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: <20230623225513.2732256-1-dhowells@redhat.com> <20230623225513.2732256-4-dhowells@redhat.com>
 <CAOi1vP9vjLfk3W+AJFeexC93jqPaPUn2dD_4NrzxwoZTbYfOnw@mail.gmail.com> <3070680.1687792331@warthog.procyon.org.uk>
In-Reply-To: <3070680.1687792331@warthog.procyon.org.uk>
From:   Ilya Dryomov <idryomov@gmail.com>
Date:   Mon, 26 Jun 2023 17:52:34 +0200
Message-ID: <CAOi1vP93bAhYbgEz+oBbWaB4nzm=vBhfRu0Dj5Cs6_GHJrxWpQ@mail.gmail.com>
Subject: Re: [PATCH net-next v5 03/16] ceph: Use sendmsg(MSG_SPLICE_PAGES)
 rather than sendpage
To:     David Howells <dhowells@redhat.com>
Cc:     netdev@vger.kernel.org,
        Alexander Duyck <alexander.duyck@gmail.com>,
        "David S. Miller" <davem@davemloft.net>,
        Eric Dumazet <edumazet@google.com>,
        Jakub Kicinski <kuba@kernel.org>,
        Paolo Abeni <pabeni@redhat.com>,
        Willem de Bruijn <willemdebruijn.kernel@gmail.com>,
        David Ahern <dsahern@kernel.org>,
        Matthew Wilcox <willy@infradead.org>,
        Jens Axboe <axboe@kernel.dk>, linux-mm@kvack.org,
        linux-kernel@vger.kernel.org, Xiubo Li <xiubli@redhat.com>,
        Jeff Layton <jlayton@kernel.org>, ceph-devel@vger.kernel.org
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
Precedence: bulk

On Mon, Jun 26, 2023 at 5:12=E2=80=AFPM David Howells <dhowells@redhat.com>=
 wrote:
>
> Ilya Dryomov <idryomov@gmail.com> wrote:
>
> > > -       int flags =3D MSG_DONTWAIT | MSG_NOSIGNAL | more;
>
> Btw, why are you setting MSG_DONTWAIT?  If you're in the middle of
> transmitting a message on a TCP socket, surely you can't just switch to
> transmitting a different message on the same socket without doing some so=
rt of
> reframing?

We don't want to hog kworker threads.  You are correct that we can't
switch to transmitting a different message on the same socket but Ceph
is massively parallel and there can be dozens or even hundreds of other
sockets to work on.

Thanks,

                Ilya