[linux-2.6-block.git] / virt / kvm / arm / arch_timer.c

/*
 * Copyright (C) 2012 ARM Ltd.
 * Author: Marc Zyngier <marc.zyngier@arm.com>
 *
 * This program is free software; you can redistribute it and/or modify
 * it under the terms of the GNU General Public License version 2 as
 * published by the Free Software Foundation.
 *
 * 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, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
 */

#include <linux/cpu.h>
#include <linux/of_irq.h>
#include <linux/kvm.h>
#include <linux/kvm_host.h>
#include <linux/interrupt.h>

#include <clocksource/arm_arch_timer.h>
#include <asm/arch_timer.h>

#include <kvm/arm_vgic.h>
#include <kvm/arm_arch_timer.h>

static struct timecounter *timecounter;
static struct workqueue_struct *wqueue;
static unsigned int host_vtimer_irq;

static cycle_t kvm_phys_timer_read(void)
{
	return timecounter->cc->read(timecounter->cc);
}

static bool timer_is_armed(struct arch_timer_cpu *timer)
{
	return timer->armed;
}

/* timer_arm: as in "arm the timer", not as in ARM the company */
static void timer_arm(struct arch_timer_cpu *timer, u64 ns)
{
	timer->armed = true;
	hrtimer_start(&timer->timer, ktime_add_ns(ktime_get(), ns),
		      HRTIMER_MODE_ABS);
}

static void timer_disarm(struct arch_timer_cpu *timer)
{
	if (timer_is_armed(timer)) {
		hrtimer_cancel(&timer->timer);
		cancel_work_sync(&timer->expired);
		timer->armed = false;
	}
}

static void kvm_timer_inject_irq(struct kvm_vcpu *vcpu)
{
	int ret;
	struct arch_timer_cpu *timer = &vcpu->arch.timer_cpu;

	timer->cntv_ctl |= ARCH_TIMER_CTRL_IT_MASK;
	ret = kvm_vgic_inject_irq(vcpu->kvm, vcpu->vcpu_id,
				  timer->irq->irq,
				  timer->irq->level);
	WARN_ON(ret);
}

static irqreturn_t kvm_arch_timer_handler(int irq, void *dev_id)
{
	struct kvm_vcpu *vcpu = *(struct kvm_vcpu **)dev_id;

	/*
	 * We disable the timer in the world switch and let it be
	 * handled by kvm_timer_sync_hwstate(). Getting a timer
	 * interrupt at this point is a sure sign of some major
	 * breakage.
	 */
	pr_warn("Unexpected interrupt %d on vcpu %p\n", irq, vcpu);
	return IRQ_HANDLED;
}

static void kvm_timer_inject_irq_work(struct work_struct *work)
{
	struct kvm_vcpu *vcpu;

	vcpu = container_of(work, struct kvm_vcpu, arch.timer_cpu.expired);
	vcpu->arch.timer_cpu.armed = false;
	kvm_timer_inject_irq(vcpu);
}

static enum hrtimer_restart kvm_timer_expire(struct hrtimer *hrt)
{
	struct arch_timer_cpu *timer;
	timer = container_of(hrt, struct arch_timer_cpu, timer);
	queue_work(wqueue, &timer->expired);
	return HRTIMER_NORESTART;
}

/**
 * kvm_timer_flush_hwstate - prepare to move the virt timer to the cpu
 * @vcpu: The vcpu pointer
 *
 * Disarm any pending soft timers, since the world-switch code will write the
 * virtual timer state back to the physical CPU.
 */
void kvm_timer_flush_hwstate(struct kvm_vcpu *vcpu)
{
	struct arch_timer_cpu *timer = &vcpu->arch.timer_cpu;

	/*
	 * We're about to run this vcpu again, so there is no need to
	 * keep the background timer running, as we're about to
	 * populate the CPU timer again.
	 */
	timer_disarm(timer);
}

/**
 * kvm_timer_sync_hwstate - sync timer state from cpu
 * @vcpu: The vcpu pointer
 *
 * Check if the virtual timer was armed and either schedule a corresponding
 * soft timer or inject directly if already expired.
 */
void kvm_timer_sync_hwstate(struct kvm_vcpu *vcpu)
{
	struct arch_timer_cpu *timer = &vcpu->arch.timer_cpu;
	cycle_t cval, now;
	u64 ns;

	if ((timer->cntv_ctl & ARCH_TIMER_CTRL_IT_MASK) ||
		!(timer->cntv_ctl & ARCH_TIMER_CTRL_ENABLE))
		return;

	cval = timer->cntv_cval;
	now = kvm_phys_timer_read() - vcpu->kvm->arch.timer.cntvoff;

	BUG_ON(timer_is_armed(timer));

	if (cval <= now) {
		/*
		 * Timer has already expired while we were not
		 * looking. Inject the interrupt and carry on.
		 */
		kvm_timer_inject_irq(vcpu);
		return;
	}

	ns = cyclecounter_cyc2ns(timecounter->cc, cval - now, timecounter->mask,
				 &timecounter->frac);
	timer_arm(timer, ns);
}

void kvm_timer_vcpu_reset(struct kvm_vcpu *vcpu,
			  const struct kvm_irq_level *irq)
{
	struct arch_timer_cpu *timer = &vcpu->arch.timer_cpu;

	/*
	 * The vcpu timer irq number cannot be determined in
	 * kvm_timer_vcpu_init() because it is called much before
	 * kvm_vcpu_set_target(). To handle this, we determine
	 * vcpu timer irq number when the vcpu is reset.
	 */
	timer->irq = irq;
}

void kvm_timer_vcpu_init(struct kvm_vcpu *vcpu)
{
	struct arch_timer_cpu *timer = &vcpu->arch.timer_cpu;

	INIT_WORK(&timer->expired, kvm_timer_inject_irq_work);
	hrtimer_init(&timer->timer, CLOCK_MONOTONIC, HRTIMER_MODE_ABS);
	timer->timer.function = kvm_timer_expire;
}

static void kvm_timer_init_interrupt(void *info)
{
	enable_percpu_irq(host_vtimer_irq, 0);
}

int kvm_arm_timer_set_reg(struct kvm_vcpu *vcpu, u64 regid, u64 value)
{
	struct arch_timer_cpu *timer = &vcpu->arch.timer_cpu;

	switch (regid) {
	case KVM_REG_ARM_TIMER_CTL:
		timer->cntv_ctl = value;
		break;
	case KVM_REG_ARM_TIMER_CNT:
		vcpu->kvm->arch.timer.cntvoff = kvm_phys_timer_read() - value;
		break;
	case KVM_REG_ARM_TIMER_CVAL:
		timer->cntv_cval = value;
		break;
	default:
		return -1;
	}
	return 0;
}

u64 kvm_arm_timer_get_reg(struct kvm_vcpu *vcpu, u64 regid)
{
	struct arch_timer_cpu *timer = &vcpu->arch.timer_cpu;

	switch (regid) {
	case KVM_REG_ARM_TIMER_CTL:
		return timer->cntv_ctl;
	case KVM_REG_ARM_TIMER_CNT:
		return kvm_phys_timer_read() - vcpu->kvm->arch.timer.cntvoff;
	case KVM_REG_ARM_TIMER_CVAL:
		return timer->cntv_cval;
	}
	return (u64)-1;
}

static int kvm_timer_cpu_notify(struct notifier_block *self,
				unsigned long action, void *cpu)
{
	switch (action) {
	case CPU_STARTING:
	case CPU_STARTING_FROZEN:
		kvm_timer_init_interrupt(NULL);
		break;
	case CPU_DYING:
	case CPU_DYING_FROZEN:
		disable_percpu_irq(host_vtimer_irq);
		break;
	}

	return NOTIFY_OK;
}

static struct notifier_block kvm_timer_cpu_nb = {
	.notifier_call = kvm_timer_cpu_notify,
};

static const struct of_device_id arch_timer_of_match[] = {
	{ .compatible	= "arm,armv7-timer",	},
	{ .compatible	= "arm,armv8-timer",	},
	{},
};

int kvm_timer_hyp_init(void)
{
	struct device_node *np;
	unsigned int ppi;
	int err;

	timecounter = arch_timer_get_timecounter();
	if (!timecounter)
		return -ENODEV;

	np = of_find_matching_node(NULL, arch_timer_of_match);
	if (!np) {
		kvm_err("kvm_arch_timer: can't find DT node\n");
		return -ENODEV;
	}

	ppi = irq_of_parse_and_map(np, 2);
	if (!ppi) {
		kvm_err("kvm_arch_timer: no virtual timer interrupt\n");
		err = -EINVAL;
		goto out;
	}

	err = request_percpu_irq(ppi, kvm_arch_timer_handler,
				 "kvm guest timer", kvm_get_running_vcpus());
	if (err) {
		kvm_err("kvm_arch_timer: can't request interrupt %d (%d)\n",
			ppi, err);
		goto out;
	}

	host_vtimer_irq = ppi;

	err = __register_cpu_notifier(&kvm_timer_cpu_nb);
	if (err) {
		kvm_err("Cannot register timer CPU notifier\n");
		goto out_free;
	}

	wqueue = create_singlethread_workqueue("kvm_arch_timer");
	if (!wqueue) {
		err = -ENOMEM;
		goto out_free;
	}

	kvm_info("%s IRQ%d\n", np->name, ppi);
	on_each_cpu(kvm_timer_init_interrupt, NULL, 1);

	goto out;
out_free:
	free_percpu_irq(ppi, kvm_get_running_vcpus());
out:
	of_node_put(np);
	return err;
}

void kvm_timer_vcpu_terminate(struct kvm_vcpu *vcpu)
{
	struct arch_timer_cpu *timer = &vcpu->arch.timer_cpu;

	timer_disarm(timer);
}

void kvm_timer_enable(struct kvm *kvm)
{
	if (kvm->arch.timer.enabled)
		return;

	/*
	 * There is a potential race here between VCPUs starting for the first
	 * time, which may be enabling the timer multiple times.  That doesn't
	 * hurt though, because we're just setting a variable to the same
	 * variable that it already was.  The important thing is that all
	 * VCPUs have the enabled variable set, before entering the guest, if
	 * the arch timers are enabled.
	 */
	if (timecounter && wqueue)
		kvm->arch.timer.enabled = 1;
}

void kvm_timer_init(struct kvm *kvm)
{
	kvm->arch.timer.cntvoff = kvm_phys_timer_read();
}
Commit	Line	Data
53e72406 MZ	1	/*
	2	* Copyright (C) 2012 ARM Ltd.
	3	* Author: Marc Zyngier <marc.zyngier@arm.com>
	4	*
	5	* This program is free software; you can redistribute it and/or modify
	6	* it under the terms of the GNU General Public License version 2 as
	7	* published by the Free Software Foundation.
	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, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
	17	*/
	18
	19	#include <linux/cpu.h>
	20	#include <linux/of_irq.h>
	21	#include <linux/kvm.h>
	22	#include <linux/kvm_host.h>
	23	#include <linux/interrupt.h>
	24
372b7c1b	25	#include <clocksource/arm_arch_timer.h>
53e72406 MZ	26	#include <asm/arch_timer.h>
53e72406 MZ	27
7275acdf MZ	28	#include <kvm/arm_vgic.h>
7275acdf MZ	29	#include <kvm/arm_arch_timer.h>
53e72406 MZ	30
	31	static struct timecounter *timecounter;
	32	static struct workqueue_struct *wqueue;
5ae7f87a	33	static unsigned int host_vtimer_irq;
53e72406 MZ	34
	35	static cycle_t kvm_phys_timer_read(void)
	36	{
	37	return timecounter->cc->read(timecounter->cc);
	38	}
	39
	40	static bool timer_is_armed(struct arch_timer_cpu *timer)
	41	{
	42	return timer->armed;
	43	}
	44
	45	/* timer_arm: as in "arm the timer", not as in ARM the company */
	46	static void timer_arm(struct arch_timer_cpu *timer, u64 ns)
	47	{
	48	timer->armed = true;
	49	hrtimer_start(&timer->timer, ktime_add_ns(ktime_get(), ns),
	50	HRTIMER_MODE_ABS);
	51	}
	52
	53	static void timer_disarm(struct arch_timer_cpu *timer)
	54	{
	55	if (timer_is_armed(timer)) {
	56	hrtimer_cancel(&timer->timer);
	57	cancel_work_sync(&timer->expired);
	58	timer->armed = false;
	59	}
	60	}
	61
	62	static void kvm_timer_inject_irq(struct kvm_vcpu *vcpu)
	63	{
05971120	64	int ret;
53e72406 MZ	65	struct arch_timer_cpu *timer = &vcpu->arch.timer_cpu;
53e72406 MZ	66
372b7c1b	67	timer->cntv_ctl \|= ARCH_TIMER_CTRL_IT_MASK;
05971120 CD	68	ret = kvm_vgic_inject_irq(vcpu->kvm, vcpu->vcpu_id,
	69	timer->irq->irq,
	70	timer->irq->level);
	71	WARN_ON(ret);
53e72406 MZ	72	}
	73
	74	static irqreturn_t kvm_arch_timer_handler(int irq, void *dev_id)
	75	{
	76	struct kvm_vcpu vcpu = (struct kvm_vcpu **)dev_id;
	77
	78	/*
	79	* We disable the timer in the world switch and let it be
	80	* handled by kvm_timer_sync_hwstate(). Getting a timer
	81	* interrupt at this point is a sure sign of some major
	82	* breakage.
	83	*/
	84	pr_warn("Unexpected interrupt %d on vcpu %p\n", irq, vcpu);
	85	return IRQ_HANDLED;
	86	}
	87
	88	static void kvm_timer_inject_irq_work(struct work_struct *work)
	89	{
	90	struct kvm_vcpu *vcpu;
	91
	92	vcpu = container_of(work, struct kvm_vcpu, arch.timer_cpu.expired);
	93	vcpu->arch.timer_cpu.armed = false;
	94	kvm_timer_inject_irq(vcpu);
	95	}
	96
	97	static enum hrtimer_restart kvm_timer_expire(struct hrtimer *hrt)
	98	{
	99	struct arch_timer_cpu *timer;
	100	timer = container_of(hrt, struct arch_timer_cpu, timer);
	101	queue_work(wqueue, &timer->expired);
	102	return HRTIMER_NORESTART;
	103	}
	104
	105	/**
	106	* kvm_timer_flush_hwstate - prepare to move the virt timer to the cpu
	107	* @vcpu: The vcpu pointer
	108	*
	109	* Disarm any pending soft timers, since the world-switch code will write the
	110	* virtual timer state back to the physical CPU.
	111	*/
	112	void kvm_timer_flush_hwstate(struct kvm_vcpu *vcpu)
	113	{
	114	struct arch_timer_cpu *timer = &vcpu->arch.timer_cpu;
	115
	116	/*
	117	* We're about to run this vcpu again, so there is no need to
	118	* keep the background timer running, as we're about to
	119	* populate the CPU timer again.
	120	*/
	121	timer_disarm(timer);
	122	}
	123
	124	/**
	125	* kvm_timer_sync_hwstate - sync timer state from cpu
	126	* @vcpu: The vcpu pointer
	127	*
	128	* Check if the virtual timer was armed and either schedule a corresponding
	129	* soft timer or inject directly if already expired.
	130	*/
	131	void kvm_timer_sync_hwstate(struct kvm_vcpu *vcpu)
	132	{
	133	struct arch_timer_cpu *timer = &vcpu->arch.timer_cpu;
	134	cycle_t cval, now;
	135	u64 ns;
136
372b7c1b MR	137	if ((timer->cntv_ctl & ARCH_TIMER_CTRL_IT_MASK) \|\|
372b7c1b MR	138	!(timer->cntv_ctl & ARCH_TIMER_CTRL_ENABLE))
53e72406 MZ	139	return;
	140
	141	cval = timer->cntv_cval;
	142	now = kvm_phys_timer_read() - vcpu->kvm->arch.timer.cntvoff;
	143
	144	BUG_ON(timer_is_armed(timer));
	145
	146	if (cval <= now) {
	147	/*
	148	* Timer has already expired while we were not
	149	* looking. Inject the interrupt and carry on.
	150	*/
	151	kvm_timer_inject_irq(vcpu);
	152	return;
	153	}
	154
2eebdde6 RC	155	ns = cyclecounter_cyc2ns(timecounter->cc, cval - now, timecounter->mask,
2eebdde6 RC	156	&timecounter->frac);
53e72406 MZ	157	timer_arm(timer, ns);
	158	}
	159
5ae7f87a AP	160	void kvm_timer_vcpu_reset(struct kvm_vcpu *vcpu,
	161	const struct kvm_irq_level *irq)
	162	{
	163	struct arch_timer_cpu *timer = &vcpu->arch.timer_cpu;
	164
	165	/*
	166	* The vcpu timer irq number cannot be determined in
	167	* kvm_timer_vcpu_init() because it is called much before
	168	* kvm_vcpu_set_target(). To handle this, we determine
	169	* vcpu timer irq number when the vcpu is reset.
	170	*/
	171	timer->irq = irq;
	172	}
	173
53e72406 MZ	174	void kvm_timer_vcpu_init(struct kvm_vcpu *vcpu)
	175	{
	176	struct arch_timer_cpu *timer = &vcpu->arch.timer_cpu;
	177
	178	INIT_WORK(&timer->expired, kvm_timer_inject_irq_work);
	179	hrtimer_init(&timer->timer, CLOCK_MONOTONIC, HRTIMER_MODE_ABS);
	180	timer->timer.function = kvm_timer_expire;
53e72406 MZ	181	}
	182
	183	static void kvm_timer_init_interrupt(void *info)
	184	{
5ae7f87a	185	enable_percpu_irq(host_vtimer_irq, 0);
53e72406 MZ	186	}
53e72406 MZ	187
39735a3a AP	188	int kvm_arm_timer_set_reg(struct kvm_vcpu *vcpu, u64 regid, u64 value)
	189	{
	190	struct arch_timer_cpu *timer = &vcpu->arch.timer_cpu;
	191
	192	switch (regid) {
	193	case KVM_REG_ARM_TIMER_CTL:
	194	timer->cntv_ctl = value;
	195	break;
	196	case KVM_REG_ARM_TIMER_CNT:
	197	vcpu->kvm->arch.timer.cntvoff = kvm_phys_timer_read() - value;
	198	break;
	199	case KVM_REG_ARM_TIMER_CVAL:
	200	timer->cntv_cval = value;
	201	break;
	202	default:
	203	return -1;
	204	}
	205	return 0;
	206	}
	207
	208	u64 kvm_arm_timer_get_reg(struct kvm_vcpu *vcpu, u64 regid)
	209	{
	210	struct arch_timer_cpu *timer = &vcpu->arch.timer_cpu;
	211
	212	switch (regid) {
	213	case KVM_REG_ARM_TIMER_CTL:
	214	return timer->cntv_ctl;
	215	case KVM_REG_ARM_TIMER_CNT:
	216	return kvm_phys_timer_read() - vcpu->kvm->arch.timer.cntvoff;
	217	case KVM_REG_ARM_TIMER_CVAL:
	218	return timer->cntv_cval;
	219	}
	220	return (u64)-1;
	221	}
53e72406 MZ	222
	223	static int kvm_timer_cpu_notify(struct notifier_block *self,
	224	unsigned long action, void *cpu)
	225	{
	226	switch (action) {
	227	case CPU_STARTING:
	228	case CPU_STARTING_FROZEN:
	229	kvm_timer_init_interrupt(NULL);
	230	break;
	231	case CPU_DYING:
	232	case CPU_DYING_FROZEN:
5ae7f87a	233	disable_percpu_irq(host_vtimer_irq);
53e72406 MZ	234	break;
	235	}
	236
	237	return NOTIFY_OK;
	238	}
	239
	240	static struct notifier_block kvm_timer_cpu_nb = {
	241	.notifier_call = kvm_timer_cpu_notify,
	242	};
	243
	244	static const struct of_device_id arch_timer_of_match[] = {
	245	{ .compatible = "arm,armv7-timer", },
f61701e0	246	{ .compatible = "arm,armv8-timer", },
53e72406 MZ	247	{},
	248	};
	249
	250	int kvm_timer_hyp_init(void)
	251	{
	252	struct device_node *np;
	253	unsigned int ppi;
	254	int err;
	255
	256	timecounter = arch_timer_get_timecounter();
	257	if (!timecounter)
	258	return -ENODEV;
	259
	260	np = of_find_matching_node(NULL, arch_timer_of_match);
	261	if (!np) {
	262	kvm_err("kvm_arch_timer: can't find DT node\n");
	263	return -ENODEV;
	264	}
	265
	266	ppi = irq_of_parse_and_map(np, 2);
	267	if (!ppi) {
	268	kvm_err("kvm_arch_timer: no virtual timer interrupt\n");
	269	err = -EINVAL;
	270	goto out;
	271	}
	272
	273	err = request_percpu_irq(ppi, kvm_arch_timer_handler,
	274	"kvm guest timer", kvm_get_running_vcpus());
	275	if (err) {
	276	kvm_err("kvm_arch_timer: can't request interrupt %d (%d)\n",
	277	ppi, err);
	278	goto out;
	279	}
	280
5ae7f87a	281	host_vtimer_irq = ppi;
53e72406	282
553f809e	283	err = __register_cpu_notifier(&kvm_timer_cpu_nb);
53e72406 MZ	284	if (err) {
	285	kvm_err("Cannot register timer CPU notifier\n");
	286	goto out_free;
	287	}
	288
	289	wqueue = create_singlethread_workqueue("kvm_arch_timer");
	290	if (!wqueue) {
	291	err = -ENOMEM;
	292	goto out_free;
	293	}
	294
	295	kvm_info("%s IRQ%d\n", np->name, ppi);
	296	on_each_cpu(kvm_timer_init_interrupt, NULL, 1);
	297
	298	goto out;
	299	out_free:
	300	free_percpu_irq(ppi, kvm_get_running_vcpus());
	301	out:
	302	of_node_put(np);
	303	return err;
	304	}
	305
	306	void kvm_timer_vcpu_terminate(struct kvm_vcpu *vcpu)
	307	{
	308	struct arch_timer_cpu *timer = &vcpu->arch.timer_cpu;
	309
	310	timer_disarm(timer);
	311	}
	312
05971120	313	void kvm_timer_enable(struct kvm *kvm)
53e72406	314	{
05971120 CD	315	if (kvm->arch.timer.enabled)
	316	return;
	317
	318	/*
	319	* There is a potential race here between VCPUs starting for the first
	320	* time, which may be enabling the timer multiple times. That doesn't
	321	* hurt though, because we're just setting a variable to the same
	322	* variable that it already was. The important thing is that all
	323	* VCPUs have the enabled variable set, before entering the guest, if
	324	* the arch timers are enabled.
	325	*/
	326	if (timecounter && wqueue)
53e72406	327	kvm->arch.timer.enabled = 1;
05971120	328	}
53e72406	329
05971120 CD	330	void kvm_timer_init(struct kvm *kvm)
	331	{
	332	kvm->arch.timer.cntvoff = kvm_phys_timer_read();
53e72406	333	}