> On 13 Feb 2018, at 5:37 pm, Christian Mauderer
> <christian.maude...@embedded-brains.de> wrote:
>> Am 13.02.2018 um 00:34 schrieb Chris Johns:
>>> On 13/02/2018 00:39, Christian Mauderer wrote:
>>> These patches add various improvements for the atsam BSP and for the
>>> sc16is752 SPI UART controller.
>>> Regarding patch 1: Note that I have thought about an auto detection
>>> instead of the option. But I had a board where something responded to
>>> the oscillator so that I could initialize it despite the fact that no
>>> crystall has been connected. So I think an auto detection would be
>> I agree with this approach.
>> Auto-detection is OK in systems with users however in embedded systems it is
>> dangerous. I have a rule, any auto-detection requires a system level module
>> audits the part of a system that has auto-detection and raises any unexpected
>> differences as errors. Auto-detection can be hard to avoid in some systems,
>> PC motherboards, PCI etc.
>> I remember back in the mid-90s working on an important system with RAM size
>> detection where one of two static RAM chips had started to fail and the
>> adjusted assuming this was all the available memory so no one noticed until
>> many months into operation the application was reconfigured and it ran out of
>> memory when it should not have. At the start of the project I had been asked
>> add the memory detection to handle different build variants of the hardware.
>> feature was disabled and I have not been a fan of it ever since.
> Hello Chris,
> thanks for the support for this approach. I mostly wanted to give a
> reason why I thought it would be necessary to add another option to the BSP.
> Any other remarks regarding the patches? Can I apply them?
Yes, they look fine. I do not know the hardware so I assume that part is fine.
devel mailing list