Even though the eventfd is released after the KVM SRCU grace period elapses, the conn_to_evt data structure itself is not; it uses RCU internally, instead. Fix the read-side critical section to happen under rcu_read_lock/unlock; the result is still protected by vcpu->kvm->srcu. Signed-off-by: Paolo Bonzini <pbonzini@xxxxxxxxxx> --- arch/x86/kvm/hyperv.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) diff --git a/arch/x86/kvm/hyperv.c b/arch/x86/kvm/hyperv.c index 14dd5e5010a2..5708e951a5c6 100644 --- a/arch/x86/kvm/hyperv.c +++ b/arch/x86/kvm/hyperv.c @@ -1296,8 +1296,10 @@ static u16 kvm_hvcall_signal_event(struct kvm_vcpu *vcpu, bool fast, u64 param) if (param & ~KVM_HYPERV_CONN_ID_MASK) return HV_STATUS_INVALID_HYPERCALL_INPUT; - /* conn_to_evt is protected by vcpu->kvm->srcu */ + /* the eventfd is protected by vcpu->kvm->srcu, but conn_to_evt isn't */ + rcu_read_lock(); eventfd = idr_find(&vcpu->kvm->arch.hyperv.conn_to_evt, param); + rcu_read_unlock(); if (!eventfd) return HV_STATUS_INVALID_PORT_ID; -- 1.8.3.1
