From: Herbert Xu <herbert@gondor.apana.org.au>
Date: Mon, 18 Aug 2008 11:25:16 +1000

So what's the issue with dev_queue_xmit? That should be taken care
of by something like rcu_barrier_bh, no?

The code in dev_queue_xmit() used to resample the pointer.  It
relied upon the fact that we always used the same top-level
spinlock to set the qdisc.

But now that the lock is in the qdisc itself instead of the
netdevice or netdev_queue, that no longer works.

That's why I got rid of the "resample" code in these places
and tried to move everything into RCU.

I think I see another way out of this:

1) Add __QDISC_STATE_DEACTIVATE.

2) Set it right before dev_deactivate() swaps resets the qdisc
   pointer.

3) Test it in dev_queue_xmit() et al. once the qdisc root lock is
   acquired, and drop lock and resample ->qdisc if
   __QDISC_STATE_DEACTIVATE is set.