Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
MIME-Version: 1.0
References: <20220731050342.56513-1-khuey@kylehuey.com> <Yuo59tV071/i6yhf@gmail.com>
 <CAP045ArF0SX84tDr=iZoK=EnXK2LsXYut3-KMkCxQO2OOhn=0A@mail.gmail.com> <Yuqvkufu7Hu4drL6@gmail.com>
In-Reply-To: <Yuqvkufu7Hu4drL6@gmail.com>
From:   Kyle Huey <me@kylehuey.com>
Date:   Wed, 3 Aug 2022 10:35:36 -0700
Message-ID: <CAP045AoBZvcacHM=rEKC+Lc8XeQQVzCvnNvsU-o=-ZHi=pE=mA@mail.gmail.com>
Subject: Re: [PATCH] x86/fpu: Allow PKRU to be (once again) written by ptrace.
To:     Ingo Molnar <mingo@kernel.org>
Cc:     Dave Hansen <dave.hansen@linux.intel.com>,
        Thomas Gleixner <tglx@linutronix.de>,
        Borislav Petkov <bp@alien8.de>, Ingo Molnar <mingo@redhat.com>,
        x86@kernel.org, "H. Peter Anvin" <hpa@zytor.com>,
        Paolo Bonzini <pbonzini@redhat.com>,
        Andy Lutomirski <luto@kernel.org>,
        Peter Zijlstra <peterz@infradead.org>,
        linux-kernel@vger.kernel.org,
        "Robert O'Callahan" <robert@ocallahan.org>,
        David Manouchehri <david.manouchehri@riseup.net>,
        kvm@vger.kernel.org, stable@vger.kernel.org
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

On Wed, Aug 3, 2022 at 10:25 AM Ingo Molnar <mingo@kernel.org> wrote:
>
>
> * Kyle Huey <me@kylehuey.com> wrote:
>
> > > Also, what's the security model for this register, do we trust all
> > > input values user-space provides for the PKRU field in the XSTATE? I
> > > realize that WRPKRU already gives user-space write access to the
> > > register - but does the CPU write it all into the XSTATE, with no
> > > restrictions on content whatsoever?
> >
> > There is no security model for this register. The CPU does write whatever
> > is given to WRPKRU (or XRSTOR) into the PKRU register. The pkeys(7) man
> > page notes:
> >
> > Protection keys have the potential to add a layer of security and
> > reliability to applications. But they have not been primarily designed as
> > a security feature. For instance, WRPKRU is a completely unprivileged
> > instruction, so pkeys are useless in any case that an attacker controls
> > the PKRU register or can execute arbitrary instructions.
>
> Ok - allowing ptrace to set the full 32 bits of the PKRU register seems OK
> then, and is 100% equivalent to using WRPKRU, right? So there's no implicit
> masking/clearing of bits depending on how many keys are available, or other
> details where WRPKRU might differ from a pure 32-bit per thread write,
> correct?

Right. The hardware doesn't have any concept of what keys are
available or not, that exists entirely in the kernel.

- Kyle

> Thanks,
>
>         Ingo