[net-next-2.6.git] / arch / x86 / xen / spinlock.c

/*
 * Split spinlock implementation out into its own file, so it can be
 * compiled in a FTRACE-compatible way.
 */
#include <linux/kernel_stat.h>
#include <linux/spinlock.h>
#include <linux/debugfs.h>
#include <linux/log2.h>

#include <asm/paravirt.h>

#include <xen/interface/xen.h>
#include <xen/events.h>

#include "xen-ops.h"
#include "debugfs.h"

#ifdef CONFIG_XEN_DEBUG_FS
static struct xen_spinlock_stats
{
	u64 taken;
	u32 taken_slow;
	u32 taken_slow_nested;
	u32 taken_slow_pickup;
	u32 taken_slow_spurious;
	u32 taken_slow_irqenable;

	u64 released;
	u32 released_slow;
	u32 released_slow_kicked;

#define HISTO_BUCKETS	30
	u32 histo_spin_total[HISTO_BUCKETS+1];
	u32 histo_spin_spinning[HISTO_BUCKETS+1];
	u32 histo_spin_blocked[HISTO_BUCKETS+1];

	u64 time_total;
	u64 time_spinning;
	u64 time_blocked;
} spinlock_stats;

static u8 zero_stats;

static unsigned lock_timeout = 1 << 10;
#define TIMEOUT lock_timeout

static inline void check_zero(void)
{
	if (unlikely(zero_stats)) {
		memset(&spinlock_stats, 0, sizeof(spinlock_stats));
		zero_stats = 0;
	}
}

#define ADD_STATS(elem, val)			\
	do { check_zero(); spinlock_stats.elem += (val); } while(0)

static inline u64 spin_time_start(void)
{
	return xen_clocksource_read();
}

static void __spin_time_accum(u64 delta, u32 *array)
{
	unsigned index = ilog2(delta);

	check_zero();

	if (index < HISTO_BUCKETS)
		array[index]++;
	else
		array[HISTO_BUCKETS]++;
}

static inline void spin_time_accum_spinning(u64 start)
{
	u32 delta = xen_clocksource_read() - start;

	__spin_time_accum(delta, spinlock_stats.histo_spin_spinning);
	spinlock_stats.time_spinning += delta;
}

static inline void spin_time_accum_total(u64 start)
{
	u32 delta = xen_clocksource_read() - start;

	__spin_time_accum(delta, spinlock_stats.histo_spin_total);
	spinlock_stats.time_total += delta;
}

static inline void spin_time_accum_blocked(u64 start)
{
	u32 delta = xen_clocksource_read() - start;

	__spin_time_accum(delta, spinlock_stats.histo_spin_blocked);
	spinlock_stats.time_blocked += delta;
}
#else  /* !CONFIG_XEN_DEBUG_FS */
#define TIMEOUT			(1 << 10)
#define ADD_STATS(elem, val)	do { (void)(val); } while(0)

static inline u64 spin_time_start(void)
{
	return 0;
}

static inline void spin_time_accum_total(u64 start)
{
}
static inline void spin_time_accum_spinning(u64 start)
{
}
static inline void spin_time_accum_blocked(u64 start)
{
}
#endif  /* CONFIG_XEN_DEBUG_FS */

struct xen_spinlock {
	unsigned char lock;		/* 0 -> free; 1 -> locked */
	unsigned short spinners;	/* count of waiting cpus */
};

static int xen_spin_is_locked(struct arch_spinlock *lock)
{
	struct xen_spinlock *xl = (struct xen_spinlock *)lock;

	return xl->lock != 0;
}

static int xen_spin_is_contended(struct arch_spinlock *lock)
{
	struct xen_spinlock *xl = (struct xen_spinlock *)lock;

	/* Not strictly true; this is only the count of contended
	   lock-takers entering the slow path. */
	return xl->spinners != 0;
}

static int xen_spin_trylock(struct arch_spinlock *lock)
{
	struct xen_spinlock *xl = (struct xen_spinlock *)lock;
	u8 old = 1;

	asm("xchgb %b0,%1"
	    : "+q" (old), "+m" (xl->lock) : : "memory");

	return old == 0;
}

static DEFINE_PER_CPU(int, lock_kicker_irq) = -1;
static DEFINE_PER_CPU(struct xen_spinlock *, lock_spinners);

/*
 * Mark a cpu as interested in a lock.  Returns the CPU's previous
 * lock of interest, in case we got preempted by an interrupt.
 */
static inline struct xen_spinlock *spinning_lock(struct xen_spinlock *xl)
{
	struct xen_spinlock *prev;

	prev = __get_cpu_var(lock_spinners);
	__get_cpu_var(lock_spinners) = xl;

	wmb();			/* set lock of interest before count */

	asm(LOCK_PREFIX " incw %0"
	    : "+m" (xl->spinners) : : "memory");

	return prev;
}

/*
 * Mark a cpu as no longer interested in a lock.  Restores previous
 * lock of interest (NULL for none).
 */
static inline void unspinning_lock(struct xen_spinlock *xl, struct xen_spinlock *prev)
{
	asm(LOCK_PREFIX " decw %0"
	    : "+m" (xl->spinners) : : "memory");
	wmb();			/* decrement count before restoring lock */
	__get_cpu_var(lock_spinners) = prev;
}

static noinline int xen_spin_lock_slow(struct arch_spinlock *lock, bool irq_enable)
{
	struct xen_spinlock *xl = (struct xen_spinlock *)lock;
	struct xen_spinlock *prev;
	int irq = __get_cpu_var(lock_kicker_irq);
	int ret;
	u64 start;

	/* If kicker interrupts not initialized yet, just spin */
	if (irq == -1)
		return 0;

	start = spin_time_start();

	/* announce we're spinning */
	prev = spinning_lock(xl);

	ADD_STATS(taken_slow, 1);
	ADD_STATS(taken_slow_nested, prev != NULL);

	do {
		unsigned long flags;

		/* clear pending */
		xen_clear_irq_pending(irq);

		/* check again make sure it didn't become free while
		   we weren't looking  */
		ret = xen_spin_trylock(lock);
		if (ret) {
			ADD_STATS(taken_slow_pickup, 1);

			/*
			 * If we interrupted another spinlock while it
			 * was blocking, make sure it doesn't block
			 * without rechecking the lock.
			 */
			if (prev != NULL)
				xen_set_irq_pending(irq);
			goto out;
		}

		flags = __raw_local_save_flags();
		if (irq_enable) {
			ADD_STATS(taken_slow_irqenable, 1);
			raw_local_irq_enable();
		}

		/*
		 * Block until irq becomes pending.  If we're
		 * interrupted at this point (after the trylock but
		 * before entering the block), then the nested lock
		 * handler guarantees that the irq will be left
		 * pending if there's any chance the lock became free;
		 * xen_poll_irq() returns immediately if the irq is
		 * pending.
		 */
		xen_poll_irq(irq);

		raw_local_irq_restore(flags);

		ADD_STATS(taken_slow_spurious, !xen_test_irq_pending(irq));
	} while (!xen_test_irq_pending(irq)); /* check for spurious wakeups */

	kstat_incr_irqs_this_cpu(irq, irq_to_desc(irq));

out:
	unspinning_lock(xl, prev);
	spin_time_accum_blocked(start);

	return ret;
}

static inline void __xen_spin_lock(struct arch_spinlock *lock, bool irq_enable)
{
	struct xen_spinlock *xl = (struct xen_spinlock *)lock;
	unsigned timeout;
	u8 oldval;
	u64 start_spin;

	ADD_STATS(taken, 1);

	start_spin = spin_time_start();

	do {
		u64 start_spin_fast = spin_time_start();

		timeout = TIMEOUT;

		asm("1: xchgb %1,%0\n"
		    "   testb %1,%1\n"
		    "   jz 3f\n"
		    "2: rep;nop\n"
		    "   cmpb $0,%0\n"
		    "   je 1b\n"
		    "   dec %2\n"
		    "   jnz 2b\n"
		    "3:\n"
		    : "+m" (xl->lock), "=q" (oldval), "+r" (timeout)
		    : "1" (1)
		    : "memory");

		spin_time_accum_spinning(start_spin_fast);

	} while (unlikely(oldval != 0 &&
			  (TIMEOUT == ~0 || !xen_spin_lock_slow(lock, irq_enable))));

	spin_time_accum_total(start_spin);
}

static void xen_spin_lock(struct arch_spinlock *lock)
{
	__xen_spin_lock(lock, false);
}

static void xen_spin_lock_flags(struct arch_spinlock *lock, unsigned long flags)
{
	__xen_spin_lock(lock, !raw_irqs_disabled_flags(flags));
}

static noinline void xen_spin_unlock_slow(struct xen_spinlock *xl)
{
	int cpu;

	ADD_STATS(released_slow, 1);

	for_each_online_cpu(cpu) {
		/* XXX should mix up next cpu selection */
		if (per_cpu(lock_spinners, cpu) == xl) {
			ADD_STATS(released_slow_kicked, 1);
			xen_send_IPI_one(cpu, XEN_SPIN_UNLOCK_VECTOR);
			break;
		}
	}
}

static void xen_spin_unlock(struct arch_spinlock *lock)
{
	struct xen_spinlock *xl = (struct xen_spinlock *)lock;

	ADD_STATS(released, 1);

	smp_wmb();		/* make sure no writes get moved after unlock */
	xl->lock = 0;		/* release lock */

	/*
	 * Make sure unlock happens before checking for waiting
	 * spinners.  We need a strong barrier to enforce the
	 * write-read ordering to different memory locations, as the
	 * CPU makes no implied guarantees about their ordering.
	 */
	mb();

	if (unlikely(xl->spinners))
		xen_spin_unlock_slow(xl);
}

static irqreturn_t dummy_handler(int irq, void *dev_id)
{
	BUG();
	return IRQ_HANDLED;
}

void __cpuinit xen_init_lock_cpu(int cpu)
{
	int irq;
	const char *name;

	name = kasprintf(GFP_KERNEL, "spinlock%d", cpu);
	irq = bind_ipi_to_irqhandler(XEN_SPIN_UNLOCK_VECTOR,
				     cpu,
				     dummy_handler,
				     IRQF_DISABLED|IRQF_PERCPU|IRQF_NOBALANCING,
				     name,
				     NULL);

	if (irq >= 0) {
		disable_irq(irq); /* make sure it's never delivered */
		per_cpu(lock_kicker_irq, cpu) = irq;
	}

	printk("cpu %d spinlock event irq %d\n", cpu, irq);
}

void xen_uninit_lock_cpu(int cpu)
{
	unbind_from_irqhandler(per_cpu(lock_kicker_irq, cpu), NULL);
}

void __init xen_init_spinlocks(void)
{
	pv_lock_ops.spin_is_locked = xen_spin_is_locked;
	pv_lock_ops.spin_is_contended = xen_spin_is_contended;
	pv_lock_ops.spin_lock = xen_spin_lock;
	pv_lock_ops.spin_lock_flags = xen_spin_lock_flags;
	pv_lock_ops.spin_trylock = xen_spin_trylock;
	pv_lock_ops.spin_unlock = xen_spin_unlock;
}

#ifdef CONFIG_XEN_DEBUG_FS

static struct dentry *d_spin_debug;

static int __init xen_spinlock_debugfs(void)
{
	struct dentry *d_xen = xen_init_debugfs();

	if (d_xen == NULL)
		return -ENOMEM;

	d_spin_debug = debugfs_create_dir("spinlocks", d_xen);

	debugfs_create_u8("zero_stats", 0644, d_spin_debug, &zero_stats);

	debugfs_create_u32("timeout", 0644, d_spin_debug, &lock_timeout);

	debugfs_create_u64("taken", 0444, d_spin_debug, &spinlock_stats.taken);
	debugfs_create_u32("taken_slow", 0444, d_spin_debug,
			   &spinlock_stats.taken_slow);
	debugfs_create_u32("taken_slow_nested", 0444, d_spin_debug,
			   &spinlock_stats.taken_slow_nested);
	debugfs_create_u32("taken_slow_pickup", 0444, d_spin_debug,
			   &spinlock_stats.taken_slow_pickup);
	debugfs_create_u32("taken_slow_spurious", 0444, d_spin_debug,
			   &spinlock_stats.taken_slow_spurious);
	debugfs_create_u32("taken_slow_irqenable", 0444, d_spin_debug,
			   &spinlock_stats.taken_slow_irqenable);

	debugfs_create_u64("released", 0444, d_spin_debug, &spinlock_stats.released);
	debugfs_create_u32("released_slow", 0444, d_spin_debug,
			   &spinlock_stats.released_slow);
	debugfs_create_u32("released_slow_kicked", 0444, d_spin_debug,
			   &spinlock_stats.released_slow_kicked);

	debugfs_create_u64("time_spinning", 0444, d_spin_debug,
			   &spinlock_stats.time_spinning);
	debugfs_create_u64("time_blocked", 0444, d_spin_debug,
			   &spinlock_stats.time_blocked);
	debugfs_create_u64("time_total", 0444, d_spin_debug,
			   &spinlock_stats.time_total);

	xen_debugfs_create_u32_array("histo_total", 0444, d_spin_debug,
				     spinlock_stats.histo_spin_total, HISTO_BUCKETS + 1);
	xen_debugfs_create_u32_array("histo_spinning", 0444, d_spin_debug,
				     spinlock_stats.histo_spin_spinning, HISTO_BUCKETS + 1);
	xen_debugfs_create_u32_array("histo_blocked", 0444, d_spin_debug,
				     spinlock_stats.histo_spin_blocked, HISTO_BUCKETS + 1);

	return 0;
}
fs_initcall(xen_spinlock_debugfs);

#endif	/* CONFIG_XEN_DEBUG_FS */
Commit	Line	Data
d5de8841 JF	1	/*
	2	* Split spinlock implementation out into its own file, so it can be
	3	* compiled in a FTRACE-compatible way.
	4	*/
	5	#include <linux/kernel_stat.h>
	6	#include <linux/spinlock.h>
994025ca JF	7	#include <linux/debugfs.h>
994025ca JF	8	#include <linux/log2.h>
d5de8841 JF	9
	10	#include <asm/paravirt.h>
	11
	12	#include <xen/interface/xen.h>
	13	#include <xen/events.h>
	14
	15	#include "xen-ops.h"
994025ca JF	16	#include "debugfs.h"
	17
	18	#ifdef CONFIG_XEN_DEBUG_FS
	19	static struct xen_spinlock_stats
	20	{
	21	u64 taken;
	22	u32 taken_slow;
	23	u32 taken_slow_nested;
	24	u32 taken_slow_pickup;
	25	u32 taken_slow_spurious;
1e696f63	26	u32 taken_slow_irqenable;
994025ca JF	27
	28	u64 released;
	29	u32 released_slow;
	30	u32 released_slow_kicked;
	31
f8eca41f JF	32	#define HISTO_BUCKETS 30
	33	u32 histo_spin_total[HISTO_BUCKETS+1];
	34	u32 histo_spin_spinning[HISTO_BUCKETS+1];
	35	u32 histo_spin_blocked[HISTO_BUCKETS+1];
	36
	37	u64 time_total;
	38	u64 time_spinning;
	39	u64 time_blocked;
994025ca JF	40	} spinlock_stats;
	41
	42	static u8 zero_stats;
	43
	44	static unsigned lock_timeout = 1 << 10;
	45	#define TIMEOUT lock_timeout
	46
	47	static inline void check_zero(void)
	48	{
	49	if (unlikely(zero_stats)) {
	50	memset(&spinlock_stats, 0, sizeof(spinlock_stats));
	51	zero_stats = 0;
	52	}
	53	}
	54
	55	#define ADD_STATS(elem, val) \
	56	do { check_zero(); spinlock_stats.elem += (val); } while(0)
	57
	58	static inline u64 spin_time_start(void)
	59	{
	60	return xen_clocksource_read();
	61	}
	62
	63	static void __spin_time_accum(u64 delta, u32 *array)
	64	{
	65	unsigned index = ilog2(delta);
	66
	67	check_zero();
	68
	69	if (index < HISTO_BUCKETS)
	70	array[index]++;
	71	else
	72	array[HISTO_BUCKETS]++;
	73	}
	74
f8eca41f JF	75	static inline void spin_time_accum_spinning(u64 start)
	76	{
	77	u32 delta = xen_clocksource_read() - start;
	78
	79	__spin_time_accum(delta, spinlock_stats.histo_spin_spinning);
	80	spinlock_stats.time_spinning += delta;
	81	}
	82
	83	static inline void spin_time_accum_total(u64 start)
994025ca JF	84	{
	85	u32 delta = xen_clocksource_read() - start;
	86
f8eca41f JF	87	__spin_time_accum(delta, spinlock_stats.histo_spin_total);
f8eca41f JF	88	spinlock_stats.time_total += delta;
994025ca JF	89	}
994025ca JF	90
f8eca41f	91	static inline void spin_time_accum_blocked(u64 start)
994025ca JF	92	{
	93	u32 delta = xen_clocksource_read() - start;
	94
f8eca41f JF	95	__spin_time_accum(delta, spinlock_stats.histo_spin_blocked);
f8eca41f JF	96	spinlock_stats.time_blocked += delta;
994025ca JF	97	}
	98	#else /* !CONFIG_XEN_DEBUG_FS */
	99	#define TIMEOUT (1 << 10)
	100	#define ADD_STATS(elem, val) do { (void)(val); } while(0)
	101
	102	static inline u64 spin_time_start(void)
	103	{
	104	return 0;
	105	}
	106
f8eca41f JF	107	static inline void spin_time_accum_total(u64 start)
	108	{
	109	}
	110	static inline void spin_time_accum_spinning(u64 start)
994025ca JF	111	{
994025ca JF	112	}
f8eca41f	113	static inline void spin_time_accum_blocked(u64 start)
994025ca JF	114	{
	115	}
	116	#endif /* CONFIG_XEN_DEBUG_FS */
d5de8841 JF	117
	118	struct xen_spinlock {
	119	unsigned char lock; /* 0 -> free; 1 -> locked */
	120	unsigned short spinners; /* count of waiting cpus */
	121	};
	122
445c8951	123	static int xen_spin_is_locked(struct arch_spinlock *lock)
d5de8841 JF	124	{
	125	struct xen_spinlock xl = (struct xen_spinlock )lock;
	126
	127	return xl->lock != 0;
	128	}
	129
445c8951	130	static int xen_spin_is_contended(struct arch_spinlock *lock)
d5de8841 JF	131	{
	132	struct xen_spinlock xl = (struct xen_spinlock )lock;
	133
	134	/* Not strictly true; this is only the count of contended
	135	lock-takers entering the slow path. */
	136	return xl->spinners != 0;
	137	}
	138
445c8951	139	static int xen_spin_trylock(struct arch_spinlock *lock)
d5de8841 JF	140	{
	141	struct xen_spinlock xl = (struct xen_spinlock )lock;
	142	u8 old = 1;
	143
	144	asm("xchgb %b0,%1"
	145	: "+q" (old), "+m" (xl->lock) : : "memory");
	146
	147	return old == 0;
	148	}
	149
	150	static DEFINE_PER_CPU(int, lock_kicker_irq) = -1;
	151	static DEFINE_PER_CPU(struct xen_spinlock *, lock_spinners);
	152
168d2f46 JF	153	/*
	154	* Mark a cpu as interested in a lock. Returns the CPU's previous
	155	* lock of interest, in case we got preempted by an interrupt.
	156	*/
	157	static inline struct xen_spinlock spinning_lock(struct xen_spinlock xl)
d5de8841	158	{
168d2f46 JF	159	struct xen_spinlock *prev;
	160
	161	prev = __get_cpu_var(lock_spinners);
d5de8841	162	__get_cpu_var(lock_spinners) = xl;
168d2f46	163
d5de8841	164	wmb(); /* set lock of interest before count */
168d2f46	165
d5de8841 JF	166	asm(LOCK_PREFIX " incw %0"
d5de8841 JF	167	: "+m" (xl->spinners) : : "memory");
168d2f46 JF	168
168d2f46 JF	169	return prev;
d5de8841 JF	170	}
d5de8841 JF	171
168d2f46 JF	172	/*
	173	* Mark a cpu as no longer interested in a lock. Restores previous
	174	* lock of interest (NULL for none).
	175	*/
	176	static inline void unspinning_lock(struct xen_spinlock xl, struct xen_spinlock prev)
d5de8841 JF	177	{
	178	asm(LOCK_PREFIX " decw %0"
	179	: "+m" (xl->spinners) : : "memory");
168d2f46 JF	180	wmb(); /* decrement count before restoring lock */
168d2f46 JF	181	__get_cpu_var(lock_spinners) = prev;
d5de8841 JF	182	}
d5de8841 JF	183
445c8951	184	static noinline int xen_spin_lock_slow(struct arch_spinlock *lock, bool irq_enable)
d5de8841 JF	185	{
d5de8841 JF	186	struct xen_spinlock xl = (struct xen_spinlock )lock;
168d2f46	187	struct xen_spinlock *prev;
d5de8841 JF	188	int irq = __get_cpu_var(lock_kicker_irq);
d5de8841 JF	189	int ret;
f8eca41f	190	u64 start;
d5de8841 JF	191
	192	/* If kicker interrupts not initialized yet, just spin */
	193	if (irq == -1)
	194	return 0;
	195
f8eca41f JF	196	start = spin_time_start();
f8eca41f JF	197
d5de8841	198	/* announce we're spinning */
168d2f46	199	prev = spinning_lock(xl);
d5de8841	200
994025ca JF	201	ADD_STATS(taken_slow, 1);
	202	ADD_STATS(taken_slow_nested, prev != NULL);
	203
168d2f46	204	do {
4d576b57 JF	205	unsigned long flags;
4d576b57 JF	206
168d2f46 JF	207	/* clear pending */
	208	xen_clear_irq_pending(irq);
	209
	210	/* check again make sure it didn't become free while
	211	we weren't looking */
	212	ret = xen_spin_trylock(lock);
	213	if (ret) {
994025ca JF	214	ADD_STATS(taken_slow_pickup, 1);
994025ca JF	215
168d2f46 JF	216	/*
	217	* If we interrupted another spinlock while it
	218	* was blocking, make sure it doesn't block
	219	* without rechecking the lock.
	220	*/
	221	if (prev != NULL)
	222	xen_set_irq_pending(irq);
	223	goto out;
	224	}
d5de8841	225
4d576b57 JF	226	flags = __raw_local_save_flags();
	227	if (irq_enable) {
	228	ADD_STATS(taken_slow_irqenable, 1);
	229	raw_local_irq_enable();
	230	}
	231
168d2f46 JF	232	/*
	233	* Block until irq becomes pending. If we're
	234	* interrupted at this point (after the trylock but
	235	* before entering the block), then the nested lock
	236	* handler guarantees that the irq will be left
	237	* pending if there's any chance the lock became free;
	238	* xen_poll_irq() returns immediately if the irq is
	239	* pending.
	240	*/
	241	xen_poll_irq(irq);
4d576b57 JF	242
	243	raw_local_irq_restore(flags);
	244
994025ca	245	ADD_STATS(taken_slow_spurious, !xen_test_irq_pending(irq));
168d2f46	246	} while (!xen_test_irq_pending(irq)); /* check for spurious wakeups */
d5de8841	247
d6c88a50	248	kstat_incr_irqs_this_cpu(irq, irq_to_desc(irq));
d5de8841 JF	249
d5de8841 JF	250	out:
168d2f46	251	unspinning_lock(xl, prev);
f8eca41f JF	252	spin_time_accum_blocked(start);
f8eca41f JF	253
d5de8841 JF	254	return ret;
	255	}
	256
445c8951	257	static inline void __xen_spin_lock(struct arch_spinlock *lock, bool irq_enable)
d5de8841 JF	258	{
d5de8841 JF	259	struct xen_spinlock xl = (struct xen_spinlock )lock;
994025ca	260	unsigned timeout;
d5de8841	261	u8 oldval;
994025ca JF	262	u64 start_spin;
	263
	264	ADD_STATS(taken, 1);
	265
	266	start_spin = spin_time_start();
d5de8841 JF	267
d5de8841 JF	268	do {
994025ca JF	269	u64 start_spin_fast = spin_time_start();
	270
	271	timeout = TIMEOUT;
d5de8841 JF	272
	273	asm("1: xchgb %1,%0\n"
	274	" testb %1,%1\n"
	275	" jz 3f\n"
	276	"2: rep;nop\n"
	277	" cmpb $0,%0\n"
	278	" je 1b\n"
	279	" dec %2\n"
	280	" jnz 2b\n"
	281	"3:\n"
	282	: "+m" (xl->lock), "=q" (oldval), "+r" (timeout)
	283	: "1" (1)
	284	: "memory");
	285
f8eca41f	286	spin_time_accum_spinning(start_spin_fast);
1e696f63 JF	287
	288	} while (unlikely(oldval != 0 &&
	289	(TIMEOUT == ~0 \|\| !xen_spin_lock_slow(lock, irq_enable))));
994025ca	290
f8eca41f	291	spin_time_accum_total(start_spin);
d5de8841 JF	292	}
d5de8841 JF	293
445c8951	294	static void xen_spin_lock(struct arch_spinlock *lock)
1e696f63 JF	295	{
	296	__xen_spin_lock(lock, false);
	297	}
	298
445c8951	299	static void xen_spin_lock_flags(struct arch_spinlock *lock, unsigned long flags)
1e696f63 JF	300	{
	301	__xen_spin_lock(lock, !raw_irqs_disabled_flags(flags));
	302	}
	303
d5de8841 JF	304	static noinline void xen_spin_unlock_slow(struct xen_spinlock *xl)
	305	{
	306	int cpu;
	307
994025ca JF	308	ADD_STATS(released_slow, 1);
994025ca JF	309
d5de8841 JF	310	for_each_online_cpu(cpu) {
	311	/* XXX should mix up next cpu selection */
	312	if (per_cpu(lock_spinners, cpu) == xl) {
994025ca	313	ADD_STATS(released_slow_kicked, 1);
d5de8841 JF	314	xen_send_IPI_one(cpu, XEN_SPIN_UNLOCK_VECTOR);
	315	break;
	316	}
	317	}
	318	}
	319
445c8951	320	static void xen_spin_unlock(struct arch_spinlock *lock)
d5de8841 JF	321	{
	322	struct xen_spinlock xl = (struct xen_spinlock )lock;
	323
994025ca JF	324	ADD_STATS(released, 1);
994025ca JF	325
d5de8841 JF	326	smp_wmb(); /* make sure no writes get moved after unlock */
	327	xl->lock = 0; /* release lock */
	328
2496afbf YX	329	/*
	330	* Make sure unlock happens before checking for waiting
	331	* spinners. We need a strong barrier to enforce the
	332	* write-read ordering to different memory locations, as the
	333	* CPU makes no implied guarantees about their ordering.
	334	*/
	335	mb();
d5de8841 JF	336
	337	if (unlikely(xl->spinners))
	338	xen_spin_unlock_slow(xl);
	339	}
	340
	341	static irqreturn_t dummy_handler(int irq, void *dev_id)
	342	{
	343	BUG();
	344	return IRQ_HANDLED;
	345	}
	346
	347	void __cpuinit xen_init_lock_cpu(int cpu)
	348	{
	349	int irq;
	350	const char *name;
	351
	352	name = kasprintf(GFP_KERNEL, "spinlock%d", cpu);
	353	irq = bind_ipi_to_irqhandler(XEN_SPIN_UNLOCK_VECTOR,
	354	cpu,
	355	dummy_handler,
	356	IRQF_DISABLED\|IRQF_PERCPU\|IRQF_NOBALANCING,
	357	name,
	358	NULL);
	359
	360	if (irq >= 0) {
	361	disable_irq(irq); /* make sure it's never delivered */
	362	per_cpu(lock_kicker_irq, cpu) = irq;
	363	}
	364
	365	printk("cpu %d spinlock event irq %d\n", cpu, irq);
	366	}
	367
d68d82af AN	368	void xen_uninit_lock_cpu(int cpu)
	369	{
	370	unbind_from_irqhandler(per_cpu(lock_kicker_irq, cpu), NULL);
	371	}
	372
d5de8841 JF	373	void __init xen_init_spinlocks(void)
	374	{
	375	pv_lock_ops.spin_is_locked = xen_spin_is_locked;
	376	pv_lock_ops.spin_is_contended = xen_spin_is_contended;
	377	pv_lock_ops.spin_lock = xen_spin_lock;
1e696f63	378	pv_lock_ops.spin_lock_flags = xen_spin_lock_flags;
d5de8841 JF	379	pv_lock_ops.spin_trylock = xen_spin_trylock;
	380	pv_lock_ops.spin_unlock = xen_spin_unlock;
	381	}
994025ca JF	382
	383	#ifdef CONFIG_XEN_DEBUG_FS
	384
	385	static struct dentry *d_spin_debug;
	386
	387	static int __init xen_spinlock_debugfs(void)
	388	{
	389	struct dentry *d_xen = xen_init_debugfs();
	390
	391	if (d_xen == NULL)
	392	return -ENOMEM;
	393
	394	d_spin_debug = debugfs_create_dir("spinlocks", d_xen);
	395
	396	debugfs_create_u8("zero_stats", 0644, d_spin_debug, &zero_stats);
	397
	398	debugfs_create_u32("timeout", 0644, d_spin_debug, &lock_timeout);
	399
	400	debugfs_create_u64("taken", 0444, d_spin_debug, &spinlock_stats.taken);
	401	debugfs_create_u32("taken_slow", 0444, d_spin_debug,
	402	&spinlock_stats.taken_slow);
	403	debugfs_create_u32("taken_slow_nested", 0444, d_spin_debug,
	404	&spinlock_stats.taken_slow_nested);
	405	debugfs_create_u32("taken_slow_pickup", 0444, d_spin_debug,
	406	&spinlock_stats.taken_slow_pickup);
	407	debugfs_create_u32("taken_slow_spurious", 0444, d_spin_debug,
	408	&spinlock_stats.taken_slow_spurious);
1e696f63 JF	409	debugfs_create_u32("taken_slow_irqenable", 0444, d_spin_debug,
1e696f63 JF	410	&spinlock_stats.taken_slow_irqenable);
994025ca JF	411
	412	debugfs_create_u64("released", 0444, d_spin_debug, &spinlock_stats.released);
	413	debugfs_create_u32("released_slow", 0444, d_spin_debug,
	414	&spinlock_stats.released_slow);
	415	debugfs_create_u32("released_slow_kicked", 0444, d_spin_debug,
	416	&spinlock_stats.released_slow_kicked);
	417
	418	debugfs_create_u64("time_spinning", 0444, d_spin_debug,
f8eca41f JF	419	&spinlock_stats.time_spinning);
	420	debugfs_create_u64("time_blocked", 0444, d_spin_debug,
	421	&spinlock_stats.time_blocked);
994025ca	422	debugfs_create_u64("time_total", 0444, d_spin_debug,
f8eca41f	423	&spinlock_stats.time_total);
994025ca JF	424
994025ca JF	425	xen_debugfs_create_u32_array("histo_total", 0444, d_spin_debug,
f8eca41f	426	spinlock_stats.histo_spin_total, HISTO_BUCKETS + 1);
994025ca	427	xen_debugfs_create_u32_array("histo_spinning", 0444, d_spin_debug,
f8eca41f JF	428	spinlock_stats.histo_spin_spinning, HISTO_BUCKETS + 1);
	429	xen_debugfs_create_u32_array("histo_blocked", 0444, d_spin_debug,
	430	spinlock_stats.histo_spin_blocked, HISTO_BUCKETS + 1);
994025ca JF	431
	432	return 0;
	433	}
	434	fs_initcall(xen_spinlock_debugfs);
	435
	436	#endif /* CONFIG_XEN_DEBUG_FS */