[PATCHv2 1/2] ARM: dts: socfpga: Fix SD card detect

[Dinh Nguyen]

On 10/20/2014 02:48 PM, Doug Anderson wrote:
> Dinh,
> 
> On Mon, Oct 20, 2014 at 12:30 PM, Doug Anderson <dianders at chromium.org> wrote:
>> Dinh,
>>
>> On Mon, Oct 20, 2014 at 9:31 AM, Dinh Nguyen
>> <dinguyen at opensource.altera.com> wrote:
>>>>>  &mmc0 {
>>>>> -       cd-gpios = <&gpio1 18 0>;
>>>>> +       cd = <&gpio1 18 0>;
>>>>
>>>> This doesn't look right to me.  What was the error that was passed back?
>>>>
>>>> I think your change has the same net effect as just deleting the
>>>> "cd-gpios" line.  ...or is there some code somewhere that is parsing
>>>> the "cd" property.
>>>>
>>>
>>> It just hangs here:
>>>
>>> dw_mmc ff704000.dwmmc0: Using PIO mode.
>>> dw_mmc ff704000.dwmmc0: Version ID is 240a
>>> dw_mmc ff704000.dwmmc0: DW MMC controller at irq 171, 32 bit host data
>>> width, 1024 deep fifo
>>> platform ff704000.dwmmc0: Driver dw_mmc requests probe deferral
>>>
>>>
>>> Without this patch :
>>> mmc_of_parse ret=-517 (EPROBE_DEFER)
>>
>> I think you need to dig more into this error.  Why are you getting an
>> -EPROBE_DEFER when you asked for this GPIO?
>>
>> The -EPROBE_DEFER tells the system that a resource is not available
>> yet and that it should try to re-run the dw_mmc init later once the
>> resource becomes available.  I believe that some other bug is causing
>> the resource to never become available.
>>
>> Guesses:
>>
>> * The GPIO specifier is wrong in the DTB and is pointing to a GPIO
>> that would never exist.
>>
>> * Something is happening in the GPIO driver that is causing it to fail.
>>
>>
>> ...so we need to dig in further.
> 
> One odd thing is that it looks like the GPIO controller you're
> referencing is disabled.  On today's linuxnext, you can see the
> "disabled":
> 
> arch/arm/boot/dts/socfpga.dtsi:         gpio at ff709000 {
> arch/arm/boot/dts/socfpga.dtsi-                 #address-cells = <1>;
> arch/arm/boot/dts/socfpga.dtsi-                 #size-cells = <0>;
> arch/arm/boot/dts/socfpga.dtsi-                 compatible = "snps,dw-apb-gpio";
> arch/arm/boot/dts/socfpga.dtsi-                 reg = <0xff709000 0x1000>;
> arch/arm/boot/dts/socfpga.dtsi-                 clocks = <&per_base_clk>;
> arch/arm/boot/dts/socfpga.dtsi-                 status = "disabled";
> arch/arm/boot/dts/socfpga.dtsi-
> arch/arm/boot/dts/socfpga.dtsi-                 gpio1: gpio-controller at 0 {
> arch/arm/boot/dts/socfpga.dtsi-                         compatible =
> "snps,dw-apb-gpio-port";
> arch/arm/boot/dts/socfpga.dtsi-                         gpio-controller;
> 
> I don't see anyone else referencing that node to enable it.  ...to me
> that means that you can't use GPIOs on GPIO1 (???).
> 
> 
> ...or did I find something wrong?
> 

Ah yes, yes! Thanks so much for find this Doug!

With the following patch, boots fine with SD driver loading.


--- a/arch/arm/boot/dts/socfpga_cyclone5_socdk.dts
+++ b/arch/arm/boot/dts/socfpga_cyclone5_socdk.dts
@@ -37,6 +37,12 @@
                 */
                ethernet0 = &gmac1;
        };
+
+       soc {
+               gpio at ff709000 {
+                       status = "okay";
+               };
+       };

Thanks,
Dinh