[PATCH rtems-libbsd] rtemsbsd: Use a separate header for test devices

[Chris Johns]

On 9/7/21 3:39 pm, Sebastian Huber wrote:
> On 08/07/2021 03:11, Kinsey Moore wrote:
>> I wonder why this never came up with Zynq or QorIQ. Maybe no one wanted to run
>> network tests on the alternate interfaces because dev boards with those
>> interfaces configured didn't exist? It's possible that the ukphy driver could
>> be improved and this entire problem just goes away or we ban that driver from
>> the default configuration for multi-interface BSPs and the problem goes away.
> 
> The QorIQ BSP uses a device tree.
> 
> The chips will get more and more complex. Managing this complexity with hand
> written C preprocessor defines is a dead end from my point of view. Device trees
> allow you to provide a generic BSP which is initialized for a particular board
> using the device tree. Compared to other operating systems, the device tree
> support and generic device enumeration in RTEMS is a bit under developed. See
> for example:
> 
> https://docs.zephyrproject.org/latest/guides/dts/index.html#dt-guide
> 

I agree and this is well said. I am concerned an ad-hock implementation leaves
us with fragmented and hard to maintain support at the BSP level.

Chris