Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:18 as permitted sender) client-ip=2620:137:e000::1:18;
Sender: Paolo Bonzini <paolo.bonzini@gmail.com>
Message-ID: <2d2a9789-083c-7e58-4628-8ebd024bee1f@redhat.com>
Date:   Fri, 27 May 2022 19:38:57 +0200
MIME-Version: 1.0
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:91.0) Gecko/20100101
 Thunderbird/91.8.0
Subject: Re: [RFC PATCH v6 047/104] KVM: x86/tdp_mmu: Support TDX private
 mapping for TDP MMU
Content-Language: en-US
To:     isaku.yamahata@intel.com, kvm@vger.kernel.org,
        linux-kernel@vger.kernel.org
Cc:     isaku.yamahata@gmail.com, erdemaktas@google.com,
        Sean Christopherson <seanjc@google.com>,
        Sagi Shahar <sagis@google.com>
References: <cover.1651774250.git.isaku.yamahata@intel.com>
 <653230043fdb2d20e871e79e73f757134ca92eeb.1651774250.git.isaku.yamahata@intel.com>
From:   Paolo Bonzini <pbonzini@redhat.com>
In-Reply-To: <653230043fdb2d20e871e79e73f757134ca92eeb.1651774250.git.isaku.yamahata@intel.com>
Content-Type: text/plain; charset=UTF-8; format=flowed
Content-Transfer-Encoding: 7bit
Precedence: bulk

On 5/5/22 20:14, isaku.yamahata@intel.com wrote:
> +/*
> + * Private page can't be release on mmu_notifier without losing page contents.
> + * The help, callback, from backing store is needed to allow page migration.
> + * For now, pin the page.
> + */
> +static bool kvm_faultin_pfn_private(struct kvm_vcpu *vcpu,
> +				    struct kvm_page_fault *fault, int *r)
> +{
> +	hva_t hva = gfn_to_hva_memslot(fault->slot, fault->gfn);
> +	struct page *page[1];
> +	unsigned int flags;
> +	int npages;
> +
> +	fault->map_writable = false;
> +	fault->pfn = KVM_PFN_ERR_FAULT;
> +	*r = -1;
> +	if (hva == KVM_HVA_ERR_RO_BAD || hva == KVM_HVA_ERR_BAD)
> +		return true;
> +
> +	/* TDX allows only RWX.  Read-only isn't supported. */
> +	WARN_ON_ONCE(!fault->write);
> +	flags = FOLL_WRITE | FOLL_LONGTERM;
> +
> +	npages = pin_user_pages_fast(hva, 1, flags, page);
> +	if (npages != 1)
> +		return true;
> +
> +	fault->map_writable = true;
> +	fault->pfn = page_to_pfn(page[0]);
> +	return false;
> +}
> +

This function is present also in the memfd notifier series.  For the 
next postings, can you remove all the KVM bits from the series and 
include it yourself?

Thanks,

Paolo