Re: [PATCH] Fix tun: wake up waitqueues after IFF_UP is set

[Jason Wang]

On 2019/6/17 äå4:10, Jason Wang wrote:
> On 2019/6/17 äå3:33, Fei Li wrote:
> > Currently after setting tap0 link up, the tun code wakes tx/rx waited
> > queues up in tun_net_open() when .ndo_open() is called, however the
> > IFF_UP flag has not been set yet. If there's already a wait queue, it
> > would fail to transmit when checking the IFF_UP flag in tun_sendmsg().
> > Then the saving vhost_poll_start() will add the wq into wqh until it
> > is waken up again. Although this works when IFF_UP flag has been set
> > when tun_chr_poll detects; this is not true if IFF_UP flag has not
> > been set at that time. Sadly the latter case is a fatal error, as
> > the wq will never be waken up in future unless later manually
> > setting link up on purpose.
> >
> > Fix this by moving the wakeup process into the NETDEV_UP event
> > notifying process, this makes sure IFF_UP has been set before all
> > waited queues been waken up.


Btw, the title needs some tweak. E.g you need use "net" as prefix since 
it's a fix for net.git and "Fix" could be removed like:

[PATCH net] tun: wake up waitqueues after IFF_UP is set.

Thanks


> > Signed-off-by: Fei Li <lifei.shirley@xxxxxxxxxxxxx>
> > ---
> > Â drivers/net/tun.c | 17 +++++++++--------
> > Â 1 file changed, 9 insertions(+), 8 deletions(-)
> >
> > diff --git a/drivers/net/tun.c b/drivers/net/tun.c
> > index c452d6d831dd..a3c9cab5a4d0 100644
> > --- a/drivers/net/tun.c
> > +++ b/drivers/net/tun.c
> > @@ -1015,17 +1015,9 @@ static void tun_net_uninit(struct net_device 
> > *dev)
> > Â static int tun_net_open(struct net_device *dev)
> > Â {
> > ÂÂÂÂÂ struct tun_struct *tun = netdev_priv(dev);
> > -ÂÂÂ int i;
> > Â ÂÂÂÂÂ netif_tx_start_all_queues(dev);
> > Â -ÂÂÂ for (i = 0; i < tun->numqueues; i++) {
> > -ÂÂÂÂÂÂÂ struct tun_file *tfile;
> > -
> > -ÂÂÂÂÂÂÂ tfile = rtnl_dereference(tun->tfiles[i]);
> > - tfile->socket.sk->sk_write_space(tfile->socket.sk);
> > -ÂÂÂ }
> > -
> > ÂÂÂÂÂ return 0;
> > Â }
> > Â @@ -3634,6 +3626,7 @@ static int tun_device_event(struct 
> > notifier_block *unused,
> > Â {
> > ÂÂÂÂÂ struct net_device *dev = netdev_notifier_info_to_dev(ptr);
> > ÂÂÂÂÂ struct tun_struct *tun = netdev_priv(dev);
> > +ÂÂÂ int i;
> > Â ÂÂÂÂÂ if (dev->rtnl_link_ops != &tun_link_ops)
> > ÂÂÂÂÂÂÂÂÂ return NOTIFY_DONE;
> > @@ -3643,6 +3636,14 @@ static int tun_device_event(struct 
> > notifier_block *unused,
> > ÂÂÂÂÂÂÂÂÂ if (tun_queue_resize(tun))
> > ÂÂÂÂÂÂÂÂÂÂÂÂÂ return NOTIFY_BAD;
> > ÂÂÂÂÂÂÂÂÂ break;
> > +ÂÂÂ case NETDEV_UP:
> > +ÂÂÂÂÂÂÂ for (i = 0; i < tun->numqueues; i++) {
> > +ÂÂÂÂÂÂÂÂÂÂÂ struct tun_file *tfile;
> > +
> > +ÂÂÂÂÂÂÂÂÂÂÂ tfile = rtnl_dereference(tun->tfiles[i]);
> > + tfile->socket.sk->sk_write_space(tfile->socket.sk);
> > +ÂÂÂÂÂÂÂ }
> > +ÂÂÂÂÂÂÂ break;
> > ÂÂÂÂÂ default:
> > ÂÂÂÂÂÂÂÂÂ break;
> > ÂÂÂÂÂ }
>
>
> Acked-by: Jason Wang <jasowang@xxxxxxxxxx)