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;
MIME-Version: 1.0
References: <20230519074047.1739879-1-dhowells@redhat.com> <20230519074047.1739879-10-dhowells@redhat.com>
 <ZGcvfLWAqmOLrnLj@infradead.org>
In-Reply-To: <ZGcvfLWAqmOLrnLj@infradead.org>
From:   Linus Torvalds <torvalds@linux-foundation.org>
Date:   Fri, 19 May 2023 09:22:22 -0700
Message-ID: <CAHk-=wgzSLXt38J_B2=QWCDi7A1c5B0_cJ3XYRj9rYn+YXbjQA@mail.gmail.com>
Subject: Re: [PATCH v20 09/32] tty, proc, kernfs, random: Use direct_splice_read()
To:     Christoph Hellwig <hch@infradead.org>
Cc:     David Howells <dhowells@redhat.com>, Jens Axboe <axboe@kernel.dk>,
        Al Viro <viro@zeniv.linux.org.uk>,
        Matthew Wilcox <willy@infradead.org>, Jan Kara <jack@suse.cz>,
        Jeff Layton <jlayton@kernel.org>,
        David Hildenbrand <david@redhat.com>,
        Jason Gunthorpe <jgg@nvidia.com>,
        Logan Gunthorpe <logang@deltatee.com>,
        Hillf Danton <hdanton@sina.com>,
        Christian Brauner <brauner@kernel.org>,
        linux-fsdevel@vger.kernel.org, linux-block@vger.kernel.org,
        linux-kernel@vger.kernel.org, linux-mm@kvack.org,
        Greg Kroah-Hartman <gregkh@linuxfoundation.org>,
        Christoph Hellwig <hch@lst.de>,
        John Hubbard <jhubbard@nvidia.com>,
        Miklos Szeredi <miklos@szeredi.hu>,
        Arnd Bergmann <arnd@arndb.de>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
Precedence: bulk

On Fri, May 19, 2023 at 1:12=E2=80=AFAM Christoph Hellwig <hch@infradead.or=
g> wrote:
>
> Pinging Al (and maybe Linus): is there any good reason to not simply
> default to direct_splice_read if ->read_iter is implemented and
> ->splice_read is not once you remove ITER_PIPE?

For me, the reason isn't so much technical as "historical pain".

We've had *so* many problems with splice on random file descriptors
that I at some point decided "no splice by default".

Now, admittedly most of the problems were due to the whole set_fs()
ambiguity, which you fixed and no longer exists. So maybe we could go
back to "implement splice by default".

I agree that as long as the default implementation is obviously safe,
it should be ok, and maybe direct_splice_read is that obvious..

           Linus