KVM: PPC: BookE: fix sleep with interrupts disabled
authorScott Wood <scottwood@freescale.com>
Thu, 30 Sep 2010 19:28:50 +0000 (14:28 -0500)
committerMarcelo Tosatti <mtosatti@redhat.com>
Fri, 5 Nov 2010 16:42:28 +0000 (14:42 -0200)
It is not legal to call mutex_lock() with interrupts disabled.
This will assert with debug checks enabled.

If there's a real need to disable interrupts here, it could be done
after the mutex is acquired -- but I don't see why it's needed at all.

Signed-off-by: Scott Wood <scottwood@freescale.com>
Reviewed-by: Christian Ehrhardt <ehrhardt@linux.vnet.ibm.com>
Signed-off-by: Alexander Graf <agraf@suse.de>
arch/powerpc/kvm/timing.c

index 46fa04f12a9b979c7cbe50c703d43e2c2705604e..a021f5827a336ce97b6c62ed5b8d0ca625d571bf 100644 (file)
@@ -35,7 +35,6 @@ void kvmppc_init_timing_stats(struct kvm_vcpu *vcpu)
        int i;
 
        /* pause guest execution to avoid concurrent updates */
-       local_irq_disable();
        mutex_lock(&vcpu->mutex);
 
        vcpu->arch.last_exit_type = 0xDEAD;
@@ -51,7 +50,6 @@ void kvmppc_init_timing_stats(struct kvm_vcpu *vcpu)
        vcpu->arch.timing_last_enter.tv64 = 0;
 
        mutex_unlock(&vcpu->mutex);
-       local_irq_enable();
 }
 
 static void add_exit_timing(struct kvm_vcpu *vcpu, u64 duration, int type)