Received-SPF: pass (google.com: domain of linux-crypto-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: <20210610134459.28541-1-tianjia.zhang@linux.alibaba.com>
 <20210610134459.28541-2-tianjia.zhang@linux.alibaba.com> <Yh32tEhUgGeSXf/A@zx2c4.com>
 <52be961d-a00d-785d-8fb1-15b1a17bd74e@linux.alibaba.com>
In-Reply-To: <52be961d-a00d-785d-8fb1-15b1a17bd74e@linux.alibaba.com>
From:   "Jason A. Donenfeld" <Jason@zx2c4.com>
Date:   Tue, 1 Mar 2022 14:22:14 +0100
Message-ID: <CAHmME9pvbeGhWD+0N4_nOh-pkWmt7=Q9PDW40QjK8NHiJi7D2w@mail.gmail.com>
Subject: Re: [PATCH 1/3] crypto: sm4 - create SM4 library based on sm4 generic code
To:     Tianjia Zhang <tianjia.zhang@linux.alibaba.com>
Cc:     Herbert Xu <herbert@gondor.apana.org.au>,
        "David S. Miller" <davem@davemloft.net>,
        Catalin Marinas <catalin.marinas@arm.com>,
        Will Deacon <will@kernel.org>,
        Thomas Gleixner <tglx@linutronix.de>,
        Ingo Molnar <mingo@redhat.com>, Borislav Petkov <bp@alien8.de>,
        "H. Peter Anvin" <hpa@zytor.com>,
        Gilad Ben-Yossef <gilad@benyossef.com>,
        Ard Biesheuvel <ard.biesheuvel@linaro.org>,
        "Markku-Juhani O . Saarinen" <mjos@iki.fi>,
        Jussi Kivilinna <jussi.kivilinna@iki.fi>,
        X86 ML <x86@kernel.org>,
        Linux Crypto Mailing List <linux-crypto@vger.kernel.org>,
        linux-arm-kernel <linux-arm-kernel@lists.infradead.org>,
        LKML <linux-kernel@vger.kernel.org>
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

Hi Tianjia,

On Tue, Mar 1, 2022 at 12:50 PM Tianjia Zhang
<tianjia.zhang@linux.alibaba.com> wrote:
>
> Hi Jason,
>
> On 3/1/22 6:34 PM, Jason A. Donenfeld wrote:
> >>   lib/crypto/Kconfig   |   3 +
> >>   lib/crypto/Makefile  |   3 +
> >>   lib/crypto/sm4.c     | 184 +++++++++++++++++++++++++++++++++++++++++++
> >
> > If this is only used by the crypto API, it does not belong in
> > lib/crypto. I understand you want fallback generic code for the SIMD
> > implementation, but we've generally done that in crypto/ when the use
> > case is only the crypto API. Can you move this to the right place?
>
> This is not only used by the crypto API, but also used for SIMD
> acceleration under the x86 and arm architectures, mainly for processing
> the remaining blocks after SIMD acceleration. In general, the
> performance of SIMD processing a single block is not as good as that of
> general software implementations.

Yes, and those accelerated implementations are part of the crypto API,
and are not used by anything except the crypto API. Hence this should
be in crypto/, just like everything else that is /only/ used for the
cryto API. lib/crypto/ is for in-kernel users of crypto via normal
code paths. sm4.c does not belong in lib/crypto/ and should be moved.

You additional export symbols of those SIMD implementations in
arch/crypto/, which is not correct either, since nothing in the tree
uses those symbols. Please remove those EXPORT_SYMBOL directives as
well. Those functions can be static, and do not need to be declared in
the .h file.

Thanks,
Jason