[deliverable/linux.git] / arch / x86 / kernel / kvm.c

/*
 * KVM paravirt_ops implementation
 *
 * This program is free software; you can redistribute it and/or modify
 * it under the terms of the GNU General Public License as published by
 * the Free Software Foundation; either version 2 of the License, or
 * (at your option) any later version.
 *
 * This program is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 * GNU General Public License for more details.
 *
 * You should have received a copy of the GNU General Public License
 * along with this program; if not, write to the Free Software
 * Foundation, 51 Franklin Street, Fifth Floor, Boston, MA  02110-1301, USA.
 *
 * Copyright (C) 2007, Red Hat, Inc., Ingo Molnar <mingo@redhat.com>
 * Copyright IBM Corporation, 2007
 *   Authors: Anthony Liguori <aliguori@us.ibm.com>
 */

#include <linux/module.h>
#include <linux/kernel.h>
#include <linux/kvm_para.h>
#include <linux/cpu.h>
#include <linux/mm.h>
#include <linux/highmem.h>
#include <linux/hardirq.h>
#include <linux/notifier.h>
#include <linux/reboot.h>
#include <linux/hash.h>
#include <linux/sched.h>
#include <linux/slab.h>
#include <linux/kprobes.h>
#include <asm/timer.h>
#include <asm/cpu.h>
#include <asm/traps.h>
#include <asm/desc.h>
#include <asm/tlbflush.h>
#include <asm/idle.h>
#include <asm/apic.h>
#include <asm/apicdef.h>
#include <asm/hypervisor.h>
#include <asm/kvm_guest.h>

static int kvmapf = 1;

static int parse_no_kvmapf(char *arg)
{
        kvmapf = 0;
        return 0;
}

early_param("no-kvmapf", parse_no_kvmapf);

static int steal_acc = 1;
static int parse_no_stealacc(char *arg)
{
        steal_acc = 0;
        return 0;
}

early_param("no-steal-acc", parse_no_stealacc);

static int kvmclock_vsyscall = 1;
static int parse_no_kvmclock_vsyscall(char *arg)
{
        kvmclock_vsyscall = 0;
        return 0;
}

early_param("no-kvmclock-vsyscall", parse_no_kvmclock_vsyscall);

static DEFINE_PER_CPU(struct kvm_vcpu_pv_apf_data, apf_reason) __aligned(64);
static DEFINE_PER_CPU(struct kvm_steal_time, steal_time) __aligned(64);
static int has_steal_clock = 0;

/*
 * No need for any "IO delay" on KVM
 */
static void kvm_io_delay(void)
{
}

#define KVM_TASK_SLEEP_HASHBITS 8
#define KVM_TASK_SLEEP_HASHSIZE (1<<KVM_TASK_SLEEP_HASHBITS)

struct kvm_task_sleep_node {
	struct hlist_node link;
	wait_queue_head_t wq;
	u32 token;
	int cpu;
	bool halted;
};

static struct kvm_task_sleep_head {
	spinlock_t lock;
	struct hlist_head list;
} async_pf_sleepers[KVM_TASK_SLEEP_HASHSIZE];

static struct kvm_task_sleep_node *_find_apf_task(struct kvm_task_sleep_head *b,
						  u32 token)
{
	struct hlist_node *p;

	hlist_for_each(p, &b->list) {
		struct kvm_task_sleep_node *n =
			hlist_entry(p, typeof(*n), link);
		if (n->token == token)
			return n;
	}

	return NULL;
}

void kvm_async_pf_task_wait(u32 token)
{
	u32 key = hash_32(token, KVM_TASK_SLEEP_HASHBITS);
	struct kvm_task_sleep_head *b = &async_pf_sleepers[key];
	struct kvm_task_sleep_node n, *e;
	DEFINE_WAIT(wait);

	spin_lock(&b->lock);
	e = _find_apf_task(b, token);
	if (e) {
		/* dummy entry exist -> wake up was delivered ahead of PF */
		hlist_del(&e->link);
		kfree(e);
		spin_unlock(&b->lock);
		return;
	}

	n.token = token;
	n.cpu = smp_processor_id();
	n.halted = is_idle_task(current) || preempt_count() > 1;
	init_waitqueue_head(&n.wq);
	hlist_add_head(&n.link, &b->list);
	spin_unlock(&b->lock);

	for (;;) {
		if (!n.halted)
			prepare_to_wait(&n.wq, &wait, TASK_UNINTERRUPTIBLE);
		if (hlist_unhashed(&n.link))
			break;

		if (!n.halted) {
			local_irq_enable();
			schedule();
			local_irq_disable();
		} else {
			/*
			 * We cannot reschedule. So halt.
			 */
			native_safe_halt();
			local_irq_disable();
		}
	}
	if (!n.halted)
		finish_wait(&n.wq, &wait);

	return;
}
EXPORT_SYMBOL_GPL(kvm_async_pf_task_wait);

static void apf_task_wake_one(struct kvm_task_sleep_node *n)
{
	hlist_del_init(&n->link);
	if (n->halted)
		smp_send_reschedule(n->cpu);
	else if (waitqueue_active(&n->wq))
		wake_up(&n->wq);
}

static void apf_task_wake_all(void)
{
	int i;

	for (i = 0; i < KVM_TASK_SLEEP_HASHSIZE; i++) {
		struct hlist_node *p, *next;
		struct kvm_task_sleep_head *b = &async_pf_sleepers[i];
		spin_lock(&b->lock);
		hlist_for_each_safe(p, next, &b->list) {
			struct kvm_task_sleep_node *n =
				hlist_entry(p, typeof(*n), link);
			if (n->cpu == smp_processor_id())
				apf_task_wake_one(n);
		}
		spin_unlock(&b->lock);
	}
}

void kvm_async_pf_task_wake(u32 token)
{
	u32 key = hash_32(token, KVM_TASK_SLEEP_HASHBITS);
	struct kvm_task_sleep_head *b = &async_pf_sleepers[key];
	struct kvm_task_sleep_node *n;

	if (token == ~0) {
		apf_task_wake_all();
		return;
	}

again:
	spin_lock(&b->lock);
	n = _find_apf_task(b, token);
	if (!n) {
		/*
		 * async PF was not yet handled.
		 * Add dummy entry for the token.
		 */
		n = kzalloc(sizeof(*n), GFP_ATOMIC);
		if (!n) {
			/*
			 * Allocation failed! Busy wait while other cpu
			 * handles async PF.
			 */
			spin_unlock(&b->lock);
			cpu_relax();
			goto again;
		}
		n->token = token;
		n->cpu = smp_processor_id();
		init_waitqueue_head(&n->wq);
		hlist_add_head(&n->link, &b->list);
	} else
		apf_task_wake_one(n);
	spin_unlock(&b->lock);
	return;
}
EXPORT_SYMBOL_GPL(kvm_async_pf_task_wake);

u32 kvm_read_and_reset_pf_reason(void)
{
	u32 reason = 0;

	if (__get_cpu_var(apf_reason).enabled) {
		reason = __get_cpu_var(apf_reason).reason;
		__get_cpu_var(apf_reason).reason = 0;
	}

	return reason;
}
EXPORT_SYMBOL_GPL(kvm_read_and_reset_pf_reason);

dotraplinkage void __kprobes
do_async_page_fault(struct pt_regs *regs, unsigned long error_code)
{
	switch (kvm_read_and_reset_pf_reason()) {
	default:
		do_page_fault(regs, error_code);
		break;
	case KVM_PV_REASON_PAGE_NOT_PRESENT:
		/* page is swapped out by the host. */
		rcu_irq_enter();
		exit_idle();
		kvm_async_pf_task_wait((u32)read_cr2());
		rcu_irq_exit();
		break;
	case KVM_PV_REASON_PAGE_READY:
		rcu_irq_enter();
		exit_idle();
		kvm_async_pf_task_wake((u32)read_cr2());
		rcu_irq_exit();
		break;
	}
}

static void __init paravirt_ops_setup(void)
{
	pv_info.name = "KVM";
	pv_info.paravirt_enabled = 1;

	if (kvm_para_has_feature(KVM_FEATURE_NOP_IO_DELAY))
		pv_cpu_ops.io_delay = kvm_io_delay;

#ifdef CONFIG_X86_IO_APIC
	no_timer_check = 1;
#endif
}

static void kvm_register_steal_time(void)
{
	int cpu = smp_processor_id();
	struct kvm_steal_time *st = &per_cpu(steal_time, cpu);

	if (!has_steal_clock)
		return;

	memset(st, 0, sizeof(*st));

	wrmsrl(MSR_KVM_STEAL_TIME, (__pa(st) | KVM_MSR_ENABLED));
	printk(KERN_INFO "kvm-stealtime: cpu %d, msr %lx\n",
		cpu, __pa(st));
}

static DEFINE_PER_CPU(unsigned long, kvm_apic_eoi) = KVM_PV_EOI_DISABLED;

static void kvm_guest_apic_eoi_write(u32 reg, u32 val)
{
	/**
	 * This relies on __test_and_clear_bit to modify the memory
	 * in a way that is atomic with respect to the local CPU.
	 * The hypervisor only accesses this memory from the local CPU so
	 * there's no need for lock or memory barriers.
	 * An optimization barrier is implied in apic write.
	 */
	if (__test_and_clear_bit(KVM_PV_EOI_BIT, &__get_cpu_var(kvm_apic_eoi)))
		return;
	apic_write(APIC_EOI, APIC_EOI_ACK);
}

void __cpuinit kvm_guest_cpu_init(void)
{
	if (!kvm_para_available())
		return;

	if (kvm_para_has_feature(KVM_FEATURE_ASYNC_PF) && kvmapf) {
		u64 pa = __pa(&__get_cpu_var(apf_reason));

#ifdef CONFIG_PREEMPT
		pa |= KVM_ASYNC_PF_SEND_ALWAYS;
#endif
		wrmsrl(MSR_KVM_ASYNC_PF_EN, pa | KVM_ASYNC_PF_ENABLED);
		__get_cpu_var(apf_reason).enabled = 1;
		printk(KERN_INFO"KVM setup async PF for cpu %d\n",
		       smp_processor_id());
	}

	if (kvm_para_has_feature(KVM_FEATURE_PV_EOI)) {
		unsigned long pa;
		/* Size alignment is implied but just to make it explicit. */
		BUILD_BUG_ON(__alignof__(kvm_apic_eoi) < 4);
		__get_cpu_var(kvm_apic_eoi) = 0;
		pa = __pa(&__get_cpu_var(kvm_apic_eoi)) | KVM_MSR_ENABLED;
		wrmsrl(MSR_KVM_PV_EOI_EN, pa);
	}

	if (has_steal_clock)
		kvm_register_steal_time();
}

static void kvm_pv_disable_apf(void)
{
	if (!__get_cpu_var(apf_reason).enabled)
		return;

	wrmsrl(MSR_KVM_ASYNC_PF_EN, 0);
	__get_cpu_var(apf_reason).enabled = 0;

	printk(KERN_INFO"Unregister pv shared memory for cpu %d\n",
	       smp_processor_id());
}

static void kvm_pv_guest_cpu_reboot(void *unused)
{
	/*
	 * We disable PV EOI before we load a new kernel by kexec,
	 * since MSR_KVM_PV_EOI_EN stores a pointer into old kernel's memory.
	 * New kernel can re-enable when it boots.
	 */
	if (kvm_para_has_feature(KVM_FEATURE_PV_EOI))
		wrmsrl(MSR_KVM_PV_EOI_EN, 0);
	kvm_pv_disable_apf();
	kvm_disable_steal_time();
}

static int kvm_pv_reboot_notify(struct notifier_block *nb,
				unsigned long code, void *unused)
{
	if (code == SYS_RESTART)
		on_each_cpu(kvm_pv_guest_cpu_reboot, NULL, 1);
	return NOTIFY_DONE;
}

static struct notifier_block kvm_pv_reboot_nb = {
	.notifier_call = kvm_pv_reboot_notify,
};

static u64 kvm_steal_clock(int cpu)
{
	u64 steal;
	struct kvm_steal_time *src;
	int version;

	src = &per_cpu(steal_time, cpu);
	do {
		version = src->version;
		rmb();
		steal = src->steal;
		rmb();
	} while ((version & 1) || (version != src->version));

	return steal;
}

void kvm_disable_steal_time(void)
{
	if (!has_steal_clock)
		return;

	wrmsr(MSR_KVM_STEAL_TIME, 0, 0);
}

#ifdef CONFIG_SMP
static void __init kvm_smp_prepare_boot_cpu(void)
{
	WARN_ON(kvm_register_clock("primary cpu clock"));
	kvm_guest_cpu_init();
	native_smp_prepare_boot_cpu();
}

static void __cpuinit kvm_guest_cpu_online(void *dummy)
{
	kvm_guest_cpu_init();
}

static void kvm_guest_cpu_offline(void *dummy)
{
	kvm_disable_steal_time();
	if (kvm_para_has_feature(KVM_FEATURE_PV_EOI))
		wrmsrl(MSR_KVM_PV_EOI_EN, 0);
	kvm_pv_disable_apf();
	apf_task_wake_all();
}

static int __cpuinit kvm_cpu_notify(struct notifier_block *self,
				    unsigned long action, void *hcpu)
{
	int cpu = (unsigned long)hcpu;
	switch (action) {
	case CPU_ONLINE:
	case CPU_DOWN_FAILED:
	case CPU_ONLINE_FROZEN:
		smp_call_function_single(cpu, kvm_guest_cpu_online, NULL, 0);
		break;
	case CPU_DOWN_PREPARE:
	case CPU_DOWN_PREPARE_FROZEN:
		smp_call_function_single(cpu, kvm_guest_cpu_offline, NULL, 1);
		break;
	default:
		break;
	}
	return NOTIFY_OK;
}

static struct notifier_block __cpuinitdata kvm_cpu_notifier = {
        .notifier_call  = kvm_cpu_notify,
};
#endif

static void __init kvm_apf_trap_init(void)
{
	set_intr_gate(14, &async_page_fault);
}

void __init kvm_guest_init(void)
{
	int i;

	if (!kvm_para_available())
		return;

	paravirt_ops_setup();
	register_reboot_notifier(&kvm_pv_reboot_nb);
	for (i = 0; i < KVM_TASK_SLEEP_HASHSIZE; i++)
		spin_lock_init(&async_pf_sleepers[i].lock);
	if (kvm_para_has_feature(KVM_FEATURE_ASYNC_PF))
		x86_init.irqs.trap_init = kvm_apf_trap_init;

	if (kvm_para_has_feature(KVM_FEATURE_STEAL_TIME)) {
		has_steal_clock = 1;
		pv_time_ops.steal_clock = kvm_steal_clock;
	}

	if (kvm_para_has_feature(KVM_FEATURE_PV_EOI))
		apic_set_eoi_write(kvm_guest_apic_eoi_write);

	if (kvmclock_vsyscall)
		kvm_setup_vsyscall_timeinfo();

#ifdef CONFIG_SMP
	smp_ops.smp_prepare_boot_cpu = kvm_smp_prepare_boot_cpu;
	register_cpu_notifier(&kvm_cpu_notifier);
#else
	kvm_guest_cpu_init();
#endif
}

static bool __init kvm_detect(void)
{
	if (!kvm_para_available())
		return false;
	return true;
}

const struct hypervisor_x86 x86_hyper_kvm __refconst = {
	.name			= "KVM",
	.detect			= kvm_detect,
};
EXPORT_SYMBOL_GPL(x86_hyper_kvm);

static __init int activate_jump_labels(void)
{
	if (has_steal_clock) {
		static_key_slow_inc(&paravirt_steal_enabled);
		if (steal_acc)
			static_key_slow_inc(&paravirt_steal_rq_enabled);
	}

	return 0;
}
arch_initcall(activate_jump_labels);
Commit	Line	Data
0cf1bfd2 MT	1	/*
	2	* KVM paravirt_ops implementation
	3	*
	4	* This program is free software; you can redistribute it and/or modify
	5	* it under the terms of the GNU General Public License as published by
	6	* the Free Software Foundation; either version 2 of the License, or
	7	* (at your option) any later version.
	8	*
	9	* This program is distributed in the hope that it will be useful,
	10	* but WITHOUT ANY WARRANTY; without even the implied warranty of
	11	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
	12	* GNU General Public License for more details.
	13	*
	14	* You should have received a copy of the GNU General Public License
	15	* along with this program; if not, write to the Free Software
	16	* Foundation, 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
	17	*
	18	* Copyright (C) 2007, Red Hat, Inc., Ingo Molnar <mingo@redhat.com>
	19	* Copyright IBM Corporation, 2007
	20	* Authors: Anthony Liguori <aliguori@us.ibm.com>
	21	*/
	22
	23	#include <linux/module.h>
	24	#include <linux/kernel.h>
	25	#include <linux/kvm_para.h>
	26	#include <linux/cpu.h>
	27	#include <linux/mm.h>
1da8a77b	28	#include <linux/highmem.h>
096d14a3	29	#include <linux/hardirq.h>
fd10cde9 GN	30	#include <linux/notifier.h>
fd10cde9 GN	31	#include <linux/reboot.h>
631bc487 GN	32	#include <linux/hash.h>
	33	#include <linux/sched.h>
	34	#include <linux/slab.h>
	35	#include <linux/kprobes.h>
a90ede7b	36	#include <asm/timer.h>
fd10cde9	37	#include <asm/cpu.h>
631bc487 GN	38	#include <asm/traps.h>
631bc487 GN	39	#include <asm/desc.h>
6c047cd9	40	#include <asm/tlbflush.h>
e0875921	41	#include <asm/idle.h>
ab9cf499 MT	42	#include <asm/apic.h>
ab9cf499 MT	43	#include <asm/apicdef.h>
fc73373b	44	#include <asm/hypervisor.h>
3dc4f7cf	45	#include <asm/kvm_guest.h>
096d14a3	46
fd10cde9 GN	47	static int kvmapf = 1;
	48
	49	static int parse_no_kvmapf(char *arg)
	50	{
	51	kvmapf = 0;
	52	return 0;
	53	}
	54
	55	early_param("no-kvmapf", parse_no_kvmapf);
	56
d910f5c1 GC	57	static int steal_acc = 1;
	58	static int parse_no_stealacc(char *arg)
	59	{
	60	steal_acc = 0;
	61	return 0;
	62	}
	63
	64	early_param("no-steal-acc", parse_no_stealacc);
	65
3dc4f7cf MT	66	static int kvmclock_vsyscall = 1;
	67	static int parse_no_kvmclock_vsyscall(char *arg)
	68	{
	69	kvmclock_vsyscall = 0;
	70	return 0;
	71	}
	72
	73	early_param("no-kvmclock-vsyscall", parse_no_kvmclock_vsyscall);
	74
fd10cde9	75	static DEFINE_PER_CPU(struct kvm_vcpu_pv_apf_data, apf_reason) __aligned(64);
d910f5c1 GC	76	static DEFINE_PER_CPU(struct kvm_steal_time, steal_time) __aligned(64);
d910f5c1 GC	77	static int has_steal_clock = 0;
096d14a3	78
0cf1bfd2 MT	79	/*
	80	* No need for any "IO delay" on KVM
	81	*/
	82	static void kvm_io_delay(void)
	83	{
	84	}
	85
631bc487 GN	86	#define KVM_TASK_SLEEP_HASHBITS 8
	87	#define KVM_TASK_SLEEP_HASHSIZE (1<<KVM_TASK_SLEEP_HASHBITS)
	88
	89	struct kvm_task_sleep_node {
	90	struct hlist_node link;
	91	wait_queue_head_t wq;
	92	u32 token;
	93	int cpu;
6c047cd9	94	bool halted;
631bc487 GN	95	};
	96
	97	static struct kvm_task_sleep_head {
	98	spinlock_t lock;
	99	struct hlist_head list;
	100	} async_pf_sleepers[KVM_TASK_SLEEP_HASHSIZE];
	101
	102	static struct kvm_task_sleep_node _find_apf_task(struct kvm_task_sleep_head b,
	103	u32 token)
	104	{
	105	struct hlist_node *p;
	106
	107	hlist_for_each(p, &b->list) {
	108	struct kvm_task_sleep_node *n =
	109	hlist_entry(p, typeof(*n), link);
	110	if (n->token == token)
	111	return n;
	112	}
	113
	114	return NULL;
	115	}
	116
	117	void kvm_async_pf_task_wait(u32 token)
	118	{
	119	u32 key = hash_32(token, KVM_TASK_SLEEP_HASHBITS);
	120	struct kvm_task_sleep_head *b = &async_pf_sleepers[key];
	121	struct kvm_task_sleep_node n, *e;
	122	DEFINE_WAIT(wait);
	123
	124	spin_lock(&b->lock);
	125	e = _find_apf_task(b, token);
	126	if (e) {
	127	/* dummy entry exist -> wake up was delivered ahead of PF */
	128	hlist_del(&e->link);
	129	kfree(e);
	130	spin_unlock(&b->lock);
	131	return;
	132	}
	133
	134	n.token = token;
	135	n.cpu = smp_processor_id();
859f8450	136	n.halted = is_idle_task(current) \|\| preempt_count() > 1;
631bc487 GN	137	init_waitqueue_head(&n.wq);
	138	hlist_add_head(&n.link, &b->list);
	139	spin_unlock(&b->lock);
	140
	141	for (;;) {
6c047cd9 GN	142	if (!n.halted)
6c047cd9 GN	143	prepare_to_wait(&n.wq, &wait, TASK_UNINTERRUPTIBLE);
631bc487 GN	144	if (hlist_unhashed(&n.link))
631bc487 GN	145	break;
6c047cd9 GN	146
	147	if (!n.halted) {
	148	local_irq_enable();
	149	schedule();
	150	local_irq_disable();
	151	} else {
	152	/*
	153	* We cannot reschedule. So halt.
	154	*/
	155	native_safe_halt();
	156	local_irq_disable();
	157	}
631bc487	158	}
6c047cd9 GN	159	if (!n.halted)
6c047cd9 GN	160	finish_wait(&n.wq, &wait);
631bc487 GN	161
	162	return;
	163	}
	164	EXPORT_SYMBOL_GPL(kvm_async_pf_task_wait);
	165
	166	static void apf_task_wake_one(struct kvm_task_sleep_node *n)
	167	{
	168	hlist_del_init(&n->link);
6c047cd9 GN	169	if (n->halted)
	170	smp_send_reschedule(n->cpu);
	171	else if (waitqueue_active(&n->wq))
631bc487 GN	172	wake_up(&n->wq);
	173	}
	174
	175	static void apf_task_wake_all(void)
	176	{
	177	int i;
	178
	179	for (i = 0; i < KVM_TASK_SLEEP_HASHSIZE; i++) {
	180	struct hlist_node p, next;
	181	struct kvm_task_sleep_head *b = &async_pf_sleepers[i];
	182	spin_lock(&b->lock);
	183	hlist_for_each_safe(p, next, &b->list) {
	184	struct kvm_task_sleep_node *n =
	185	hlist_entry(p, typeof(*n), link);
	186	if (n->cpu == smp_processor_id())
	187	apf_task_wake_one(n);
	188	}
	189	spin_unlock(&b->lock);
	190	}
	191	}
	192
	193	void kvm_async_pf_task_wake(u32 token)
	194	{
	195	u32 key = hash_32(token, KVM_TASK_SLEEP_HASHBITS);
	196	struct kvm_task_sleep_head *b = &async_pf_sleepers[key];
	197	struct kvm_task_sleep_node *n;
	198
	199	if (token == ~0) {
	200	apf_task_wake_all();
	201	return;
	202	}
	203
	204	again:
	205	spin_lock(&b->lock);
	206	n = _find_apf_task(b, token);
	207	if (!n) {
	208	/*
	209	* async PF was not yet handled.
	210	* Add dummy entry for the token.
	211	*/
62c49cc9	212	n = kzalloc(sizeof(*n), GFP_ATOMIC);
631bc487 GN	213	if (!n) {
	214	/*
	215	* Allocation failed! Busy wait while other cpu
	216	* handles async PF.
	217	*/
	218	spin_unlock(&b->lock);
	219	cpu_relax();
	220	goto again;
	221	}
	222	n->token = token;
	223	n->cpu = smp_processor_id();
	224	init_waitqueue_head(&n->wq);
	225	hlist_add_head(&n->link, &b->list);
	226	} else
	227	apf_task_wake_one(n);
	228	spin_unlock(&b->lock);
	229	return;
	230	}
	231	EXPORT_SYMBOL_GPL(kvm_async_pf_task_wake);
	232
	233	u32 kvm_read_and_reset_pf_reason(void)
	234	{
	235	u32 reason = 0;
	236
	237	if (__get_cpu_var(apf_reason).enabled) {
	238	reason = __get_cpu_var(apf_reason).reason;
	239	__get_cpu_var(apf_reason).reason = 0;
	240	}
	241
	242	return reason;
	243	}
	244	EXPORT_SYMBOL_GPL(kvm_read_and_reset_pf_reason);
	245
	246	dotraplinkage void __kprobes
	247	do_async_page_fault(struct pt_regs *regs, unsigned long error_code)
	248	{
	249	switch (kvm_read_and_reset_pf_reason()) {
	250	default:
	251	do_page_fault(regs, error_code);
	252	break;
	253	case KVM_PV_REASON_PAGE_NOT_PRESENT:
	254	/* page is swapped out by the host. */
c5e015d4 SL	255	rcu_irq_enter();
c5e015d4 SL	256	exit_idle();
631bc487	257	kvm_async_pf_task_wait((u32)read_cr2());
c5e015d4	258	rcu_irq_exit();
631bc487 GN	259	break;
631bc487 GN	260	case KVM_PV_REASON_PAGE_READY:
e0875921 GN	261	rcu_irq_enter();
e0875921 GN	262	exit_idle();
631bc487	263	kvm_async_pf_task_wake((u32)read_cr2());
e0875921	264	rcu_irq_exit();
631bc487 GN	265	break;
	266	}
	267	}
	268
d3ac8815	269	static void __init paravirt_ops_setup(void)
0cf1bfd2 MT	270	{
	271	pv_info.name = "KVM";
	272	pv_info.paravirt_enabled = 1;
	273
	274	if (kvm_para_has_feature(KVM_FEATURE_NOP_IO_DELAY))
	275	pv_cpu_ops.io_delay = kvm_io_delay;
	276
a90ede7b MT	277	#ifdef CONFIG_X86_IO_APIC
	278	no_timer_check = 1;
	279	#endif
0cf1bfd2 MT	280	}
0cf1bfd2 MT	281
d910f5c1 GC	282	static void kvm_register_steal_time(void)
	283	{
	284	int cpu = smp_processor_id();
	285	struct kvm_steal_time *st = &per_cpu(steal_time, cpu);
	286
	287	if (!has_steal_clock)
	288	return;
	289
	290	memset(st, 0, sizeof(*st));
	291
	292	wrmsrl(MSR_KVM_STEAL_TIME, (__pa(st) \| KVM_MSR_ENABLED));
	293	printk(KERN_INFO "kvm-stealtime: cpu %d, msr %lx\n",
	294	cpu, __pa(st));
	295	}
	296
ab9cf499 MT	297	static DEFINE_PER_CPU(unsigned long, kvm_apic_eoi) = KVM_PV_EOI_DISABLED;
	298
	299	static void kvm_guest_apic_eoi_write(u32 reg, u32 val)
	300	{
	301	/**
	302	* This relies on __test_and_clear_bit to modify the memory
	303	* in a way that is atomic with respect to the local CPU.
	304	* The hypervisor only accesses this memory from the local CPU so
	305	* there's no need for lock or memory barriers.
	306	* An optimization barrier is implied in apic write.
	307	*/
	308	if (__test_and_clear_bit(KVM_PV_EOI_BIT, &__get_cpu_var(kvm_apic_eoi)))
	309	return;
90536664	310	apic_write(APIC_EOI, APIC_EOI_ACK);
ab9cf499 MT	311	}
ab9cf499 MT	312
fd10cde9 GN	313	void __cpuinit kvm_guest_cpu_init(void)
	314	{
	315	if (!kvm_para_available())
	316	return;
	317
	318	if (kvm_para_has_feature(KVM_FEATURE_ASYNC_PF) && kvmapf) {
	319	u64 pa = __pa(&__get_cpu_var(apf_reason));
	320
6adba527 GN	321	#ifdef CONFIG_PREEMPT
	322	pa \|= KVM_ASYNC_PF_SEND_ALWAYS;
	323	#endif
fd10cde9 GN	324	wrmsrl(MSR_KVM_ASYNC_PF_EN, pa \| KVM_ASYNC_PF_ENABLED);
	325	__get_cpu_var(apf_reason).enabled = 1;
	326	printk(KERN_INFO"KVM setup async PF for cpu %d\n",
	327	smp_processor_id());
	328	}
d910f5c1	329
ab9cf499 MT	330	if (kvm_para_has_feature(KVM_FEATURE_PV_EOI)) {
	331	unsigned long pa;
	332	/* Size alignment is implied but just to make it explicit. */
	333	BUILD_BUG_ON(__alignof__(kvm_apic_eoi) < 4);
	334	__get_cpu_var(kvm_apic_eoi) = 0;
	335	pa = __pa(&__get_cpu_var(kvm_apic_eoi)) \| KVM_MSR_ENABLED;
	336	wrmsrl(MSR_KVM_PV_EOI_EN, pa);
	337	}
	338
d910f5c1 GC	339	if (has_steal_clock)
d910f5c1 GC	340	kvm_register_steal_time();
fd10cde9 GN	341	}
fd10cde9 GN	342
ab9cf499	343	static void kvm_pv_disable_apf(void)
fd10cde9 GN	344	{
	345	if (!__get_cpu_var(apf_reason).enabled)
	346	return;
	347
	348	wrmsrl(MSR_KVM_ASYNC_PF_EN, 0);
	349	__get_cpu_var(apf_reason).enabled = 0;
	350
	351	printk(KERN_INFO"Unregister pv shared memory for cpu %d\n",
	352	smp_processor_id());
	353	}
	354
ab9cf499 MT	355	static void kvm_pv_guest_cpu_reboot(void *unused)
	356	{
	357	/*
	358	* We disable PV EOI before we load a new kernel by kexec,
	359	* since MSR_KVM_PV_EOI_EN stores a pointer into old kernel's memory.
	360	* New kernel can re-enable when it boots.
	361	*/
	362	if (kvm_para_has_feature(KVM_FEATURE_PV_EOI))
	363	wrmsrl(MSR_KVM_PV_EOI_EN, 0);
	364	kvm_pv_disable_apf();
8fbe6a54	365	kvm_disable_steal_time();
ab9cf499 MT	366	}
ab9cf499 MT	367
fd10cde9 GN	368	static int kvm_pv_reboot_notify(struct notifier_block *nb,
	369	unsigned long code, void *unused)
	370	{
	371	if (code == SYS_RESTART)
ab9cf499	372	on_each_cpu(kvm_pv_guest_cpu_reboot, NULL, 1);
fd10cde9 GN	373	return NOTIFY_DONE;
	374	}
	375
	376	static struct notifier_block kvm_pv_reboot_nb = {
	377	.notifier_call = kvm_pv_reboot_notify,
	378	};
	379
d910f5c1 GC	380	static u64 kvm_steal_clock(int cpu)
	381	{
	382	u64 steal;
	383	struct kvm_steal_time *src;
	384	int version;
	385
	386	src = &per_cpu(steal_time, cpu);
	387	do {
	388	version = src->version;
	389	rmb();
	390	steal = src->steal;
	391	rmb();
	392	} while ((version & 1) \|\| (version != src->version));
	393
	394	return steal;
	395	}
	396
	397	void kvm_disable_steal_time(void)
	398	{
	399	if (!has_steal_clock)
	400	return;
	401
	402	wrmsr(MSR_KVM_STEAL_TIME, 0, 0);
	403	}
	404
ca3f1017 GN	405	#ifdef CONFIG_SMP
	406	static void __init kvm_smp_prepare_boot_cpu(void)
	407	{
	408	WARN_ON(kvm_register_clock("primary cpu clock"));
fd10cde9	409	kvm_guest_cpu_init();
ca3f1017 GN	410	native_smp_prepare_boot_cpu();
ca3f1017 GN	411	}
fd10cde9	412
775077a0	413	static void __cpuinit kvm_guest_cpu_online(void *dummy)
fd10cde9 GN	414	{
	415	kvm_guest_cpu_init();
	416	}
	417
	418	static void kvm_guest_cpu_offline(void *dummy)
	419	{
d910f5c1	420	kvm_disable_steal_time();
ab9cf499 MT	421	if (kvm_para_has_feature(KVM_FEATURE_PV_EOI))
	422	wrmsrl(MSR_KVM_PV_EOI_EN, 0);
	423	kvm_pv_disable_apf();
631bc487	424	apf_task_wake_all();
fd10cde9 GN	425	}
	426
	427	static int __cpuinit kvm_cpu_notify(struct notifier_block *self,
	428	unsigned long action, void *hcpu)
	429	{
	430	int cpu = (unsigned long)hcpu;
	431	switch (action) {
	432	case CPU_ONLINE:
	433	case CPU_DOWN_FAILED:
	434	case CPU_ONLINE_FROZEN:
	435	smp_call_function_single(cpu, kvm_guest_cpu_online, NULL, 0);
	436	break;
	437	case CPU_DOWN_PREPARE:
	438	case CPU_DOWN_PREPARE_FROZEN:
	439	smp_call_function_single(cpu, kvm_guest_cpu_offline, NULL, 1);
	440	break;
	441	default:
	442	break;
	443	}
	444	return NOTIFY_OK;
	445	}
	446
	447	static struct notifier_block __cpuinitdata kvm_cpu_notifier = {
	448	.notifier_call = kvm_cpu_notify,
	449	};
ca3f1017 GN	450	#endif
ca3f1017 GN	451
631bc487 GN	452	static void __init kvm_apf_trap_init(void)
	453	{
	454	set_intr_gate(14, &async_page_fault);
	455	}
	456
0cf1bfd2 MT	457	void __init kvm_guest_init(void)
0cf1bfd2 MT	458	{
631bc487 GN	459	int i;
631bc487 GN	460
0cf1bfd2 MT	461	if (!kvm_para_available())
	462	return;
	463
	464	paravirt_ops_setup();
fd10cde9	465	register_reboot_notifier(&kvm_pv_reboot_nb);
631bc487 GN	466	for (i = 0; i < KVM_TASK_SLEEP_HASHSIZE; i++)
	467	spin_lock_init(&async_pf_sleepers[i].lock);
	468	if (kvm_para_has_feature(KVM_FEATURE_ASYNC_PF))
	469	x86_init.irqs.trap_init = kvm_apf_trap_init;
	470
d910f5c1 GC	471	if (kvm_para_has_feature(KVM_FEATURE_STEAL_TIME)) {
	472	has_steal_clock = 1;
	473	pv_time_ops.steal_clock = kvm_steal_clock;
	474	}
	475
90536664 MT	476	if (kvm_para_has_feature(KVM_FEATURE_PV_EOI))
90536664 MT	477	apic_set_eoi_write(kvm_guest_apic_eoi_write);
ab9cf499	478
3dc4f7cf MT	479	if (kvmclock_vsyscall)
	480	kvm_setup_vsyscall_timeinfo();
	481
ca3f1017 GN	482	#ifdef CONFIG_SMP
ca3f1017 GN	483	smp_ops.smp_prepare_boot_cpu = kvm_smp_prepare_boot_cpu;
fd10cde9 GN	484	register_cpu_notifier(&kvm_cpu_notifier);
	485	#else
	486	kvm_guest_cpu_init();
ca3f1017	487	#endif
0cf1bfd2	488	}
d910f5c1	489
fc73373b PB	490	static bool __init kvm_detect(void)
	491	{
	492	if (!kvm_para_available())
	493	return false;
	494	return true;
	495	}
	496
	497	const struct hypervisor_x86 x86_hyper_kvm __refconst = {
	498	.name = "KVM",
	499	.detect = kvm_detect,
	500	};
	501	EXPORT_SYMBOL_GPL(x86_hyper_kvm);
	502
d910f5c1 GC	503	static __init int activate_jump_labels(void)
	504	{
	505	if (has_steal_clock) {
c5905afb	506	static_key_slow_inc(&paravirt_steal_enabled);
d910f5c1	507	if (steal_acc)
c5905afb	508	static_key_slow_inc(&paravirt_steal_rq_enabled);
d910f5c1 GC	509	}
	510
	511	return 0;
	512	}
	513	arch_initcall(activate_jump_labels);