From mboxrd@z Thu Jan 1 00:00:00 1970 From: Eric Dumazet Subject: Re: [PATCHv7 3/3] vhost_net: a kernel-level virtio server Date: Tue, 03 Nov 2009 19:51:35 +0100 Message-ID: <4AF07BB7.1020802@gmail.com> References: <20091103172422.GD5591@redhat.com> <4AF0708B.4020406@gmail.com> <4AF07199.2020601@gmail.com> <4AF072EE.9020202@gmail.com> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable Cc: "Michael S. Tsirkin" , netdev@vger.kernel.org, virtualization@lists.linux-foundation.org, kvm@vger.kernel.org, linux-kernel@vger.kernel.org, mingo@elte.hu, linux-mm@kvack.org, akpm@linux-foundation.org, hpa@zytor.com, Rusty Russell , s.hetze@linux-ag.com, "Paul E. McKenney" To: Gregory Haskins Return-path: In-Reply-To: <4AF072EE.9020202@gmail.com> Sender: owner-linux-mm@kvack.org List-Id: netdev.vger.kernel.org Gregory Haskins a =E9crit : > Gregory Haskins wrote: >> Eric Dumazet wrote: >>> Michael S. Tsirkin a =E9crit : >>>> +static void handle_tx(struct vhost_net *net) >>>> +{ >>>> + struct vhost_virtqueue *vq =3D &net->dev.vqs[VHOST_NET_VQ_TX]; >>>> + unsigned head, out, in, s; >>>> + struct msghdr msg =3D { >>>> + .msg_name =3D NULL, >>>> + .msg_namelen =3D 0, >>>> + .msg_control =3D NULL, >>>> + .msg_controllen =3D 0, >>>> + .msg_iov =3D vq->iov, >>>> + .msg_flags =3D MSG_DONTWAIT, >>>> + }; >>>> + size_t len, total_len =3D 0; >>>> + int err, wmem; >>>> + size_t hdr_size; >>>> + struct socket *sock =3D rcu_dereference(vq->private_data); >>>> + if (!sock) >>>> + return; >>>> + >>>> + wmem =3D atomic_read(&sock->sk->sk_wmem_alloc); >>>> + if (wmem >=3D sock->sk->sk_sndbuf) >>>> + return; >>>> + >>>> + use_mm(net->dev.mm); >>>> + mutex_lock(&vq->mutex); >>>> + vhost_no_notify(vq); >>>> + >>> using rcu_dereference() and mutex_lock() at the same time seems wrong= , I suspect >>> that your use of RCU is not correct. >>> >>> 1) rcu_dereference() should be done inside a read_rcu_lock() section,= and >>> we are not allowed to sleep in such a section. >>> (Quoting Documentation/RCU/whatisRCU.txt : >>> It is illegal to block while in an RCU read-side critical sectio= n, ) >>> >>> 2) mutex_lock() can sleep (ie block) >>> >> >> Michael, >> I warned you that this needed better documentation ;) >> >> Eric, >> I think I flagged this once before, but Michael convinced me that it >> was indeed "ok", if but perhaps a bit unconventional. I will try to >> find the thread. >> >> Kind Regards, >> -Greg >> >=20 > Here it is: >=20 > http://lkml.org/lkml/2009/8/12/173 >=20 Yes, this doesnt convince me at all, and could be a precedent for a wrong= RCU use. People wanting to use RCU do a grep on kernel sources to find how to corr= ectly use RCU. Michael, please use existing locking/barrier mechanisms, and not pretend = to use RCU. Some automatic tools might barf later. For example, we could add a debugging facility to check that rcu_derefere= nce() is used in an appropriate context, ie conflict with existing mutex_lock() debuggi= ng facility. -- To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to majordomo@kvack.org. For more info on Linux MM, see: http://www.linux-mm.org/ . Don't email: email@kvack.org