linux-kernel - Re: [PATCH v2 2/3] KVM: nSVM: properly call kvm_mmu_new

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives
Hash Suite: Windows password security audit tool. GUI, reports in PDF.
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <b7989497-562e-c9a1-3f62-dd5afb9fd3d5@redhat.com>
Date:   Wed, 8 Jul 2020 13:49:55 +0200
From:   Paolo Bonzini <pbonzini@...hat.com>
To:     Vitaly Kuznetsov <vkuznets@...hat.com>, kvm@...r.kernel.org
Cc:     Sean Christopherson <sean.j.christopherson@...el.com>,
        Wanpeng Li <wanpengli@...cent.com>,
        Jim Mattson <jmattson@...gle.com>,
        Junaid Shahid <junaids@...gle.com>,
        linux-kernel@...r.kernel.org
Subject: Re: [PATCH v2 2/3] KVM: nSVM: properly call kvm_mmu_new_pgd() upon
 switching to guest

On 08/07/20 11:36, Vitaly Kuznetsov wrote:
> Undesired triple fault gets injected to L1 guest on SVM when L2 is
> launched with certain CR3 values. #TF is raised by mmu_check_root()
> check in fast_pgd_switch() and the root cause is that when
> kvm_set_cr3() is called from nested_prepare_vmcb_save() with NPT
> enabled CR3 points to a nGPA so we can't check it with
> kvm_is_visible_gfn().
> 
> Calling kvm_mmu_new_pgd() with L2's CR3 idea when NPT is in use
> seems to be wrong, an acceptable place for it seems to be
> kvm_init_shadow_npt_mmu(). This also matches nVMX code.
> 
> Fixes: 7c390d350f8b ("kvm: x86: Add fast CR3 switch code path")
> Signed-off-by: Vitaly Kuznetsov <vkuznets@...hat.com>
> ---
>  arch/x86/include/asm/kvm_host.h | 7 ++++++-
>  arch/x86/kvm/mmu/mmu.c          | 2 ++
>  arch/x86/kvm/svm/nested.c       | 2 +-
>  arch/x86/kvm/x86.c              | 8 +++++---
>  4 files changed, 14 insertions(+), 5 deletions(-)
> 
> diff --git a/arch/x86/include/asm/kvm_host.h b/arch/x86/include/asm/kvm_host.h
> index be5363b21540..49b62f024f51 100644
> --- a/arch/x86/include/asm/kvm_host.h
> +++ b/arch/x86/include/asm/kvm_host.h
> @@ -1459,7 +1459,12 @@ int kvm_task_switch(struct kvm_vcpu *vcpu, u16 tss_selector, int idt_index,
>  		    int reason, bool has_error_code, u32 error_code);
>  
>  int kvm_set_cr0(struct kvm_vcpu *vcpu, unsigned long cr0);
> -int kvm_set_cr3(struct kvm_vcpu *vcpu, unsigned long cr3);
> +int __kvm_set_cr3(struct kvm_vcpu *vcpu, unsigned long cr3, bool cr3_is_nested);
> +static inline int kvm_set_cr3(struct kvm_vcpu *vcpu, unsigned long cr3)
> +{
> +	return __kvm_set_cr3(vcpu, cr3, false);
> +}
> +
>  int kvm_set_cr4(struct kvm_vcpu *vcpu, unsigned long cr4);
>  int kvm_set_cr8(struct kvm_vcpu *vcpu, unsigned long cr8);
>  int kvm_set_dr(struct kvm_vcpu *vcpu, int dr, unsigned long val);
> diff --git a/arch/x86/kvm/mmu/mmu.c b/arch/x86/kvm/mmu/mmu.c
> index 167d12ab957a..ebf0cb3f1ce0 100644
> --- a/arch/x86/kvm/mmu/mmu.c
> +++ b/arch/x86/kvm/mmu/mmu.c
> @@ -4987,6 +4987,8 @@ void kvm_init_shadow_npt_mmu(struct kvm_vcpu *vcpu, u32 cr0, u32 cr4, u32 efer,
>  	union kvm_mmu_role new_role =
>  		kvm_calc_shadow_mmu_root_page_role(vcpu, false);
>  
> +	__kvm_mmu_new_pgd(vcpu, nested_cr3, new_role.base, true, true);
> +
>  	if (new_role.as_u64 != context->mmu_role.as_u64)
>  		shadow_mmu_init_context(vcpu, cr0, cr4, efer, new_role);
>  }
> diff --git a/arch/x86/kvm/svm/nested.c b/arch/x86/kvm/svm/nested.c
> index e424bce13e6c..b467917a9784 100644
> --- a/arch/x86/kvm/svm/nested.c
> +++ b/arch/x86/kvm/svm/nested.c
> @@ -324,7 +324,7 @@ static void nested_prepare_vmcb_save(struct vcpu_svm *svm, struct vmcb *nested_v
>  	svm_set_efer(&svm->vcpu, nested_vmcb->save.efer);
>  	svm_set_cr0(&svm->vcpu, nested_vmcb->save.cr0);
>  	svm_set_cr4(&svm->vcpu, nested_vmcb->save.cr4);
> -	(void)kvm_set_cr3(&svm->vcpu, nested_vmcb->save.cr3);
> +	(void)__kvm_set_cr3(&svm->vcpu, nested_vmcb->save.cr3, npt_enabled);
>  
>  	svm->vmcb->save.cr2 = svm->vcpu.arch.cr2 = nested_vmcb->save.cr2;
>  	kvm_rax_write(&svm->vcpu, nested_vmcb->save.rax);
> diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
> index 3b92db412335..3761135eb052 100644
> --- a/arch/x86/kvm/x86.c
> +++ b/arch/x86/kvm/x86.c
> @@ -1004,7 +1004,7 @@ int kvm_set_cr4(struct kvm_vcpu *vcpu, unsigned long cr4)
>  }
>  EXPORT_SYMBOL_GPL(kvm_set_cr4);
>  
> -int kvm_set_cr3(struct kvm_vcpu *vcpu, unsigned long cr3)
> +int __kvm_set_cr3(struct kvm_vcpu *vcpu, unsigned long cr3, bool cr3_is_nested)
>  {
>  	bool skip_tlb_flush = false;
>  #ifdef CONFIG_X86_64
> @@ -1031,13 +1031,15 @@ int kvm_set_cr3(struct kvm_vcpu *vcpu, unsigned long cr3)
>  		 !load_pdptrs(vcpu, vcpu->arch.walk_mmu, cr3))
>  		return 1;
>  
> -	kvm_mmu_new_pgd(vcpu, cr3, skip_tlb_flush, skip_tlb_flush);
> +	if (!cr3_is_nested)
> +		kvm_mmu_new_pgd(vcpu, cr3, skip_tlb_flush, skip_tlb_flush);
> +
>  	vcpu->arch.cr3 = cr3;
>  	kvm_register_mark_available(vcpu, VCPU_EXREG_CR3);
>  
>  	return 0;
>  }
> -EXPORT_SYMBOL_GPL(kvm_set_cr3);
> +EXPORT_SYMBOL_GPL(__kvm_set_cr3);
>  
>  int kvm_set_cr8(struct kvm_vcpu *vcpu, unsigned long cr8)
>  {
> 

Instead of the new argument (which is not really named right since it's
never true for !NPT) you could perhaps check vcpu->arch.mmu.  But also,
for NPT=1 the kvm_mmu_new_pgd is also unnecessary on vmexit, because the
old roots are still valid (or has been invalidated otherwise) while L2
was running.

I'm also not sure if skip_tlb_flush can use X86_CR3_PCID_NOFLUSH the way
kvm_set_cr3 does, so I wouldn't mind duplicating the code completely as
is already the case for nested_vmx_load_cr3.  It would introduce some
code duplication, but overall the code would be better.  For now, there
need not be an equivalent to nested_vmx_transition_mmu_sync, ASID
handling can be left for later.

Paolo