On Mon, Sep 23, 2019 at 7:47 PM Hayes Wang <hayesw...@realtek.com> wrote:
>
> Prashant Malani [mailto:pmal...@chromium.org]
> > Sent: Tuesday, September 24, 2019 6:27 AM
> > To: Hayes Wang
> [...]
> > - do {
> > + while (1) {
> > struct tx_agg *agg;
> > + struct net_device *netdev = tp->netdev;
> >
> > if (skb_queue_empty(&tp->tx_queue))
> > break;
> > @@ -2188,26 +2189,25 @@ static void tx_bottom(struct r8152 *tp)
> > break;
> >
> > res = r8152_tx_agg_fill(tp, agg);
> > - if (res) {
> > - struct net_device *netdev = tp->netdev;
> > + if (!res)
> > + break;
>
> I let the loop run continually until an error occurs or the queue is empty.
> However, you stop the loop when r8152_tx_agg_fill() is successful.
Hayes,
Are you sure about both assertions?
The do/while loop exits if "res == 0". Isn't that the same as "!res"?
> If an error occurs continually, the loop may not be broken.
And what prevents that from happening with the current code?
Should current code break out of the loop in -ENODEV case, right?
That would be more obvious if the code inside the loop were:
...
res = r8152_tx_agg_fill(tp, agg);
if (res == -ENODEV) {
...
break;
}
if (!res)
break;
...
(Or whatever the right code is to "loop until an error occurs or queue
is empty").
cheers,
grant
>
> > - if (res == -ENODEV) {
> > - rtl_set_unplug(tp);
> > - netif_device_detach(netdev);
> > - } else {
> > - struct net_device_stats *stats =
> > &netdev->stats;
> > - unsigned long flags;
> > + if (res == -ENODEV) {
> > + rtl_set_unplug(tp);
> > + netif_device_detach(netdev);
> > + } else {
> > + struct net_device_stats *stats = &netdev->stats;
> > + unsigned long flags;
> >
> > - netif_warn(tp, tx_err, netdev,
> > - "failed tx_urb %d\n", res);
> > - stats->tx_dropped += agg->skb_num;
> > + netif_warn(tp, tx_err, netdev,
> > + "failed tx_urb %d\n", res);
> > + stats->tx_dropped += agg->skb_num;
> >
> > - spin_lock_irqsave(&tp->tx_lock, flags);
> > - list_add_tail(&agg->list, &tp->tx_free);
> > - spin_unlock_irqrestore(&tp->tx_lock, flags);
> > - }
> > + spin_lock_irqsave(&tp->tx_lock, flags);
> > + list_add_tail(&agg->list, &tp->tx_free);
> > + spin_unlock_irqrestore(&tp->tx_lock, flags);
> > }
> > - } while (res == 0);
> > + }
>
> I think the behavior is different from the current one.
>
> Best Regards,
> Hayes
>
