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

/*  KVM paravirtual clock driver. A clocksource implementation
    Copyright (C) 2008 Glauber de Oliveira Costa, Red Hat Inc.

    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, Inc., 51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
*/

#include <linux/clocksource.h>
#include <linux/kvm_para.h>
#include <asm/pvclock.h>
#include <asm/msr.h>
#include <asm/apic.h>
#include <linux/percpu.h>
#include <linux/hardirq.h>
#include <linux/memblock.h>

#include <asm/x86_init.h>
#include <asm/reboot.h>

static int kvmclock = 1;
static int msr_kvm_system_time = MSR_KVM_SYSTEM_TIME;
static int msr_kvm_wall_clock = MSR_KVM_WALL_CLOCK;

static int parse_no_kvmclock(char *arg)
{
	kvmclock = 0;
	return 0;
}
early_param("no-kvmclock", parse_no_kvmclock);

/* The hypervisor will put information about time periodically here */
struct pvclock_aligned_vcpu_time_info {
	struct pvclock_vcpu_time_info clock;
} __attribute__((__aligned__(SMP_CACHE_BYTES)));

static struct pvclock_aligned_vcpu_time_info *hv_clock;
static struct pvclock_wall_clock wall_clock;

/*
 * The wallclock is the time of day when we booted. Since then, some time may
 * have elapsed since the hypervisor wrote the data. So we try to account for
 * that with system time
 */
static unsigned long kvm_get_wallclock(void)
{
	struct pvclock_vcpu_time_info *vcpu_time;
	struct timespec ts;
	int low, high;
	int cpu;

	low = (int)__pa_symbol(&wall_clock);
	high = ((u64)__pa_symbol(&wall_clock) >> 32);

	native_write_msr(msr_kvm_wall_clock, low, high);

	preempt_disable();
	cpu = smp_processor_id();

	vcpu_time = &hv_clock[cpu].clock;
	pvclock_read_wallclock(&wall_clock, vcpu_time, &ts);

	preempt_enable();

	return ts.tv_sec;
}

static int kvm_set_wallclock(unsigned long now)
{
	return -1;
}

static cycle_t kvm_clock_read(void)
{
	struct pvclock_vcpu_time_info *src;
	cycle_t ret;
	int cpu;

	preempt_disable_notrace();
	cpu = smp_processor_id();
	src = &hv_clock[cpu].clock;
	ret = pvclock_clocksource_read(src);
	preempt_enable_notrace();
	return ret;
}

static cycle_t kvm_clock_get_cycles(struct clocksource *cs)
{
	return kvm_clock_read();
}

/*
 * If we don't do that, there is the possibility that the guest
 * will calibrate under heavy load - thus, getting a lower lpj -
 * and execute the delays themselves without load. This is wrong,
 * because no delay loop can finish beforehand.
 * Any heuristics is subject to fail, because ultimately, a large
 * poll of guests can be running and trouble each other. So we preset
 * lpj here
 */
static unsigned long kvm_get_tsc_khz(void)
{
	struct pvclock_vcpu_time_info *src;
	int cpu;
	unsigned long tsc_khz;

	preempt_disable();
	cpu = smp_processor_id();
	src = &hv_clock[cpu].clock;
	tsc_khz = pvclock_tsc_khz(src);
	preempt_enable();
	return tsc_khz;
}

static void kvm_get_preset_lpj(void)
{
	unsigned long khz;
	u64 lpj;

	khz = kvm_get_tsc_khz();

	lpj = ((u64)khz * 1000);
	do_div(lpj, HZ);
	preset_lpj = lpj;
}

bool kvm_check_and_clear_guest_paused(void)
{
	bool ret = false;
	struct pvclock_vcpu_time_info *src;
	int cpu = smp_processor_id();

	if (!hv_clock)
		return ret;

	src = &hv_clock[cpu].clock;
	if ((src->flags & PVCLOCK_GUEST_STOPPED) != 0) {
		src->flags &= ~PVCLOCK_GUEST_STOPPED;
		ret = true;
	}

	return ret;
}

static struct clocksource kvm_clock = {
	.name = "kvm-clock",
	.read = kvm_clock_get_cycles,
	.rating = 400,
	.mask = CLOCKSOURCE_MASK(64),
	.flags = CLOCK_SOURCE_IS_CONTINUOUS,
};

int kvm_register_clock(char *txt)
{
	int cpu = smp_processor_id();
	int low, high, ret;
	struct pvclock_vcpu_time_info *src = &hv_clock[cpu].clock;

	low = (int)__pa(src) | 1;
	high = ((u64)__pa(src) >> 32);
	ret = native_write_msr_safe(msr_kvm_system_time, low, high);
	printk(KERN_INFO "kvm-clock: cpu %d, msr %x:%x, %s\n",
	       cpu, high, low, txt);

	return ret;
}

static void kvm_save_sched_clock_state(void)
{
}

static void kvm_restore_sched_clock_state(void)
{
	kvm_register_clock("primary cpu clock, resume");
}

#ifdef CONFIG_X86_LOCAL_APIC
static void __cpuinit kvm_setup_secondary_clock(void)
{
	/*
	 * Now that the first cpu already had this clocksource initialized,
	 * we shouldn't fail.
	 */
	WARN_ON(kvm_register_clock("secondary cpu clock"));
}
#endif

/*
 * After the clock is registered, the host will keep writing to the
 * registered memory location. If the guest happens to shutdown, this memory
 * won't be valid. In cases like kexec, in which you install a new kernel, this
 * means a random memory location will be kept being written. So before any
 * kind of shutdown from our side, we unregister the clock by writting anything
 * that does not have the 'enable' bit set in the msr
 */
#ifdef CONFIG_KEXEC
static void kvm_crash_shutdown(struct pt_regs *regs)
{
	native_write_msr(msr_kvm_system_time, 0, 0);
	kvm_disable_steal_time();
	native_machine_crash_shutdown(regs);
}
#endif

static void kvm_shutdown(void)
{
	native_write_msr(msr_kvm_system_time, 0, 0);
	kvm_disable_steal_time();
	native_machine_shutdown();
}

void __init kvmclock_init(void)
{
	unsigned long mem;

	if (!kvm_para_available())
		return;

	if (kvmclock && kvm_para_has_feature(KVM_FEATURE_CLOCKSOURCE2)) {
		msr_kvm_system_time = MSR_KVM_SYSTEM_TIME_NEW;
		msr_kvm_wall_clock = MSR_KVM_WALL_CLOCK_NEW;
	} else if (!(kvmclock && kvm_para_has_feature(KVM_FEATURE_CLOCKSOURCE)))
		return;

	printk(KERN_INFO "kvm-clock: Using msrs %x and %x",
		msr_kvm_system_time, msr_kvm_wall_clock);

	mem = memblock_alloc(sizeof(struct pvclock_aligned_vcpu_time_info) * NR_CPUS,
			     PAGE_SIZE);
	if (!mem)
		return;
	hv_clock = __va(mem);

	if (kvm_register_clock("boot clock")) {
		hv_clock = NULL;
		memblock_free(mem,
			sizeof(struct pvclock_aligned_vcpu_time_info)*NR_CPUS);
		return;
	}
	pv_time_ops.sched_clock = kvm_clock_read;
	x86_platform.calibrate_tsc = kvm_get_tsc_khz;
	x86_platform.get_wallclock = kvm_get_wallclock;
	x86_platform.set_wallclock = kvm_set_wallclock;
#ifdef CONFIG_X86_LOCAL_APIC
	x86_cpuinit.early_percpu_clock_init =
		kvm_setup_secondary_clock;
#endif
	x86_platform.save_sched_clock_state = kvm_save_sched_clock_state;
	x86_platform.restore_sched_clock_state = kvm_restore_sched_clock_state;
	machine_ops.shutdown  = kvm_shutdown;
#ifdef CONFIG_KEXEC
	machine_ops.crash_shutdown  = kvm_crash_shutdown;
#endif
	kvm_get_preset_lpj();
	clocksource_register_hz(&kvm_clock, NSEC_PER_SEC);
	pv_info.paravirt_enabled = 1;
	pv_info.name = "KVM";

	if (kvm_para_has_feature(KVM_FEATURE_CLOCKSOURCE_STABLE_BIT))
		pvclock_set_flags(PVCLOCK_TSC_STABLE_BIT);
}
Commit	Line	Data
790c73f6 GOC	1	/* KVM paravirtual clock driver. A clocksource implementation
	2	Copyright (C) 2008 Glauber de Oliveira Costa, Red Hat Inc.
	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, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
	17	*/
	18
	19	#include <linux/clocksource.h>
	20	#include <linux/kvm_para.h>
f6e16d5a	21	#include <asm/pvclock.h>
790c73f6 GOC	22	#include <asm/msr.h>
	23	#include <asm/apic.h>
	24	#include <linux/percpu.h>
3b5d56b9	25	#include <linux/hardirq.h>
7069ed67	26	#include <linux/memblock.h>
736decac TG	27
736decac TG	28	#include <asm/x86_init.h>
1e977aa1	29	#include <asm/reboot.h>
790c73f6	30
790c73f6	31	static int kvmclock = 1;
838815a7 GC	32	static int msr_kvm_system_time = MSR_KVM_SYSTEM_TIME;
838815a7 GC	33	static int msr_kvm_wall_clock = MSR_KVM_WALL_CLOCK;
790c73f6 GOC	34
	35	static int parse_no_kvmclock(char *arg)
	36	{
	37	kvmclock = 0;
	38	return 0;
	39	}
	40	early_param("no-kvmclock", parse_no_kvmclock);
	41
	42	/* The hypervisor will put information about time periodically here */
7069ed67 MT	43	struct pvclock_aligned_vcpu_time_info {
	44	struct pvclock_vcpu_time_info clock;
	45	} __attribute__((__aligned__(SMP_CACHE_BYTES)));
	46
	47	static struct pvclock_aligned_vcpu_time_info *hv_clock;
f6e16d5a	48	static struct pvclock_wall_clock wall_clock;
790c73f6	49
790c73f6 GOC	50	/*
	51	* The wallclock is the time of day when we booted. Since then, some time may
	52	* have elapsed since the hypervisor wrote the data. So we try to account for
	53	* that with system time
	54	*/
2ddfd20e	55	static unsigned long kvm_get_wallclock(void)
790c73f6	56	{
f6e16d5a	57	struct pvclock_vcpu_time_info *vcpu_time;
790c73f6	58	struct timespec ts;
790c73f6	59	int low, high;
7069ed67	60	int cpu;
790c73f6	61
a20316d2 GC	62	low = (int)__pa_symbol(&wall_clock);
a20316d2 GC	63	high = ((u64)__pa_symbol(&wall_clock) >> 32);
838815a7 GC	64
838815a7 GC	65	native_write_msr(msr_kvm_wall_clock, low, high);
790c73f6	66
7069ed67 MT	67	preempt_disable();
	68	cpu = smp_processor_id();
	69
	70	vcpu_time = &hv_clock[cpu].clock;
f6e16d5a	71	pvclock_read_wallclock(&wall_clock, vcpu_time, &ts);
7069ed67 MT	72
7069ed67 MT	73	preempt_enable();
790c73f6	74
f6e16d5a	75	return ts.tv_sec;
790c73f6 GOC	76	}
790c73f6 GOC	77
2ddfd20e	78	static int kvm_set_wallclock(unsigned long now)
790c73f6	79	{
f6e16d5a	80	return -1;
790c73f6 GOC	81	}
790c73f6 GOC	82
790c73f6 GOC	83	static cycle_t kvm_clock_read(void)
790c73f6 GOC	84	{
f6e16d5a GH	85	struct pvclock_vcpu_time_info *src;
f6e16d5a GH	86	cycle_t ret;
7069ed67	87	int cpu;
790c73f6	88
95ef1e52	89	preempt_disable_notrace();
7069ed67 MT	90	cpu = smp_processor_id();
7069ed67 MT	91	src = &hv_clock[cpu].clock;
f6e16d5a	92	ret = pvclock_clocksource_read(src);
95ef1e52	93	preempt_enable_notrace();
f6e16d5a	94	return ret;
790c73f6	95	}
f6e16d5a	96
8e19608e MD	97	static cycle_t kvm_clock_get_cycles(struct clocksource *cs)
	98	{
	99	return kvm_clock_read();
	100	}
	101
0293615f GC	102	/*
	103	* If we don't do that, there is the possibility that the guest
	104	* will calibrate under heavy load - thus, getting a lower lpj -
	105	* and execute the delays themselves without load. This is wrong,
	106	* because no delay loop can finish beforehand.
	107	* Any heuristics is subject to fail, because ultimately, a large
	108	* poll of guests can be running and trouble each other. So we preset
	109	* lpj here
	110	*/
	111	static unsigned long kvm_get_tsc_khz(void)
	112	{
e93353c9	113	struct pvclock_vcpu_time_info *src;
7069ed67 MT	114	int cpu;
	115	unsigned long tsc_khz;
	116
	117	preempt_disable();
	118	cpu = smp_processor_id();
	119	src = &hv_clock[cpu].clock;
	120	tsc_khz = pvclock_tsc_khz(src);
	121	preempt_enable();
	122	return tsc_khz;
0293615f GC	123	}
	124
	125	static void kvm_get_preset_lpj(void)
	126	{
0293615f GC	127	unsigned long khz;
	128	u64 lpj;
	129
e93353c9	130	khz = kvm_get_tsc_khz();
0293615f GC	131
	132	lpj = ((u64)khz * 1000);
	133	do_div(lpj, HZ);
	134	preset_lpj = lpj;
	135	}
	136
3b5d56b9 EM	137	bool kvm_check_and_clear_guest_paused(void)
	138	{
	139	bool ret = false;
	140	struct pvclock_vcpu_time_info *src;
7069ed67 MT	141	int cpu = smp_processor_id();
	142
	143	if (!hv_clock)
	144	return ret;
3b5d56b9	145
7069ed67	146	src = &hv_clock[cpu].clock;
3b5d56b9	147	if ((src->flags & PVCLOCK_GUEST_STOPPED) != 0) {
7069ed67	148	src->flags &= ~PVCLOCK_GUEST_STOPPED;
3b5d56b9 EM	149	ret = true;
	150	}
	151
	152	return ret;
	153	}
3b5d56b9	154
790c73f6 GOC	155	static struct clocksource kvm_clock = {
790c73f6 GOC	156	.name = "kvm-clock",
8e19608e	157	.read = kvm_clock_get_cycles,
790c73f6 GOC	158	.rating = 400,
790c73f6 GOC	159	.mask = CLOCKSOURCE_MASK(64),
790c73f6 GOC	160	.flags = CLOCK_SOURCE_IS_CONTINUOUS,
	161	};
	162
ca3f1017	163	int kvm_register_clock(char *txt)
790c73f6 GOC	164	{
790c73f6 GOC	165	int cpu = smp_processor_id();
19b6a85b	166	int low, high, ret;
7069ed67	167	struct pvclock_vcpu_time_info *src = &hv_clock[cpu].clock;
19b6a85b	168
7069ed67 MT	169	low = (int)__pa(src) \| 1;
7069ed67 MT	170	high = ((u64)__pa(src) >> 32);
19b6a85b	171	ret = native_write_msr_safe(msr_kvm_system_time, low, high);
f6e16d5a GH	172	printk(KERN_INFO "kvm-clock: cpu %d, msr %x:%x, %s\n",
f6e16d5a GH	173	cpu, high, low, txt);
838815a7	174
19b6a85b	175	return ret;
790c73f6 GOC	176	}
790c73f6 GOC	177
b74f05d6 MT	178	static void kvm_save_sched_clock_state(void)
	179	{
	180	}
	181
	182	static void kvm_restore_sched_clock_state(void)
	183	{
	184	kvm_register_clock("primary cpu clock, resume");
	185	}
	186
b8ba5f10	187	#ifdef CONFIG_X86_LOCAL_APIC
23a14b9e	188	static void __cpuinit kvm_setup_secondary_clock(void)
790c73f6 GOC	189	{
	190	/*
	191	* Now that the first cpu already had this clocksource initialized,
	192	* we shouldn't fail.
	193	*/
f6e16d5a	194	WARN_ON(kvm_register_clock("secondary cpu clock"));
790c73f6	195	}
b8ba5f10	196	#endif
790c73f6	197
1e977aa1 GC	198	/*
	199	* After the clock is registered, the host will keep writing to the
	200	* registered memory location. If the guest happens to shutdown, this memory
	201	* won't be valid. In cases like kexec, in which you install a new kernel, this
	202	* means a random memory location will be kept being written. So before any
	203	* kind of shutdown from our side, we unregister the clock by writting anything
	204	* that does not have the 'enable' bit set in the msr
	205	*/
	206	#ifdef CONFIG_KEXEC
	207	static void kvm_crash_shutdown(struct pt_regs *regs)
	208	{
838815a7	209	native_write_msr(msr_kvm_system_time, 0, 0);
d910f5c1	210	kvm_disable_steal_time();
1e977aa1 GC	211	native_machine_crash_shutdown(regs);
	212	}
	213	#endif
	214
	215	static void kvm_shutdown(void)
	216	{
838815a7	217	native_write_msr(msr_kvm_system_time, 0, 0);
d910f5c1	218	kvm_disable_steal_time();
1e977aa1 GC	219	native_machine_shutdown();
	220	}
	221
790c73f6 GOC	222	void __init kvmclock_init(void)
790c73f6 GOC	223	{
7069ed67 MT	224	unsigned long mem;
7069ed67 MT	225
790c73f6 GOC	226	if (!kvm_para_available())
	227	return;
	228
838815a7 GC	229	if (kvmclock && kvm_para_has_feature(KVM_FEATURE_CLOCKSOURCE2)) {
	230	msr_kvm_system_time = MSR_KVM_SYSTEM_TIME_NEW;
	231	msr_kvm_wall_clock = MSR_KVM_WALL_CLOCK_NEW;
	232	} else if (!(kvmclock && kvm_para_has_feature(KVM_FEATURE_CLOCKSOURCE)))
	233	return;
	234
	235	printk(KERN_INFO "kvm-clock: Using msrs %x and %x",
	236	msr_kvm_system_time, msr_kvm_wall_clock);
	237
7069ed67 MT	238	mem = memblock_alloc(sizeof(struct pvclock_aligned_vcpu_time_info) * NR_CPUS,
	239	PAGE_SIZE);
	240	if (!mem)
	241	return;
	242	hv_clock = __va(mem);
	243
	244	if (kvm_register_clock("boot clock")) {
	245	hv_clock = NULL;
	246	memblock_free(mem,
	247	sizeof(struct pvclock_aligned_vcpu_time_info)*NR_CPUS);
838815a7	248	return;
7069ed67	249	}
838815a7 GC	250	pv_time_ops.sched_clock = kvm_clock_read;
	251	x86_platform.calibrate_tsc = kvm_get_tsc_khz;
	252	x86_platform.get_wallclock = kvm_get_wallclock;
	253	x86_platform.set_wallclock = kvm_set_wallclock;
b8ba5f10	254	#ifdef CONFIG_X86_LOCAL_APIC
df156f90	255	x86_cpuinit.early_percpu_clock_init =
838815a7	256	kvm_setup_secondary_clock;
b8ba5f10	257	#endif
b74f05d6 MT	258	x86_platform.save_sched_clock_state = kvm_save_sched_clock_state;
b74f05d6 MT	259	x86_platform.restore_sched_clock_state = kvm_restore_sched_clock_state;
838815a7	260	machine_ops.shutdown = kvm_shutdown;
1e977aa1	261	#ifdef CONFIG_KEXEC
838815a7	262	machine_ops.crash_shutdown = kvm_crash_shutdown;
1e977aa1	263	#endif
838815a7	264	kvm_get_preset_lpj();
b01cc1b0	265	clocksource_register_hz(&kvm_clock, NSEC_PER_SEC);
838815a7 GC	266	pv_info.paravirt_enabled = 1;
838815a7 GC	267	pv_info.name = "KVM";
3a0d7256 GC	268
	269	if (kvm_para_has_feature(KVM_FEATURE_CLOCKSOURCE_STABLE_BIT))
	270	pvclock_set_flags(PVCLOCK_TSC_STABLE_BIT);
790c73f6	271	}