netdev - Re: [PATCH bpf] xsk: fix crash in poll when device does not support ndo_xsk

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-ID: <CAJ8uoz2kfVSWnOkNzbL+D7tcOaoGxE3WnemY=QrbXyiDiO-bBg@mail.gmail.com>
Date:   Wed, 2 Oct 2019 08:06:37 +0200
From:   Magnus Karlsson <magnus.karlsson@...il.com>
To:     Daniel Borkmann <daniel@...earbox.net>
Cc:     Magnus Karlsson <magnus.karlsson@...el.com>,
        Björn Töpel <bjorn.topel@...el.com>,
        Alexei Starovoitov <ast@...nel.org>,
        Network Development <netdev@...r.kernel.org>,
        Jonathan Lemon <jonathan.lemon@...il.com>,
        bpf <bpf@...r.kernel.org>
Subject: Re: [PATCH bpf] xsk: fix crash in poll when device does not support ndo_xsk_wakeup

On Wed, Oct 2, 2019 at 7:57 AM Daniel Borkmann <daniel@...earbox.net> wrote:
>
> On Mon, Sep 30, 2019 at 03:30:12PM +0200, Magnus Karlsson wrote:
> > Fixes a crash in poll() when an AF_XDP socket is opened in copy mode
> > with the XDP_USE_NEED_WAKEUP flag set and the bound device does not
> > have ndo_xsk_wakeup defined. Avoid trying to call the non-existing ndo
> > and instead call the internal xsk sendmsg functionality to send
> > packets in the same way (from the application's point of view) as
> > calling sendmsg() in any mode or poll() in zero-copy mode would have
> > done. The application should behave in the same way independent on if
> > zero-copy mode or copy-mode is used.
> >
> > Fixes: 77cd0d7b3f25 ("xsk: add support for need_wakeup flag in AF_XDP rings")
> > Reported-by: syzbot+a5765ed8cdb1cca4d249@...kaller.appspotmail.com
> > Signed-off-by: Magnus Karlsson <magnus.karlsson@...el.com>
>
> Overall looks good, two very small nits:
>
> > ---
> >  net/xdp/xsk.c | 33 +++++++++++++++++++++++----------
> >  1 file changed, 23 insertions(+), 10 deletions(-)
> >
> > diff --git a/net/xdp/xsk.c b/net/xdp/xsk.c
> > index c2f1af3..a478d8ec 100644
> > --- a/net/xdp/xsk.c
> > +++ b/net/xdp/xsk.c
> > @@ -327,8 +327,7 @@ static void xsk_destruct_skb(struct sk_buff *skb)
> >       sock_wfree(skb);
> >  }
> >
> > -static int xsk_generic_xmit(struct sock *sk, struct msghdr *m,
> > -                         size_t total_len)
> > +static int xsk_generic_xmit(struct sock *sk)
> >  {
> >       u32 max_batch = TX_BATCH_SIZE;
> >       struct xdp_sock *xs = xdp_sk(sk);
> > @@ -394,22 +393,31 @@ static int xsk_generic_xmit(struct sock *sk, struct msghdr *m,
> >       return err;
> >  }
> >
> > -static int xsk_sendmsg(struct socket *sock, struct msghdr *m, size_t total_len)
> > +static int __xsk_sendmsg(struct socket *sock)
> >  {
> > -     bool need_wait = !(m->msg_flags & MSG_DONTWAIT);
> >       struct sock *sk = sock->sk;
> >       struct xdp_sock *xs = xdp_sk(sk);
>
> Can't we just only pass xs as an argument to __xsk_sendmsg()? From
> below xsk_sendmsg() we eventually fetch xs anyway same as for the
> xsk_poll(), yet in __xsk_sendmsg() we pass sock (which is otherwise
> unused) and then we need to refetch sk and xs once again.

xs is enough for the zerocopy path, but I need sk for the generic skb
path. In any way, sock can be exterminated and the code simplified, as
you suggest. Will fix.

> > -     if (unlikely(!xsk_is_bound(xs)))
> > -             return -ENXIO;
> >       if (unlikely(!(xs->dev->flags & IFF_UP)))
> >               return -ENETDOWN;
> >       if (unlikely(!xs->tx))
> >               return -ENOBUFS;
> > +
> > +     return xs->zc ? xsk_zc_xmit(sk) : xsk_generic_xmit(sk);
> > +}
> > +
> > +static int xsk_sendmsg(struct socket *sock, struct msghdr *m, size_t total_len)
> > +{
> > +     bool need_wait = !(m->msg_flags & MSG_DONTWAIT);
> > +     struct sock *sk = sock->sk;
> > +     struct xdp_sock *xs = xdp_sk(sk);
> > +
> > +     if (unlikely(!xsk_is_bound(xs)))
> > +             return -ENXIO;
> >       if (need_wait)
>
> Not directly related but since you touch this code, need_wait should
> be marked unlikely() as well.

Will fix.

Thanks: Magnus

> >               return -EOPNOTSUPP;
> >
> > -     return (xs->zc) ? xsk_zc_xmit(sk) : xsk_generic_xmit(sk, m, total_len);
> > +     return __xsk_sendmsg(sock);
> >  }
> >
> >  static unsigned int xsk_poll(struct file *file, struct socket *sock,
> > @@ -426,9 +434,14 @@ static unsigned int xsk_poll(struct file *file, struct socket *sock,
> >       dev = xs->dev;
> >       umem = xs->umem;
> >
> > -     if (umem->need_wakeup)
> > -             dev->netdev_ops->ndo_xsk_wakeup(dev, xs->queue_id,
> > -                                             umem->need_wakeup);
> > +     if (umem->need_wakeup) {
> > +             if (dev->netdev_ops->ndo_xsk_wakeup)
> > +                     dev->netdev_ops->ndo_xsk_wakeup(dev, xs->queue_id,
> > +                                                     umem->need_wakeup);
> > +             else
> > +                     /* Poll needs to drive Tx also in copy mode */
> > +                     __xsk_sendmsg(sock);
> > +     }
> >
> >       if (xs->rx && !xskq_empty_desc(xs->rx))
> >               mask |= POLLIN | POLLRDNORM;
> > --
> > 2.7.4
> >