[deliverable/linux.git] / arch / x86 / kernel / vsyscall_64.c

/*
 *  Copyright (C) 2001 Andrea Arcangeli <andrea@suse.de> SuSE
 *  Copyright 2003 Andi Kleen, SuSE Labs.
 *
 *  Thanks to hpa@transmeta.com for some useful hint.
 *  Special thanks to Ingo Molnar for his early experience with
 *  a different vsyscall implementation for Linux/IA32 and for the name.
 *
 *  vsyscall 1 is located at -10Mbyte, vsyscall 2 is located
 *  at virtual address -10Mbyte+1024bytes etc... There are at max 4
 *  vsyscalls. One vsyscall can reserve more than 1 slot to avoid
 *  jumping out of line if necessary. We cannot add more with this
 *  mechanism because older kernels won't return -ENOSYS.
 *  If we want more than four we need a vDSO.
 *
 *  Note: the concept clashes with user mode linux. If you use UML and
 *  want per guest time just set the kernel.vsyscall64 sysctl to 0.
 */

#include <linux/time.h>
#include <linux/init.h>
#include <linux/kernel.h>
#include <linux/timer.h>
#include <linux/seqlock.h>
#include <linux/jiffies.h>
#include <linux/sysctl.h>
#include <linux/clocksource.h>
#include <linux/getcpu.h>
#include <linux/cpu.h>
#include <linux/smp.h>
#include <linux/notifier.h>

#include <asm/vsyscall.h>
#include <asm/pgtable.h>
#include <asm/page.h>
#include <asm/unistd.h>
#include <asm/fixmap.h>
#include <asm/errno.h>
#include <asm/io.h>
#include <asm/segment.h>
#include <asm/desc.h>
#include <asm/topology.h>
#include <asm/vgtod.h>

#define __vsyscall(nr) __attribute__ ((unused,__section__(".vsyscall_" #nr)))
#define __syscall_clobber "r11","cx","memory"
#define __pa_vsymbol(x)			\
	({unsigned long v;  		\
	extern char __vsyscall_0; 	\
	  asm("" : "=r" (v) : "0" (x)); \
	  ((v - VSYSCALL_START) + __pa_symbol(&__vsyscall_0)); })

/*
 * vsyscall_gtod_data contains data that is :
 * - readonly from vsyscalls
 * - written by timer interrupt or systcl (/proc/sys/kernel/vsyscall64)
 * Try to keep this structure as small as possible to avoid cache line ping pongs
 */
int __vgetcpu_mode __section_vgetcpu_mode;

struct vsyscall_gtod_data __vsyscall_gtod_data __section_vsyscall_gtod_data =
{
	.lock = SEQLOCK_UNLOCKED,
	.sysctl_enabled = 1,
};

void update_vsyscall_tz(void)
{
	unsigned long flags;

	write_seqlock_irqsave(&vsyscall_gtod_data.lock, flags);
	/* sys_tz has changed */
	vsyscall_gtod_data.sys_tz = sys_tz;
	write_sequnlock_irqrestore(&vsyscall_gtod_data.lock, flags);
}

void update_vsyscall(struct timespec *wall_time, struct clocksource *clock)
{
	unsigned long flags;

	write_seqlock_irqsave(&vsyscall_gtod_data.lock, flags);
	/* copy vsyscall data */
	vsyscall_gtod_data.clock.vread = clock->vread;
	vsyscall_gtod_data.clock.cycle_last = clock->cycle_last;
	vsyscall_gtod_data.clock.mask = clock->mask;
	vsyscall_gtod_data.clock.mult = clock->mult;
	vsyscall_gtod_data.clock.shift = clock->shift;
	vsyscall_gtod_data.wall_time_sec = wall_time->tv_sec;
	vsyscall_gtod_data.wall_time_nsec = wall_time->tv_nsec;
	vsyscall_gtod_data.wall_to_monotonic = wall_to_monotonic;
	write_sequnlock_irqrestore(&vsyscall_gtod_data.lock, flags);
}

/* RED-PEN may want to readd seq locking, but then the variable should be
 * write-once.
 */
static __always_inline void do_get_tz(struct timezone * tz)
{
	*tz = __vsyscall_gtod_data.sys_tz;
}

static __always_inline int gettimeofday(struct timeval *tv, struct timezone *tz)
{
	int ret;
	asm volatile("vsysc2: syscall"
		: "=a" (ret)
		: "0" (__NR_gettimeofday),"D" (tv),"S" (tz)
		: __syscall_clobber );
	return ret;
}

static __always_inline long time_syscall(long *t)
{
	long secs;
	asm volatile("vsysc1: syscall"
		: "=a" (secs)
		: "0" (__NR_time),"D" (t) : __syscall_clobber);
	return secs;
}

static __always_inline void do_vgettimeofday(struct timeval * tv)
{
	cycle_t now, base, mask, cycle_delta;
	unsigned seq;
	unsigned long mult, shift, nsec;
	cycle_t (*vread)(void);
	do {
		seq = read_seqbegin(&__vsyscall_gtod_data.lock);

		vread = __vsyscall_gtod_data.clock.vread;
		if (unlikely(!__vsyscall_gtod_data.sysctl_enabled || !vread)) {
			gettimeofday(tv,NULL);
			return;
		}
		now = vread();
		base = __vsyscall_gtod_data.clock.cycle_last;
		mask = __vsyscall_gtod_data.clock.mask;
		mult = __vsyscall_gtod_data.clock.mult;
		shift = __vsyscall_gtod_data.clock.shift;

		tv->tv_sec = __vsyscall_gtod_data.wall_time_sec;
		nsec = __vsyscall_gtod_data.wall_time_nsec;
	} while (read_seqretry(&__vsyscall_gtod_data.lock, seq));

	/* calculate interval: */
	cycle_delta = (now - base) & mask;
	/* convert to nsecs: */
	nsec += (cycle_delta * mult) >> shift;

	while (nsec >= NSEC_PER_SEC) {
		tv->tv_sec += 1;
		nsec -= NSEC_PER_SEC;
	}
	tv->tv_usec = nsec / NSEC_PER_USEC;
}

int __vsyscall(0) vgettimeofday(struct timeval * tv, struct timezone * tz)
{
	if (tv)
		do_vgettimeofday(tv);
	if (tz)
		do_get_tz(tz);
	return 0;
}

/* This will break when the xtime seconds get inaccurate, but that is
 * unlikely */
time_t __vsyscall(1) vtime(time_t *t)
{
	struct timeval tv;
	time_t result;
	if (unlikely(!__vsyscall_gtod_data.sysctl_enabled))
		return time_syscall(t);

	vgettimeofday(&tv, NULL);
	result = tv.tv_sec;
	if (t)
		*t = result;
	return result;
}

/* Fast way to get current CPU and node.
   This helps to do per node and per CPU caches in user space.
   The result is not guaranteed without CPU affinity, but usually
   works out because the scheduler tries to keep a thread on the same
   CPU.

   tcache must point to a two element sized long array.
   All arguments can be NULL. */
long __vsyscall(2)
vgetcpu(unsigned *cpu, unsigned *node, struct getcpu_cache *tcache)
{
	unsigned int p;
	unsigned long j = 0;

	/* Fast cache - only recompute value once per jiffies and avoid
	   relatively costly rdtscp/cpuid otherwise.
	   This works because the scheduler usually keeps the process
	   on the same CPU and this syscall doesn't guarantee its
	   results anyways.
	   We do this here because otherwise user space would do it on
	   its own in a likely inferior way (no access to jiffies).
	   If you don't like it pass NULL. */
	if (tcache && tcache->blob[0] == (j = __jiffies)) {
		p = tcache->blob[1];
	} else if (__vgetcpu_mode == VGETCPU_RDTSCP) {
		/* Load per CPU data from RDTSCP */
		native_read_tscp(&p);
	} else {
		/* Load per CPU data from GDT */
		asm("lsl %1,%0" : "=r" (p) : "r" (__PER_CPU_SEG));
	}
	if (tcache) {
		tcache->blob[0] = j;
		tcache->blob[1] = p;
	}
	if (cpu)
		*cpu = p & 0xfff;
	if (node)
		*node = p >> 12;
	return 0;
}

long __vsyscall(3) venosys_1(void)
{
	return -ENOSYS;
}

#ifdef CONFIG_SYSCTL

#define SYSCALL 0x050f
#define NOP2    0x9090

/*
 * NOP out syscall in vsyscall page when not needed.
 */
static int vsyscall_sysctl_change(ctl_table *ctl, int write, struct file * filp,
                        void __user *buffer, size_t *lenp, loff_t *ppos)
{
	extern u16 vsysc1, vsysc2;
	u16 __iomem *map1;
	u16 __iomem *map2;
	int ret = proc_dointvec(ctl, write, filp, buffer, lenp, ppos);
	if (!write)
		return ret;
	/* gcc has some trouble with __va(__pa()), so just do it this
	   way. */
	map1 = ioremap(__pa_vsymbol(&vsysc1), 2);
	if (!map1)
		return -ENOMEM;
	map2 = ioremap(__pa_vsymbol(&vsysc2), 2);
	if (!map2) {
		ret = -ENOMEM;
		goto out;
	}
	if (!vsyscall_gtod_data.sysctl_enabled) {
		writew(SYSCALL, map1);
		writew(SYSCALL, map2);
	} else {
		writew(NOP2, map1);
		writew(NOP2, map2);
	}
	iounmap(map2);
out:
	iounmap(map1);
	return ret;
}

static ctl_table kernel_table2[] = {
	{ .procname = "vsyscall64",
	  .data = &vsyscall_gtod_data.sysctl_enabled, .maxlen = sizeof(int),
	  .mode = 0644,
	  .proc_handler = vsyscall_sysctl_change },
	{}
};

static ctl_table kernel_root_table2[] = {
	{ .ctl_name = CTL_KERN, .procname = "kernel", .mode = 0555,
	  .child = kernel_table2 },
	{}
};

#endif

/* Assume __initcall executes before all user space. Hopefully kmod
   doesn't violate that. We'll find out if it does. */
static void __cpuinit vsyscall_set_cpu(int cpu)
{
	unsigned long *d;
	unsigned long node = 0;
#ifdef CONFIG_NUMA
	node = cpu_to_node(cpu);
#endif
	if (cpu_has(&cpu_data(cpu), X86_FEATURE_RDTSCP))
		write_rdtscp_aux((node << 12) | cpu);

	/* Store cpu number in limit so that it can be loaded quickly
	   in user space in vgetcpu.
	   12 bits for the CPU and 8 bits for the node. */
	d = (unsigned long *)(get_cpu_gdt_table(cpu) + GDT_ENTRY_PER_CPU);
	*d = 0x0f40000000000ULL;
	*d |= cpu;
	*d |= (node & 0xf) << 12;
	*d |= (node >> 4) << 48;
}

static void __cpuinit cpu_vsyscall_init(void *arg)
{
	/* preemption should be already off */
	vsyscall_set_cpu(raw_smp_processor_id());
}

static int __cpuinit
cpu_vsyscall_notifier(struct notifier_block *n, unsigned long action, void *arg)
{
	long cpu = (long)arg;
	if (action == CPU_ONLINE || action == CPU_ONLINE_FROZEN)
		smp_call_function_single(cpu, cpu_vsyscall_init, NULL, 0, 1);
	return NOTIFY_DONE;
}

void __init map_vsyscall(void)
{
	extern char __vsyscall_0;
	unsigned long physaddr_page0 = __pa_symbol(&__vsyscall_0);

	/* Note that VSYSCALL_MAPPED_PAGES must agree with the code below. */
	__set_fixmap(VSYSCALL_FIRST_PAGE, physaddr_page0, PAGE_KERNEL_VSYSCALL);
}

static int __init vsyscall_init(void)
{
	BUG_ON(((unsigned long) &vgettimeofday !=
			VSYSCALL_ADDR(__NR_vgettimeofday)));
	BUG_ON((unsigned long) &vtime != VSYSCALL_ADDR(__NR_vtime));
	BUG_ON((VSYSCALL_ADDR(0) != __fix_to_virt(VSYSCALL_FIRST_PAGE)));
	BUG_ON((unsigned long) &vgetcpu != VSYSCALL_ADDR(__NR_vgetcpu));
#ifdef CONFIG_SYSCTL
	register_sysctl_table(kernel_root_table2);
#endif
	on_each_cpu(cpu_vsyscall_init, NULL, 0, 1);
	hotcpu_notifier(cpu_vsyscall_notifier, 0);
	return 0;
}

__initcall(vsyscall_init);
Commit	Line	Data
1da177e4	1	/*
1da177e4 LT	2	* Copyright (C) 2001 Andrea Arcangeli <andrea@suse.de> SuSE
	3	* Copyright 2003 Andi Kleen, SuSE Labs.
	4	*
	5	* Thanks to hpa@transmeta.com for some useful hint.
	6	* Special thanks to Ingo Molnar for his early experience with
	7	* a different vsyscall implementation for Linux/IA32 and for the name.
	8	*
	9	* vsyscall 1 is located at -10Mbyte, vsyscall 2 is located
	10	* at virtual address -10Mbyte+1024bytes etc... There are at max 4
	11	* vsyscalls. One vsyscall can reserve more than 1 slot to avoid
	12	* jumping out of line if necessary. We cannot add more with this
	13	* mechanism because older kernels won't return -ENOSYS.
	14	* If we want more than four we need a vDSO.
	15	*
	16	* Note: the concept clashes with user mode linux. If you use UML and
	17	* want per guest time just set the kernel.vsyscall64 sysctl to 0.
	18	*/
	19
	20	#include <linux/time.h>
	21	#include <linux/init.h>
	22	#include <linux/kernel.h>
	23	#include <linux/timer.h>
	24	#include <linux/seqlock.h>
	25	#include <linux/jiffies.h>
	26	#include <linux/sysctl.h>
7460ed28	27	#include <linux/clocksource.h>
c08c8205	28	#include <linux/getcpu.h>
8c131af1 AK	29	#include <linux/cpu.h>
	30	#include <linux/smp.h>
	31	#include <linux/notifier.h>
1da177e4 LT	32
	33	#include <asm/vsyscall.h>
	34	#include <asm/pgtable.h>
	35	#include <asm/page.h>
7460ed28	36	#include <asm/unistd.h>
1da177e4 LT	37	#include <asm/fixmap.h>
	38	#include <asm/errno.h>
	39	#include <asm/io.h>
c08c8205 VP	40	#include <asm/segment.h>
	41	#include <asm/desc.h>
	42	#include <asm/topology.h>
2aae950b	43	#include <asm/vgtod.h>
1da177e4 LT	44
1da177e4 LT	45	#define __vsyscall(nr) __attribute__ ((unused,__section__(".vsyscall_" #nr)))
65ea5b03	46	#define __syscall_clobber "r11","cx","memory"
0dbf7028 VG	47	#define __pa_vsymbol(x) \
	48	({unsigned long v; \
	49	extern char __vsyscall_0; \
	50	asm("" : "=r" (v) : "0" (x)); \
957ff882	51	((v - VSYSCALL_START) + __pa_symbol(&__vsyscall_0)); })
1da177e4	52
c8118c6c ED	53	/*
	54	* vsyscall_gtod_data contains data that is :
	55	* - readonly from vsyscalls
676b1855	56	* - written by timer interrupt or systcl (/proc/sys/kernel/vsyscall64)
c8118c6c ED	57	* Try to keep this structure as small as possible to avoid cache line ping pongs
c8118c6c ED	58	*/
c08c8205	59	int __vgetcpu_mode __section_vgetcpu_mode;
1da177e4	60
2aae950b	61	struct vsyscall_gtod_data __vsyscall_gtod_data __section_vsyscall_gtod_data =
1da177e4	62	{
7460ed28	63	.lock = SEQLOCK_UNLOCKED,
	64	.sysctl_enabled = 1,
	65	};
1da177e4	66
2c622148 TB	67	void update_vsyscall_tz(void)
	68	{
	69	unsigned long flags;
	70
	71	write_seqlock_irqsave(&vsyscall_gtod_data.lock, flags);
	72	/* sys_tz has changed */
	73	vsyscall_gtod_data.sys_tz = sys_tz;
	74	write_sequnlock_irqrestore(&vsyscall_gtod_data.lock, flags);
	75	}
	76
7460ed28	77	void update_vsyscall(struct timespec wall_time, struct clocksource clock)
1da177e4	78	{
7460ed28	79	unsigned long flags;
1da177e4	80
7460ed28	81	write_seqlock_irqsave(&vsyscall_gtod_data.lock, flags);
7460ed28	82	/* copy vsyscall data */
c8118c6c ED	83	vsyscall_gtod_data.clock.vread = clock->vread;
	84	vsyscall_gtod_data.clock.cycle_last = clock->cycle_last;
	85	vsyscall_gtod_data.clock.mask = clock->mask;
	86	vsyscall_gtod_data.clock.mult = clock->mult;
	87	vsyscall_gtod_data.clock.shift = clock->shift;
	88	vsyscall_gtod_data.wall_time_sec = wall_time->tv_sec;
	89	vsyscall_gtod_data.wall_time_nsec = wall_time->tv_nsec;
2aae950b	90	vsyscall_gtod_data.wall_to_monotonic = wall_to_monotonic;
7460ed28	91	write_sequnlock_irqrestore(&vsyscall_gtod_data.lock, flags);
1da177e4 LT	92	}
1da177e4 LT	93
7460ed28	94	/* RED-PEN may want to readd seq locking, but then the variable should be
	95	* write-once.
	96	*/
2c8bc944	97	static __always_inline void do_get_tz(struct timezone * tz)
1da177e4	98	{
7460ed28	99	*tz = __vsyscall_gtod_data.sys_tz;
1da177e4 LT	100	}
1da177e4 LT	101
2c8bc944	102	static __always_inline int gettimeofday(struct timeval tv, struct timezone tz)
1da177e4 LT	103	{
	104	int ret;
	105	asm volatile("vsysc2: syscall"
	106	: "=a" (ret)
7460ed28	107	: "0" (__NR_gettimeofday),"D" (tv),"S" (tz)
7460ed28	108	: __syscall_clobber );
1da177e4 LT	109	return ret;
	110	}
	111
2c8bc944	112	static __always_inline long time_syscall(long *t)
1da177e4 LT	113	{
	114	long secs;
	115	asm volatile("vsysc1: syscall"
	116	: "=a" (secs)
	117	: "0" (__NR_time),"D" (t) : __syscall_clobber);
	118	return secs;
	119	}
	120
7460ed28	121	static __always_inline void do_vgettimeofday(struct timeval * tv)
	122	{
	123	cycle_t now, base, mask, cycle_delta;
c8118c6c ED	124	unsigned seq;
c8118c6c ED	125	unsigned long mult, shift, nsec;
7460ed28	126	cycle_t (*vread)(void);
	127	do {
	128	seq = read_seqbegin(&__vsyscall_gtod_data.lock);
	129
	130	vread = __vsyscall_gtod_data.clock.vread;
	131	if (unlikely(!__vsyscall_gtod_data.sysctl_enabled \|\| !vread)) {
89952d13	132	gettimeofday(tv,NULL);
7460ed28	133	return;
	134	}
	135	now = vread();
	136	base = __vsyscall_gtod_data.clock.cycle_last;
	137	mask = __vsyscall_gtod_data.clock.mask;
	138	mult = __vsyscall_gtod_data.clock.mult;
	139	shift = __vsyscall_gtod_data.clock.shift;
	140
c8118c6c ED	141	tv->tv_sec = __vsyscall_gtod_data.wall_time_sec;
c8118c6c ED	142	nsec = __vsyscall_gtod_data.wall_time_nsec;
7460ed28	143	} while (read_seqretry(&__vsyscall_gtod_data.lock, seq));
	144
	145	/* calculate interval: */
	146	cycle_delta = (now - base) & mask;
	147	/* convert to nsecs: */
c8118c6c	148	nsec += (cycle_delta * mult) >> shift;
7460ed28	149
c8118c6c	150	while (nsec >= NSEC_PER_SEC) {
7460ed28	151	tv->tv_sec += 1;
c8118c6c	152	nsec -= NSEC_PER_SEC;
7460ed28	153	}
c8118c6c	154	tv->tv_usec = nsec / NSEC_PER_USEC;
7460ed28	155	}
7460ed28	156
2e8ad43e	157	int __vsyscall(0) vgettimeofday(struct timeval * tv, struct timezone * tz)
1da177e4	158	{
1da177e4 LT	159	if (tv)
	160	do_vgettimeofday(tv);
	161	if (tz)
	162	do_get_tz(tz);
	163	return 0;
	164	}
	165
	166	/* This will break when the xtime seconds get inaccurate, but that is
	167	* unlikely */
2e8ad43e	168	time_t __vsyscall(1) vtime(time_t *t)
1da177e4	169	{
d0aff6e6	170	struct timeval tv;
272a3713	171	time_t result;
7460ed28	172	if (unlikely(!__vsyscall_gtod_data.sysctl_enabled))
1da177e4	173	return time_syscall(t);
d0aff6e6	174
c80544dc	175	vgettimeofday(&tv, NULL);
d0aff6e6	176	result = tv.tv_sec;
272a3713 ED	177	if (t)
	178	*t = result;
	179	return result;
1da177e4 LT	180	}
1da177e4 LT	181
c08c8205 VP	182	/* Fast way to get current CPU and node.
	183	This helps to do per node and per CPU caches in user space.
	184	The result is not guaranteed without CPU affinity, but usually
	185	works out because the scheduler tries to keep a thread on the same
	186	CPU.
	187
	188	tcache must point to a two element sized long array.
	189	All arguments can be NULL. */
	190	long __vsyscall(2)
	191	vgetcpu(unsigned cpu, unsigned node, struct getcpu_cache *tcache)
1da177e4	192	{
8f12dea6	193	unsigned int p;
c08c8205 VP	194	unsigned long j = 0;
	195
	196	/* Fast cache - only recompute value once per jiffies and avoid
	197	relatively costly rdtscp/cpuid otherwise.
	198	This works because the scheduler usually keeps the process
	199	on the same CPU and this syscall doesn't guarantee its
	200	results anyways.
	201	We do this here because otherwise user space would do it on
	202	its own in a likely inferior way (no access to jiffies).
	203	If you don't like it pass NULL. */
34596dc9 AK	204	if (tcache && tcache->blob[0] == (j = __jiffies)) {
34596dc9 AK	205	p = tcache->blob[1];
c08c8205 VP	206	} else if (__vgetcpu_mode == VGETCPU_RDTSCP) {
c08c8205 VP	207	/* Load per CPU data from RDTSCP */
8f12dea6	208	native_read_tscp(&p);
c08c8205 VP	209	} else {
	210	/* Load per CPU data from GDT */
	211	asm("lsl %1,%0" : "=r" (p) : "r" (__PER_CPU_SEG));
	212	}
	213	if (tcache) {
34596dc9 AK	214	tcache->blob[0] = j;
34596dc9 AK	215	tcache->blob[1] = p;
c08c8205 VP	216	}
	217	if (cpu)
	218	*cpu = p & 0xfff;
	219	if (node)
	220	*node = p >> 12;
	221	return 0;
1da177e4 LT	222	}
1da177e4 LT	223
2e8ad43e	224	long __vsyscall(3) venosys_1(void)
1da177e4 LT	225	{
	226	return -ENOSYS;
	227	}
	228
	229	#ifdef CONFIG_SYSCTL
	230
	231	#define SYSCALL 0x050f
	232	#define NOP2 0x9090
	233
	234	/*
	235	* NOP out syscall in vsyscall page when not needed.
	236	*/
	237	static int vsyscall_sysctl_change(ctl_table ctl, int write, struct file filp,
	238	void __user buffer, size_t lenp, loff_t *ppos)
	239	{
	240	extern u16 vsysc1, vsysc2;
131cfd7b AK	241	u16 __iomem *map1;
131cfd7b AK	242	u16 __iomem *map2;
1da177e4 LT	243	int ret = proc_dointvec(ctl, write, filp, buffer, lenp, ppos);
	244	if (!write)
	245	return ret;
	246	/* gcc has some trouble with __va(__pa()), so just do it this
	247	way. */
0dbf7028	248	map1 = ioremap(__pa_vsymbol(&vsysc1), 2);
1da177e4 LT	249	if (!map1)
1da177e4 LT	250	return -ENOMEM;
0dbf7028	251	map2 = ioremap(__pa_vsymbol(&vsysc2), 2);
1da177e4 LT	252	if (!map2) {
	253	ret = -ENOMEM;
	254	goto out;
	255	}
7460ed28	256	if (!vsyscall_gtod_data.sysctl_enabled) {
131cfd7b AK	257	writew(SYSCALL, map1);
131cfd7b AK	258	writew(SYSCALL, map2);
1da177e4	259	} else {
131cfd7b AK	260	writew(NOP2, map1);
131cfd7b AK	261	writew(NOP2, map2);
1da177e4 LT	262	}
	263	iounmap(map2);
	264	out:
	265	iounmap(map1);
	266	return ret;
	267	}
	268
1da177e4	269	static ctl_table kernel_table2[] = {
282a821f	270	{ .procname = "vsyscall64",
7460ed28	271	.data = &vsyscall_gtod_data.sysctl_enabled, .maxlen = sizeof(int),
7460ed28	272	.mode = 0644,
1da177e4	273	.proc_handler = vsyscall_sysctl_change },
7a44d37d	274	{}
1da177e4 LT	275	};
	276
	277	static ctl_table kernel_root_table2[] = {
	278	{ .ctl_name = CTL_KERN, .procname = "kernel", .mode = 0555,
	279	.child = kernel_table2 },
7a44d37d	280	{}
1da177e4 LT	281	};
	282
	283	#endif
	284
8c131af1 AK	285	/* Assume __initcall executes before all user space. Hopefully kmod
	286	doesn't violate that. We'll find out if it does. */
	287	static void __cpuinit vsyscall_set_cpu(int cpu)
c08c8205 VP	288	{
	289	unsigned long *d;
	290	unsigned long node = 0;
	291	#ifdef CONFIG_NUMA
98c9e27a	292	node = cpu_to_node(cpu);
c08c8205	293	#endif
92cb7612	294	if (cpu_has(&cpu_data(cpu), X86_FEATURE_RDTSCP))
8c131af1	295	write_rdtscp_aux((node << 12) \| cpu);
c08c8205 VP	296
	297	/* Store cpu number in limit so that it can be loaded quickly
	298	in user space in vgetcpu.
	299	12 bits for the CPU and 8 bits for the node. */
f6dc247c	300	d = (unsigned long *)(get_cpu_gdt_table(cpu) + GDT_ENTRY_PER_CPU);
c08c8205 VP	301	*d = 0x0f40000000000ULL;
	302	*d \|= cpu;
	303	*d \|= (node & 0xf) << 12;
	304	*d \|= (node >> 4) << 48;
	305	}
	306
8c131af1 AK	307	static void __cpuinit cpu_vsyscall_init(void *arg)
	308	{
	309	/* preemption should be already off */
	310	vsyscall_set_cpu(raw_smp_processor_id());
	311	}
	312
	313	static int __cpuinit
	314	cpu_vsyscall_notifier(struct notifier_block n, unsigned long action, void arg)
	315	{
	316	long cpu = (long)arg;
8bb78442	317	if (action == CPU_ONLINE \|\| action == CPU_ONLINE_FROZEN)
8c131af1 AK	318	smp_call_function_single(cpu, cpu_vsyscall_init, NULL, 0, 1);
	319	return NOTIFY_DONE;
	320	}
	321
e4026440	322	void __init map_vsyscall(void)
1da177e4 LT	323	{
	324	extern char __vsyscall_0;
	325	unsigned long physaddr_page0 = __pa_symbol(&__vsyscall_0);
	326
103efcd9	327	/* Note that VSYSCALL_MAPPED_PAGES must agree with the code below. */
1da177e4 LT	328	__set_fixmap(VSYSCALL_FIRST_PAGE, physaddr_page0, PAGE_KERNEL_VSYSCALL);
	329	}
	330
	331	static int __init vsyscall_init(void)
	332	{
	333	BUG_ON(((unsigned long) &vgettimeofday !=
	334	VSYSCALL_ADDR(__NR_vgettimeofday)));
	335	BUG_ON((unsigned long) &vtime != VSYSCALL_ADDR(__NR_vtime));
	336	BUG_ON((VSYSCALL_ADDR(0) != __fix_to_virt(VSYSCALL_FIRST_PAGE)));
c08c8205	337	BUG_ON((unsigned long) &vgetcpu != VSYSCALL_ADDR(__NR_vgetcpu));
f3c5f5e7	338	#ifdef CONFIG_SYSCTL
0b4d4147	339	register_sysctl_table(kernel_root_table2);
f3c5f5e7	340	#endif
8c131af1 AK	341	on_each_cpu(cpu_vsyscall_init, NULL, 0, 1);
8c131af1 AK	342	hotcpu_notifier(cpu_vsyscall_notifier, 0);
1da177e4 LT	343	return 0;
	344	}
	345
	346	__initcall(vsyscall_init);