Received-SPF: softfail (google.com: domain of transitioning linux-kernel-owner@vger.kernel.org does not designate 23.128.96.19 as permitted sender) client-ip=23.128.96.19;
Date:   Wed, 1 Jun 2022 20:25:20 +0100
From:   Matthew Wilcox <willy@infradead.org>
To:     Linus Torvalds <torvalds@linux-foundation.org>
Cc:     Alexey Gladkov <legion@kernel.org>,
        LKML <linux-kernel@vger.kernel.org>,
        "Eric W . Biederman" <ebiederm@xmission.com>,
        Andrew Morton <akpm@linux-foundation.org>,
        Christian Brauner <brauner@kernel.org>,
        Iurii Zaikin <yzaikin@google.com>,
        Kees Cook <keescook@chromium.org>,
        Linux Containers <containers@lists.linux.dev>,
        linux-fsdevel <linux-fsdevel@vger.kernel.org>,
        Luis Chamberlain <mcgrof@kernel.org>,
        Vasily Averin <vvs@virtuozzo.com>
Subject: Re: [RFC PATCH 1/4] sysctl: API extension for handling sysctl
Message-ID: <Ype9ILKm+8WLOq9W@casper.infradead.org>
References: <CAHk-=whi2SzU4XT_FsdTCAuK2qtYmH+-hwi1cbSdG8zu0KXL=g@mail.gmail.com>
 <cover.1654086665.git.legion@kernel.org>
 <5ec6759ab3b617f9c12449a9606b6f0b5a7582d0.1654086665.git.legion@kernel.org>
 <Ype7skNJzEQ1W96v@casper.infradead.org>
 <CAHk-=wiTtYMia0FR4h7_nV2RZ5pq=wR-7oMMK3o8o=EgAxMsmg@mail.gmail.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <CAHk-=wiTtYMia0FR4h7_nV2RZ5pq=wR-7oMMK3o8o=EgAxMsmg@mail.gmail.com>
Precedence: bulk

On Wed, Jun 01, 2022 at 12:23:06PM -0700, Linus Torvalds wrote:
> On Wed, Jun 1, 2022 at 12:19 PM Matthew Wilcox <willy@infradead.org> wrote:
> >
> > Why not pass the iocb in ->read and ->write?  We're still regretting not
> > doing that with file_operations.
> 
> No, all the actual "io" is done by the caller.
> 
> There is no way in hell I want the sysctl callbacks to actually
> possibly do user space accesses etc.
> 
> They get a kernel buffer that has already been set up. There is no
> iocb or iovec left for them.

I wasn't suggesting the iovec.  Just the iocb, instead of passing in the
ki_filp and the ki_pos.

> (That also means that they can take whatever locks they need,
> including spinlocks, because there's not going to be any random user
> accesses or complex pipe buffer lookups or whatever).
> 
>                 Linus