From: "Rémi Denis-Courmont" <remi.denis-courmont@xxxxxxxxx>
Date: Mon, 20 Jul 2009 09:35:18 +0300
> In the racy case, tx_complete() fires, incrementation in
> usbpn_xmit() and decrementation in tx_complete() will cancel each
> other. So, regardless of their respective order, tx_queue will be
> unchanged, and the assertion remains valid. A fortiori, it works
> fine if usbpn_xmit() races with more than one call of tx_complete().
You can only do the wakeup race free without deadlocking the TX queue
state if you grab the TX path lock, as TG3 does:
/* Need to make the tx_cons update visible to tg3_start_xmit()
* before checking for netif_queue_stopped(). Without the
* memory barrier, there is a small possibility that tg3_start_xmit()
* will miss it and cause the queue to be stopped forever.
*/
smp_mb();
if (unlikely(netif_queue_stopped(tp->dev) &&
(tg3_tx_avail(tp) > TG3_TX_WAKEUP_THRESH(tp)))) {
netif_tx_lock(tp->dev);
if (netif_queue_stopped(tp->dev) &&
(tg3_tx_avail(tp) > TG3_TX_WAKEUP_THRESH(tp)))
netif_wake_queue(tp->dev);
netif_tx_unlock(tp->dev);
}
--
To unsubscribe from this list: send the line "unsubscribe linux-usb" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
