Skip to contentSkip to conversation

KVM: x86: fix deadlock in clock-in-progress request handling

PATCH Marcelo Tosatti 2 participants 2 messages Mar 18, 2013

Subsystems (3)
x86 KVM KVM x86
lore
Share & tools
lore

Marcelo Tosatti

author
#0 lore raw

There is a deadlock in pvclock handling:

cpu0: cpu1:
kvm_gen_update_masterclock()
kvm_guest_time_update()
spin_lock(pvclock_gtod_sync_lock)
local_irq_save(flags)

spin_lock(pvclock_gtod_sync_lock)

kvm_make_mclock_inprogress_request(kvm)
make_all_cpus_request()
smp_call_function_many()

Now if smp_call_function_many() called by cpu0 tries to call function on cpu1 there will be a deadlock.

Fix by moving pvclock_gtod_sync_lock protected section outside irq disabled section.

Analyzed by Gleb Natapov <[email protected]>

Reported-and-Tested-by Yongjie Ren <[email protected]>
Signed-off-by Marcelo Tosatti <[email protected]>
arch/x86/kvm/x86.c +9 -9
index f71500a..f7c850b 100644
@@ -1416,15 +1416,6 @@ static int kvm_guest_time_update(struct kvm_vcpu *v)
 	kernel_ns = 0;
 	host_tsc = 0;
 
-	/* Keep irq disabled to prevent changes to the clock */
-	local_irq_save(flags);
-	this_tsc_khz = __get_cpu_var(cpu_tsc_khz);
-	if (unlikely(this_tsc_khz == 0)) {
-		local_irq_restore(flags);
-		kvm_make_request(KVM_REQ_CLOCK_UPDATE, v);
-		return 1;
-	}
-
 	/*
 	 * If the host uses TSC clock, then passthrough TSC as stable
 	 * to the guest.

@@ -1436,6 +1427,15 @@ static int kvm_guest_time_update(struct kvm_vcpu *v)
 		kernel_ns = ka->master_kernel_ns;
 	}
 	spin_unlock(&ka->pvclock_gtod_sync_lock);
+
+	/* Keep irq disabled to prevent changes to the clock */
+	local_irq_save(flags);
+	this_tsc_khz = __get_cpu_var(cpu_tsc_khz);
+	if (unlikely(this_tsc_khz == 0)) {
+		local_irq_restore(flags);
+		kvm_make_request(KVM_REQ_CLOCK_UPDATE, v);
+		return 1;
+	}
 	if (!use_master_clock) {
 		host_tsc = native_read_tsc();
 		kernel_ns = get_kernel_ns();
Acked-by
#1 lore raw
Quoted 18 lines · Marcelo Tosatti

There is a deadlock in pvclock handling:

cpu0: cpu1:
kvm_gen_update_masterclock()
kvm_guest_time_update()
spin_lock(pvclock_gtod_sync_lock)
local_irq_save(flags)

spin_lock(pvclock_gtod_sync_lock)

kvm_make_mclock_inprogress_request(kvm)
make_all_cpus_request()
smp_call_function_many()

Now if smp_call_function_many() called by cpu0 tries to call function on cpu1 there will be a deadlock.

Quoted 7 lines

Fix by moving pvclock_gtod_sync_lock protected section outside irq disabled section.

Analyzed by Gleb Natapov <[email protected]>

Reported-and-Tested-by Yongjie Ren <[email protected]>
Signed-off-by Marcelo Tosatti <[email protected]>
Acked-by Gleb Natapov <[email protected]>
Quoted 26 lines
arch/x86/kvm/x86.c +2 -9
index f71500a..f7c850b 100644
@@ -1416,15 +1416,6 @@ static int kvm_guest_time_update(struct kvm_vcpu *v)
 	kernel_ns = 0;
 	host_tsc = 0;
 
-	/* Keep irq disabled to prevent changes to the clock */
-	local_irq_save(flags);
-	this_tsc_khz = __get_cpu_var(cpu_tsc_khz);
-	if (unlikely(this_tsc_khz == 0)) {
-		local_irq_restore(flags);
-		kvm_make_request(KVM_REQ_CLOCK_UPDATE, v);
-		return 1;
-	}
-
 	/*
 	 * If the host uses TSC clock, then passthrough TSC as stable
 	 * to the guest.

@@ -1436,6 +1427,15 @@ static int kvm_guest_time_update(struct kvm_vcpu *v)
 		kernel_ns = ka->master_kernel_ns;
 	}
 	spin_unlock(&ka->pvclock_gtod_sync_lock);
+
+	/* Keep irq disabled to prevent changes to the clock */

arch/x86/kvm/x86.c static int kvm_guest_time_update(struct kvm_vcpu *v) · line 1441

Quoted 10 lines
arch/x86/kvm/x86.c +7 -0
+	local_irq_save(flags);
+	this_tsc_khz = __get_cpu_var(cpu_tsc_khz);
+	if (unlikely(this_tsc_khz == 0)) {
+		local_irq_restore(flags);
+		kvm_make_request(KVM_REQ_CLOCK_UPDATE, v);
+		return 1;
+	}
 	if (!use_master_clock) {
 		host_tsc = native_read_tsc();
 		kernel_ns = get_kernel_ns();

--
Gleb.