Stephen Hemminger <[EMAIL PROTECTED]> :
> Francois Romieu <[EMAIL PROTECTED]> wrote:
> > Stephen Hemminger <[EMAIL PROTECTED]> :
> > [...]
> > > @@ -1682,12 +1685,11 @@ static void rtl8139_tx_timeout_task (voi
> > > rtl8139_tx_clear (tp);
> > > spin_unlock_irq(&tp->lock);
> > >
> > > + netif_poll_enable();
> > ^ -> dev
> > > +
> > > /* ...and finally, reset everything */
> > > - if (netif_running(dev)) {
> > > - rtl8139_hw_start (dev);
> > > - netif_wake_queue (dev);
> > > - }
> > > - spin_unlock_bh(&tp->rx_lock);
> > > + rtl8139_hw_start (dev);
> > > + netif_wake_queue (dev);
> > > }
> >
> > rtl8139_hw_start() may mess with cur_rx, whence a race with rtl8139_rx()
> > if an in-flight interruption enables it a bit too fast. I'd rather go
> > with:
>
> but rt8139_rx is not possible here because we have blocked the poll
> routine from starting. Basically it uses the NAPI rx scheduler bit
> to replace the rx_lock.
1 - the irq handler is waiting for tp->lock
2 - rtl8139_tx_timeout_task releases the lock
3 - rtl8139_tx_timeout_task issues netif_poll_enable
4 - the irq handler schedules ->poll(), returns
5 - rtl8139_hw_start() races with ->poll(), aka rtl8139_rx(), for cur_rx
--
Ueimor
-
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to [EMAIL PROTECTED]
More majordomo info at http://vger.kernel.org/majordomo-info.html
