[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <202301121441.1E38EE308@keescook>
Date: Thu, 12 Jan 2023 14:42:56 -0800
From: Kees Cook <keescook@...omium.org>
To: Sean Christopherson <seanjc@...gle.com>
Cc: 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>,
"Gustavo A. R. Silva" <gustavoars@...nel.org>, x86@...nel.org,
"H. Peter Anvin" <hpa@...or.com>, kvm@...r.kernel.org,
linux-kernel@...r.kernel.org, linux-hardening@...r.kernel.org
Subject: Re: [PATCH] KVM: x86: Replace 0-length arrays with flexible arrays
On Mon, Jan 09, 2023 at 05:30:48PM +0000, Sean Christopherson wrote:
> On Thu, Jan 05, 2023, Kees Cook wrote:
> > Zero-length arrays are deprecated[1]. Replace struct kvm_nested_state's
> > "data" union 0-length arrays with flexible arrays. (How are the
> > sizes of these arrays verified?)
>
> It's not really interpreted as an array, it's a mandatory single-entry "array".
>
> if (copy_from_user(vmcs12, user_vmx_nested_state->vmcs12, sizeof(*vmcs12)))
> return -EFAULT;
If it's mandatory, why is it [0] instead of just a single struct? i.e.
why is it not:
union {
struct kvm_vmx_nested_state_data vmx;
struct kvm_svm_nested_state_data svm;
};
>
> > Detected with GCC 13, using -fstrict-flex-arrays=3:
> >
> > arch/x86/kvm/svm/nested.c: In function 'svm_get_nested_state':
> > arch/x86/kvm/svm/nested.c:1536:17: error: array subscript 0 is outside array bounds of 'struct kvm_svm_nested_state_data[0]' [-Werror=array-bounds=]
> > 1536 | &user_kvm_nested_state->data.svm[0];
> > | ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
> > In file included from include/uapi/linux/kvm.h:15,
> > from include/linux/kvm_host.h:40,
> > from arch/x86/kvm/svm/nested.c:18:
> > arch/x86/include/uapi/asm/kvm.h:511:50: note: while referencing 'svm'
> > 511 | struct kvm_svm_nested_state_data svm[0];
> > | ^~~
> >
> > [1] https://www.kernel.org/doc/html/latest/process/deprecated.html#zero-length-and-one-element-arrays
> >
> > Cc: Sean Christopherson <seanjc@...gle.com>
> > Cc: Paolo Bonzini <pbonzini@...hat.com>
> > Cc: Thomas Gleixner <tglx@...utronix.de>
> > Cc: Ingo Molnar <mingo@...hat.com>
> > Cc: Borislav Petkov <bp@...en8.de>
> > Cc: Dave Hansen <dave.hansen@...ux.intel.com>
> > Cc: "Gustavo A. R. Silva" <gustavoars@...nel.org>
> > Cc: x86@...nel.org
> > Cc: "H. Peter Anvin" <hpa@...or.com>
> > Cc: kvm@...r.kernel.org
> > Signed-off-by: Kees Cook <keescook@...omium.org>
> > ---
>
> Nit on the comment aside,
>
> Reviewed-by: Sean Christopherson <seanjc@...gle.com>
>
> > arch/x86/include/uapi/asm/kvm.h | 6 +++---
> > 1 file changed, 3 insertions(+), 3 deletions(-)
> >
> > diff --git a/arch/x86/include/uapi/asm/kvm.h b/arch/x86/include/uapi/asm/kvm.h
> > index e48deab8901d..8ec3dfd641b0 100644
> > --- a/arch/x86/include/uapi/asm/kvm.h
> > +++ b/arch/x86/include/uapi/asm/kvm.h
> > @@ -502,13 +502,13 @@ struct kvm_nested_state {
> > } hdr;
> >
> > /*
> > - * Define data region as 0 bytes to preserve backwards-compatability
> > + * Define union of flexible arrays to preserve backwards-compatability
>
> I think I'd actually prefer the "as 0 bytes" comment. The important part is that
> the size of "data" be zero, how that happens is immaterial.
Okay, I'll drop this part.
>
> > * to old definition of kvm_nested_state in order to avoid changing
> > * KVM_{GET,PUT}_NESTED_STATE ioctl values.
> > */
> > union {
> > - struct kvm_vmx_nested_state_data vmx[0];
> > - struct kvm_svm_nested_state_data svm[0];
> > + __DECLARE_FLEX_ARRAY(struct kvm_vmx_nested_state_data, vmx);
> > + __DECLARE_FLEX_ARRAY(struct kvm_svm_nested_state_data, svm);
> > } data;
> > };
> >
> > --
> > 2.34.1
> >
--
Kees Cook
Powered by blists - more mailing lists