As Scott put it:

> If we get a signal after the check, we want to be sure that we don't
> receive the reschedule IPI until after we're in the guest, so that it
> will cause another signal check.

we need to have interrupts disabled from the point we do signal_check()
all the way until we actually enter the guest.

This patch fixes potential signal loss races.

Reported-by: Scott Wood <scottw...@freescale.com>
Signed-off-by: Alexander Graf <ag...@suse.de>
---
 arch/powerpc/kvm/book3s_pr.c |   20 ++++++++++++++++++++
 1 files changed, 20 insertions(+), 0 deletions(-)

diff --git a/arch/powerpc/kvm/book3s_pr.c b/arch/powerpc/kvm/book3s_pr.c
index 3e1f68f..c1d5878 100644
--- a/arch/powerpc/kvm/book3s_pr.c
+++ b/arch/powerpc/kvm/book3s_pr.c
@@ -51,6 +51,8 @@ static int kvmppc_handle_ext(struct kvm_vcpu *vcpu, unsigned 
int exit_nr,
 #define MSR_USER32 MSR_USER
 #define MSR_USER64 MSR_USER
 #define HW_PAGE_SIZE PAGE_SIZE
+#define __hard_irq_disable local_irq_disable
+#define __hard_irq_enable local_irq_enable
 #endif
 
 void kvmppc_core_vcpu_load(struct kvm_vcpu *vcpu, int cpu)
@@ -778,7 +780,16 @@ program_interrupt:
                /* To avoid clobbering exit_reason, only check for signals if
                 * we aren't already exiting to userspace for some other
                 * reason. */
+
+               /*
+                * Interrupts could be timers for the guest which we have to
+                * inject again, so let's postpone them until we're in the guest
+                * and if we really did time things so badly, then we just exit
+                * again due to a host external interrupt.
+                */
+               __hard_irq_disable();
                if (signal_pending(current)) {
+                       __hard_irq_enable();
 #ifdef EXIT_DEBUG
                        printk(KERN_EMERG "KVM: Going back to host\n");
 #endif
@@ -961,8 +972,17 @@ int kvmppc_vcpu_run(struct kvm_run *kvm_run, struct 
kvm_vcpu *vcpu)
 
        kvmppc_core_prepare_to_enter(vcpu);
 
+       /*
+        * Interrupts could be timers for the guest which we have to inject
+        * again, so let's postpone them until we're in the guest and if we
+        * really did time things so badly, then we just exit again due to
+        * a host external interrupt.
+        */
+       __hard_irq_disable();
+
        /* No need to go into the guest when all we do is going out */
        if (signal_pending(current)) {
+               __hard_irq_enable();
                kvm_run->exit_reason = KVM_EXIT_INTR;
                ret = -EINTR;
                goto out;
-- 
1.6.0.2

--
To unsubscribe from this list: send the line "unsubscribe kvm" in
the body of a message to majord...@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Reply via email to