Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
MIME-Version: 1.0
References: <20200904165216.1799796-1-hch@lst.de> <20200904165216.1799796-4-hch@lst.de>
In-Reply-To: <20200904165216.1799796-4-hch@lst.de>
From:   Arnd Bergmann <arnd@arndb.de>
Date:   Fri, 4 Sep 2020 20:04:34 +0200
Message-ID: <CAK8P3a1=udSEq8kpsmLk-b7ye0p=6eKTvBV74jBoGYmufL2oEw@mail.gmail.com>
Subject: Re: [PATCH 3/8] asm-generic: fix unaligned access hamdling in raw_copy_{from,to}_user
To:     Christoph Hellwig <hch@lst.de>
Cc:     Paul Walmsley <paul.walmsley@sifive.com>,
        Palmer Dabbelt <palmer@dabbelt.com>,
        Alexander Viro <viro@zeniv.linux.org.uk>,
        linux-riscv <linux-riscv@lists.infradead.org>,
        "linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>,
        linux-arch <linux-arch@vger.kernel.org>
Content-Type: text/plain; charset="UTF-8"
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Fri, Sep 4, 2020 at 6:52 PM Christoph Hellwig <hch@lst.de> wrote:
>
> Use get_unaligned and put_unaligned for the small constant size cases
> in the generic uaccess routines.  This ensures they can be used for
> architectures that do not support unaligned loads and stores, while
> being a no-op for those that do.
>
> Signed-off-by: Christoph Hellwig <hch@lst.de>
> ---
>  include/asm-generic/uaccess.h | 20 ++++++++------------
>  1 file changed, 8 insertions(+), 12 deletions(-)
>
> diff --git a/include/asm-generic/uaccess.h b/include/asm-generic/uaccess.h
> index cc3b2c8b68fab4..768502bbfb154e 100644
> --- a/include/asm-generic/uaccess.h
> +++ b/include/asm-generic/uaccess.h
> @@ -36,19 +36,17 @@ raw_copy_from_user(void *to, const void __user * from, unsigned long n)
>         if (__builtin_constant_p(n)) {
>                 switch(n) {
>                 case 1:
> -                       *(u8 *)to = *(u8 __force *)from;
> +                       *(u8 *)to = get_unaligned((u8 __force *)from);
>                         return 0;
>                 case 2:
> -                       *(u16 *)to = *(u16 __force *)from;
> +                       *(u16 *)to = get_unaligned((u16 __force *)from);
>                         return 0;

The change look correct and necessary, but I wonder if this could be done
in a way that is a little easier on the compiler than the nested switch/case.

If I see it right, __put_user() and __get_user() can probably
be reduced to a plain put_unaligned() and get_unaligned() here,
which would simplify these a lot.

In turn it seems that the generic raw_copy_to_user() can just be the
a plain memcpy(), IIRC the optimization for small sizes should also
be done by modern compilers whenever they can.

     Arnd