Rearrange the tests in unbind_from_irq() so that we can still unbind
an irq even if the underlying event channel is bad. This allows a
device driver to shuffle its irqs on save/restore before the
underlying event channels have been fixed up.
Signed-off-by: Jeremy Fitzhardinge <jeremy.fitzhardinge@citrix.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
spin_lock(&irq_mapping_update_lock);
- if (VALID_EVTCHN(evtchn) && (--irq_bindcount[irq] == 0)) {
+ if ((--irq_bindcount[irq] == 0) && VALID_EVTCHN(evtchn)) {
close.port = evtchn;
if (HYPERVISOR_event_channel_op(EVTCHNOP_close, &close) != 0)
BUG();
evtchn_to_irq[evtchn] = -1;
irq_info[irq] = IRQ_UNBOUND;
- dynamic_irq_init(irq);
+ dynamic_irq_cleanup(irq);
}
spin_unlock(&irq_mapping_update_lock);