[PATCH] ARM: sun7i: add support for A20-OLinuXino-Lime2

[Javier Martinez Canillas]

Hello Maxim, Iain,

On Thu, Oct 2, 2014 at 2:48 PM, Maxime Ripard
<maxime.ripard at free-electrons.com> wrote:
> On Mon, Sep 29, 2014 at 03:32:43PM +0100, Iain Paton wrote:
>> On 29/09/14 14:25, Maxime Ripard wrote:
>> > On Mon, Sep 29, 2014 at 01:56:49PM +0100, Iain Paton wrote:
>>
>> >> What additional patches are we talking about?  The following:
>> >> regulator: axp20x: Use parent device as regulator configuration device
>> >> i2c: mv64xxx: continue probe when clock-frequency is missing
>> >> or are there others I'd need to pick out of linux-next?
>> >
>> > http://lists.infradead.org/pipermail/linux-arm-kernel/2014-September/288901.html
>>
>> Ok, I'll pull those on top of something not quite as bleeding edge as linux-next
>>
>> > Leave it out then. If you have no way to test it, it's going to be
>> > very bug prone, and we can always add it later.
>>
>> I moved my root disk to a SATA drive, so only a minor inconvenience.
>>
>> dmesg shows the following with the regulators node as in the v2 patch,
>> the bottom half being mostly what I'd expect
>>
>> [    0.868607] sunxi-rtc 1c20d00.rtc: RTC enabled
>> [    0.877062] Adding alias for supply acin,(null) -> acin,0-0034
>> [    0.882948] Adding alias for supply vin2,(null) -> vin2,0-0034
>> [    0.888775] Adding alias for supply vin3,(null) -> vin3,0-0034
>> [    0.894612] Adding alias for supply ldo24in,(null) -> ldo24in,0-0034
>> [    0.900956] Adding alias for supply ldo3in,(null) -> ldo3in,0-0034
>> [    0.907160] Adding alias for supply ldo5in,(null) -> ldo5in,0-0034
>> [    0.914029] LDO1: 1300 mV
>> [    0.917428] LDO2: 1800 <--> 3300 mV at 3000 mV
>> [    0.922483] LDO3: 700 <--> 3500 mV at 2275 mV
>> [    0.927136] LDO4: 1250 <--> 3300 mV at 2800 mV
>> [    0.932334] LDO5: 1800 <--> 3300 mV at 2800 mV
>> [    0.937308] DCDC2: 1400 mV
>> [    0.940552] DCDC3: 1250 mV
>> [    0.943474] axp20x 0-0034: AXP20X driver loaded
>
> No, it's not. You don't have the parenting relationship between the
> regulators, and you shouldn't use the aliases anymore.
>
> Like I said, you'd better drop that part until things settle down.
>
>> > The SD card issue is more troublesome. We'll have to look into it.
>>
>> dmesg shows the mmc driver loading:
>>
>> [    0.957716] sdhci: Secure Digital Host Controller Interface driver
>> [    0.963963] sdhci: Copyright(c) Pierre Ossman
>> [    0.968843] sunxi-mmc 1c0f000.mmc: No vqmmc regulator found
>> [    0.974918] sunxi-mmc 1c0f000.mmc: Got CD GPIO
>> [    1.012036] sunxi-mmc 1c0f000.mmc: base:0xf00ac000 irq:64
>> [    1.017623] sdhci-pltfm: SDHCI platform and OF driver helper
>>
>> however on an older, working, kernel I get a number on the end of
>> the 'Got CD GPIO' line like this:
>>
>> [    0.896271] sunxi-mmc 1c0f000.mmc: Got CD GPIO #225.
>
> That's interesting. It seems like it might be related to the recent
> breakages in the boot farm on the A31. These breakages were not
> happening on the arm-soc branch, but on linux-next only, so it must be
> coming from somewhere else, which is not that great, but rules out the
> DTS.
>
> I'll bisect it this weekend, but don't worry too much about it.
>

The problem you are facing is because commit 98e90de ("mmc: host:
switch OF parser to use gpio descriptors") changed the semantics when
the mmc cd and wp properties were converted to use the GPIO descriptor
based API so card detection is not working on your board.

I reported the issue to Linus W. and he provided a fix [0] which
should land in linux-next soon.

> Maxime
>

Best regards,
Javier

[0]: https://www.mail-archive.com/linux-mmc@vger.kernel.org/msg28751.html