Received-SPF: pass (google.com: domain of linux-kernel+bounces-94781-linux.lists.archive=gmail.com@vger.kernel.org designates 2604:1380:45e3:2400::1 as permitted sender) client-ip=2604:1380:45e3:2400::1;
Date: Wed, 6 Mar 2024 16:30:58 -0800
From: David Matlack <dmatlack@google.com>
To: isaku.yamahata@intel.com
Cc: kvm@vger.kernel.org, isaku.yamahata@gmail.com,
	linux-kernel@vger.kernel.org,
	Sean Christopherson <seanjc@google.com>,
	Paolo Bonzini <pbonzini@redhat.com>,
	Michael Roth <michael.roth@amd.com>,
	Federico Parola <federico.parola@polito.it>
Subject: Re: [RFC PATCH 6/8] KVM: x86: Implement kvm_arch_{,
 pre_}vcpu_map_memory()
Message-ID: <ZekKwlLdf6vm5e5u@google.com>
References: <cover.1709288671.git.isaku.yamahata@intel.com>
 <66a957f4ec4a8591d2ff2550686e361ec648b308.1709288671.git.isaku.yamahata@intel.com>
Precedence: bulk
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <66a957f4ec4a8591d2ff2550686e361ec648b308.1709288671.git.isaku.yamahata@intel.com>

On 2024-03-01 09:28 AM, isaku.yamahata@intel.com wrote:
> From: Isaku Yamahata <isaku.yamahata@intel.com>
> 
> Wire KVM_MAP_MEMORY ioctl to kvm_mmu_map_tdp_page() to populate guest
> memory.
> 
> Signed-off-by: Isaku Yamahata <isaku.yamahata@intel.com>
> ---
>  arch/x86/kvm/x86.c | 49 ++++++++++++++++++++++++++++++++++++++++++++++
>  1 file changed, 49 insertions(+)
> 
> diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
> index 3b8cb69b04fa..6025c0e12d89 100644
> --- a/arch/x86/kvm/x86.c
> +++ b/arch/x86/kvm/x86.c
> @@ -4660,6 +4660,7 @@ int kvm_vm_ioctl_check_extension(struct kvm *kvm, long ext)
>  	case KVM_CAP_VM_DISABLE_NX_HUGE_PAGES:
>  	case KVM_CAP_IRQFD_RESAMPLE:
>  	case KVM_CAP_MEMORY_FAULT_INFO:
> +	case KVM_CAP_MAP_MEMORY:
>  		r = 1;
>  		break;
>  	case KVM_CAP_EXIT_HYPERCALL:
> @@ -5805,6 +5806,54 @@ static int kvm_vcpu_ioctl_enable_cap(struct kvm_vcpu *vcpu,
>  	}
>  }
>  
> +int kvm_arch_vcpu_pre_map_memory(struct kvm_vcpu *vcpu)
> +{
> +	return kvm_mmu_reload(vcpu);
> +}

Why is the here and not kvm_arch_vcpu_map_memory()?

> +
> +int kvm_arch_vcpu_map_memory(struct kvm_vcpu *vcpu,
> +			     struct kvm_memory_mapping *mapping)
> +{
> +	u8 max_level, goal_level = PG_LEVEL_4K;
> +	u32 error_code;
> +	int r;
> +
> +	error_code = 0;
> +	if (mapping->flags & KVM_MEMORY_MAPPING_FLAG_WRITE)
> +		error_code |= PFERR_WRITE_MASK;
> +	if (mapping->flags & KVM_MEMORY_MAPPING_FLAG_EXEC)
> +		error_code |= PFERR_FETCH_MASK;
> +	if (mapping->flags & KVM_MEMORY_MAPPING_FLAG_USER)
> +		error_code |= PFERR_USER_MASK;
> +	if (mapping->flags & KVM_MEMORY_MAPPING_FLAG_PRIVATE) {
> +#ifdef PFERR_PRIVATE_ACCESS
> +		error_code |= PFERR_PRIVATE_ACCESS;
> +#else
> +		return -OPNOTSUPP;

-EOPNOTSUPP

> +#endif
> +	}
> +
> +	if (IS_ALIGNED(mapping->base_gfn, KVM_PAGES_PER_HPAGE(PG_LEVEL_1G)) &&
> +	    mapping->nr_pages >= KVM_PAGES_PER_HPAGE(PG_LEVEL_1G))
> +		max_level = PG_LEVEL_1G;
> +	else if (IS_ALIGNED(mapping->base_gfn, KVM_PAGES_PER_HPAGE(PG_LEVEL_2M)) &&
> +		 mapping->nr_pages >= KVM_PAGES_PER_HPAGE(PG_LEVEL_2M))
> +		max_level = PG_LEVEL_2M;
> +	else
> +		max_level = PG_LEVEL_4K;

Is there a requirement that KVM must not map memory outside of the
requested region?

> +
> +	r = kvm_mmu_map_page(vcpu, gfn_to_gpa(mapping->base_gfn), error_code,
> +			     max_level, &goal_level);
> +	if (r)
> +		return r;
> +
> +	if (mapping->source)
> +		mapping->source += KVM_HPAGE_SIZE(goal_level);
> +	mapping->base_gfn += KVM_PAGES_PER_HPAGE(goal_level);
> +	mapping->nr_pages -= KVM_PAGES_PER_HPAGE(goal_level);
> +	return r;
> +}
> +
>  long kvm_arch_vcpu_ioctl(struct file *filp,
>  			 unsigned int ioctl, unsigned long arg)
>  {
> -- 
> 2.25.1
>