[GitHub/mt8127/android_kernel_alcatel_ttab.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/context_tracking.h>
#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);

	rcu_irq_enter();

	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);

		rcu_irq_exit();
		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.
			 */
			rcu_irq_exit();
			native_safe_halt();
			local_irq_disable();
			rcu_irq_enter();
		}
	}
	if (!n.halted)
		finish_wait(&n.wq, &wait);

	rcu_irq_exit();
	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)
{
	enum ctx_state prev_state;

	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. */
		prev_state = exception_enter();
		exit_idle();
		kvm_async_pf_task_wait((u32)read_cr2());
		exception_exit(prev_state);
		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";

	/*
	 * KVM isn't paravirt in the sense of paravirt_enabled.  A KVM
	 * guest kernel works like a bare metal kernel with additional
	 * features, and paravirt_enabled is about features that are
	 * missing.
	 */
	pv_info.paravirt_enabled = 0;

	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, (slow_virt_to_phys(st) | KVM_MSR_ENABLED));
	pr_info("kvm-stealtime: cpu %d, msr %llx\n",
		cpu, (unsigned long long) slow_virt_to_phys(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 = slow_virt_to_phys(&__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 = slow_virt_to_phys(&__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,
	.x2apic_available	= kvm_para_available,
};
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
56dd9470	23	#include <linux/context_tracking.h>
0cf1bfd2 MT	24	#include <linux/module.h>
	25	#include <linux/kernel.h>
	26	#include <linux/kvm_para.h>
	27	#include <linux/cpu.h>
	28	#include <linux/mm.h>
1da8a77b	29	#include <linux/highmem.h>
096d14a3	30	#include <linux/hardirq.h>
fd10cde9 GN	31	#include <linux/notifier.h>
fd10cde9 GN	32	#include <linux/reboot.h>
631bc487 GN	33	#include <linux/hash.h>
	34	#include <linux/sched.h>
	35	#include <linux/slab.h>
	36	#include <linux/kprobes.h>
a90ede7b	37	#include <asm/timer.h>
fd10cde9	38	#include <asm/cpu.h>
631bc487 GN	39	#include <asm/traps.h>
631bc487 GN	40	#include <asm/desc.h>
6c047cd9	41	#include <asm/tlbflush.h>
e0875921	42	#include <asm/idle.h>
ab9cf499 MT	43	#include <asm/apic.h>
ab9cf499 MT	44	#include <asm/apicdef.h>
fc73373b	45	#include <asm/hypervisor.h>
3dc4f7cf	46	#include <asm/kvm_guest.h>
096d14a3	47
fd10cde9 GN	48	static int kvmapf = 1;
	49
	50	static int parse_no_kvmapf(char *arg)
	51	{
	52	kvmapf = 0;
	53	return 0;
	54	}
	55
	56	early_param("no-kvmapf", parse_no_kvmapf);
	57
d910f5c1 GC	58	static int steal_acc = 1;
	59	static int parse_no_stealacc(char *arg)
	60	{
	61	steal_acc = 0;
	62	return 0;
	63	}
	64
	65	early_param("no-steal-acc", parse_no_stealacc);
	66
3dc4f7cf MT	67	static int kvmclock_vsyscall = 1;
	68	static int parse_no_kvmclock_vsyscall(char *arg)
	69	{
	70	kvmclock_vsyscall = 0;
	71	return 0;
	72	}
	73
	74	early_param("no-kvmclock-vsyscall", parse_no_kvmclock_vsyscall);
	75
fd10cde9	76	static DEFINE_PER_CPU(struct kvm_vcpu_pv_apf_data, apf_reason) __aligned(64);
d910f5c1 GC	77	static DEFINE_PER_CPU(struct kvm_steal_time, steal_time) __aligned(64);
d910f5c1 GC	78	static int has_steal_clock = 0;
096d14a3	79
0cf1bfd2 MT	80	/*
	81	* No need for any "IO delay" on KVM
	82	*/
	83	static void kvm_io_delay(void)
	84	{
	85	}
	86
631bc487 GN	87	#define KVM_TASK_SLEEP_HASHBITS 8
	88	#define KVM_TASK_SLEEP_HASHSIZE (1<<KVM_TASK_SLEEP_HASHBITS)
	89
	90	struct kvm_task_sleep_node {
	91	struct hlist_node link;
	92	wait_queue_head_t wq;
	93	u32 token;
	94	int cpu;
6c047cd9	95	bool halted;
631bc487 GN	96	};
	97
	98	static struct kvm_task_sleep_head {
	99	spinlock_t lock;
	100	struct hlist_head list;
	101	} async_pf_sleepers[KVM_TASK_SLEEP_HASHSIZE];
	102
	103	static struct kvm_task_sleep_node _find_apf_task(struct kvm_task_sleep_head b,
	104	u32 token)
	105	{
	106	struct hlist_node *p;
	107
	108	hlist_for_each(p, &b->list) {
	109	struct kvm_task_sleep_node *n =
	110	hlist_entry(p, typeof(*n), link);
	111	if (n->token == token)
	112	return n;
	113	}
	114
	115	return NULL;
	116	}
	117
	118	void kvm_async_pf_task_wait(u32 token)
	119	{
	120	u32 key = hash_32(token, KVM_TASK_SLEEP_HASHBITS);
	121	struct kvm_task_sleep_head *b = &async_pf_sleepers[key];
	122	struct kvm_task_sleep_node n, *e;
	123	DEFINE_WAIT(wait);
	124
9b132fbe LZ	125	rcu_irq_enter();
9b132fbe LZ	126
631bc487 GN	127	spin_lock(&b->lock);
	128	e = _find_apf_task(b, token);
	129	if (e) {
	130	/* dummy entry exist -> wake up was delivered ahead of PF */
	131	hlist_del(&e->link);
	132	kfree(e);
	133	spin_unlock(&b->lock);
9b132fbe LZ	134
9b132fbe LZ	135	rcu_irq_exit();
631bc487 GN	136	return;
	137	}
	138
	139	n.token = token;
	140	n.cpu = smp_processor_id();
859f8450	141	n.halted = is_idle_task(current) \|\| preempt_count() > 1;
631bc487 GN	142	init_waitqueue_head(&n.wq);
	143	hlist_add_head(&n.link, &b->list);
	144	spin_unlock(&b->lock);
	145
	146	for (;;) {
6c047cd9 GN	147	if (!n.halted)
6c047cd9 GN	148	prepare_to_wait(&n.wq, &wait, TASK_UNINTERRUPTIBLE);
631bc487 GN	149	if (hlist_unhashed(&n.link))
631bc487 GN	150	break;
6c047cd9 GN	151
	152	if (!n.halted) {
	153	local_irq_enable();
	154	schedule();
	155	local_irq_disable();
	156	} else {
	157	/*
	158	* We cannot reschedule. So halt.
	159	*/
9b132fbe	160	rcu_irq_exit();
6c047cd9 GN	161	native_safe_halt();
6c047cd9 GN	162	local_irq_disable();
ca58e312	163	rcu_irq_enter();
6c047cd9	164	}
631bc487	165	}
6c047cd9 GN	166	if (!n.halted)
6c047cd9 GN	167	finish_wait(&n.wq, &wait);
631bc487	168
9b132fbe	169	rcu_irq_exit();
631bc487 GN	170	return;
	171	}
	172	EXPORT_SYMBOL_GPL(kvm_async_pf_task_wait);
	173
	174	static void apf_task_wake_one(struct kvm_task_sleep_node *n)
	175	{
	176	hlist_del_init(&n->link);
6c047cd9 GN	177	if (n->halted)
	178	smp_send_reschedule(n->cpu);
	179	else if (waitqueue_active(&n->wq))
631bc487 GN	180	wake_up(&n->wq);
	181	}
	182
	183	static void apf_task_wake_all(void)
	184	{
	185	int i;
	186
	187	for (i = 0; i < KVM_TASK_SLEEP_HASHSIZE; i++) {
	188	struct hlist_node p, next;
	189	struct kvm_task_sleep_head *b = &async_pf_sleepers[i];
	190	spin_lock(&b->lock);
	191	hlist_for_each_safe(p, next, &b->list) {
	192	struct kvm_task_sleep_node *n =
	193	hlist_entry(p, typeof(*n), link);
	194	if (n->cpu == smp_processor_id())
	195	apf_task_wake_one(n);
	196	}
	197	spin_unlock(&b->lock);
	198	}
	199	}
	200
	201	void kvm_async_pf_task_wake(u32 token)
	202	{
	203	u32 key = hash_32(token, KVM_TASK_SLEEP_HASHBITS);
	204	struct kvm_task_sleep_head *b = &async_pf_sleepers[key];
	205	struct kvm_task_sleep_node *n;
	206
	207	if (token == ~0) {
	208	apf_task_wake_all();
	209	return;
	210	}
	211
	212	again:
	213	spin_lock(&b->lock);
	214	n = _find_apf_task(b, token);
	215	if (!n) {
	216	/*
	217	* async PF was not yet handled.
	218	* Add dummy entry for the token.
	219	*/
62c49cc9	220	n = kzalloc(sizeof(*n), GFP_ATOMIC);
631bc487 GN	221	if (!n) {
	222	/*
	223	* Allocation failed! Busy wait while other cpu
	224	* handles async PF.
	225	*/
	226	spin_unlock(&b->lock);
	227	cpu_relax();
	228	goto again;
	229	}
	230	n->token = token;
	231	n->cpu = smp_processor_id();
	232	init_waitqueue_head(&n->wq);
	233	hlist_add_head(&n->link, &b->list);
	234	} else
	235	apf_task_wake_one(n);
	236	spin_unlock(&b->lock);
	237	return;
	238	}
	239	EXPORT_SYMBOL_GPL(kvm_async_pf_task_wake);
	240
	241	u32 kvm_read_and_reset_pf_reason(void)
	242	{
	243	u32 reason = 0;
	244
	245	if (__get_cpu_var(apf_reason).enabled) {
	246	reason = __get_cpu_var(apf_reason).reason;
	247	__get_cpu_var(apf_reason).reason = 0;
	248	}
	249
	250	return reason;
	251	}
	252	EXPORT_SYMBOL_GPL(kvm_read_and_reset_pf_reason);
	253
	254	dotraplinkage void __kprobes
	255	do_async_page_fault(struct pt_regs *regs, unsigned long error_code)
	256	{
6c1e0256 FW	257	enum ctx_state prev_state;
6c1e0256 FW	258
631bc487 GN	259	switch (kvm_read_and_reset_pf_reason()) {
	260	default:
	261	do_page_fault(regs, error_code);
	262	break;
	263	case KVM_PV_REASON_PAGE_NOT_PRESENT:
	264	/* page is swapped out by the host. */
6c1e0256	265	prev_state = exception_enter();
c5e015d4	266	exit_idle();
631bc487	267	kvm_async_pf_task_wait((u32)read_cr2());
6c1e0256	268	exception_exit(prev_state);
631bc487 GN	269	break;
631bc487 GN	270	case KVM_PV_REASON_PAGE_READY:
e0875921 GN	271	rcu_irq_enter();
e0875921 GN	272	exit_idle();
631bc487	273	kvm_async_pf_task_wake((u32)read_cr2());
e0875921	274	rcu_irq_exit();
631bc487 GN	275	break;
	276	}
	277	}
	278
d3ac8815	279	static void __init paravirt_ops_setup(void)
0cf1bfd2 MT	280	{
0cf1bfd2 MT	281	pv_info.name = "KVM";
9d2b6132 AL	282
	283	/*
	284	* KVM isn't paravirt in the sense of paravirt_enabled. A KVM
	285	* guest kernel works like a bare metal kernel with additional
	286	* features, and paravirt_enabled is about features that are
	287	* missing.
	288	*/
	289	pv_info.paravirt_enabled = 0;
0cf1bfd2 MT	290
	291	if (kvm_para_has_feature(KVM_FEATURE_NOP_IO_DELAY))
	292	pv_cpu_ops.io_delay = kvm_io_delay;
	293
a90ede7b MT	294	#ifdef CONFIG_X86_IO_APIC
	295	no_timer_check = 1;
	296	#endif
0cf1bfd2 MT	297	}
0cf1bfd2 MT	298
d910f5c1 GC	299	static void kvm_register_steal_time(void)
	300	{
	301	int cpu = smp_processor_id();
	302	struct kvm_steal_time *st = &per_cpu(steal_time, cpu);
	303
	304	if (!has_steal_clock)
	305	return;
	306
	307	memset(st, 0, sizeof(*st));
	308
5dfd486c	309	wrmsrl(MSR_KVM_STEAL_TIME, (slow_virt_to_phys(st) \| KVM_MSR_ENABLED));
136867f5 SK	310	pr_info("kvm-stealtime: cpu %d, msr %llx\n",
136867f5 SK	311	cpu, (unsigned long long) slow_virt_to_phys(st));
d910f5c1 GC	312	}
d910f5c1 GC	313
ab9cf499 MT	314	static DEFINE_PER_CPU(unsigned long, kvm_apic_eoi) = KVM_PV_EOI_DISABLED;
	315
	316	static void kvm_guest_apic_eoi_write(u32 reg, u32 val)
	317	{
	318	/**
	319	* This relies on __test_and_clear_bit to modify the memory
	320	* in a way that is atomic with respect to the local CPU.
	321	* The hypervisor only accesses this memory from the local CPU so
	322	* there's no need for lock or memory barriers.
	323	* An optimization barrier is implied in apic write.
	324	*/
	325	if (__test_and_clear_bit(KVM_PV_EOI_BIT, &__get_cpu_var(kvm_apic_eoi)))
	326	return;
90536664	327	apic_write(APIC_EOI, APIC_EOI_ACK);
ab9cf499 MT	328	}
ab9cf499 MT	329
fd10cde9 GN	330	void __cpuinit kvm_guest_cpu_init(void)
	331	{
	332	if (!kvm_para_available())
	333	return;
	334
	335	if (kvm_para_has_feature(KVM_FEATURE_ASYNC_PF) && kvmapf) {
5dfd486c	336	u64 pa = slow_virt_to_phys(&__get_cpu_var(apf_reason));
fd10cde9	337
6adba527 GN	338	#ifdef CONFIG_PREEMPT
	339	pa \|= KVM_ASYNC_PF_SEND_ALWAYS;
	340	#endif
fd10cde9 GN	341	wrmsrl(MSR_KVM_ASYNC_PF_EN, pa \| KVM_ASYNC_PF_ENABLED);
	342	__get_cpu_var(apf_reason).enabled = 1;
	343	printk(KERN_INFO"KVM setup async PF for cpu %d\n",
	344	smp_processor_id());
	345	}
d910f5c1	346
ab9cf499 MT	347	if (kvm_para_has_feature(KVM_FEATURE_PV_EOI)) {
	348	unsigned long pa;
	349	/* Size alignment is implied but just to make it explicit. */
	350	BUILD_BUG_ON(__alignof__(kvm_apic_eoi) < 4);
	351	__get_cpu_var(kvm_apic_eoi) = 0;
5dfd486c DH	352	pa = slow_virt_to_phys(&__get_cpu_var(kvm_apic_eoi))
5dfd486c DH	353	\| KVM_MSR_ENABLED;
ab9cf499 MT	354	wrmsrl(MSR_KVM_PV_EOI_EN, pa);
	355	}
	356
d910f5c1 GC	357	if (has_steal_clock)
d910f5c1 GC	358	kvm_register_steal_time();
fd10cde9 GN	359	}
fd10cde9 GN	360
ab9cf499	361	static void kvm_pv_disable_apf(void)
fd10cde9 GN	362	{
	363	if (!__get_cpu_var(apf_reason).enabled)
	364	return;
	365
	366	wrmsrl(MSR_KVM_ASYNC_PF_EN, 0);
	367	__get_cpu_var(apf_reason).enabled = 0;
	368
	369	printk(KERN_INFO"Unregister pv shared memory for cpu %d\n",
	370	smp_processor_id());
	371	}
	372
ab9cf499 MT	373	static void kvm_pv_guest_cpu_reboot(void *unused)
	374	{
	375	/*
	376	* We disable PV EOI before we load a new kernel by kexec,
	377	* since MSR_KVM_PV_EOI_EN stores a pointer into old kernel's memory.
	378	* New kernel can re-enable when it boots.
	379	*/
	380	if (kvm_para_has_feature(KVM_FEATURE_PV_EOI))
	381	wrmsrl(MSR_KVM_PV_EOI_EN, 0);
	382	kvm_pv_disable_apf();
8fbe6a54	383	kvm_disable_steal_time();
ab9cf499 MT	384	}
ab9cf499 MT	385
fd10cde9 GN	386	static int kvm_pv_reboot_notify(struct notifier_block *nb,
	387	unsigned long code, void *unused)
	388	{
	389	if (code == SYS_RESTART)
ab9cf499	390	on_each_cpu(kvm_pv_guest_cpu_reboot, NULL, 1);
fd10cde9 GN	391	return NOTIFY_DONE;
	392	}
	393
	394	static struct notifier_block kvm_pv_reboot_nb = {
	395	.notifier_call = kvm_pv_reboot_notify,
	396	};
	397
d910f5c1 GC	398	static u64 kvm_steal_clock(int cpu)
	399	{
	400	u64 steal;
	401	struct kvm_steal_time *src;
	402	int version;
	403
	404	src = &per_cpu(steal_time, cpu);
	405	do {
	406	version = src->version;
	407	rmb();
	408	steal = src->steal;
	409	rmb();
	410	} while ((version & 1) \|\| (version != src->version));
	411
	412	return steal;
	413	}
	414
	415	void kvm_disable_steal_time(void)
	416	{
	417	if (!has_steal_clock)
	418	return;
	419
	420	wrmsr(MSR_KVM_STEAL_TIME, 0, 0);
	421	}
	422
ca3f1017 GN	423	#ifdef CONFIG_SMP
	424	static void __init kvm_smp_prepare_boot_cpu(void)
	425	{
	426	WARN_ON(kvm_register_clock("primary cpu clock"));
fd10cde9	427	kvm_guest_cpu_init();
ca3f1017 GN	428	native_smp_prepare_boot_cpu();
ca3f1017 GN	429	}
fd10cde9	430
775077a0	431	static void __cpuinit kvm_guest_cpu_online(void *dummy)
fd10cde9 GN	432	{
	433	kvm_guest_cpu_init();
	434	}
	435
	436	static void kvm_guest_cpu_offline(void *dummy)
	437	{
d910f5c1	438	kvm_disable_steal_time();
ab9cf499 MT	439	if (kvm_para_has_feature(KVM_FEATURE_PV_EOI))
	440	wrmsrl(MSR_KVM_PV_EOI_EN, 0);
	441	kvm_pv_disable_apf();
631bc487	442	apf_task_wake_all();
fd10cde9 GN	443	}
	444
	445	static int __cpuinit kvm_cpu_notify(struct notifier_block *self,
	446	unsigned long action, void *hcpu)
	447	{
	448	int cpu = (unsigned long)hcpu;
	449	switch (action) {
	450	case CPU_ONLINE:
	451	case CPU_DOWN_FAILED:
	452	case CPU_ONLINE_FROZEN:
	453	smp_call_function_single(cpu, kvm_guest_cpu_online, NULL, 0);
	454	break;
	455	case CPU_DOWN_PREPARE:
	456	case CPU_DOWN_PREPARE_FROZEN:
	457	smp_call_function_single(cpu, kvm_guest_cpu_offline, NULL, 1);
	458	break;
	459	default:
	460	break;
	461	}
	462	return NOTIFY_OK;
	463	}
	464
	465	static struct notifier_block __cpuinitdata kvm_cpu_notifier = {
	466	.notifier_call = kvm_cpu_notify,
	467	};
ca3f1017 GN	468	#endif
ca3f1017 GN	469
631bc487 GN	470	static void __init kvm_apf_trap_init(void)
	471	{
	472	set_intr_gate(14, &async_page_fault);
	473	}
	474
0cf1bfd2 MT	475	void __init kvm_guest_init(void)
0cf1bfd2 MT	476	{
631bc487 GN	477	int i;
631bc487 GN	478
0cf1bfd2 MT	479	if (!kvm_para_available())
	480	return;
	481
	482	paravirt_ops_setup();
fd10cde9	483	register_reboot_notifier(&kvm_pv_reboot_nb);
631bc487 GN	484	for (i = 0; i < KVM_TASK_SLEEP_HASHSIZE; i++)
	485	spin_lock_init(&async_pf_sleepers[i].lock);
	486	if (kvm_para_has_feature(KVM_FEATURE_ASYNC_PF))
	487	x86_init.irqs.trap_init = kvm_apf_trap_init;
	488
d910f5c1 GC	489	if (kvm_para_has_feature(KVM_FEATURE_STEAL_TIME)) {
	490	has_steal_clock = 1;
	491	pv_time_ops.steal_clock = kvm_steal_clock;
	492	}
	493
90536664 MT	494	if (kvm_para_has_feature(KVM_FEATURE_PV_EOI))
90536664 MT	495	apic_set_eoi_write(kvm_guest_apic_eoi_write);
ab9cf499	496
3dc4f7cf MT	497	if (kvmclock_vsyscall)
	498	kvm_setup_vsyscall_timeinfo();
	499
ca3f1017 GN	500	#ifdef CONFIG_SMP
ca3f1017 GN	501	smp_ops.smp_prepare_boot_cpu = kvm_smp_prepare_boot_cpu;
fd10cde9 GN	502	register_cpu_notifier(&kvm_cpu_notifier);
	503	#else
	504	kvm_guest_cpu_init();
ca3f1017	505	#endif
0cf1bfd2	506	}
d910f5c1	507
fc73373b PB	508	static bool __init kvm_detect(void)
	509	{
	510	if (!kvm_para_available())
	511	return false;
	512	return true;
	513	}
	514
	515	const struct hypervisor_x86 x86_hyper_kvm __refconst = {
	516	.name = "KVM",
	517	.detect = kvm_detect,
4cca6ea0	518	.x2apic_available = kvm_para_available,
fc73373b PB	519	};
	520	EXPORT_SYMBOL_GPL(x86_hyper_kvm);
	521
d910f5c1 GC	522	static __init int activate_jump_labels(void)
	523	{
	524	if (has_steal_clock) {
c5905afb	525	static_key_slow_inc(&paravirt_steal_enabled);
d910f5c1	526	if (steal_acc)
c5905afb	527	static_key_slow_inc(&paravirt_steal_rq_enabled);
d910f5c1 GC	528	}
	529
	530	return 0;
	531	}
	532	arch_initcall(activate_jump_labels);