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: <202211171422.7A7A7A9@keescook> <CANn89iLQcLNX+x_gJCMy5kD5GW3Xg8U4s0VGHtSuN8iegmhjxQ@mail.gmail.com>
 <202211171513.28D070E@keescook> <CANn89iKgMvhLbTi=SHn41R--rBQ8As=E52Hnecch6nOhXVYGrg@mail.gmail.com>
 <202211171624.963F44FCE@keescook>
In-Reply-To: <202211171624.963F44FCE@keescook>
From:   Eric Dumazet <edumazet@google.com>
Date:   Thu, 17 Nov 2022 17:24:36 -0800
Message-ID: <CANn89iJ1ciQkv5nt5XgRXAXPVzEW6J=GdiUYvqrYgjUU440OtQ@mail.gmail.com>
Subject: Re: Coverity: __sock_gen_cookie(): Error handling issues
To:     Kees Cook <keescook@chromium.org>
Cc:     linux-kernel@vger.kernel.org,
        Marc Kleine-Budde <mkl@pengutronix.de>,
        Paolo Abeni <pabeni@redhat.com>,
        Stefano Garzarella <sgarzare@redhat.com>,
        Jakub Kicinski <kuba@kernel.org>,
        Nikolay Aleksandrov <nikolay@nvidia.com>,
        "David S. Miller" <davem@davemloft.net>,
        Florian Fainelli <f.fainelli@gmail.com>,
        netdev@vger.kernel.org,
        "Gustavo A. R. Silva" <gustavo@embeddedor.com>,
        linux-next@vger.kernel.org, linux-hardening@vger.kernel.org
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

On Thu, Nov 17, 2022 at 4:25 PM Kees Cook <keescook@chromium.org> wrote:
>
> On Thu, Nov 17, 2022 at 03:22:22PM -0800, Eric Dumazet wrote:
> > On Thu, Nov 17, 2022 at 3:14 PM Kees Cook <keescook@chromium.org> wrote:
> > >
> > > On Thu, Nov 17, 2022 at 02:49:55PM -0800, Eric Dumazet wrote:
> > > > On Thu, Nov 17, 2022 at 2:22 PM coverity-bot <keescook@chromium.org> wrote:
> > > > >
> > > > > Hello!
> > > > >
> > > > > This is an experimental semi-automated report about issues detected by
> > > > > Coverity from a scan of next-20221117 as part of the linux-next scan project:
> > > > > https://scan.coverity.com/projects/linux-next-weekly-scan
> > > > >
> > > > > You're getting this email because you were associated with the identified
> > > > > lines of code (noted below) that were touched by commits:
> > > > >
> > > > >   Wed Nov 16 12:42:01 2022 +0000
> > > > >     4ebf802cf1c6 ("net: __sock_gen_cookie() cleanup")
> > > > >
> > > > > Coverity reported the following:
> > > > >
> > > > > *** CID 1527347:  Error handling issues  (CHECKED_RETURN)
> > > > > net/core/sock_diag.c:33 in __sock_gen_cookie()
> > > > > 27     {
> > > > > 28      u64 res = atomic64_read(&sk->sk_cookie);
> > > > > 29
> > > > > 30      if (!res) {
> > > > > 31              u64 new = gen_cookie_next(&sock_cookie);
> > > > > 32
> > > > > vvv     CID 1527347:  Error handling issues  (CHECKED_RETURN)
> > > > > vvv     Calling "atomic64_try_cmpxchg" without checking return value (as is done elsewhere 8 out of 9 times).
> > > > > 33              atomic64_try_cmpxchg(&sk->sk_cookie, &res, new);
> > > >
> > > >
> > > > Hmmm. for some reason I thought @res was always updated...
> > > >
> > > > A fix would be to read sk->sk_cookie, but I guess your tool will still
> > > > complain we do not care
> > > > of  atomic64_try_cmpxchg() return value ?
> > > >
> > > > diff --git a/net/core/sock_diag.c b/net/core/sock_diag.c
> > > > index b11593cae5a09b15a10d6ba35bccc22263cb8fc8..58efb9c1c8dd4f8e5a3009a0176e1b96487daaff
> > > > 100644
> > > > --- a/net/core/sock_diag.c
> > > > +++ b/net/core/sock_diag.c
> > > > @@ -31,6 +31,10 @@ u64 __sock_gen_cookie(struct sock *sk)
> > > >                 u64 new = gen_cookie_next(&sock_cookie);
> > > >
> > > >                 atomic64_try_cmpxchg(&sk->sk_cookie, &res, new);
> > > > +               /* Another cpu/thread might have won the race,
> > > > +                * reload the final value.
> > > > +                */
> > > > +               res = atomic64_read(&sk->sk_cookie);
> > > >         }
> > > >         return res;
> > > >  }
> > >
> > > I think it's saying it was expecting an update loop -- i.e. to make sure
> > > the value actually got swapped (the "try" part...)?
> >
> > The value has been updated, either by us or someone else.
> >
> > We do not particularly care who won the race, since the value is
> > updated once only.
>
> Ah! Okay, now I understand the added comment. Thanks :)

I guess we could simply go back to atomic64_cmpxchg() to avoid a false positive.

This boils to avoid the loop we had prior to 4ebf802cf1c6

diff --git a/net/core/sock_diag.c b/net/core/sock_diag.c
index b11593cae5a09b15a10d6ba35bccc22263cb8fc8..7b9e321e0f6b15f2fb7af9f53fceb874439cbd02
100644
--- a/net/core/sock_diag.c
+++ b/net/core/sock_diag.c
@@ -30,7 +30,11 @@ u64 __sock_gen_cookie(struct sock *sk)
        if (!res) {
                u64 new = gen_cookie_next(&sock_cookie);

-               atomic64_try_cmpxchg(&sk->sk_cookie, &res, new);
+               atomic64_cmpxchg(&sk->sk_cookie, res, new);
+               /* Another cpu/thread might have won the race,
+                * load the final value.
+                */
+               res = atomic64_read(&sk->sk_cookie);
        }
        return res;
 }