xen: fix unmask_evtchn for HVM guests

When unmask_evtchn is called, if we already have an event pending, we
just set evtchn_pending_sel waiting for irq to be re-enabled. That is
because x86 pv guests overwrite the irq_enable pvops with
xen_irq_enable_direct that also handles pending events.

However x86 HVM guests and ARM guests do not change or do not have the
irq_enable pvop, so this scheme doesn't work properly for them.
Considering that having a pending irq when unmask_evtchn is called is
not very common, and it is better to keep the native_irq_enable
implementation for HVM guests and ARM guests, the best thing to do is
just using the EVTCHNOP_unmask callback (Xen re-injects pending
events in response).

Considering that this patch fixes a bug on x86 for current PV on HVM
guests, I'll resend it separately.

Signed-off-by: Stefano Stabellini <stefano.stabellini@eu.citrix.com>

diff --git a/drivers/xen/events.c b/drivers/xen/events.c
index eae0d0bbb0c1419011275d8cca7464533ecfb657..0132505dc13fd9882acd46ea6b8ee6d496322ded 100644 (file)
--- a/drivers/xen/events.c
+++ b/drivers/xen/events.c
@@ -372,8 +372,11 @@ static void unmask_evtchn(int port)
 
        BUG_ON(!irqs_disabled());
 
-       /* Slow path (hypercall) if this is a non-local port. */
-       if (unlikely(cpu != cpu_from_evtchn(port))) {
+       /* Slow path (hypercall) if this is a non-local port or if this is
+        * an hvm domain and an event is pending (hvm domains don't have
+        * their own implementation of irq_enable). */
+       if (unlikely((cpu != cpu_from_evtchn(port)) ||
+                       (xen_hvm_domain() && sync_test_bit(port, &s->evtchn_pending[0])))) {
                struct evtchn_unmask unmask = { .port = port };
                (void)HYPERVISOR_event_channel_op(EVTCHNOP_unmask, &unmask);
        } else {