Messages in this thread | | | Date | Wed, 22 Jan 2020 11:40:09 +1100 (AEDT) | From | Finn Thain <> | Subject | Re: [PATCH net v2 01/12] net/sonic: Add mutual exclusion for accessing shared state |
| |
On Tue, 21 Jan 2020, Eric Dumazet wrote:
> On 1/21/20 3:33 PM, Finn Thain wrote: > > On Tue, 21 Jan 2020, Eric Dumazet wrote: > > > >> On 1/21/20 1:22 PM, Finn Thain wrote: > >>> The netif_stop_queue() call in sonic_send_packet() races with the > >>> netif_wake_queue() call in sonic_interrupt(). This causes issues > >>> like "NETDEV WATCHDOG: eth0 (macsonic): transmit queue 0 timed out". > >>> Fix this by disabling interrupts when accessing tx_skb[] and next_tx. > >>> Update a comment to clarify the synchronization properties. > >>> > >>> Fixes: efcce839360f ("[PATCH] macsonic/jazzsonic network drivers update") > >>> Tested-by: Stan Johnson <userm57@yahoo.com> > >>> Signed-off-by: Finn Thain <fthain@telegraphics.com.au> > >> > >>> @@ -284,9 +287,16 @@ static irqreturn_t sonic_interrupt(int irq, void *dev_id) > >>> struct net_device *dev = dev_id; > >>> struct sonic_local *lp = netdev_priv(dev); > >>> int status; > >>> + unsigned long flags; > >>> + > >>> + spin_lock_irqsave(&lp->lock, flags); > >> > >> > >> This is a hard irq handler, no need to block hard irqs. > >> > >> spin_lock() here is enough. > >> > > > > Well, yes, assuming we're dealing with SMP [1]. Probably just disabling > > pre-emption is all that will ever be needed. > > > > Anyway, the real problem solved by disabling irqs is that macsonic must > > avoid re-entrance of sonic_interrupt(). [2] > > > > [1] > > https://lore.kernel.org/netdev/alpine.LNX.2.21.1.2001211026190.8@nippy.intranet/T/#m0523c8b2a26a410ed56889d9230c37ba1160d40a > > > > [2] > > https://lore.kernel.org/netdev/alpine.LNX.2.21.1.2001211026190.8@nippy.intranet/T/#m1c8ca580d2b45e61a628d17839978d0bd5aaf061 > > >
I was overlooking the fact that sonic_send_packet() really does have to exclude sonic_interrupt(). So disabling irqs is mandatory here, right?
> Oh well... > > I would rather keep the m68k specific wrapper... >
If disabling irqs is unavoidable, the wrapper is redundant.
> Please add a fat comment of why spin_lock_irqsave() is used, to avoid a > future 'cleanup', since average network developer wont be aware of m68k > oddities. >
OK.
| |