Re: [PATCH 5/6] KVM: s390: wire up seb feature

[Christian Borntraeger <borntraeger@de.ibm.com>]

On 01/17/2018 12:28 PM, Christian Borntraeger wrote:
> 
> 
> On 01/17/2018 12:22 PM, Paolo Bonzini wrote:
>>> while this is kvm code, my current plan is to submit the "final"
>>> version after review and probably some fixes/renames via Martin
>>> together with the other patches.  Are you ok with that? Right now it
>>> seems that the CAP number is still fine.
>> Sure, though there will be a capability introduced by PPC for similar
>> purposes, so check for conflicts.
>>
>> On 17/01/2018 12:18, Christian Borntraeger wrote:
>>> diff --git a/arch/s390/kvm/kvm-s390.c b/arch/s390/kvm/kvm-s390.c
>>> index 2c93cbb..0c18f73 100644
>>> --- a/arch/s390/kvm/kvm-s390.c
>>> +++ b/arch/s390/kvm/kvm-s390.c
>>> @@ -421,6 +421,9 @@ int kvm_vm_ioctl_check_extension(struct kvm *kvm, long ext)
>>>  	case KVM_CAP_S390_GS:
>>>  		r = test_facility(133);
>>>  		break;
>>> +	case KVM_CAP_S390_SEB:
>>> +		r = test_facility(82);
>>> +		break;
>>>  	default:
>>>  		r = 0;
>>
>> Can you add a generic "test facility" capability and ioctl?
> 
> The problem is not that I announce the facility, I in fact announce that the
> programmatic interface is available (the sebc sync reg and the usage of that field).
> (So the CAP is part of this patch to have both in lockstep)
> A non-existing facility will then just disable that programmatic interface.

To put it differently. CAP_S390_GS and CAP_S390_SEB could also just
do a 

return 1;

and the QEMU has to check both (which it probably does anyway)

Christian