> On Mon, 27 May 2019 21:25:41 +0200
>>> "H. Nikolaus Schaller" <hns at goldelico.com> wrote:
>>> 
>>>>> Am 27.05.2019 um 14:46 schrieb H. Nikolaus Schaller <hns at goldelico.com>:
>>>>> 
>>>>> Hi,
>>>>> unfortunately every year or so some upstream change breaks
>>>>> our (still out-of tree) DSI panel driver for the Pyra(Phone).
>>>>> This time it happened with v5.2-rc1.
>>>>> 
>>>>> v5.2-rc1 also broke the GTA04 display and the L3704/L7004 displays,
>>>>> but those are DPI panels and the reasons were two separate
>>>>> issues which have nothing to do with the Pyra DSI panel.
>>>>> 
>>>>> There have been some API changes for DSI panel drivers and
>>>>> I have done a
>>>>> 
>>>>> 	git diff v5.2-rc1 v5.1 drivers/gpu/drm/omapdrm/displays/panel-dsi-cm.c
>>>>> 
>>>>> to identify them and tried to apply similar patches to our
>>>>> panel. But still the driver is successfully probed, but shows
>>>>> not activity. No calls to enable() or connect(). Backlight stays
>>>>> black. And as a result there is also no /dev/fb0, text console
>>>>> or X11.
>>>>> 
>>>>> Unfortunately, it has not improved after upgrading to v5.2-rc2.
>>>>> 
>>>>> So once again it appears that I have to git bisect upstream
>>>>> patches to find out what is going on...    
>>>> 
>>>> After spending more hours... it does not seem to be in our patches
>>>> and the panel driver is likely almost correct now.
>>>> 
>>>> I believe this because I also found not working:
>>>> * dwc3 (usb3/otg port)
>>>> * bq24297 battery charger driver (my Pyra fortunately runs well enough with U-Boot initialization)
>>>> * tca6424 (gpio expander - we use the pcal9524 variant)
>>>> 
>>>> The tca6424 driver seems to be broken (not loaded).
>>>> 
>>> [    6.581432] ts3a227e 4-003b: Cannot request irq 0 (-22)
>>> [    6.597252] ts3a227e: probe of 4-003b failed with error -22
>>> 
>>> so somehow it insists of having an irq and we are not providing one in
>>> the dtb, at least not in pyra-mainboard-v5.1.dtsi  
>> 
>> The reference is set in pyra-mainboard-v5.2.dtsi
>> 
> ... which is as far as I understand not used on a 5.1 board.
> 
>> 		interrupt-parent = <&gpio99>;
>> 		interrupts = <14 IRQ_TYPE_EDGE_RISING>;
>> 
>> But: this is another symptom of the tca6424 (gpio99) not being available.
>> It is not only a gpio expander but also an interrupt expander.
> 
> so it an independant problem.
>> 
>> So we should focus on finding out why the tca6424 / pcal6524 does
>> not probe.
> 
> yes, I disabled the hdmi stuff completely. Result: Display works.

Ok, fine! So we at least have an idea that the broken display is this
time just a symptom and not a bug...

Strangely I do not see a printk() added to pca953x_probe()...

BR and thanks,
Nikolaus