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: <20210315135720.002213995@linuxfoundation.org> <20210315135720.384809636@linuxfoundation.org>
 <20210316094137.GA12946@amd>
In-Reply-To: <20210316094137.GA12946@amd>
From:   Eric Dumazet <edumazet@google.com>
Date:   Tue, 16 Mar 2021 10:49:13 +0100
Message-ID: <CANn89iKvhq3Bpu5eaZ5mrmRNBqNm9OwoGzCSg3-OaBL8CrAPfA@mail.gmail.com>
Subject: Re: [PATCH 4.19 011/120] tcp: annotate tp->copied_seq lockless reads
To:     Pavel Machek <pavel@denx.de>
Cc:     Greg Kroah-Hartman <gregkh@linuxfoundation.org>,
        LKML <linux-kernel@vger.kernel.org>, stable@vger.kernel.org,
        "David S. Miller" <davem@davemloft.net>
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

On Tue, Mar 16, 2021 at 10:41 AM Pavel Machek <pavel@denx.de> wrote:
>
> Hi!
>
> > From: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
> >
> > From: Eric Dumazet <edumazet@google.com>
>
> Two From: fields here.
>
> > [ Upstream commit 7db48e983930285b765743ebd665aecf9850582b ]
> >
> > There are few places where we fetch tp->copied_seq while
> > this field can change from IRQ or other cpu.
>
> And there are few such places even after the patch is applied; I
> quoted them below.
>
> Doing addition to variable without locking... is kind of
> interesting. Are you sure it is okay?

We are holding the socket lock here.

The WRITE_ONCE() here is paired with sides doing READ_ONCE() while
socket lock is _not_ held.

We want to make sure compiler won't write into this variable one byte at a time,
or using stupid things.


>
> > @@ -2112,7 +2112,7 @@ int tcp_recvmsg(struct sock *sk, struct
> >                       if (urg_offset < used) {
> >                               if (!urg_offset) {
> >                                       if (!sock_flag(sk, SOCK_URGINLINE)) {
> > -                                             ++*seq;
> > +                                             WRITE_ONCE(*seq, *seq + 1);
> >                                               urg_hole++;
> >                                               offset++;
> >                                               used--;
> > @@ -2134,7 +2134,7 @@ int tcp_recvmsg(struct sock *sk, struct
> >                       }
> >               }
> >
> > -             *seq += used;
> > +             WRITE_ONCE(*seq, *seq + used);
> >               copied += used;
> >               len -= used;
> >
> > @@ -2163,7 +2163,7 @@ skip_copy:
> >
> >       found_fin_ok:
> >               /* Process the FIN. */
> > -             ++*seq;
> > +             WRITE_ONCE(*seq, *seq + 1);
> >               if (!(flags & MSG_PEEK))
> >                       sk_eat_skb(sk, skb);
> >               break;
>
> Best regards,
>                                                                 Pavel
> --
> DENX Software Engineering GmbH,      Managing Director: Wolfgang Denk
> HRB 165235 Munich, Office: Kirchenstr.5, D-82194 Groebenzell, Germany