[MinnowBoard] Need to disable both UART on LSE?

[Darren Hart]

On 4/29/15, 3:11 PM, "Lucas De Marchi" <lucas.de.marchi at gmail.com> wrote:

>On Wed, Apr 29, 2015 at 6:12 PM, Darren Hart <dvhart at linux.intel.com>
>wrote:
>> On 4/29/15, 1:35 PM, "Lucas De Marchi" <lucas.de.marchi at gmail.com>
>>wrote:
>>
>>>Hi John,
>>>
>>>On Wed, Apr 29, 2015 at 3:25 PM, John Hawley <john.hawley at intel.com>
>>>wrote:
>>>> I assume all you are trying to do is flip the UARTs over to GPIO, this
>>>> can be done in the bios configuration menu here:
>>>>
>>>> http://elinux.org/Minnowboard:MaxBios#LPSS_.26_SCC_Configuration
>>>>
>>>> you'll notice
>>>>
>>>> * LPSS 1 Configuration
>>>> ** LPSS HSUART #1 Support
>>>> ** LPSS HSUART #1 FlowCtrl
>>>> ** LPSS HSUART #2 Support
>>>>
>>>> Just flip them to disabled and the GPIO pins should work as you are
>>>> expecting.
>>>
>>>Yeah I saw that works but I'd like to use the other UART.  Do I really
>>>need to disable both?
>>>
>>>Ideally I could even use UART1 if I just disabled FlowCtrl.
>>
>> Correct. "Support" means the RX/TX signals are used for the UART,
>>Disabled
>> means GPIO.
>> FlowCTRL refers to the CTS/RTS signals.
>
>The thing I'm trying to say is that doing this doesn't work. I already
>flipped the config in the bios and even then I can't use the GPIO.
>Tested: kernel 3.19.5 with firmware 0.71, 0.77 and 0.79
>
>So, I decided to check if it's a kernel regression, and it is... in
>3.14.4  I can successfully export the pin as GPIO.  I'll try to bisect
>this and check the offending commit. Tomorrow. If it's known to work
>for a more recent kernel version, bisecting it would be much easier.

Ah, great catch. That looks all the world like a firmware menu bug... But
your data suggests not.

Can you send the output of /sys/kernel/debug/gpio? That will provide the
mux state for the relevant lines.


-- 
Darren Hart
Intel Open Source Technology Center