Message ID | 1394712342-15778-64-Taiwan-albertk@realtek.com |
---|---|
State | Superseded, archived |
Delegated to: | David Miller |
Headers | show |
Hayes Wang <hayeswang@realtek.com> writes: > It would cause dead lock for runtime suspend, when the workqueue > is running and runtime suspend occurs before the workqueue wakes > up the device. The rtl8152_suspend() waits the workqueue to finish > because of calling cancel_delayed_work_sync(). The workqueue waits > the suspend function to finish for waking up the device because of > calling usb_autopm_get_interface(). > > Signed-off-by: Hayes Wang <hayeswang@realtek.com> > --- > drivers/net/usb/r8152.c | 3 ++- > 1 file changed, 2 insertions(+), 1 deletion(-) > > diff --git a/drivers/net/usb/r8152.c b/drivers/net/usb/r8152.c > index 864159e..7d4e55a 100644 > --- a/drivers/net/usb/r8152.c > +++ b/drivers/net/usb/r8152.c > @@ -3200,12 +3200,13 @@ static int rtl8152_suspend(struct usb_interface *intf, pm_message_t message) > if (netif_running(tp->netdev)) { > clear_bit(WORK_ENABLE, &tp->flags); > usb_kill_urb(tp->intr_urb); > - cancel_delayed_work_sync(&tp->schedule); > tasklet_disable(&tp->tl); > if (test_bit(SELECTIVE_SUSPEND, &tp->flags)) { > + cancel_delayed_work(&tp->schedule); > rtl_stop_rx(tp); > rtl_runtime_suspend_enable(tp, true); > } else { > + cancel_delayed_work_sync(&tp->schedule); > tp->rtl_ops.down(tp); > } > tasklet_enable(&tp->tl); This looks strange to me. The delayed work will cause an immediate resume due to the usb_autopm_get_interface() it starts with. Wouldn't it be better to just prevent runtime suspending by returning -EBUSY if there is any delayed work scheduled? Bjørn -- To unsubscribe from this list: send the line "unsubscribe netdev" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On Fri, 2014-10-17 at 13:55 +0800, Hayes Wang wrote: > It would cause dead lock for runtime suspend, when the workqueue > is running and runtime suspend occurs before the workqueue wakes > up the device. The rtl8152_suspend() waits the workqueue to finish > because of calling cancel_delayed_work_sync(). The workqueue waits > the suspend function to finish for waking up the device because of > calling usb_autopm_get_interface(). The diagnosis is good, the fix is not good. It opens a race during which the queued work can touch a suspended device. Regards Oliver -- To unsubscribe from this list: send the line "unsubscribe netdev" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
Oliver Neukum [mailto:oliver@neukum.org] > Sent: Sunday, October 19, 2014 3:48 AM [...] > The diagnosis is good, the fix is not good. It opens a race > during which the queued work can touch a suspended device. The delayed work would wake up the device by calling usb_autopm_get_interface() before accessing the device. Besides, there is a mutex to avoid the race. Best Regards, Hayes -- To unsubscribe from this list: send the line "unsubscribe netdev" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
diff --git a/drivers/net/usb/r8152.c b/drivers/net/usb/r8152.c index 864159e..7d4e55a 100644 --- a/drivers/net/usb/r8152.c +++ b/drivers/net/usb/r8152.c @@ -3200,12 +3200,13 @@ static int rtl8152_suspend(struct usb_interface *intf, pm_message_t message) if (netif_running(tp->netdev)) { clear_bit(WORK_ENABLE, &tp->flags); usb_kill_urb(tp->intr_urb); - cancel_delayed_work_sync(&tp->schedule); tasklet_disable(&tp->tl); if (test_bit(SELECTIVE_SUSPEND, &tp->flags)) { + cancel_delayed_work(&tp->schedule); rtl_stop_rx(tp); rtl_runtime_suspend_enable(tp, true); } else { + cancel_delayed_work_sync(&tp->schedule); tp->rtl_ops.down(tp); } tasklet_enable(&tp->tl);
It would cause dead lock for runtime suspend, when the workqueue is running and runtime suspend occurs before the workqueue wakes up the device. The rtl8152_suspend() waits the workqueue to finish because of calling cancel_delayed_work_sync(). The workqueue waits the suspend function to finish for waking up the device because of calling usb_autopm_get_interface(). Signed-off-by: Hayes Wang <hayeswang@realtek.com> --- drivers/net/usb/r8152.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)