[meta-freescale] Device tree question

[Gary Thomas]

On 2015-07-13 09:18, Gary Thomas wrote:
> On 2015-07-13 09:14, Robin Findley wrote:
>> On 07/13/2015 10:37 AM, Gary Thomas wrote:
>>> A bit off topic, but perhaps someone here knows the answer :-)
>>>
>>> If my device tree has a device/element that is enabled, why would
>>> that device be disabled when I boot?  I have this on my (LS1021) board:
>>>           quadspi at 1550000 {
>>>               compatible = "fsl,ls1-qspi";
>>>               #address-cells = <0x00000001>;
>>>               #size-cells = <0x00000000>;
>>>               reg = <0x00000000 0x01550000 0x00000000 0x00010000 0x00000000
>> 0x40000000 0x00000000 0x04000000>;
>>>               reg-names = "QuadSPI", "QuadSPI-memory";
>>>               interrupts = <0x00000000 0x00000083 0x00000004>;
>>>               clock-names = "qspi_en", "qspi";
>>>               clocks = <0x00000003 0x00000001 0x00000003 0x00000001>;
>>>               big-endian;
>>>               amba-base = <0x40000000>;
>>>               num-cs = <0x00000002>;
>>>               status = "okay";
>>>               s70fl01gs at 0 {
>>>                   #address-cells = <0x00000001>;
>>>                   #size-cells = <0x00000001>;
>>>                   compatible = "spansion,s70fl01gs";
>>>                   spi-max-frequency = <0x02faf080>;
>>>                   reg = <0x00000000>;
>>>                   partition at 0 {
>>>                       label = "s70fl01gs-0";
>>>                       reg = <0x00000000 0x04000000>;
>>>                   };
>>>               };
>>>           };
>>>
>>> However when I boot the system, this device is disabled.
>>>     # cat /proc/device-tree/soc/quadspi at 1550000/status
>>>     disabled
>>> I know this must happen very early on as the device driver
>>> for this device is never even probed.
>>>
>>> Any ideas where/why this becomes disabled and how I keep that
>>> from happening?
>>
>>
>> Have you checked the rest of your device tree files (top-level and includes)
>> to see if quadspi is referenced anywhere else?  A later assignment can
>> override an earlier one.
>
> Yes and no, there are no overrides.
>
> Besides, the listing above is from dumping the compiled device
> tree using fdtdump, so includes and multiple sections addressing
> the same element have been collapsed.

I found the culprit!  My quadspi element was being disabled
by U-Boot (not sure why, this was just part of the LS102x common code)

-- 
------------------------------------------------------------
Gary Thomas                 |  Consulting for the
MLB Associates              |    Embedded world
------------------------------------------------------------