Re: [Qemu-devel] BUG: KASAN: use-after-free in free_old_xmit_skbs

[Jason Wang]

On 2017年06月23日 02:53, Michael S. Tsirkin wrote:
> On Thu, Jun 22, 2017 at 08:15:58AM +0200, jean-philippe menil wrote:
> > 2017-06-06 1:52 GMT+02:00 Michael S. Tsirkin <address@hidden>:
> >
> >      On Mon, Jun 05, 2017 at 05:08:25AM +0300, Michael S. Tsirkin wrote:
> >      > On Mon, Jun 05, 2017 at 12:48:53AM +0200, Jean-Philippe Menil wrote:
> >      > > Hi,
> >      > >
> >      > > while playing with xdp and ebpf, i'm hitting the following:
> >      > >
> >      > > [  309.993136]
> >      > > ==================================================================
> >      > > [  309.994735] BUG: KASAN: use-after-free in
> >      > > free_old_xmit_skbs.isra.29+0x2b7/0x2e0 [virtio_net]
> >      > > [  309.998396] Read of size 8 at addr ffff88006aa64220 by task sshd/323
> >      > > [  310.000650]
> >      > > [  310.002305] CPU: 1 PID: 323 Comm: sshd Not tainted 4.12.0-rc3+ #2
> >      > > [  310.004018] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996),
> >      BIOS
> >      > > 1.10.2-20170228_101828-anatol 04/01/2014
> ...
>
> >      >
> >      > Since commit 680557cf79f82623e2c4fd42733077d60a843513
> >      >     virtio_net: rework mergeable buffer handling
> >      >
> >      > we no longer must do the resets, we now have enough space
> >      > to store a bit saying whether a buffer is xdp one or not.
> >      >
> >      > And that's probably a cleaner way to fix these issues than
> >      > try to find and fix the race condition.
> >      >
> >      > John?
> >      >
> >      > --
> >      > MST
> >
> >
> >      I think I see the source of the race. virtio net calls
> >      netif_device_detach and assumes no packets will be sent after
> >      this point. However, all it does is stop all queues so
> >      no new packets will be transmitted.
> >
> >      Try locking with HARD_TX_LOCK?
> >     
> >
> >      --
> >      MST
> >
> >
> > Hi Michael,
> >
> > from what i see, the race appear when we hit virtnet_reset in virtnet_xdp_set.
> > virtnet_reset
> >    _remove_vq_common
> >      virtnet_del_vqs
> >        virtnet_free_queues
> >          kfree(vi->sq)
> > when the xdp program (with two instances of the program to trigger it faster)
> > is added or removed.
> >
> > It's easily repeatable, with 2 cpus and 4 queues on the qemu command line,
> > running the xdp_ttl tool from Jesper.
> >
> > For now, i'm able to continue my qualification, testing if xdp_qp is not null,
> > but do not seem to be a sustainable trick.
> > if (xdp_qp && vi->xdp_queues_pairs != xdp_qp)
> >
> > Maybe it will be more clear to you with theses informations.
> >
> > Best regards.
> >
> > Jean-Philippe
>
> I'm pretty clear about the issue here, I was trying to figure out a fix.
> Jason, any thoughts?

Hi Jean:

Does the following fix this issue? (I can't reproduce it locally through 
xdp_ttl)

Thanks

diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
index 1f8c15c..3e65c3f 100644
--- a/drivers/net/virtio_net.c
+++ b/drivers/net/virtio_net.c
@@ -1801,7 +1801,9 @@ static void virtnet_freeze_down(struct 
virtio_device *vdev)
        /* Make sure no work handler is accessing the device */
        flush_work(&vi->config_work);

+       netif_tx_lock_bh(vi->dev);
        netif_device_detach(vi->dev);
+       netif_tx_unlock_bh(vi->dev);
        cancel_delayed_work_sync(&vi->refill);