linux-kernel - Re: [PATCH] KVM: VMX: Don't register posted interrupt wakeup handler if alloc_kvm

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: <20260114031139.GA107826@k08j02272.eu95sqa>
Date: Wed, 14 Jan 2026 11:11:39 +0800
From: Hou Wenlong <houwenlong.hwl@...group.com>
To: Sean Christopherson <seanjc@...gle.com>
Cc: kvm@...r.kernel.org, Paolo Bonzini <pbonzini@...hat.com>,
	Thomas Gleixner <tglx@...utronix.de>,
	Ingo Molnar <mingo@...hat.com>, Borislav Petkov <bp@...en8.de>,
	Dave Hansen <dave.hansen@...ux.intel.com>, x86@...nel.org,
	"H. Peter Anvin" <hpa@...or.com>, linux-kernel@...r.kernel.org
Subject: Re: [PATCH] KVM: VMX: Don't register posted interrupt wakeup handler
 if alloc_kvm_area() fails

On Tue, Jan 13, 2026 at 08:17:23AM -0800, Sean Christopherson wrote:
> On Tue, Jan 13, 2026, Hou Wenlong wrote:
> > Unregistering the posted interrupt wakeup handler only happens during
> > hardware unsetup. Therefore, if alloc_kvm_area() fails and continue to
> > register the posted interrupt wakeup handler, this will leave the global
> > posted interrupt wakeup handler pointer in an incorrect state. Although
> > it should not be an issue, it's still better to change it.
> 
> Ouch, yeah, that's ugly.  It's not entirely benign, as a failed allocation followed
> by a spurious notification vector IRQ would trigger UAF.  So it's probably worth
> adding:
> 
>   Fixes: ec5a4919fa7b ("KVM: VMX: Unregister posted interrupt wakeup handler on hardware unsetup")
>   Cc: stable@...r.kernel.org
>
Actually, I'm not sure which commit is better as the fix tag:
'bf9f6ac8d749' or 'ec5a4919fa7b'. Before commit 'ec5a4919fa7b', the
handler was registered before alloc_kvm_areas() and was not unregistered
if alloc_kvm_areas() failed. However, it seems my commit message
description is more suitable for fixing 'ec5a4919fa7b'.

> even though I agree it's extremely unlikely to be an issue in practice.
> 
> > Signed-off-by: Hou Wenlong <houwenlong.hwl@...group.com>
> > ---
> >  arch/x86/kvm/vmx/vmx.c | 7 +++++--
> >  1 file changed, 5 insertions(+), 2 deletions(-)
> > 
> > diff --git a/arch/x86/kvm/vmx/vmx.c b/arch/x86/kvm/vmx/vmx.c
> > index 9b92f672ccfe..676f32aa72bb 100644
> > --- a/arch/x86/kvm/vmx/vmx.c
> > +++ b/arch/x86/kvm/vmx/vmx.c
> > @@ -8829,8 +8829,11 @@ __init int vmx_hardware_setup(void)
> >  	}
> >  
> >  	r = alloc_kvm_area();
> > -	if (r && nested)
> > -		nested_vmx_hardware_unsetup();
> > +	if (r) {
> > +		if (nested)
> > +			nested_vmx_hardware_unsetup();
> > +		return r;
> > +	}
> 
> I'm leaning towards using a goto with an explicit "return 0" in the happy case,
> to make it less likely that a similar bug is introduced in the future.  Any
> preference on your end?
> 
I don't have a strong preference either way. However, I agree that using
a goto statement could help prevent potential bugs in the future. Do I
need to send a v2?

Thanks!

> E.g. (untested)
> 
> diff --git a/arch/x86/kvm/vmx/vmx.c b/arch/x86/kvm/vmx/vmx.c
> index 9b92f672ccfe..cecaaeb3f82a 100644
> --- a/arch/x86/kvm/vmx/vmx.c
> +++ b/arch/x86/kvm/vmx/vmx.c
> @@ -8829,8 +8829,8 @@ __init int vmx_hardware_setup(void)
>         }
>  
>         r = alloc_kvm_area();
> -       if (r && nested)
> -               nested_vmx_hardware_unsetup();
> +       if (r)
> +               goto err_kvm_area;
>  
>         kvm_set_posted_intr_wakeup_handler(pi_wakeup_handler);
>  
> @@ -8857,6 +8857,11 @@ __init int vmx_hardware_setup(void)
>  
>         kvm_caps.inapplicable_quirks &= ~KVM_X86_QUIRK_IGNORE_GUEST_PAT;
>  
> +       return 0;
> +
> +err_kvm_area:
> +       if (nested)
> +               nested_vmx_hardware_unsetup();
>         return r;
>  }