Re: [RFC PATCH 5/8] KVM: x86/mmu: Introduce kvm_mmu_map_page() for prepopulating guest memory

From: Isaku Yamahata <isaku.yamahata@intel.com>
To: Sean Christopherson <seanjc@google.com>
Cc: isaku.yamahata@intel.com, kvm@vger.kernel.org,
	isaku.yamahata@gmail.com, linux-kernel@vger.kernel.org,
	Paolo Bonzini <pbonzini@redhat.com>,
	Michael Roth <michael.roth@amd.com>,
	David Matlack <dmatlack@google.com>,
	Federico Parola <federico.parola@polito.it>,
	isaku.yamahata@linux.intel.com
Subject: Re: [RFC PATCH 5/8] KVM: x86/mmu: Introduce kvm_mmu_map_page() for prepopulating guest memory
Date: Mon, 11 Mar 2024 15:57:18 -0700	[thread overview]
Message-ID: <20240311225718.GC935089@ls.amr.corp.intel.com> (raw)
In-Reply-To: <Ze8_XfMN_mZBabKP@google.com>

On Mon, Mar 11, 2024 at 10:29:01AM -0700,
Sean Christopherson <seanjc@google.com> wrote:

> On Fri, Mar 01, 2024, isaku.yamahata@intel.com wrote:
> > From: Isaku Yamahata <isaku.yamahata@intel.com>
> > 
> > Introduce a helper function to call kvm fault handler.  This allows
> > a new ioctl to invoke kvm fault handler to populate without seeing
> > RET_PF_* enums or other KVM MMU internal definitions.
> > 
> > Signed-off-by: Isaku Yamahata <isaku.yamahata@intel.com>
> > ---
> >  arch/x86/kvm/mmu.h     |  3 +++
> >  arch/x86/kvm/mmu/mmu.c | 30 ++++++++++++++++++++++++++++++
> >  2 files changed, 33 insertions(+)
> > 
> > diff --git a/arch/x86/kvm/mmu.h b/arch/x86/kvm/mmu.h
> > index 60f21bb4c27b..48870c5e08ec 100644
> > --- a/arch/x86/kvm/mmu.h
> > +++ b/arch/x86/kvm/mmu.h
> > @@ -183,6 +183,9 @@ static inline void kvm_mmu_refresh_passthrough_bits(struct kvm_vcpu *vcpu,
> >  	__kvm_mmu_refresh_passthrough_bits(vcpu, mmu);
> >  }
> >  
> > +int kvm_mmu_map_page(struct kvm_vcpu *vcpu, gpa_t gpa, u32 error_code,
> > +		     u8 max_level, u8 *goal_level);
> > +
> >  /*
> >   * Check if a given access (described through the I/D, W/R and U/S bits of a
> >   * page fault error code pfec) causes a permission fault with the given PTE
> > diff --git a/arch/x86/kvm/mmu/mmu.c b/arch/x86/kvm/mmu/mmu.c
> > index e4cc7f764980..7d5e80d17977 100644
> > --- a/arch/x86/kvm/mmu/mmu.c
> > +++ b/arch/x86/kvm/mmu/mmu.c
> > @@ -4659,6 +4659,36 @@ int kvm_tdp_page_fault(struct kvm_vcpu *vcpu, struct kvm_page_fault *fault)
> >  	return direct_page_fault(vcpu, fault);
> >  }
> >  
> > +int kvm_mmu_map_page(struct kvm_vcpu *vcpu, gpa_t gpa, u32 error_code,
> > +		     u8 max_level, u8 *goal_level)
> > +{
> > +	struct kvm_page_fault fault = KVM_PAGE_FAULT_INIT(vcpu, gpa, error_code,
> > +							  false, max_level);
> > +	int r;
> > +
> > +	r = __kvm_mmu_do_page_fault(vcpu, &fault);
> > +
> > +	if (is_error_noslot_pfn(fault.pfn) || vcpu->kvm->vm_bugged)
> > +		return -EFAULT;
> 
> This clobbers a non-zero 'r'.  And KVM return -EIO if the VM is bugged/dead, not
> -EFAULT.  I also don't see why KVM needs to explicitly check is_error_noslot_pfn(),
> that should be funneled to RET_PF_EMULATE.

I'll drop this check.

> > +
> > +	switch (r) {
> > +	case RET_PF_RETRY:
> > +		return -EAGAIN;
> > +
> > +	case RET_PF_FIXED:
> > +	case RET_PF_SPURIOUS:
> > +		*goal_level = fault.goal_level;
> > +		return 0;
> > +
> > +	case RET_PF_CONTINUE:
> > +	case RET_PF_EMULATE:
> 
> -EINVAL woud be more appropriate for RET_PF_EMULATE.
> 
> > +	case RET_PF_INVALID:
> 
> CONTINUE and INVALID should be WARN conditions.

Will update them.
-- 
Isaku Yamahata <isaku.yamahata@intel.com>