RE: [PATCH stable] KVM: x86: Fix lost interrupt on irr_pending race

["Saso Slavicic" <saso.linux@astim.si>]

> From: Paolo Bonzini
> Sent: Tuesday, April 21, 2015 10:48 AM


Hi, big thanks to all involved in this and to Brad for endless reboots ;-)


>> Signed-off-by: Nadav Amit <namit@cs.technion.ac.il>
>> Fixes: 33e4c68656a2e461b296ce714ec322978de85412
>> Cc: stable@vger.kernel.org # 2.6.32+
>> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
>> ---
>> 	The race was reported in 3.17+ by Brad Campbell and in
>> 	2.6.32 by Saso Slavicic, so it qualifies for stable.

> Patch for kernels before 3.17:

This will probably end up in RHEL6 sooner or later (but probably not before
6.7)?
As I like to experiment a bit, would this patch do for -2.6.32-504.12.2.el6
kernel? The code in that function is somewhat different...

--- a/arch/x86/kvm/lapic.c
+++ b/arch/x86/kvm/lapic.c
@@ -230,8 +230,13 @@
 
 static inline int apic_test_and_set_irr(int vec, struct kvm_lapic *apic)
 {
+	int ret = apic_test_and_set_vector(vec, apic->regs + APIC_IRR);
+	/*
+	 * irr_pending must be true if any interrupt is pending; set it
after
+	 * APIC_IRR to avoid race with apic_clear_irr
+	 */
 	apic->irr_pending = true;
-	return apic_test_and_set_vector(vec, apic->regs + APIC_IRR);
+	return ret;
 }
 
 static inline int apic_search_irr(struct kvm_lapic *apic)


Regards,
Saso Slavicic