And I think I see what the problem is:

quoted

+	if (unlikely(netif_queue_stopped(dev) &&
+		     TX_BUFFS_AVAIL(gp) > (MAX_SKB_FRAGS + 1))) {
+		netif_tx_lock(dev);
+		if (netif_queue_stopped(dev) &&
+		    TX_BUFFS_AVAIL(gp) > (MAX_SKB_FRAGS + 1))
+			netif_wake_queue(dev);
+		netif_tx_unlock(dev);
+	}
 }

Don't use netif_tx_lock(), that has a loop and multiple atomics :-)

It's going to grab a special global TX lock, and then grab a lock for
TX queue zero, and finally set an atomic state bit in TX queue zero.

Take a look at the implementation in netdevice.h

Ah good point ! I think I stole that from another driver (or I just had
a brain fart), indeed, it's bad.

It's a special "lock everything TX", a mechanism for multiqueue
drivers to shut quiesce all TX queue activity safely in one operation.

Instead, do something like:

	struct netdev_queue *txq = netdev_get_tx_queue(dev, 0);

	__netif_tx_lock(txq, smp_processor_id();
	...
	__netif_tx_unlock(txq);

and I bet your TX numbers improve a bit.

Right, I'll give that a go. With the assistance of the other Ben H I've
been able to simplify the driver a lot more now too. mutex and remaining
lock are gone, rtnl lock does the job fine for synchronizing vs. reset
task and I cleared up a ton more of unused bits and pieces now that we
don't deal with link timer when the thing is off anymore.

I'll have a new patch later today hopefully with new numbers.

Cheers,
Ben.