[patch] preserve int status through call to kvm_insert_vmm_mapping

[Jes Sorensen <jes@xxxxxxx>]

Hi,

Here's another one, this one actually should be safe as it is, but with
this change it's safe, even if the code is moved around later.

Cheers,
Jes

Preserve interrupt status around call to kvm_insert_vmm_mappin()
in kvm_vcpu_pre_transition().

Signed-off-by: Jes Sorensen <jes@xxxxxxx>

---
 arch/ia64/kvm/kvm-ia64.c |    7 ++++++-
 1 file changed, 6 insertions(+), 1 deletion(-)

Index: linux-2.6.git/arch/ia64/kvm/kvm-ia64.c
===================================================================
--- linux-2.6.git.orig/arch/ia64/kvm/kvm-ia64.c
+++ linux-2.6.git/arch/ia64/kvm/kvm-ia64.c
@@ -567,6 +567,8 @@
 
 static int kvm_vcpu_pre_transition(struct kvm_vcpu *vcpu)
 {
+	unsigned long psr;
+	int r;
 	int cpu = smp_processor_id();
 
 	if (vcpu->arch.last_run_cpu != cpu ||
@@ -578,7 +580,10 @@
 
 	vcpu->arch.host_rr6 = ia64_get_rr(RR6);
 	vti_set_rr6(vcpu->arch.vmm_rr);
-	return kvm_insert_vmm_mapping(vcpu);
+	local_irq_save(psr);
+	r = kvm_insert_vmm_mapping(vcpu);
+	local_irq_restore(psr);
+	return r;
 }
 
 static void kvm_vcpu_post_transition(struct kvm_vcpu *vcpu)