Merge tag 'kvm-s390-master-20150730' of git://git.kernel.org/pub/scm/linux/kernel...

KVM: s390: bugfix for kvm/master (4.2)

Here is a bugfix for a regression that was introduced after 4.1
with the commit commit 785dbef407d8 ("KVM: s390: optimize round
trip time in request handling"). After lots of cpu hotplugs in the
guest (online/offline) sometimes a guest CPU did loop within host
KVM code. Reason was that PROG_REQUEST was set in the sie control
block, but no request was pending. This made commit 785dbef407d8
the suspect and changing that area to always reset PROG_REQUEST
did indeed fix the problem.

Special thanks to David Hildenbrand, who helped understanding the
exact sequence that led to the problem.

diff --git a/arch/s390/kvm/kvm-s390.c b/arch/s390/kvm/kvm-s390.c
index 2078f92d15ac90adcfec617b46df750e073b76d0..f32f843a3631359e49b88169ab8a1eed2b76b946 100644 (file)
--- a/arch/s390/kvm/kvm-s390.c
+++ b/arch/s390/kvm/kvm-s390.c
@@ -1742,10 +1742,10 @@ static bool ibs_enabled(struct kvm_vcpu *vcpu)
 
 static int kvm_s390_handle_requests(struct kvm_vcpu *vcpu)
 {
-       if (!vcpu->requests)
-               return 0;
 retry:
        kvm_s390_vcpu_request_handled(vcpu);
+       if (!vcpu->requests)
+               return 0;
        /*
         * We use MMU_RELOAD just to re-arm the ipte notifier for the
         * guest prefix page. gmap_ipte_notify will wait on the ptl lock.