Re: [PATCH v9 41/43] KVM: arm64: Expose support for private memory

[Steven Price <steven.price@arm.com>]

On 12/06/2025 16:14, Joey Gouly wrote:
> Hi Steven,
> 
> On Wed, Jun 11, 2025 at 11:48:38AM +0100, Steven Price wrote:
>> Select KVM_GENERIC_PRIVATE_MEM and provide the necessary support
>> functions.
>>
>> Signed-off-by: Steven Price <steven.price@arm.com>
>> Reviewed-by: Gavin Shan <gshan@redhat.com>
>> ---
>> Changes since v2:
>>  * Switch kvm_arch_has_private_mem() to a macro to avoid overhead of a
>>    function call.
>>  * Guard definitions of kvm_arch_{pre,post}_set_memory_attributes() with
>>    #ifdef CONFIG_KVM_GENERIC_MEMORY_ATTRIBUTES.
>>  * Early out in kvm_arch_post_set_memory_attributes() if the WARN_ON
>>    should trigger.
>> ---
>>  arch/arm64/include/asm/kvm_host.h |  6 ++++++
>>  arch/arm64/kvm/Kconfig            |  1 +
>>  arch/arm64/kvm/mmu.c              | 24 ++++++++++++++++++++++++
>>  3 files changed, 31 insertions(+)
>>
>> diff --git a/arch/arm64/include/asm/kvm_host.h b/arch/arm64/include/asm/kvm_host.h
>> index a1857802db64..9903b0e8ef3f 100644
>> --- a/arch/arm64/include/asm/kvm_host.h
>> +++ b/arch/arm64/include/asm/kvm_host.h
>> @@ -1514,6 +1514,12 @@ struct kvm *kvm_arch_alloc_vm(void);
>>  
>>  #define vcpu_is_protected(vcpu)		kvm_vm_is_protected((vcpu)->kvm)
>>  
>> +#ifdef CONFIG_KVM_PRIVATE_MEM
>> +#define kvm_arch_has_private_mem(kvm) ((kvm)->arch.is_realm)
>> +#else
>> +#define kvm_arch_has_private_mem(kvm) false
>> +#endif
> 
> I don't understand the ifdef here (or below). In the Kconfig you 'select
> KVM_GENERIC_PRIVATE_MEM', so it will always be on/defined? Unless I'm
> misunderstanding something.

I have to admit this is somewhat cargo-culted from x86. And I think they
have more build configurations which don't include KVM_GENERIC_PRIVATE_MEM.

It is possible to build without KVM_GENERIC_PRIVATE_MEM (by disabling
CONFIG_KVM). But that's probably not very interesting here - the
definition shouldn't actually matter in that case.

So I think you're right - there's no need for the #ifdeffery here.

Thanks,
Steve

>> +
>>  int kvm_arm_vcpu_finalize(struct kvm_vcpu *vcpu, int feature);
>>  bool kvm_arm_vcpu_is_finalized(struct kvm_vcpu *vcpu);
>>  
>> diff --git a/arch/arm64/kvm/Kconfig b/arch/arm64/kvm/Kconfig
>> index 713248f240e0..3a04b040869d 100644
>> --- a/arch/arm64/kvm/Kconfig
>> +++ b/arch/arm64/kvm/Kconfig
>> @@ -37,6 +37,7 @@ menuconfig KVM
>>  	select HAVE_KVM_VCPU_RUN_PID_CHANGE
>>  	select SCHED_INFO
>>  	select GUEST_PERF_EVENTS if PERF_EVENTS
>> +	select KVM_GENERIC_PRIVATE_MEM
>>  	help
>>  	  Support hosting virtualized guest machines.
>>  
>> diff --git a/arch/arm64/kvm/mmu.c b/arch/arm64/kvm/mmu.c
>> index 580ed362833c..c866891fd8f9 100644
>> --- a/arch/arm64/kvm/mmu.c
>> +++ b/arch/arm64/kvm/mmu.c
>> @@ -2384,6 +2384,30 @@ int kvm_arch_prepare_memory_region(struct kvm *kvm,
>>  	return ret;
>>  }
>>  
>> +#ifdef CONFIG_KVM_GENERIC_MEMORY_ATTRIBUTES
>> +bool kvm_arch_pre_set_memory_attributes(struct kvm *kvm,
>> +					struct kvm_gfn_range *range)
>> +{
>> +	WARN_ON_ONCE(!kvm_arch_has_private_mem(kvm));
>> +	return false;
>> +}
>> +
>> +bool kvm_arch_post_set_memory_attributes(struct kvm *kvm,
>> +					 struct kvm_gfn_range *range)
>> +{
>> +	if (WARN_ON_ONCE(!kvm_arch_has_private_mem(kvm)))
>> +		return false;
>> +
>> +	if (range->arg.attributes & KVM_MEMORY_ATTRIBUTE_PRIVATE)
>> +		range->attr_filter = KVM_FILTER_SHARED;
>> +	else
>> +		range->attr_filter = KVM_FILTER_PRIVATE;
>> +	kvm_unmap_gfn_range(kvm, range);
>> +
>> +	return false;
>> +}
>> +#endif
>> +
>>  void kvm_arch_free_memslot(struct kvm *kvm, struct kvm_memory_slot *slot)
>>  {
>>  }
>> -- 
>> 2.43.0
>>
> 
> Thanks,
> Joey