Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.33 as permitted sender) client-ip=23.128.96.33;
MIME-Version: 1.0
References: <20231010224630.238254-1-singhabhinav9051571833@gmail.com>
 <2023101136-irritate-shrine-cde6@gregkh> <3073e9a6-9f10-4326-9734-7e203d509888@gmail.com>
In-Reply-To: <3073e9a6-9f10-4326-9734-7e203d509888@gmail.com>
From:   Eric Dumazet <edumazet@google.com>
Date:   Wed, 11 Oct 2023 19:43:20 +0200
Message-ID: <CANn89i+hMbhLqXKCF2P=HVeeRSCxvgH_xY1b=T=udLFJjG3ZwA@mail.gmail.com>
Subject: Re: [PATCH] Remove extra unlock for the mutex
To:     Abhinav Singh <singhabhinav9051571833@gmail.com>
Cc:     Greg KH <gregkh@linuxfoundation.org>, davem@davemloft.net,
        dsahern@kernel.org, kuba@kernel.org, pabeni@redhat.com,
        netdev@vger.kernel.org,
        linux-kernel-mentees@lists.linuxfoundation.org,
        linux-kernel@vger.kernel.org
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
Precedence: bulk

On Wed, Oct 11, 2023 at 5:50=E2=80=AFPM Abhinav Singh
<singhabhinav9051571833@gmail.com> wrote:
>
> On 10/11/23 12:00, Greg KH wrote:
> > On Wed, Oct 11, 2023 at 04:16:30AM +0530, Abhinav Singh wrote:
> >> There is a double unlock on mutex. This can cause undefined behaviour.
> >>
> >> Signed-off-by: Abhinav Singh <singhabhinav9051571833@gmail.com>
> >> ---
> >>   net/ipv4/inet_connection_sock.c | 1 -
> >>   1 file changed, 1 deletion(-)
> >>
> >> diff --git a/net/ipv4/inet_connection_sock.c b/net/ipv4/inet_connectio=
n_sock.c
> >> index aeebe8816689..f11fe8c727a4 100644
> >> --- a/net/ipv4/inet_connection_sock.c
> >> +++ b/net/ipv4/inet_connection_sock.c
> >> @@ -597,7 +597,6 @@ int inet_csk_get_port(struct sock *sk, unsigned sh=
ort snum)
> >>      }
> >>      if (head2_lock_acquired)
> >>              spin_unlock(&head2->lock);
> >> -    spin_unlock_bh(&head->lock);
> >
> > How was this tested?
> >
> > And where is the now-needed unlock of the head->lock?
> >
> > How was this change found?
> >
> > And your subject line needs a lot of work...
> >
> > thanks,
> >
> > greg k-h
> Hello, I used sparse tool and got it this warning message "warning:
> context imbalance in 'inet_csk_get_port' - unexpected unlock"
> Due to my over excitement of sending a good patch to kernel I didnt see
> correctly and misread `head` as `head2` and thought it was double
> unlocking the mutex. I m very sorry. But on a different note think we
> should do a check for `head->lock` as well before unlocking. Unlocking a
> non locked mutex can also trigger a undefined behaviour.
>

There is no undefined behavior, only sparse that might be confused a little=
.

I do not think we can express in sparse the fact that
inet_csk_find_open_port() acquires head->lock

(head being the return value of this function...)

The following does not help.

diff --git a/net/ipv4/inet_connection_sock.c b/net/ipv4/inet_connection_soc=
k.c
index aeebe881668996057d1495c84eee0f0b644b7ad0..ed7b3993316cd1ba0b2859b0bd3=
f447e066bd3b5
100644
--- a/net/ipv4/inet_connection_sock.c
+++ b/net/ipv4/inet_connection_sock.c
@@ -323,6 +323,7 @@ static struct inet_bind_hashbucket *
 inet_csk_find_open_port(const struct sock *sk, struct
inet_bind_bucket **tb_ret,
                        struct inet_bind2_bucket **tb2_ret,
                        struct inet_bind_hashbucket **head2_ret, int *port_=
ret)
+       __acquires(head->lock)
 {
        struct inet_hashinfo *hinfo =3D tcp_or_dccp_get_hashinfo(sk);
        int i, low, high, attempt_half, port, l3mdev;