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;
Date:   Wed, 6 Jul 2022 17:24:49 +0200
From:   "Jason A. Donenfeld" <Jason@zx2c4.com>
To:     Theodore Ts'o <tytso@mit.edu>
Cc:     linux-arm-kernel@lists.infradead.org,
        linuxppc-dev@lists.ozlabs.org, linux-s390@vger.kernel.org,
        linux-kernel@vger.kernel.org, x86@kernel.org,
        Catalin Marinas <catalin.marinas@arm.com>,
        Will Deacon <will@kernel.org>,
        Michael Ellerman <mpe@ellerman.id.au>,
        Heiko Carstens <hca@linux.ibm.com>,
        Alexander Gordeev <agordeev@linux.ibm.com>,
        Thomas Gleixner <tglx@linutronix.de>,
        "H . Peter Anvin" <hpa@zytor.com>,
        Greg Kroah-Hartman <gregkh@linuxfoundation.org>,
        Arnd Bergmann <arnd@arndb.de>
Subject: Re: [PATCH] random: remove CONFIG_ARCH_RANDOM and "nordrand"
Message-ID: <YsWpQWXVeDFXvq0F@zx2c4.com>
References: <20220705190121.293703-1-Jason@zx2c4.com>
 <YsWiSH4BrY5oNJuM@mit.edu>
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Disposition: inline
In-Reply-To: <YsWiSH4BrY5oNJuM@mit.edu>
Precedence: bulk

Hi Ted,

On Wed, Jul 06, 2022 at 10:55:04AM -0400, Theodore Ts'o wrote:
> On Tue, Jul 05, 2022 at 09:01:21PM +0200, Jason A. Donenfeld wrote:
> > Later the thinking evolved. With a properly designed RNG, using RDRAND
> > values alone won't harm anything, even if the outputs are malicious.
> 
> I personally think it's totally fine to remove nordrand.  However, the
> reason why it was there was that there were some rather extreme
> tin-foil-hatters who believed that if (the completely unavailable to
> the public for auditing) RDRAND implementation *were* malicious *and*
> the microcode had access to the register file and/or the instruction
> pipeline, then in theory, a malicious CPU could subvert how the RDRAND
> is mixed into the getrandom output to force a particular output.
> 
> Personally, I've always considered it to be insane, since a much
> easier way to compromise a CPU would be to drop a Minix system hidden
> into the CPU running a web server that had massive security bugs in it
> that were only discovered years later.  And if you don't trust the CPU
> manufacture to that extent, you should probably simply not use CPU's
> from that manufacturer.  :-)

That specific attack scenario is actually something I've fixed over the
last few months, by ensuring that all RDRAND values go through the hash
function. So even if the CPU is super malicious, it'd still need a hash
preimage, which isn't considered to be computable for blake2s.

Minix in the cpu... haha.. surely that would never happen... haha
surely...

Jason