KVM: Emulate local APIC in kernel

Because lightweight exits (exits which don't involve userspace) are many
times faster than heavyweight exits, it makes sense to emulate high usage
devices in the kernel.  The local APIC is one such device, especially for
Windows and for SMP, so we add an APIC model to kvm.

It also allows in-kernel host-side drivers to inject interrupts without
going through userspace.

[compile fix on i386 from Jindrich Makovicka]
Signed-off-by: NYaozu (Eddie) Dong <Eddie.Dong@intel.com>
Signed-off-by: NQing He <qing.he@intel.com>
Signed-off-by: NAvi Kivity <avi@qumranet.com>

drivers/kvm/Makefile

@@ -2,7 +2,7 @@
 # Makefile for Kernel-based Virtual Machine module
 #
 
-kvm-objs := kvm_main.o mmu.o x86_emulate.o i8259.o irq.o
+kvm-objs := kvm_main.o mmu.o x86_emulate.o i8259.o irq.o lapic.o
 obj-$(CONFIG_KVM) += kvm.o
 kvm-intel-objs = vmx.o
 obj-$(CONFIG_KVM_INTEL) += kvm-intel.o

drivers/kvm/irq.c

@@ -30,14 +30,13 @@
  */
 int kvm_cpu_has_interrupt(struct kvm_vcpu *v)
 {
-	struct kvm_pic *s = pic_irqchip(v->kvm);
-
-	if (s->output)	/* PIC */
-		return 1;
-	/*
-	 * TODO: APIC
-	 */
-	return 0;
+	struct kvm_pic *s;
+
+	if (kvm_apic_has_interrupt(v) == -1) {	/* LAPIC */
+		s = pic_irqchip(v->kvm);	/* PIC */
+		return s->output;
+	}
+	return 1;
 }
 EXPORT_SYMBOL_GPL(kvm_cpu_has_interrupt);
 
@@ -46,16 +45,36 @@ EXPORT_SYMBOL_GPL(kvm_cpu_has_interrupt);
  */
 int kvm_cpu_get_interrupt(struct kvm_vcpu *v)
 {
-	struct kvm_pic *s = pic_irqchip(v->kvm);
+	struct kvm_pic *s;
 	int vector;
 
-	s->output = 0;
-	vector = kvm_pic_read_irq(s);
-	if (vector != -1)
-		return vector;
-	/*
-	 * TODO: APIC
-	 */
-	return -1;
+	vector = kvm_get_apic_interrupt(v);	/* APIC */
+	if (vector == -1) {
+		s = pic_irqchip(v->kvm);
+		s->output = 0;		/* PIC */
+		vector = kvm_pic_read_irq(s);
+	}
+	return vector;
 }
 EXPORT_SYMBOL_GPL(kvm_cpu_get_interrupt);
+
+static void vcpu_kick_intr(void *info)
+{
+#ifdef DEBUG
+	struct kvm_vcpu *vcpu = (struct kvm_vcpu *)info;
+	printk(KERN_DEBUG "vcpu_kick_intr %p \n", vcpu);
+#endif
+}
+
+void kvm_vcpu_kick(struct kvm_vcpu *vcpu)
+{
+	int ipi_pcpu = vcpu->cpu;
+
+	if (vcpu->guest_mode)
+		smp_call_function_single(ipi_pcpu, vcpu_kick_intr, vcpu, 0, 0);
+}
+
+void kvm_ioapic_update_eoi(struct kvm *kvm, int vector)
+{
+	/* TODO: for kernel IOAPIC */
+}

drivers/kvm/irq.h

@@ -26,7 +26,6 @@
 
 typedef void irq_request_func(void *opaque, int level);
 
-struct kvm_pic;
 struct kvm_kpic_state {
 	u8 last_irr;	/* edge detection */
 	u8 irr;		/* interrupt request register */
@@ -61,4 +60,44 @@ int kvm_pic_read_irq(struct kvm_pic *s);
 int kvm_cpu_get_interrupt(struct kvm_vcpu *v);
 int kvm_cpu_has_interrupt(struct kvm_vcpu *v);
 
+struct kvm_lapic {
+	unsigned long base_address;
+	struct kvm_io_device dev;
+	struct {
+		atomic_t pending;
+		s64 period;	/* unit: ns */
+		u32 divide_count;
+		ktime_t last_update;
+		struct hrtimer dev;
+	} timer;
+	struct kvm_vcpu *vcpu;
+	struct page *regs_page;
+	void *regs;
+};
+
+#ifdef DEBUG
+#define ASSERT(x)  							\
+do {									\
+	if (!(x)) {							\
+		printk(KERN_EMERG "assertion failed %s: %d: %s\n",	\
+		       __FILE__, __LINE__, #x);				\
+		BUG();							\
+	}								\
+} while (0)
+#else
+#define ASSERT(x) do { } while (0)
+#endif
+
+void kvm_vcpu_kick(struct kvm_vcpu *vcpu);
+int kvm_apic_has_interrupt(struct kvm_vcpu *vcpu);
+int kvm_get_apic_interrupt(struct kvm_vcpu *vcpu);
+int kvm_create_lapic(struct kvm_vcpu *vcpu);
+void kvm_free_apic(struct kvm_lapic *apic);
+u64 kvm_lapic_get_cr8(struct kvm_vcpu *vcpu);
+void kvm_lapic_set_tpr(struct kvm_vcpu *vcpu, unsigned long cr8);
+void kvm_lapic_set_base(struct kvm_vcpu *vcpu, u64 value);
+u64 kvm_get_apic_base(struct kvm_vcpu *vcpu);
+void kvm_set_apic_base(struct kvm_vcpu *vcpu, u64 data);
+void kvm_ioapic_update_eoi(struct kvm *kvm, int vector);
+
 #endif

drivers/kvm/kvm.h

@@ -324,6 +324,7 @@ struct kvm_vcpu {
 	u64 pdptrs[4]; /* pae */
 	u64 shadow_efer;
 	u64 apic_base;
+	struct kvm_lapic *apic;    /* kernel irqchip context */
 	u64 ia32_misc_enable_msr;
 
 	struct kvm_mmu mmu;
@@ -569,8 +570,6 @@ void set_cr3(struct kvm_vcpu *vcpu, unsigned long cr0);
 void set_cr4(struct kvm_vcpu *vcpu, unsigned long cr0);
 void set_cr8(struct kvm_vcpu *vcpu, unsigned long cr0);
 unsigned long get_cr8(struct kvm_vcpu *vcpu);
-u64 kvm_get_apic_base(struct kvm_vcpu *vcpu);
-void kvm_set_apic_base(struct kvm_vcpu *vcpu, u64 data);
 void lmsw(struct kvm_vcpu *vcpu, unsigned long msw);
 
 int kvm_get_msr_common(struct kvm_vcpu *vcpu, u32 msr, u64 *pdata);

drivers/kvm/kvm_main.c

@@ -281,6 +281,7 @@ EXPORT_SYMBOL_GPL(kvm_vcpu_init);
 void kvm_vcpu_uninit(struct kvm_vcpu *vcpu)
 {
 	kvm_mmu_destroy(vcpu);
+	kvm_free_apic(vcpu->apic);
 	free_page((unsigned long)vcpu->pio_data);
 	free_page((unsigned long)vcpu->run);
 }
@@ -598,25 +599,38 @@ void set_cr8(struct kvm_vcpu *vcpu, unsigned long cr8)
 		inject_gp(vcpu);
 		return;
 	}
-	vcpu->cr8 = cr8;
+	if (irqchip_in_kernel(vcpu->kvm))
+		kvm_lapic_set_tpr(vcpu, cr8);
+	else
+		vcpu->cr8 = cr8;
 }
 EXPORT_SYMBOL_GPL(set_cr8);
 
 unsigned long get_cr8(struct kvm_vcpu *vcpu)
 {
-	return vcpu->cr8;
+	if (irqchip_in_kernel(vcpu->kvm))
+		return kvm_lapic_get_cr8(vcpu);
+	else
+		return vcpu->cr8;
 }
 EXPORT_SYMBOL_GPL(get_cr8);
 
 u64 kvm_get_apic_base(struct kvm_vcpu *vcpu)
 {
-	return vcpu->apic_base;
+	if (irqchip_in_kernel(vcpu->kvm))
+		return vcpu->apic_base;
+	else
+		return vcpu->apic_base;
 }
 EXPORT_SYMBOL_GPL(kvm_get_apic_base);
 
 void kvm_set_apic_base(struct kvm_vcpu *vcpu, u64 data)
 {
-	vcpu->apic_base = data;
+	/* TODO: reserve bits check */
+	if (irqchip_in_kernel(vcpu->kvm))
+		kvm_lapic_set_base(vcpu, data);
+	else
+		vcpu->apic_base = data;
 }
 EXPORT_SYMBOL_GPL(kvm_set_apic_base);
 
@@ -986,15 +1000,31 @@ static int emulator_write_std(unsigned long addr,
 	return X86EMUL_UNHANDLEABLE;
 }
 
+/*
+ * Only apic need an MMIO device hook, so shortcut now..
+ */
+static struct kvm_io_device *vcpu_find_pervcpu_dev(struct kvm_vcpu *vcpu,
+						gpa_t addr)
+{
+	struct kvm_io_device *dev;
+
+	if (vcpu->apic) {
+		dev = &vcpu->apic->dev;
+		if (dev->in_range(dev, addr))
+			return dev;
+	}
+	return NULL;
+}
+
 static struct kvm_io_device *vcpu_find_mmio_dev(struct kvm_vcpu *vcpu,
 						gpa_t addr)
 {
-	/*
-	 * Note that its important to have this wrapper function because
-	 * in the very near future we will be checking for MMIOs against
-	 * the LAPIC as well as the general MMIO bus
-	 */
-	return kvm_io_bus_find_dev(&vcpu->kvm->mmio_bus, addr);
+	struct kvm_io_device *dev;
+
+	dev = vcpu_find_pervcpu_dev(vcpu, addr);
+	if (dev == NULL)
+		dev = kvm_io_bus_find_dev(&vcpu->kvm->mmio_bus, addr);
+	return dev;
 }
 
 static struct kvm_io_device *vcpu_find_pio_dev(struct kvm_vcpu *vcpu,
@@ -2256,6 +2286,8 @@ static int kvm_vcpu_ioctl_interrupt(struct kvm_vcpu *vcpu,
 {
 	if (irq->irq < 0 || irq->irq >= 256)
 		return -EINVAL;
+	if (irqchip_in_kernel(vcpu->kvm))
+		return -ENXIO;
 	vcpu_load(vcpu);
 
 	set_bit(irq->irq, vcpu->irq_pending);

drivers/kvm/svm.c

@@ -573,6 +573,12 @@ static struct kvm_vcpu *svm_create_vcpu(struct kvm *kvm, unsigned int id)
 	if (err)
 		goto free_svm;
 
+	if (irqchip_in_kernel(kvm)) {
+		err = kvm_create_lapic(&svm->vcpu);
+		if (err < 0)
+			goto free_svm;
+	}
+
 	page = alloc_page(GFP_KERNEL);
 	if (!page) {
 		err = -ENOMEM;

drivers/kvm/vmx.c

@@ -2390,6 +2390,12 @@ static struct kvm_vcpu *vmx_create_vcpu(struct kvm *kvm, unsigned int id)
 	if (err)
 		goto free_vcpu;
 
+	if (irqchip_in_kernel(kvm)) {
+		err = kvm_create_lapic(&vmx->vcpu);
+		if (err < 0)
+			goto free_vcpu;
+	}
+
 	vmx->guest_msrs = kmalloc(PAGE_SIZE, GFP_KERNEL);
 	if (!vmx->guest_msrs) {
 		err = -ENOMEM;

include/linux/kvm.h

@@ -34,7 +34,7 @@ struct kvm_memory_alias {
 	__u64 target_phys_addr;
 };
 
-/* for KVM_SET_IRQ_LEVEL */
+/* for KVM_IRQ_LINE */
 struct kvm_irq_level {
 	/*
 	 * ACPI gsi notion of irq.
@@ -297,7 +297,7 @@ struct kvm_signal_mask {
 #define KVM_SET_MEMORY_ALIAS      _IOW(KVMIO, 0x43, struct kvm_memory_alias)
 /* Device model IOC */
 #define KVM_CREATE_IRQCHIP	  _IO(KVMIO,  0x60)
-#define KVM_IRQ_LINE		  _IO(KVMIO,  0x61)
+#define KVM_IRQ_LINE		  _IOW(KVMIO, 0x61, struct kvm_irq_level)
 
 /*
  * ioctls for vcpu fds