User-Agent: Microsoft-MacOutlook/16.56.21121100
Date:   Wed, 05 Jan 2022 17:19:00 +0800
Subject: Re: [PATCH] af_unix: missing lock releases in af_unix.c
From:   Ryan Cai <ycaibb@gmail.com>
To:     Shoaib Rao <rao.shoaib@oracle.com>, <davem@davemloft.net>,
        <kuba@kernel.org>, <ast@kernel.org>, <daniel@iogearbox.net>,
        <andrii@kernel.org>, <kafai@fb.com>, <songliubraving@fb.com>,
        <yhs@fb.com>, <john.fastabend@gmail.com>, <kpsingh@kernel.org>,
        <cong.wang@bytedance.com>, <viro@zeniv.linux.org.uk>,
        <edumazet@google.com>, <jiang.wang@bytedance.com>,
        <christian.brauner@ubuntu.com>, <kuniyu@amazon.co.jp>
CC:     <netdev@vger.kernel.org>, <linux-kernel@vger.kernel.org>,
        <bpf@vger.kernel.org>
Message-ID: <9BADB912-9338-4798-94C4-AA6268C4CA06@gmail.com>
Thread-Topic: [PATCH] af_unix: missing lock releases in af_unix.c
References: <20220103135830.59118-1-ycaibb@gmail.com>
 <555a3e2b-3981-672d-c6cf-5ecb357d2fa6@oracle.com>
In-Reply-To: <555a3e2b-3981-672d-c6cf-5ecb357d2fa6@oracle.com>
Mime-version: 1.0
Content-type: text/plain;
        charset="UTF-8"
Content-transfer-encoding: quoted-printable
Precedence: bulk

Hi, Shoaib

          After further checking, the lock is indeed released at the out_fr=
ee. My patch is invalid. Sorry for the inconvenience.

Best,
Ryan

=EF=BB=BFOn 4/1/2022, 2:47 AM, "Shoaib Rao" <rao.shoaib@oracle.com> wrote:


    On 1/3/22 05:58, Ryan Cai wrote:
    > In method __unix_dgram_recvmsg, the lock u->iolock is not released wh=
en skb is true and loop breaks.
    >
    > Signed-off-by: Ryan Cai <ycaibb@gmail.com>
    > ---
    >   net/unix/af_unix.c | 1 +
    >   1 file changed, 1 insertion(+)
    >
    > diff --git a/net/unix/af_unix.c b/net/unix/af_unix.c
    > index b0bfc78e421c..b97972948d9d 100644
    > --- a/net/unix/af_unix.c
    > +++ b/net/unix/af_unix.c
    > @@ -2305,6 +2305,7 @@ int __unix_dgram_recvmsg(struct sock *sk, struc=
t msghdr *msg, size_t size,
    >   		if (skb) {
    >   			if (!(flags & MSG_PEEK))
    >   				scm_stat_del(sk, skb);
    > +			mutex_unlock(&u->iolock);
    >   			break;
    >   		}
    >  =20

    It seems to me that the unlock at the end will release the mutex?

    out_free:
             skb_free_datagram(sk, skb);
             mutex_unlock(&u->iolock);

    Shoaib