KVM: s390: fix external call injection without sigp interpretation

Commit ea5f49692575 ("KVM: s390: only one external call may be pending
at a time") introduced a bug on machines that don't have SIGP
interpretation facility installed.
The injection of an external call will now always fail with -EBUSY
(if none is already pending).

This leads to the following symptoms:
- An external call will be injected but with the wrong "src cpu id",
  as this id will not be remembered.
- The target vcpu will not be woken up, therefore the guest will hang if
  it cannot deal with unexpected failures of the SIGP EXTERNAL CALL
  instruction.
- If an external call is already pending, -EBUSY will not be reported.

Reviewed-by: Christian Borntraeger <borntraeger@de.ibm.com>
Reviewed-by: Jens Freimann <jfrei@linux.vnet.ibm.com>
Signed-off-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
Cc: stable@vger.kernel.org # v4.0
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>

diff --git a/arch/s390/kvm/interrupt.c b/arch/s390/kvm/interrupt.c
index 9de47265ef73da07ffd7ef37337bf2e44e59bd46..2836b25f63d3b883a33d607c0d45866af41c4270 100644 (file)
--- a/arch/s390/kvm/interrupt.c
+++ b/arch/s390/kvm/interrupt.c
@@ -1061,7 +1061,7 @@ static int __inject_extcall(struct kvm_vcpu *vcpu, struct kvm_s390_irq *irq)
        if (sclp_has_sigpif())
                return __inject_extcall_sigpif(vcpu, src_id);
 
-       if (!test_and_set_bit(IRQ_PEND_EXT_EXTERNAL, &li->pending_irqs))
+       if (test_and_set_bit(IRQ_PEND_EXT_EXTERNAL, &li->pending_irqs))
                return -EBUSY;
        *extcall = irq->u.extcall;
        atomic_set_mask(CPUSTAT_EXT_INT, li->cpuflags);