From: "Jason A. Donenfeld" <Jason@zx2c4.com>
Subject: Re: [PATCH net-next v6 07/23] zinc: ChaCha20 ARM and ARM64 implementations
Date: Wed, 26 Sep 2018 19:03:34 +0200
Message-ID: <CAHmME9pZKUMnnSZ0670rUH4H61g0x88c-B-DZYw_X-8+xgH-=g@mail.gmail.com>
References: <20180925145622.29959-1-Jason@zx2c4.com> <20180925145622.29959-8-Jason@zx2c4.com>
 <CAKv+Gu9mVAfdBvOMCFqRJj+wBiWu3JVOgPZdkcdjzqSdQQ5Jrw@mail.gmail.com>
 <CAHmME9r9KppoFwwNVpzpYbU+9dCPzb7Pit+4iRa4MY_ouJBWrA@mail.gmail.com>
 <CAKv+Gu8ih-TsASRGqK+ST_5+EQ0=Zo-zhGCadOdGyPjucMFTCg@mail.gmail.com> <D93C2C15-DC70-4A95-9350-AD4036C9556A@amacapital.net>
Mime-Version: 1.0
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
Cc: Ard Biesheuvel <ard.biesheuvel@linaro.org>,
        Herbert Xu <herbert@gondor.apana.org.au>,
        Thomas Gleixner <tglx@linutronix.de>,
        LKML <linux-kernel@vger.kernel.org>,
        Netdev <netdev@vger.kernel.org>,
        Linux Crypto Mailing List <linux-crypto@vger.kernel.org>,
        David Miller <davem@davemloft.net>,
        Greg Kroah-Hartman <gregkh@linuxfoundation.org>,
        Samuel Neves <sneves@dei.uc.pt>,
        Andrew Lutomirski <luto@kernel.org>,
        Jean-Philippe Aumasson <jeanphilippe.aumasson@gmail.com>,
        Russell King - ARM Linux <linux@armlinux.org.uk>,
        linux-arm-kernel@lists.infradead.org
To: Andy Lutomirski <luto@amacapital.net>
Return-path: <netdev-owner@vger.kernel.org>
In-Reply-To: <D93C2C15-DC70-4A95-9350-AD4036C9556A@amacapital.net>
Sender: netdev-owner@vger.kernel.org
List-Id: linux-crypto.vger.kernel.org

On Wed, Sep 26, 2018 at 6:21 PM Andy Lutomirski <luto@amacapital.net> wrote=
:
> Are, is what you=E2=80=99re saying that the Zinc chacha20 functions shoul=
d call simd_relax() every n bytes automatically for some reasonable value o=
f n?  If so, seems sensible, except that some care might be needed to make =
sure they interact with preemption correctly.
>
> What I mean is: the public Zinc entry points should either be callable in=
 an atomic context or they should not be.  I think this should be checked a=
t runtime in an appropriate place with an __might_sleep or similar.  Or sim=
d_relax should learn to *not* schedule if the result of preempt_enable() le=
aves it atomic. (And the latter needs to be done in a way that works even o=
n non-preempt kernels, and I don=E2=80=99t remember whether that=E2=80=99s =
possible.). And this should happen regardless of how many bytes are process=
ed. IOW, calling into Zinc should be equally not atomic-safe for 100 bytes =
and for 10 MB.

I'm not sure this is actually a problem. Namely:

preempt_disable();
kernel_fpu_begin();
kernel_fpu_end();
schedule(); <--- bug!

Calling kernel_fpu_end() disables preemption, but AFAIK, preemption
enabling/disabling is recursive, so kernel_fpu_end's use of
preempt_disable won't actually do anything until the outer preempt
enable is called:

preempt_disable();
kernel_fpu_begin();
kernel_fpu_end();
preempt_enable();
schedule(); <--- works!

Or am I missing some more subtle point?

Jason