linux-kernel - Re: [PATCH] KVM: nSVM: prepare guest save area while is_guest

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]

Date:   Thu, 18 Feb 2021 09:42:11 -0800
From:   Sean Christopherson <seanjc@...gle.com>
To:     Paolo Bonzini <pbonzini@...hat.com>
Cc:     linux-kernel@...r.kernel.org, kvm@...r.kernel.org, jroedel@...e.de,
        mlevitsk@...hat.com
Subject: Re: [PATCH] KVM: nSVM: prepare guest save area while is_guest_mode
 is true

On Thu, Feb 18, 2021, Paolo Bonzini wrote:
> Right now, enter_svm_guest_mode is calling nested_prepare_vmcb_save and
> nested_prepare_vmcb_control.  This results in is_guest_mode being false
> until the end of nested_prepare_vmcb_control.
> 
> This is a problem because nested_prepare_vmcb_save can in turn cause
> changes to the intercepts and these have to be applied to the "host VMCB"
> (stored in svm->nested.hsave) and then merged with the VMCB12 intercepts
> into svm->vmcb.
> 
> In particular, without this change we forget to set the CR0 read and CR0
> write intercepts when running a real mode L2 guest with NPT disabled.
> The guest is therefore able to see the CR0.PG bit that KVM sets to
> enable "paged real mode".  This patch fixes the svm.flat mode_switch
> test case with npt=0.  There are no other problematic calls in
> nested_prepare_vmcb_save.

It might be worth explicitly pointing out that get_host_vmcb() in
svm_clr_intercept() and svm_set_intercept() will grab the wrong VMCB.
 
> The bug is present since commit 06fc7772690d ("KVM: SVM: Activate nested
> state only when guest state is complete", 2010-04-25).  Unfortunately,
> it is not clear from the commit message what issue exactly led to the
> change back then.  It was probably related to svm_set_cr0 however because
> the patch series cover letter[1] mentioned lazy FPU switching.

Aha!  It was indeed related to svm_set_cr0().  Specifically, the next patch,
commit 66a562f7e257 ("KVM: SVM: Make lazy FPU switching work with nested svm"),
added is_nested() checks in update_cr0_intercept() to merge L1's intercepts with
L0's intercepts.

I dug through all other is_nested() usage, none of them were reachable via the
world switch logic.

> [1] https://lore.kernel.org/kvm/1266493115-28386-1-git-send-email-joerg.roedel@amd.com/
> 
> Signed-off-by: Paolo Bonzini <pbonzini@...hat.com>

Reviewed-by: Sean Christopherson <seanjc@...gle.com>

> ---
>  arch/x86/kvm/svm/nested.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/arch/x86/kvm/svm/nested.c b/arch/x86/kvm/svm/nested.c
> index 92d3aaaac612..35891d9a1099 100644
> --- a/arch/x86/kvm/svm/nested.c
> +++ b/arch/x86/kvm/svm/nested.c
> @@ -469,8 +469,8 @@ int enter_svm_guest_mode(struct vcpu_svm *svm, u64 vmcb12_gpa,
>  
>  	svm->nested.vmcb12_gpa = vmcb12_gpa;
>  	load_nested_vmcb_control(svm, &vmcb12->control);
> -	nested_prepare_vmcb_save(svm, vmcb12);
>  	nested_prepare_vmcb_control(svm);
> +	nested_prepare_vmcb_save(svm, vmcb12);
>  
>  	ret = nested_svm_load_cr3(&svm->vcpu, vmcb12->save.cr3,
>  				  nested_npt_enabled(svm));
> -- 
> 2.26.2
>