Issues with the first PowerPC updates for the kernel 6.1

[Michael Ellerman]

Olof Johansson <olof at lixom.net> writes:
> On Fri, Oct 14, 2022 at 9:11 AM Christian Zigotzky
> <chzigotzky at xenosoft.de> wrote:
>>
>> make oldconfig has asked because of the CPU family. I choosed GENERIC for my P.A. Semi PWRficient PA6T-1682M. Is this correct? Maybe this is the problem.
>>
>> config GENERIC_CPU
>> - bool "Generic (POWER4 and above)"
>> + bool "Generic (POWER5 and PowerPC 970 and above)"
>>   depends on PPC_BOOK3S_64 && !CPU_LITTLE_ENDIAN
>>   select PPC_64S_HASH_MMU
>>
>> There isn’t a POWER4 anymore and I used it via CONFIG_GENERIC_CPU=y before.
>>
>> Before the first PowerPC updates:
>> CONFIG_GENERIC_CPU=y
>> # CONFIG_POWER5_CPU is not set
>>
>> Link: https://github.com/torvalds/linux/blob/master/arch/powerpc/platforms/Kconfig.cputype
>
> I think the mistake here is that there are two GENERIC_CPUs now, and
> one of them is the wrong one and overrides settings. I'm surprised
> Kconfig didn't balk at it.

We've had two GENERIC_CPUs since 2018:

  a73657ea19ae ("powerpc/64: Add GENERIC_CPU support for little endian")

They're mutually exclusive because of the ENDIAN dependencies.

I don't know if that's supposed to be allowed by Kconfig, but AFAIK it
has been working.

> I have a vague memory that PA6T had a little-endian mode, but it was
> never used much and I wouldn't expect it to work.

I don't think endian is an issue here. Christian should be getting the
big endian version of GENERIC_CPU.

cheers