Skip to contentSkip to conversation

xen/events: fix binding user event channels to cpus

PATCH In review Juergen Gross 2 participants 2 messages Jun 21, 2019 – Jul 2, 2019

Subsystems (2)
lore
Share & tools
lore

Juergen Gross

author maintainer
#0 lore raw

PATCH 1/1 xen/events: fix binding user event channels to cpus

When binding an interdomain event channel to a vcpu via IOCTL_EVTCHN_BIND_INTERDOMAIN not only the event channel needs to be bound, but the affinity of the associated IRQi must be changed, too. Otherwise the IRQ and the event channel won't be moved to another vcpu in case the original vcpu they were bound to is going offline.

Cc <[email protected]> # 4.13
Fixes c48f64ab472389df ("xen-evtchn: Bind dyn evtchn:qemu-dm interrupt to next online VCPU")
---
 drivers/xen/events/events_base.c | 12 ++++++++++--
 drivers/xen/evtchn.c             |  2 +-
 include/xen/events.h             |  3 ++-
 3 files changed, 13 insertions(+), 4 deletions(-)
drivers/xen/events/events_base.c +10 -2
index ff9b51055b14..e718c8fea18b 100644
@@ -1294,7 +1294,7 @@ void rebind_evtchn_irq(int evtchn, int irq)
 }
 
 /* Rebind an evtchn so that it gets delivered to a specific cpu */
-int xen_rebind_evtchn_to_cpu(int evtchn, unsigned tcpu)
+static int xen_rebind_evtchn_to_cpu(int evtchn, unsigned int tcpu)
 {
 	struct evtchn_bind_vcpu bind_vcpu;
 	int masked;
@@ -1328,7 +1328,6 @@ int xen_rebind_evtchn_to_cpu(int evtchn, unsigned tcpu)
 
 	return 0;
 }
-EXPORT_SYMBOL_GPL(xen_rebind_evtchn_to_cpu);
 
 static int set_affinity_irq(struct irq_data *data, const struct cpumask *dest,
 			    bool force)
@@ -1342,6 +1341,15 @@ static int set_affinity_irq(struct irq_data *data, const struct cpumask *dest,
 	return ret;
 }
 
+/* To be called with desc->lock held. */
+int xen_set_affinity_evtchn(struct irq_desc *desc, unsigned int tcpu)
+{
+	struct irq_data *d = irq_desc_get_irq_data(desc);
+
+	return set_affinity_irq(d, cpumask_of(tcpu), false);
+}
+EXPORT_SYMBOL_GPL(xen_set_affinity_evtchn);
+
 static void enable_dynirq(struct irq_data *data)
 {
 	int evtchn = evtchn_from_irq(data->irq);
drivers/xen/evtchn.c +1 -1
index f341b016672f..052b55a14ebc 100644
@@ -447,7 +447,7 @@ static void evtchn_bind_interdom_next_vcpu(int evtchn)
 	this_cpu_write(bind_last_selected_cpu, selected_cpu);
 
 	/* unmask expects irqs to be disabled */
-	xen_rebind_evtchn_to_cpu(evtchn, selected_cpu);
+	xen_set_affinity_evtchn(desc, selected_cpu);
 	raw_spin_unlock_irqrestore(&desc->lock, flags);
 }
 
include/xen/events.h +2 -1
index a48897199975..c0e6a0598397 100644
@@ -3,6 +3,7 @@
 #define _XEN_EVENTS_H
 
 #include <linux/interrupt.h>
+#include <linux/irq.h>
 #ifdef CONFIG_PCI_MSI
 #include <linux/msi.h>
 #endif
@@ -59,7 +60,7 @@ void evtchn_put(unsigned int evtchn);
 
 void xen_send_IPI_one(unsigned int cpu, enum ipi_vector vector);
 void rebind_evtchn_irq(int evtchn, int irq);
-int xen_rebind_evtchn_to_cpu(int evtchn, unsigned tcpu);
+int xen_set_affinity_evtchn(struct irq_desc *desc, unsigned int tcpu);
 
 static inline void notify_remote_via_evtchn(int port)
 {
Signature · 1 line
2.16.4
Reviewed-by on v1 1/1
#1 lore raw
Quoted 9 lines · Juergen Gross

When binding an interdomain event channel to a vcpu via IOCTL_EVTCHN_BIND_INTERDOMAIN not only the event channel needs to be bound, but the affinity of the associated IRQi must be changed, too. Otherwise the IRQ and the event channel won't be moved to another vcpu in case the original vcpu they were bound to is going offline.

Cc <[email protected]> # 4.13
Fixes c48f64ab472389df ("xen-evtchn: Bind dyn evtchn:qemu-dm interrupt to next online VCPU")