linux-kernel - Re: [PATCH] KVM: x86: allow KVM_REQ_GET_NESTED_STATE

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: <YA8hwsL8SWzWEA0h@google.com>
Date:   Mon, 25 Jan 2021 11:53:38 -0800
From:   Sean Christopherson <seanjc@...gle.com>
To:     Paolo Bonzini <pbonzini@...hat.com>
Cc:     linux-kernel@...r.kernel.org, kvm@...r.kernel.org,
        stable@...r.kernel.org
Subject: Re: [PATCH] KVM: x86: allow KVM_REQ_GET_NESTED_STATE_PAGES outside
 guest mode for VMX

On Mon, Jan 25, 2021, Paolo Bonzini wrote:
> On 25/01/21 20:16, Sean Christopherson wrote:
> > >   }
> > > +static bool vmx_get_nested_state_pages(struct kvm_vcpu *vcpu)
> > > +{
> > > +	if (!nested_get_evmcs_page(vcpu))
> > > +		return false;
> > > +
> > > +	if (is_guest_mode(vcpu) && !nested_get_vmcs12_pages(vcpu))
> > > +		return false;
> > nested_get_evmcs_page() will get called twice in the common case of
> > is_guest_mode() == true.  I can't tell if that will ever be fatal, but it's
> > definitely weird.  Maybe this?
> > 
> > 	if (!is_guest_mode(vcpu))
> > 		return nested_get_evmcs_page(vcpu);
> > 
> > 	return nested_get_vmcs12_pages(vcpu);
> > 
> 
> I wouldn't say there is a common case;

Eh, I would argue that it is more common to do KVM_REQ_GET_NESTED_STATE_PAGES
with is_guest_mode() than it is with !is_guest_mode(), as the latter is valid if
and only if eVMCS is in use.  But, I think we're only vying for internet points. :-)

> however the idea was to remove the call to nested_get_evmcs_page from
> nested_get_vmcs12_pages, since that one is only needed after
> KVM_GET_NESTED_STATE and not during VMLAUNCH/VMRESUME.

I'm confused, this patch explicitly adds a call to nested_get_evmcs_page() in
nested_get_vmcs12_pages().