Re: [PATCH v2 13/17] kvm: arm/arm64: Allow tuning the physical address size for VM

[Suzuki K Poulose]

On 25/04/18 17:10, Julien Grall wrote:
> Hi Suzuki,
>
> On 27/03/18 14:15, Suzuki K Poulose wrote:
> > Allow specifying the physical address size for a new VM via
> > the kvm_type argument for KVM_CREATE_VM ioctl. This allows
> > us to finalise the stage2 page table format as early as possible
> > and hence perform the right checks on the memory slots without
> > complication. The size is encoded as Log2(PA_Siz) in the bits[7:0]
>
> s/PA_Siz/PA_Size/.

> > --- a/include/uapi/linux/kvm.h
> > +++ b/include/uapi/linux/kvm.h
> > @@ -740,6 +740,16 @@ struct kvm_ppc_resize_hpt {
> > Â #define KVM_S390_SIE_PAGE_OFFSET 1
> > Â /*
> > + * On arm/arm64, machine type can be used to request the physical
> > + * address size for the VM. Bits [7-0] have been reserved for the
> > + * PA size shift (i.e, log2(PA-Size)). For backward compatibility,
>
> I would s/PA-Size/PA_Size/ to avoid the impression that it is a substraction.

Agree.

> > + * value 0 implies the default IPA size, which is 40bits.
> > + */
> > +#define KVM_VM_TYPE_ARM_PHYS_SHIFT_MASKÂÂÂ 0xff
> > +#define KVM_VM_TYPE_ARM_PHYS_SHIFT(x)ÂÂÂÂÂÂÂ \
> > +ÂÂÂ ((x) & KVM_VM_TYPE_ARM_PHYS_SHIFT_MASK)
> > +
> > +/*
> > ÂÂ * ioctls for /dev/kvm fds:
> > ÂÂ */
> > Â #define KVM_GET_API_VERSIONÂÂÂÂÂÂ _IO(KVMIO,ÂÂ 0x00)
> > diff --git a/virt/kvm/arm/arm.c b/virt/kvm/arm/arm.c
> > index 53bb05c..13eb66f 100644
> > --- a/virt/kvm/arm/arm.c
> > +++ b/virt/kvm/arm/arm.c
> > @@ -110,6 +110,25 @@ void kvm_arch_check_processor_compat(void *rtn)
> > Â }
> > +static int kvm_arch_config_vm(struct kvm *kvm, unsigned long type)
> > +{
> > +ÂÂÂ u32 ipa_shift = (u32)type & KVM_VM_TYPE_ARM_PHYS_SHIFT_MASK;
>
> How about using KVM_VM_TYPE_ARM_PHYS_SHIFT(type) directly here?

Thanks for that, I have this already fixed in for v3.

> > Â /**
> > ÂÂ * kvm_arch_init_vm - initializes a VM data structure
> > ÂÂ * @kvm:ÂÂÂ pointer to the KVM struct
> > @@ -118,8 +137,9 @@ int kvm_arch_init_vm(struct kvm *kvm, unsigned long type)
> > Â {
> > ÂÂÂÂÂ int ret, cpu;
> > -ÂÂÂ if (type)
> > -ÂÂÂÂÂÂÂ return -EINVAL;
> > +ÂÂÂ ret = kvm_arch_config_vm(kvm, type);
> > +ÂÂÂ if (ret)
> > +ÂÂÂÂÂÂÂ return ret;
> > ÂÂÂÂÂ kvm->arch.last_vcpu_ran = alloc_percpu(typeof(*kvm->arch.last_vcpu_ran));
> > ÂÂÂÂÂ if (!kvm->arch.last_vcpu_ran)
>
> Cheers,

Thanks for the review.

Cheers.