Re: serious netpoll bug w/NAPI

[Matt Mackall <mpm@xxxxxxxxxxx>]

On Thu, Feb 10, 2005 at 10:16:08AM +0100, Martin Josefsson wrote:
> On Wed, 9 Feb 2005, Matt Mackall wrote:
> 
> > --- mm1npc.orig/net/core/dev.c      2005-02-09 14:15:11.236086000 -0800
> > +++ mm1npc/net/core/dev.c   2005-02-09 14:15:13.710042000 -0800
> > @@ -1772,6 +1772,7 @@
> >
> >             dev = list_entry(queue->poll_list.next,
> >                              struct net_device, poll_list);
> > +           netpoll_poll_lock(dev);
> >
> >             if (dev->quota <= 0 || dev->poll(dev, &budget)) {
> >                     local_irq_disable();
> > @@ -1782,9 +1783,11 @@
> >                     else
> >                             dev->quota = dev->weight;
> >             } else {
> > +                   netpoll_poll_unlock(dev);
> >                     dev_put(dev);
> >                     local_irq_disable();
> >             }
> > +           netpoll_poll_unlock(dev);
> >
> >  #ifdef CONFIG_KGDBOE
> >             kgdb_process_breakpoint();
> 
> Double unlock?

Yes. The second one should instead be up a few lines here:

                if (dev->quota <= 0 || dev->poll(dev, &budget)) {
+                       netpoll_poll_unlock(dev);
                        local_irq_disable();


-- 
Mathematics is the supreme nostalgia of our time.