KVM: X86: use paravirtualized TLB Shootdown

Remote TLB flush does a busy wait which is fine in bare-metal
scenario. But with-in the guest, the vcpus might have been pre-empted or
blocked. In this scenario, the initator vcpu would end up busy-waiting
for a long amount of time; it also consumes CPU unnecessarily to wake
up the target of the shootdown.

This patch set adds support for KVM's new paravirtualized TLB flush;
remote TLB flush does not wait for vcpus that are sleeping, instead
KVM will flush the TLB as soon as the vCPU starts running again.

The improvement is clearly visible when the host is overcommitted; in this
case, the PV TLB flush (in addition to avoiding the wait on the main CPU)
prevents preempted vCPUs from stealing precious execution time from the
running ones.

Testing on a Xeon Gold 6142 2.6GHz 2 sockets, 32 cores, 64 threads,
so 64 pCPUs, and each VM is 64 vCPUs.

ebizzy -M
              vanilla    optimized     boost
1VM            46799       48670         4%
2VM            23962       42691        78%
3VM            16152       37539       132%

Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Radim Krčmář <rkrcmar@redhat.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Signed-off-by: Wanpeng Li <wanpeng.li@hotmail.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Radim Krčmář <rkrcmar@redhat.com>

Documentation/virtual/kvm/cpuid.txt

@@ -54,6 +54,10 @@ KVM_FEATURE_PV_UNHALT              ||     7 || guest checks this feature bit
                                    ||       || before enabling paravirtualized
                                    ||       || spinlock support.
 ------------------------------------------------------------------------------
+KVM_FEATURE_PV_TLB_FLUSH           ||     9 || guest checks this feature bit
+                                   ||       || before enabling paravirtualized
+                                   ||       || tlb flush.
+------------------------------------------------------------------------------
 KVM_FEATURE_CLOCKSOURCE_STABLE_BIT ||    24 || host will warn if no guest-side
                                    ||       || per-cpu warps are expected in
                                    ||       || kvmclock.

arch/x86/include/uapi/asm/kvm_para.h

@@ -25,6 +25,7 @@
 #define KVM_FEATURE_STEAL_TIME		5
 #define KVM_FEATURE_PV_EOI		6
 #define KVM_FEATURE_PV_UNHALT		7
+#define KVM_FEATURE_PV_TLB_FLUSH	9
 
 /* The last 8 bits are used to indicate how to interpret the flags field
  * in pvclock structure. If no bits are set, all flags are ignored.
@@ -52,6 +53,7 @@ struct kvm_steal_time {
 };
 
 #define KVM_VCPU_PREEMPTED          (1 << 0)
+#define KVM_VCPU_FLUSH_TLB          (1 << 1)
 
 #define KVM_CLOCK_PAIRING_WALLCLOCK 0
 struct kvm_clock_pairing {

arch/x86/kernel/kvm.c

@@ -498,6 +498,34 @@ static void __init kvm_apf_trap_init(void)
 	update_intr_gate(X86_TRAP_PF, async_page_fault);
 }
 
+static DEFINE_PER_CPU(cpumask_var_t, __pv_tlb_mask);
+
+static void kvm_flush_tlb_others(const struct cpumask *cpumask,
+			const struct flush_tlb_info *info)
+{
+	u8 state;
+	int cpu;
+	struct kvm_steal_time *src;
+	struct cpumask *flushmask = this_cpu_cpumask_var_ptr(__pv_tlb_mask);
+
+	cpumask_copy(flushmask, cpumask);
+	/*
+	 * We have to call flush only on online vCPUs. And
+	 * queue flush_on_enter for pre-empted vCPUs
+	 */
+	for_each_cpu(cpu, flushmask) {
+		src = &per_cpu(steal_time, cpu);
+		state = READ_ONCE(src->preempted);
+		if ((state & KVM_VCPU_PREEMPTED)) {
+			if (try_cmpxchg(&src->preempted, &state,
+					state | KVM_VCPU_FLUSH_TLB))
+				__cpumask_clear_cpu(cpu, flushmask);
+		}
+	}
+
+	native_flush_tlb_others(flushmask, info);
+}
+
 static void __init kvm_guest_init(void)
 {
 	int i;
@@ -517,6 +545,9 @@ static void __init kvm_guest_init(void)
 		pv_time_ops.steal_clock = kvm_steal_clock;
 	}
 
+	if (kvm_para_has_feature(KVM_FEATURE_PV_TLB_FLUSH))
+		pv_mmu_ops.flush_tlb_others = kvm_flush_tlb_others;
+
 	if (kvm_para_has_feature(KVM_FEATURE_PV_EOI))
 		apic_set_eoi_write(kvm_guest_apic_eoi_write);
 
@@ -598,6 +629,22 @@ static __init int activate_jump_labels(void)
 }
 arch_initcall(activate_jump_labels);
 
+static __init int kvm_setup_pv_tlb_flush(void)
+{
+	int cpu;
+
+	if (kvm_para_has_feature(KVM_FEATURE_PV_TLB_FLUSH)) {
+		for_each_possible_cpu(cpu) {
+			zalloc_cpumask_var_node(per_cpu_ptr(&__pv_tlb_mask, cpu),
+				GFP_KERNEL, cpu_to_node(cpu));
+		}
+		pr_info("KVM setup pv remote TLB flush\n");
+	}
+
+	return 0;
+}
+arch_initcall(kvm_setup_pv_tlb_flush);
+
 #ifdef CONFIG_PARAVIRT_SPINLOCKS
 
 /* Kick a cpu by its apicid. Used to wake up a halted vcpu */