From:   Florian Weimer <fweimer@redhat.com>
To:     Andy Lutomirski <luto@kernel.org>
Cc:     linux-arch@vger.kernel.org, Linux API <linux-api@vger.kernel.org>,
        linux-x86_64@vger.kernel.org, kernel-hardening@lists.openwall.com,
        linux-mm@kvack.org, the arch/x86 maintainers <x86@kernel.org>,
        musl@lists.openwall.com, libc-alpha@sourceware.org,
        linux-kernel@vger.kernel.org, Dave Hansen <dave.hansen@intel.com>,
        Kees Cook <keescook@chromium.org>,
        Andrei Vagin <avagin@gmail.com>
Subject: Re: [PATCH v3 1/3] x86: Implement arch_prctl(ARCH_VSYSCALL_CONTROL)
 to disable vsyscall
References: <3a1c8280967b491bf6917a18fbff6c9b52e8df24.1641398395.git.fweimer@redhat.com>
        <e431fa42-26ec-8ac6-f954-e681b1e0e9a6@kernel.org>
Date:   Fri, 14 Jan 2022 14:36:24 +0100
In-Reply-To: <e431fa42-26ec-8ac6-f954-e681b1e0e9a6@kernel.org> (Andy
        Lutomirski's message of "Thu, 13 Jan 2022 13:47:26 -0800")
Message-ID: <87sftqtp5z.fsf@oldenburg.str.redhat.com>
User-Agent: Gnus/5.13 (Gnus v5.13) Emacs/27.2 (gnu/linux)
MIME-Version: 1.0
Content-Type: text/plain
Precedence: bulk

* Andy Lutomirski:

> Is there a reason you didn't just change the check earlier in the
> function to:
>
> if (vsyscall_mode == NONE || current->mm->context.vsyscall_disabled)

Andrei requested that I don't print anything if vsyscall was disabled.

The original patch used a different message for better diagnostics.

> Also, I still think the prctl should not be available if
> vsyscall=emulate.  Either we should fully implement it or we should
> not implement.  We could even do:
>
> pr_warn_once("userspace vsyscall hardening request ignored because you
> have vsyscall=emulate.  Unless you absolutely need vsyscall=emulate, 
> update your system to use vsyscall=xonly.\n");
>
> and thus encourage good behavior.

I think there is still some hardening applied even with
vsyscall=emulate.  The question is what is more important: the
additional hardening, or clean, easily described behavior of the
interface.

Maybe ARCH_VSYSCALL_CONTROL could return different values based on to
what degree it could disable vsyscall?

The pr_warn_once does not seem particularly useful.  Anyone who upgrades
glibc and still uses vsyscall=emulate will see that, with no way to
disable it.

Thanks,
Florian