Re: [PATCH] KVM: PPC: Update MAINTAINERS

[Michael Ellerman <mpe@ellerman.id.au>]

Sean Christopherson <seanjc@google.com> writes:
> On Thu, Jun 08, 2023, Nicholas Piggin wrote:
>> Michael is merging KVM PPC patches via the powerpc tree and KVM topic
>> branches. He doesn't necessarily have time to be across all of KVM so
>> is reluctant to call himself maintainer, but for the mechanics of how
>> patches flow upstream, it is maintained and does make sense to have
>> some contact people in MAINTAINERS.
>> 
>> So add Michael Ellerman as KVM PPC maintainer and myself as reviewer.
>> Split out the subarchs that don't get so much attention.
>> 
>> Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
>> ---
>
> Thanks for documenting the reality of things, much appreciated!
>
> Acked-by: Sean Christopherson <seanjc@google.com>
>
>>  MAINTAINERS | 6 ++++++
>>  1 file changed, 6 insertions(+)
>> 
>> diff --git a/MAINTAINERS b/MAINTAINERS
>> index 0dab9737ec16..44417acd2936 100644
>> --- a/MAINTAINERS
>> +++ b/MAINTAINERS
>> @@ -11379,7 +11379,13 @@ F:	arch/mips/include/uapi/asm/kvm*
>>  F:	arch/mips/kvm/
>>  
>>  KERNEL VIRTUAL MACHINE FOR POWERPC (KVM/powerpc)
>> +M:	Michael Ellerman <mpe@ellerman.id.au>
>> +R:	Nicholas Piggin <npiggin@gmail.com>
>>  L:	linuxppc-dev@lists.ozlabs.org
>> +L:	kvm@vger.kernel.org
>> +S:	Maintained (Book3S 64-bit HV)
>> +S:	Odd fixes (Book3S 64-bit PR)
>> +S:	Orphan (Book3E and 32-bit)
>
> Do you think there's any chance of dropping support for everything except Book3S
> 64-bit HV at some point soonish?

Nick proposed disabling BookE KVM, which prompted some users to report
they are still actively using it:

  https://lore.kernel.org/all/20221128043623.1745708-1-npiggin@gmail.com/

There are also still some KVM PR users.

In total I'd guess it's only some small 100s of users, but we don't
really know.

> There haven't been many generic KVM changes that touch PPC, but in my
> experience when such series do come along, the many flavors and layers
> of PPC incur quite a bit of development and testing cost, and have a
> high chance of being broken compared to other architectures.

Ack.

cheers