[PATCH 00/12] clocksource/timekeeping cleanup

[PATCH 00/12] clocksource/timekeeping cleanup

[Daniel Walker]

This patchset represents the most stable clocksource changes in my tree. Also
John (and others) have reviewed these changes a few times. I think it's all 
acceptable.
        
Daniel
        
-- 

[PATCH 01/12] timekeeping: create kernel/time/timekeeping.c

[Daniel Walker]

[-- Attachment #1: timekeeping_move_out_of_timer_c.patch --]
[-- Type: text/plain, Size: 38400 bytes --]

Move the generic timekeeping code from kernel/timer.c to
kernel/time/timekeeping.c . This requires some glue code which is
added to the include/linux/timekeeping.h header.

Signed-Off-By: Daniel Walker <dwalker@mvista.com> 

---
 include/linux/clocksource.h |    3 
 include/linux/timekeeping.h |   19 +
 kernel/time/Makefile        |    2 
 kernel/time/clocksource.c   |    3 
 kernel/time/timekeeping.c   |  639 ++++++++++++++++++++++++++++++++++++++++++++
 kernel/timer.c              |  630 -------------------------------------------
 6 files changed, 663 insertions(+), 633 deletions(-)

Index: linux-2.6.19/include/linux/clocksource.h
===================================================================
--- linux-2.6.19.orig/include/linux/clocksource.h
+++ linux-2.6.19/include/linux/clocksource.h
@@ -18,6 +18,9 @@
 /* clocksource cycle base type */
 typedef u64 cycle_t;
 
+/* XXX - Would like a better way for initializing curr_clocksource */
+extern struct clocksource clocksource_jiffies;
+
 /**
  * struct clocksource - hardware abstraction for a free running counter
  *	Provides mostly state-free accessors to the underlying hardware.
Index: linux-2.6.19/include/linux/timekeeping.h
===================================================================
--- /dev/null
+++ linux-2.6.19/include/linux/timekeeping.h
@@ -0,0 +1,19 @@
+#ifndef _LINUX_TIMEKEEPING_H
+#define _LINUX_TIMEKEEPING_H
+
+#include <linux/clocksource.h>
+
+extern void update_wall_time(void);
+
+#ifdef CONFIG_GENERIC_TIME
+
+extern struct clocksource *clock;
+
+#else /* CONFIG_GENERIC_TIME */
+static inline int change_clocksource(void)
+{
+	return 0;
+}
+#endif /* !CONFIG_GENERIC_TIME */
+
+#endif /* _LINUX_TIMEKEEPING_H */
Index: linux-2.6.19/kernel/time/Makefile
===================================================================
--- linux-2.6.19.orig/kernel/time/Makefile
+++ linux-2.6.19/kernel/time/Makefile
@@ -1,4 +1,4 @@
-obj-y += ntp.o clocksource.o jiffies.o timer_list.o
+obj-y += ntp.o clocksource.o jiffies.o timer_list.o timekeeping.o
 
 obj-$(CONFIG_GENERIC_CLOCKEVENTS)	+= clockevents.o
 obj-$(CONFIG_TIMER_STATS)		+= timer_stats.o
Index: linux-2.6.19/kernel/time/clocksource.c
===================================================================
--- linux-2.6.19.orig/kernel/time/clocksource.c
+++ linux-2.6.19/kernel/time/clocksource.c
@@ -29,9 +29,6 @@
 #include <linux/init.h>
 #include <linux/module.h>
 
-/* XXX - Would like a better way for initializing curr_clocksource */
-extern struct clocksource clocksource_jiffies;
-
 /*[Clocksource internal variables]---------
  * curr_clocksource:
  *	currently selected clocksource. Initialized to clocksource_jiffies.
Index: linux-2.6.19/kernel/time/timekeeping.c
===================================================================
--- /dev/null
+++ linux-2.6.19/kernel/time/timekeeping.c
@@ -0,0 +1,639 @@
+
+
+#include <linux/module.h>
+#include <linux/timekeeping.h>
+#include <linux/time.h>
+#include <linux/hrtimer.h>
+
+/*
+ * flag for if timekeeping is suspended
+ */
+static int timekeeping_suspended;
+
+/*
+ * time in seconds when suspend began
+ */
+static unsigned long timekeeping_suspend_time;
+
+/*
+ * Clock used for timekeeping
+ */
+struct clocksource *clock = &clocksource_jiffies;
+
+/*
+ * The current time
+ * wall_to_monotonic is what we need to add to xtime (or xtime corrected
+ * for sub jiffie times) to get to monotonic time.  Monotonic is pegged
+ * at zero at system boot time, so wall_to_monotonic will be negative,
+ * however, we will ALWAYS keep the tv_nsec part positive so we can use
+ * the usual normalization.
+ */
+struct timespec xtime __attribute__ ((aligned (16)));
+struct timespec wall_to_monotonic __attribute__ ((aligned (16)));
+
+EXPORT_SYMBOL(xtime);
+
+#ifdef CONFIG_GENERIC_TIME
+/**
+ * __get_nsec_offset - Returns nanoseconds since last call to periodic_hook
+ *
+ * private function, must hold xtime_lock lock when being
+ * called. Returns the number of nanoseconds since the
+ * last call to update_wall_time() (adjusted by NTP scaling)
+ */
+static inline s64 __get_nsec_offset(void)
+{
+	cycle_t cycle_now, cycle_delta;
+	s64 ns_offset;
+
+	/* read clocksource: */
+	cycle_now = clocksource_read(clock);
+
+	/* calculate the delta since the last update_wall_time: */
+	cycle_delta = (cycle_now - clock->cycle_last) & clock->mask;
+
+	/* convert to nanoseconds: */
+	ns_offset = cyc2ns(clock, cycle_delta);
+
+	return ns_offset;
+}
+
+/**
+ * __get_realtime_clock_ts - Returns the time of day in a timespec
+ * @ts:		pointer to the timespec to be set
+ *
+ * Returns the time of day in a timespec. Used by
+ * do_gettimeofday() and get_realtime_clock_ts().
+ */
+static inline void __get_realtime_clock_ts(struct timespec *ts)
+{
+	unsigned long seq;
+	s64 nsecs;
+
+	do {
+		seq = read_seqbegin(&xtime_lock);
+
+		*ts = xtime;
+		nsecs = __get_nsec_offset();
+
+	} while (read_seqretry(&xtime_lock, seq));
+
+	timespec_add_ns(ts, nsecs);
+}
+
+/**
+ * getnstimeofday - Returns the time of day in a timespec
+ * @ts:		pointer to the timespec to be set
+ *
+ * Returns the time of day in a timespec.
+ */
+void getnstimeofday(struct timespec *ts)
+{
+	__get_realtime_clock_ts(ts);
+}
+
+EXPORT_SYMBOL(getnstimeofday);
+
+/**
+ * do_gettimeofday - Returns the time of day in a timeval
+ * @tv:		pointer to the timeval to be set
+ *
+ * NOTE: Users should be converted to using get_realtime_clock_ts()
+ */
+void do_gettimeofday(struct timeval *tv)
+{
+	struct timespec now;
+
+	__get_realtime_clock_ts(&now);
+	tv->tv_sec = now.tv_sec;
+	tv->tv_usec = now.tv_nsec/1000;
+}
+
+EXPORT_SYMBOL(do_gettimeofday);
+/**
+ * do_settimeofday - Sets the time of day
+ * @tv:		pointer to the timespec variable containing the new time
+ *
+ * Sets the time of day to the new time and update NTP and notify hrtimers
+ */
+int do_settimeofday(struct timespec *tv)
+{
+	unsigned long flags;
+	time_t wtm_sec, sec = tv->tv_sec;
+	long wtm_nsec, nsec = tv->tv_nsec;
+
+	if ((unsigned long)tv->tv_nsec >= NSEC_PER_SEC)
+		return -EINVAL;
+
+	write_seqlock_irqsave(&xtime_lock, flags);
+
+	nsec -= __get_nsec_offset();
+
+	wtm_sec  = wall_to_monotonic.tv_sec + (xtime.tv_sec - sec);
+	wtm_nsec = wall_to_monotonic.tv_nsec + (xtime.tv_nsec - nsec);
+
+	set_normalized_timespec(&xtime, sec, nsec);
+	set_normalized_timespec(&wall_to_monotonic, wtm_sec, wtm_nsec);
+
+	clock->error = 0;
+	ntp_clear();
+
+	write_sequnlock_irqrestore(&xtime_lock, flags);
+
+	/* signal hrtimers about time change */
+	clock_was_set();
+
+	return 0;
+}
+
+EXPORT_SYMBOL(do_settimeofday);
+
+/**
+ * change_clocksource - Swaps clocksources if a new one is available
+ *
+ * Accumulates current time interval and initializes new clocksource
+ */
+static int change_clocksource(void)
+{
+	struct clocksource *new;
+	cycle_t now;
+	u64 nsec;
+	new = clocksource_get_next();
+	if (clock != new) {
+		now = clocksource_read(new);
+		nsec =  __get_nsec_offset();
+		timespec_add_ns(&xtime, nsec);
+
+		clock = new;
+		clock->cycle_last = now;
+		printk(KERN_INFO "Time: %s clocksource has been installed.\n",
+		       clock->name);
+		return 1;
+	} else if (clock->update_callback) {
+		return clock->update_callback();
+	}
+	return 0;
+}
+
+/**
+ * timeofday_is_continuous - check to see if timekeeping is free running
+ */
+int timekeeping_is_continuous(void)
+{
+	unsigned long seq;
+	int ret;
+
+	do {
+		seq = read_seqbegin(&xtime_lock);
+
+		ret = clock->is_continuous;
+
+	} while (read_seqretry(&xtime_lock, seq));
+
+	return ret;
+}
+#endif /* CONFIG_GENERIC_TIME */
+
+/**
+ * read_persistent_clock -  Return time in seconds from the persistent clock.
+ *
+ * Weak dummy function for arches that do not yet support it.
+ * Returns seconds from epoch using the battery backed persistent clock.
+ * Returns zero if unsupported.
+ *
+ *  XXX - Do be sure to remove it once all arches implement it.
+ */
+unsigned long __attribute__((weak)) read_persistent_clock(void)
+{
+	return 0;
+}
+
+/*
+ * timekeeping_init - Initializes the clocksource and common timekeeping values
+ */
+void __init timekeeping_init(void)
+{
+	unsigned long flags;
+	unsigned long sec = read_persistent_clock();
+
+	write_seqlock_irqsave(&xtime_lock, flags);
+
+	ntp_clear();
+
+	clock = clocksource_get_next();
+	clocksource_calculate_interval(clock, tick_nsec);
+	clock->cycle_last = clocksource_read(clock);
+
+	xtime.tv_sec = sec;
+	xtime.tv_nsec = 0;
+	set_normalized_timespec(&wall_to_monotonic,
+		-xtime.tv_sec, -xtime.tv_nsec);
+
+	write_sequnlock_irqrestore(&xtime_lock, flags);
+}
+
+
+/* flag for if timekeeping is suspended */
+static int timekeeping_suspended;
+/* time in seconds when suspend began */
+static unsigned long timekeeping_suspend_time;
+
+/**
+ * timekeeping_resume - Resumes the generic timekeeping subsystem.
+ * @dev:	unused
+ *
+ * This is for the generic clocksource timekeeping.
+ * xtime/wall_to_monotonic/jiffies/etc are
+ * still managed by arch specific suspend/resume code.
+ */
+static int timekeeping_resume(struct sys_device *dev)
+{
+	unsigned long flags;
+	unsigned long now = read_persistent_clock();
+
+	write_seqlock_irqsave(&xtime_lock, flags);
+
+	if (now && (now > timekeeping_suspend_time)) {
+		unsigned long sleep_length = now - timekeeping_suspend_time;
+		xtime.tv_sec += sleep_length;
+		jiffies_64 += (u64)sleep_length * HZ;
+	}
+	/* re-base the last cycle value */
+	clock->cycle_last = clocksource_read(clock);
+	clock->error = 0;
+	timekeeping_suspended = 0;
+	write_sequnlock_irqrestore(&xtime_lock, flags);
+
+	hrtimer_notify_resume();
+
+	return 0;
+}
+
+static int timekeeping_suspend(struct sys_device *dev, pm_message_t state)
+{
+	unsigned long flags;
+
+	write_seqlock_irqsave(&xtime_lock, flags);
+	timekeeping_suspended = 1;
+	timekeeping_suspend_time = read_persistent_clock();
+	write_sequnlock_irqrestore(&xtime_lock, flags);
+	return 0;
+}
+
+/* sysfs resume/suspend bits for timekeeping */
+static struct sysdev_class timekeeping_sysclass = {
+	.resume		= timekeeping_resume,
+	.suspend	= timekeeping_suspend,
+	set_kset_name("timekeeping"),
+};
+
+static struct sys_device device_timer = {
+	.id		= 0,
+	.cls		= &timekeeping_sysclass,
+};
+
+static int __init timekeeping_init_device(void)
+{
+	int error = sysdev_class_register(&timekeeping_sysclass);
+	if (!error)
+		error = sysdev_register(&device_timer);
+	return error;
+}
+
+device_initcall(timekeeping_init_device);
+
+/*
+ * If the error is already larger, we look ahead even further
+ * to compensate for late or lost adjustments.
+ */
+static __always_inline int clocksource_bigadjust(s64 error, s64 *interval,
+						 s64 *offset)
+{
+	s64 tick_error, i;
+	u32 look_ahead, adj;
+	s32 error2, mult;
+
+	/*
+	 * Use the current error value to determine how much to look ahead.
+	 * The larger the error the slower we adjust for it to avoid problems
+	 * with losing too many ticks, otherwise we would overadjust and
+	 * produce an even larger error.  The smaller the adjustment the
+	 * faster we try to adjust for it, as lost ticks can do less harm
+	 * here.  This is tuned so that an error of about 1 msec is adusted
+	 * within about 1 sec (or 2^20 nsec in 2^SHIFT_HZ ticks).
+	 */
+	error2 = clock->error >> (TICK_LENGTH_SHIFT + 22 - 2 * SHIFT_HZ);
+	error2 = abs(error2);
+	for (look_ahead = 0; error2 > 0; look_ahead++)
+		error2 >>= 2;
+
+	/*
+	 * Now calculate the error in (1 << look_ahead) ticks, but first
+	 * remove the single look ahead already included in the error.
+	 */
+	tick_error = current_tick_length() >>
+		(TICK_LENGTH_SHIFT - clock->shift + 1);
+	tick_error -= clock->xtime_interval >> 1;
+	error = ((error - tick_error) >> look_ahead) + tick_error;
+
+	/* Finally calculate the adjustment shift value.  */
+	i = *interval;
+	mult = 1;
+	if (error < 0) {
+		error = -error;
+		*interval = -*interval;
+		*offset = -*offset;
+		mult = -1;
+	}
+	for (adj = 0; error > i; adj++)
+		error >>= 1;
+
+	*interval <<= adj;
+	*offset <<= adj;
+	return mult << adj;
+}
+
+/*
+ * Adjust the multiplier to reduce the error value,
+ * this is optimized for the most common adjustments of -1,0,1,
+ * for other values we can do a bit more work.
+ */
+static void clocksource_adjust(struct clocksource *clock, s64 offset)
+{
+	s64 error, interval = clock->cycle_interval;
+	int adj;
+
+	error = clock->error >> (TICK_LENGTH_SHIFT - clock->shift - 1);
+	if (error > interval) {
+		error >>= 2;
+		if (likely(error <= interval))
+			adj = 1;
+		else
+			adj = clocksource_bigadjust(error, &interval, &offset);
+	} else if (error < -interval) {
+		error >>= 2;
+		if (likely(error >= -interval)) {
+			adj = -1;
+			interval = -interval;
+			offset = -offset;
+		} else
+			adj = clocksource_bigadjust(error, &interval, &offset);
+	} else
+		return;
+
+	clock->mult += adj;
+	clock->xtime_interval += interval;
+	clock->xtime_nsec -= offset;
+	clock->error -= (interval - offset) <<
+			(TICK_LENGTH_SHIFT - clock->shift);
+}
+
+/**
+ * update_wall_time - Uses the current clocksource to increment the wall time
+ *
+ * Called from the timer interrupt, must hold a write on xtime_lock.
+ */
+void update_wall_time(void)
+{
+	cycle_t offset;
+
+	/* Make sure we're fully resumed: */
+	if (unlikely(timekeeping_suspended))
+		return;
+
+#ifdef CONFIG_GENERIC_TIME
+	offset = (clocksource_read(clock) - clock->cycle_last) & clock->mask;
+#else
+	offset = clock->cycle_interval;
+#endif
+	clock->xtime_nsec += (s64)xtime.tv_nsec << clock->shift;
+
+	/* normally this loop will run just once, however in the
+	 * case of lost or late ticks, it will accumulate correctly.
+	 */
+	while (offset >= clock->cycle_interval) {
+		/* accumulate one interval */
+		clock->xtime_nsec += clock->xtime_interval;
+		clock->cycle_last += clock->cycle_interval;
+		offset -= clock->cycle_interval;
+
+		if (clock->xtime_nsec >= (u64)NSEC_PER_SEC << clock->shift) {
+			clock->xtime_nsec -= (u64)NSEC_PER_SEC << clock->shift;
+			xtime.tv_sec++;
+			second_overflow();
+		}
+
+		/* interpolator bits */
+		time_interpolator_update(clock->xtime_interval
+						>> clock->shift);
+
+		/* accumulate error between NTP and clock interval */
+		clock->error += current_tick_length();
+		clock->error -= clock->xtime_interval << (TICK_LENGTH_SHIFT - clock->shift);
+	}
+
+	/* correct the clock when NTP error is too big */
+	clocksource_adjust(clock, offset);
+
+	/* store full nanoseconds into xtime */
+	xtime.tv_nsec = (s64)clock->xtime_nsec >> clock->shift;
+	clock->xtime_nsec -= (s64)xtime.tv_nsec << clock->shift;
+
+	/* check to see if there is a new clocksource to use */
+	if (change_clocksource()) {
+		clock->error = 0;
+		clock->xtime_nsec = 0;
+		hrtimer_clock_notify();
+		clocksource_calculate_interval(clock, tick_nsec);
+	}
+	update_vsyscall(&xtime, clock);
+}
+
+#ifdef CONFIG_TIME_INTERPOLATION
+
+struct time_interpolator *time_interpolator __read_mostly;
+static struct time_interpolator *time_interpolator_list __read_mostly;
+static DEFINE_SPINLOCK(time_interpolator_lock);
+
+static inline cycles_t time_interpolator_get_cycles(unsigned int src)
+{
+	unsigned long (*x)(void);
+
+	switch (src)
+	{
+		case TIME_SOURCE_FUNCTION:
+			x = time_interpolator->addr;
+			return x();
+
+		case TIME_SOURCE_MMIO64	:
+			return readq_relaxed((void __iomem *)time_interpolator->addr);
+
+		case TIME_SOURCE_MMIO32	:
+			return readl_relaxed((void __iomem *)time_interpolator->addr);
+
+		default: return get_cycles();
+	}
+}
+
+static inline u64 time_interpolator_get_counter(int writelock)
+{
+	unsigned int src = time_interpolator->source;
+
+	if (time_interpolator->jitter)
+	{
+		cycles_t lcycle;
+		cycles_t now;
+
+		do {
+			lcycle = time_interpolator->last_cycle;
+			now = time_interpolator_get_cycles(src);
+			if (lcycle && time_after(lcycle, now))
+				return lcycle;
+
+			/* When holding the xtime write lock, there's no need
+			 * to add the overhead of the cmpxchg.  Readers are
+			 * force to retry until the write lock is released.
+			 */
+			if (writelock) {
+				time_interpolator->last_cycle = now;
+				return now;
+			}
+			/* Keep track of the last timer value returned. The use of cmpxchg here
+			 * will cause contention in an SMP environment.
+			 */
+		} while (unlikely(cmpxchg(&time_interpolator->last_cycle, lcycle, now) != lcycle));
+		return now;
+	}
+	else
+		return time_interpolator_get_cycles(src);
+}
+
+void time_interpolator_reset(void)
+{
+	time_interpolator->offset = 0;
+	time_interpolator->last_counter = time_interpolator_get_counter(1);
+}
+
+#define GET_TI_NSECS(count,i) (((((count) - i->last_counter) & (i)->mask) * (i)->nsec_per_cyc) >> (i)->shift)
+
+unsigned long time_interpolator_get_offset(void)
+{
+	/* If we do not have a time interpolator set up then just return zero */
+	if (!time_interpolator)
+		return 0;
+
+	return time_interpolator->offset +
+		GET_TI_NSECS(time_interpolator_get_counter(0), time_interpolator);
+}
+
+#define INTERPOLATOR_ADJUST 65536
+#define INTERPOLATOR_MAX_SKIP 10*INTERPOLATOR_ADJUST
+
+void time_interpolator_update(long delta_nsec)
+{
+	u64 counter;
+	unsigned long offset;
+
+	/* If there is no time interpolator set up then do nothing */
+	if (!time_interpolator)
+		return;
+
+	/*
+	 * The interpolator compensates for late ticks by accumulating the late
+	 * time in time_interpolator->offset. A tick earlier than expected will
+	 * lead to a reset of the offset and a corresponding jump of the clock
+	 * forward. Again this only works if the interpolator clock is running
+	 * slightly slower than the regular clock and the tuning logic insures
+	 * that.
+	 */
+
+	counter = time_interpolator_get_counter(1);
+	offset = time_interpolator->offset +
+			GET_TI_NSECS(counter, time_interpolator);
+
+	if (delta_nsec < 0 || (unsigned long) delta_nsec < offset)
+		time_interpolator->offset = offset - delta_nsec;
+	else {
+		time_interpolator->skips++;
+		time_interpolator->ns_skipped += delta_nsec - offset;
+		time_interpolator->offset = 0;
+	}
+	time_interpolator->last_counter = counter;
+
+	/* Tuning logic for time interpolator invoked every minute or so.
+	 * Decrease interpolator clock speed if no skips occurred and an offset is carried.
+	 * Increase interpolator clock speed if we skip too much time.
+	 */
+	if (jiffies % INTERPOLATOR_ADJUST == 0)
+	{
+		if (time_interpolator->skips == 0 && time_interpolator->offset > tick_nsec)
+			time_interpolator->nsec_per_cyc--;
+		if (time_interpolator->ns_skipped > INTERPOLATOR_MAX_SKIP && time_interpolator->offset == 0)
+			time_interpolator->nsec_per_cyc++;
+		time_interpolator->skips = 0;
+		time_interpolator->ns_skipped = 0;
+	}
+}
+
+static inline int
+is_better_time_interpolator(struct time_interpolator *new)
+{
+	if (!time_interpolator)
+		return 1;
+	return new->frequency > 2*time_interpolator->frequency ||
+	    (unsigned long)new->drift < (unsigned long)time_interpolator->drift;
+}
+
+void
+register_time_interpolator(struct time_interpolator *ti)
+{
+	unsigned long flags;
+
+	/* Sanity check */
+	BUG_ON(ti->frequency == 0 || ti->mask == 0);
+
+	ti->nsec_per_cyc = ((u64)NSEC_PER_SEC << ti->shift) / ti->frequency;
+	spin_lock(&time_interpolator_lock);
+	write_seqlock_irqsave(&xtime_lock, flags);
+	if (is_better_time_interpolator(ti)) {
+		time_interpolator = ti;
+		time_interpolator_reset();
+	}
+	write_sequnlock_irqrestore(&xtime_lock, flags);
+
+	ti->next = time_interpolator_list;
+	time_interpolator_list = ti;
+	spin_unlock(&time_interpolator_lock);
+}
+
+void
+unregister_time_interpolator(struct time_interpolator *ti)
+{
+	struct time_interpolator *curr, **prev;
+	unsigned long flags;
+
+	spin_lock(&time_interpolator_lock);
+	prev = &time_interpolator_list;
+	for (curr = *prev; curr; curr = curr->next) {
+		if (curr == ti) {
+			*prev = curr->next;
+			break;
+		}
+		prev = &curr->next;
+	}
+
+	write_seqlock_irqsave(&xtime_lock, flags);
+	if (ti == time_interpolator) {
+		/* we lost the best time-interpolator: */
+		time_interpolator = NULL;
+		/* find the next-best interpolator */
+		for (curr = time_interpolator_list; curr; curr = curr->next)
+			if (is_better_time_interpolator(curr))
+				time_interpolator = curr;
+		time_interpolator_reset();
+	}
+	write_sequnlock_irqrestore(&xtime_lock, flags);
+	spin_unlock(&time_interpolator_lock);
+}
+#endif /* CONFIG_TIME_INTERPOLATION */
+
Index: linux-2.6.19/kernel/timer.c
===================================================================
--- linux-2.6.19.orig/kernel/timer.c
+++ linux-2.6.19/kernel/timer.c
@@ -35,6 +35,7 @@
 #include <linux/syscalls.h>
 #include <linux/delay.h>
 #include <linux/kallsyms.h>
+#include <linux/timekeeping.h>
 
 #include <asm/uaccess.h>
 #include <asm/unistd.h>
@@ -761,447 +762,6 @@ unsigned long next_timer_interrupt(void)
 
 #endif
 
-/******************************************************************/
-
-/* 
- * The current time 
- * wall_to_monotonic is what we need to add to xtime (or xtime corrected 
- * for sub jiffie times) to get to monotonic time.  Monotonic is pegged
- * at zero at system boot time, so wall_to_monotonic will be negative,
- * however, we will ALWAYS keep the tv_nsec part positive so we can use
- * the usual normalization.
- */
-struct timespec xtime __attribute__ ((aligned (16)));
-struct timespec wall_to_monotonic __attribute__ ((aligned (16)));
-
-EXPORT_SYMBOL(xtime);
-
-
-/* XXX - all of this timekeeping code should be later moved to time.c */
-#include <linux/clocksource.h>
-static struct clocksource *clock; /* pointer to current clocksource */
-
-#ifdef CONFIG_GENERIC_TIME
-/**
- * __get_nsec_offset - Returns nanoseconds since last call to periodic_hook
- *
- * private function, must hold xtime_lock lock when being
- * called. Returns the number of nanoseconds since the
- * last call to update_wall_time() (adjusted by NTP scaling)
- */
-static inline s64 __get_nsec_offset(void)
-{
-	cycle_t cycle_now, cycle_delta;
-	s64 ns_offset;
-
-	/* read clocksource: */
-	cycle_now = clocksource_read(clock);
-
-	/* calculate the delta since the last update_wall_time: */
-	cycle_delta = (cycle_now - clock->cycle_last) & clock->mask;
-
-	/* convert to nanoseconds: */
-	ns_offset = cyc2ns(clock, cycle_delta);
-
-	return ns_offset;
-}
-
-/**
- * __get_realtime_clock_ts - Returns the time of day in a timespec
- * @ts:		pointer to the timespec to be set
- *
- * Returns the time of day in a timespec. Used by
- * do_gettimeofday() and get_realtime_clock_ts().
- */
-static inline void __get_realtime_clock_ts(struct timespec *ts)
-{
-	unsigned long seq;
-	s64 nsecs;
-
-	do {
-		seq = read_seqbegin(&xtime_lock);
-
-		*ts = xtime;
-		nsecs = __get_nsec_offset();
-
-	} while (read_seqretry(&xtime_lock, seq));
-
-	timespec_add_ns(ts, nsecs);
-}
-
-/**
- * getnstimeofday - Returns the time of day in a timespec
- * @ts:		pointer to the timespec to be set
- *
- * Returns the time of day in a timespec.
- */
-void getnstimeofday(struct timespec *ts)
-{
-	__get_realtime_clock_ts(ts);
-}
-
-EXPORT_SYMBOL(getnstimeofday);
-
-/**
- * do_gettimeofday - Returns the time of day in a timeval
- * @tv:		pointer to the timeval to be set
- *
- * NOTE: Users should be converted to using get_realtime_clock_ts()
- */
-void do_gettimeofday(struct timeval *tv)
-{
-	struct timespec now;
-
-	__get_realtime_clock_ts(&now);
-	tv->tv_sec = now.tv_sec;
-	tv->tv_usec = now.tv_nsec/1000;
-}
-
-EXPORT_SYMBOL(do_gettimeofday);
-/**
- * do_settimeofday - Sets the time of day
- * @tv:		pointer to the timespec variable containing the new time
- *
- * Sets the time of day to the new time and update NTP and notify hrtimers
- */
-int do_settimeofday(struct timespec *tv)
-{
-	unsigned long flags;
-	time_t wtm_sec, sec = tv->tv_sec;
-	long wtm_nsec, nsec = tv->tv_nsec;
-
-	if ((unsigned long)tv->tv_nsec >= NSEC_PER_SEC)
-		return -EINVAL;
-
-	write_seqlock_irqsave(&xtime_lock, flags);
-
-	nsec -= __get_nsec_offset();
-
-	wtm_sec  = wall_to_monotonic.tv_sec + (xtime.tv_sec - sec);
-	wtm_nsec = wall_to_monotonic.tv_nsec + (xtime.tv_nsec - nsec);
-
-	set_normalized_timespec(&xtime, sec, nsec);
-	set_normalized_timespec(&wall_to_monotonic, wtm_sec, wtm_nsec);
-
-	clock->error = 0;
-	ntp_clear();
-
-	write_sequnlock_irqrestore(&xtime_lock, flags);
-
-	/* signal hrtimers about time change */
-	clock_was_set();
-
-	return 0;
-}
-
-EXPORT_SYMBOL(do_settimeofday);
-
-/**
- * change_clocksource - Swaps clocksources if a new one is available
- *
- * Accumulates current time interval and initializes new clocksource
- */
-static int change_clocksource(void)
-{
-	struct clocksource *new;
-	cycle_t now;
-	u64 nsec;
-	new = clocksource_get_next();
-	if (clock != new) {
-		now = clocksource_read(new);
-		nsec =  __get_nsec_offset();
-		timespec_add_ns(&xtime, nsec);
-
-		clock = new;
-		clock->cycle_last = now;
-		printk(KERN_INFO "Time: %s clocksource has been installed.\n",
-		       clock->name);
-		return 1;
-	} else if (clock->update_callback) {
-		return clock->update_callback();
-	}
-	return 0;
-}
-#else
-static inline int change_clocksource(void)
-{
-	return 0;
-}
-#endif
-
-/**
- * timeofday_is_continuous - check to see if timekeeping is free running
- */
-int timekeeping_is_continuous(void)
-{
-	unsigned long seq;
-	int ret;
-
-	do {
-		seq = read_seqbegin(&xtime_lock);
-
-		ret = clock->is_continuous;
-
-	} while (read_seqretry(&xtime_lock, seq));
-
-	return ret;
-}
-
-/**
- * read_persistent_clock -  Return time in seconds from the persistent clock.
- *
- * Weak dummy function for arches that do not yet support it.
- * Returns seconds from epoch using the battery backed persistent clock.
- * Returns zero if unsupported.
- *
- *  XXX - Do be sure to remove it once all arches implement it.
- */
-unsigned long __attribute__((weak)) read_persistent_clock(void)
-{
-	return 0;
-}
-
-/*
- * timekeeping_init - Initializes the clocksource and common timekeeping values
- */
-void __init timekeeping_init(void)
-{
-	unsigned long flags;
-	unsigned long sec = read_persistent_clock();
-
-	write_seqlock_irqsave(&xtime_lock, flags);
-
-	ntp_clear();
-
-	clock = clocksource_get_next();
-	clocksource_calculate_interval(clock, tick_nsec);
-	clock->cycle_last = clocksource_read(clock);
-
-	xtime.tv_sec = sec;
-	xtime.tv_nsec = 0;
-	set_normalized_timespec(&wall_to_monotonic,
-		-xtime.tv_sec, -xtime.tv_nsec);
-
-	write_sequnlock_irqrestore(&xtime_lock, flags);
-}
-
-
-/* flag for if timekeeping is suspended */
-static int timekeeping_suspended;
-/* time in seconds when suspend began */
-static unsigned long timekeeping_suspend_time;
-
-/**
- * timekeeping_resume - Resumes the generic timekeeping subsystem.
- * @dev:	unused
- *
- * This is for the generic clocksource timekeeping.
- * xtime/wall_to_monotonic/jiffies/etc are
- * still managed by arch specific suspend/resume code.
- */
-static int timekeeping_resume(struct sys_device *dev)
-{
-	unsigned long flags;
-	unsigned long now = read_persistent_clock();
-
-	write_seqlock_irqsave(&xtime_lock, flags);
-
-	if (now && (now > timekeeping_suspend_time)) {
-		unsigned long sleep_length = now - timekeeping_suspend_time;
-		xtime.tv_sec += sleep_length;
-		jiffies_64 += (u64)sleep_length * HZ;
-	}
-	/* re-base the last cycle value */
-	clock->cycle_last = clocksource_read(clock);
-	clock->error = 0;
-	timekeeping_suspended = 0;
-	write_sequnlock_irqrestore(&xtime_lock, flags);
-
-	hrtimer_notify_resume();
-
-	return 0;
-}
-
-static int timekeeping_suspend(struct sys_device *dev, pm_message_t state)
-{
-	unsigned long flags;
-
-	write_seqlock_irqsave(&xtime_lock, flags);
-	timekeeping_suspended = 1;
-	timekeeping_suspend_time = read_persistent_clock();
-	write_sequnlock_irqrestore(&xtime_lock, flags);
-	return 0;
-}
-
-/* sysfs resume/suspend bits for timekeeping */
-static struct sysdev_class timekeeping_sysclass = {
-	.resume		= timekeeping_resume,
-	.suspend	= timekeeping_suspend,
-	set_kset_name("timekeeping"),
-};
-
-static struct sys_device device_timer = {
-	.id		= 0,
-	.cls		= &timekeeping_sysclass,
-};
-
-static int __init timekeeping_init_device(void)
-{
-	int error = sysdev_class_register(&timekeeping_sysclass);
-	if (!error)
-		error = sysdev_register(&device_timer);
-	return error;
-}
-
-device_initcall(timekeeping_init_device);
-
-/*
- * If the error is already larger, we look ahead even further
- * to compensate for late or lost adjustments.
- */
-static __always_inline int clocksource_bigadjust(s64 error, s64 *interval,
-						 s64 *offset)
-{
-	s64 tick_error, i;
-	u32 look_ahead, adj;
-	s32 error2, mult;
-
-	/*
-	 * Use the current error value to determine how much to look ahead.
-	 * The larger the error the slower we adjust for it to avoid problems
-	 * with losing too many ticks, otherwise we would overadjust and
-	 * produce an even larger error.  The smaller the adjustment the
-	 * faster we try to adjust for it, as lost ticks can do less harm
-	 * here.  This is tuned so that an error of about 1 msec is adusted
-	 * within about 1 sec (or 2^20 nsec in 2^SHIFT_HZ ticks).
-	 */
-	error2 = clock->error >> (TICK_LENGTH_SHIFT + 22 - 2 * SHIFT_HZ);
-	error2 = abs(error2);
-	for (look_ahead = 0; error2 > 0; look_ahead++)
-		error2 >>= 2;
-
-	/*
-	 * Now calculate the error in (1 << look_ahead) ticks, but first
-	 * remove the single look ahead already included in the error.
-	 */
-	tick_error = current_tick_length() >>
-		(TICK_LENGTH_SHIFT - clock->shift + 1);
-	tick_error -= clock->xtime_interval >> 1;
-	error = ((error - tick_error) >> look_ahead) + tick_error;
-
-	/* Finally calculate the adjustment shift value.  */
-	i = *interval;
-	mult = 1;
-	if (error < 0) {
-		error = -error;
-		*interval = -*interval;
-		*offset = -*offset;
-		mult = -1;
-	}
-	for (adj = 0; error > i; adj++)
-		error >>= 1;
-
-	*interval <<= adj;
-	*offset <<= adj;
-	return mult << adj;
-}
-
-/*
- * Adjust the multiplier to reduce the error value,
- * this is optimized for the most common adjustments of -1,0,1,
- * for other values we can do a bit more work.
- */
-static void clocksource_adjust(struct clocksource *clock, s64 offset)
-{
-	s64 error, interval = clock->cycle_interval;
-	int adj;
-
-	error = clock->error >> (TICK_LENGTH_SHIFT - clock->shift - 1);
-	if (error > interval) {
-		error >>= 2;
-		if (likely(error <= interval))
-			adj = 1;
-		else
-			adj = clocksource_bigadjust(error, &interval, &offset);
-	} else if (error < -interval) {
-		error >>= 2;
-		if (likely(error >= -interval)) {
-			adj = -1;
-			interval = -interval;
-			offset = -offset;
-		} else
-			adj = clocksource_bigadjust(error, &interval, &offset);
-	} else
-		return;
-
-	clock->mult += adj;
-	clock->xtime_interval += interval;
-	clock->xtime_nsec -= offset;
-	clock->error -= (interval - offset) <<
-			(TICK_LENGTH_SHIFT - clock->shift);
-}
-
-/**
- * update_wall_time - Uses the current clocksource to increment the wall time
- *
- * Called from the timer interrupt, must hold a write on xtime_lock.
- */
-static void update_wall_time(void)
-{
-	cycle_t offset;
-
-	/* Make sure we're fully resumed: */
-	if (unlikely(timekeeping_suspended))
-		return;
-
-#ifdef CONFIG_GENERIC_TIME
-	offset = (clocksource_read(clock) - clock->cycle_last) & clock->mask;
-#else
-	offset = clock->cycle_interval;
-#endif
-	clock->xtime_nsec += (s64)xtime.tv_nsec << clock->shift;
-
-	/* normally this loop will run just once, however in the
-	 * case of lost or late ticks, it will accumulate correctly.
-	 */
-	while (offset >= clock->cycle_interval) {
-		/* accumulate one interval */
-		clock->xtime_nsec += clock->xtime_interval;
-		clock->cycle_last += clock->cycle_interval;
-		offset -= clock->cycle_interval;
-
-		if (clock->xtime_nsec >= (u64)NSEC_PER_SEC << clock->shift) {
-			clock->xtime_nsec -= (u64)NSEC_PER_SEC << clock->shift;
-			xtime.tv_sec++;
-			second_overflow();
-		}
-
-		/* interpolator bits */
-		time_interpolator_update(clock->xtime_interval
-						>> clock->shift);
-
-		/* accumulate error between NTP and clock interval */
-		clock->error += current_tick_length();
-		clock->error -= clock->xtime_interval << (TICK_LENGTH_SHIFT - clock->shift);
-	}
-
-	/* correct the clock when NTP error is too big */
-	clocksource_adjust(clock, offset);
-
-	/* store full nanoseconds into xtime */
-	xtime.tv_nsec = (s64)clock->xtime_nsec >> clock->shift;
-	clock->xtime_nsec -= (s64)xtime.tv_nsec << clock->shift;
-
-	/* check to see if there is a new clocksource to use */
-	if (change_clocksource()) {
-		clock->error = 0;
-		clock->xtime_nsec = 0;
-		hrtimer_clock_notify();
-		clocksource_calculate_interval(clock, tick_nsec);
-	}
-	update_vsyscall(&xtime, clock);
-}
-
 /*
  * Called from the timer interrupt handler to charge one tick to the current 
  * process.  user_tick is 1 if the tick is user time, 0 for system.
@@ -1733,194 +1293,6 @@ void __init init_timers(void)
 	open_softirq(TIMER_SOFTIRQ, run_timer_softirq, NULL);
 }
 
-#ifdef CONFIG_TIME_INTERPOLATION
-
-struct time_interpolator *time_interpolator __read_mostly;
-static struct time_interpolator *time_interpolator_list __read_mostly;
-static DEFINE_SPINLOCK(time_interpolator_lock);
-
-static inline cycles_t time_interpolator_get_cycles(unsigned int src)
-{
-	unsigned long (*x)(void);
-
-	switch (src)
-	{
-		case TIME_SOURCE_FUNCTION:
-			x = time_interpolator->addr;
-			return x();
-
-		case TIME_SOURCE_MMIO64	:
-			return readq_relaxed((void __iomem *)time_interpolator->addr);
-
-		case TIME_SOURCE_MMIO32	:
-			return readl_relaxed((void __iomem *)time_interpolator->addr);
-
-		default: return get_cycles();
-	}
-}
-
-static inline u64 time_interpolator_get_counter(int writelock)
-{
-	unsigned int src = time_interpolator->source;
-
-	if (time_interpolator->jitter)
-	{
-		cycles_t lcycle;
-		cycles_t now;
-
-		do {
-			lcycle = time_interpolator->last_cycle;
-			now = time_interpolator_get_cycles(src);
-			if (lcycle && time_after(lcycle, now))
-				return lcycle;
-
-			/* When holding the xtime write lock, there's no need
-			 * to add the overhead of the cmpxchg.  Readers are
-			 * force to retry until the write lock is released.
-			 */
-			if (writelock) {
-				time_interpolator->last_cycle = now;
-				return now;
-			}
-			/* Keep track of the last timer value returned. The use of cmpxchg here
-			 * will cause contention in an SMP environment.
-			 */
-		} while (unlikely(cmpxchg(&time_interpolator->last_cycle, lcycle, now) != lcycle));
-		return now;
-	}
-	else
-		return time_interpolator_get_cycles(src);
-}
-
-void time_interpolator_reset(void)
-{
-	time_interpolator->offset = 0;
-	time_interpolator->last_counter = time_interpolator_get_counter(1);
-}
-
-#define GET_TI_NSECS(count,i) (((((count) - i->last_counter) & (i)->mask) * (i)->nsec_per_cyc) >> (i)->shift)
-
-unsigned long time_interpolator_get_offset(void)
-{
-	/* If we do not have a time interpolator set up then just return zero */
-	if (!time_interpolator)
-		return 0;
-
-	return time_interpolator->offset +
-		GET_TI_NSECS(time_interpolator_get_counter(0), time_interpolator);
-}
-
-#define INTERPOLATOR_ADJUST 65536
-#define INTERPOLATOR_MAX_SKIP 10*INTERPOLATOR_ADJUST
-
-void time_interpolator_update(long delta_nsec)
-{
-	u64 counter;
-	unsigned long offset;
-
-	/* If there is no time interpolator set up then do nothing */
-	if (!time_interpolator)
-		return;
-
-	/*
-	 * The interpolator compensates for late ticks by accumulating the late
-	 * time in time_interpolator->offset. A tick earlier than expected will
-	 * lead to a reset of the offset and a corresponding jump of the clock
-	 * forward. Again this only works if the interpolator clock is running
-	 * slightly slower than the regular clock and the tuning logic insures
-	 * that.
-	 */
-
-	counter = time_interpolator_get_counter(1);
-	offset = time_interpolator->offset +
-			GET_TI_NSECS(counter, time_interpolator);
-
-	if (delta_nsec < 0 || (unsigned long) delta_nsec < offset)
-		time_interpolator->offset = offset - delta_nsec;
-	else {
-		time_interpolator->skips++;
-		time_interpolator->ns_skipped += delta_nsec - offset;
-		time_interpolator->offset = 0;
-	}
-	time_interpolator->last_counter = counter;
-
-	/* Tuning logic for time interpolator invoked every minute or so.
-	 * Decrease interpolator clock speed if no skips occurred and an offset is carried.
-	 * Increase interpolator clock speed if we skip too much time.
-	 */
-	if (jiffies % INTERPOLATOR_ADJUST == 0)
-	{
-		if (time_interpolator->skips == 0 && time_interpolator->offset > tick_nsec)
-			time_interpolator->nsec_per_cyc--;
-		if (time_interpolator->ns_skipped > INTERPOLATOR_MAX_SKIP && time_interpolator->offset == 0)
-			time_interpolator->nsec_per_cyc++;
-		time_interpolator->skips = 0;
-		time_interpolator->ns_skipped = 0;
-	}
-}
-
-static inline int
-is_better_time_interpolator(struct time_interpolator *new)
-{
-	if (!time_interpolator)
-		return 1;
-	return new->frequency > 2*time_interpolator->frequency ||
-	    (unsigned long)new->drift < (unsigned long)time_interpolator->drift;
-}
-
-void
-register_time_interpolator(struct time_interpolator *ti)
-{
-	unsigned long flags;
-
-	/* Sanity check */
-	BUG_ON(ti->frequency == 0 || ti->mask == 0);
-
-	ti->nsec_per_cyc = ((u64)NSEC_PER_SEC << ti->shift) / ti->frequency;
-	spin_lock(&time_interpolator_lock);
-	write_seqlock_irqsave(&xtime_lock, flags);
-	if (is_better_time_interpolator(ti)) {
-		time_interpolator = ti;
-		time_interpolator_reset();
-	}
-	write_sequnlock_irqrestore(&xtime_lock, flags);
-
-	ti->next = time_interpolator_list;
-	time_interpolator_list = ti;
-	spin_unlock(&time_interpolator_lock);
-}
-
-void
-unregister_time_interpolator(struct time_interpolator *ti)
-{
-	struct time_interpolator *curr, **prev;
-	unsigned long flags;
-
-	spin_lock(&time_interpolator_lock);
-	prev = &time_interpolator_list;
-	for (curr = *prev; curr; curr = curr->next) {
-		if (curr == ti) {
-			*prev = curr->next;
-			break;
-		}
-		prev = &curr->next;
-	}
-
-	write_seqlock_irqsave(&xtime_lock, flags);
-	if (ti == time_interpolator) {
-		/* we lost the best time-interpolator: */
-		time_interpolator = NULL;
-		/* find the next-best interpolator */
-		for (curr = time_interpolator_list; curr; curr = curr->next)
-			if (is_better_time_interpolator(curr))
-				time_interpolator = curr;
-		time_interpolator_reset();
-	}
-	write_sequnlock_irqrestore(&xtime_lock, flags);
-	spin_unlock(&time_interpolator_lock);
-}
-#endif /* CONFIG_TIME_INTERPOLATION */
-
 /**
  * msleep - sleep safely even with waitqueue interruptions
  * @msecs: Time in milliseconds to sleep for

-- 

[PATCH 02/12] clocksource: rating sorted list

[Daniel Walker]

[-- Attachment #1: clocksource_priority_sorted_list.patch --]
[-- Type: text/plain, Size: 9248 bytes --]

Converts the original plain list into a sorted list based on the clock rating.
Later in my tree this allows some of the variables to be dropped since the
highest rated clock is always at the front of the list. This also does some
other nice things like allow the sysfs files to print the clocks in a more
interesting order. It's forward looking.

Signed-Off-By: Daniel Walker <dwalker@mvista.com>

---
 arch/i386/kernel/tsc.c      |    2 
 arch/x86_64/kernel/tsc.c    |    2 
 include/linux/clocksource.h |    9 ++-
 kernel/time/clocksource.c   |  132 +++++++++++++++++++++++++++++---------------
 4 files changed, 97 insertions(+), 48 deletions(-)

Index: linux-2.6.19/arch/i386/kernel/tsc.c
===================================================================
--- linux-2.6.19.orig/arch/i386/kernel/tsc.c
+++ linux-2.6.19/arch/i386/kernel/tsc.c
@@ -361,7 +361,7 @@ static int tsc_update_callback(void)
 	/* check to see if we should switch to the safe clocksource: */
 	if (clocksource_tsc.rating != 0 && check_tsc_unstable()) {
 		clocksource_tsc.rating = 0;
-		clocksource_reselect();
+		clocksource_rating_change(&clocksource_tsc);
 		change = 1;
 	}
 
Index: linux-2.6.19/arch/x86_64/kernel/tsc.c
===================================================================
--- linux-2.6.19.orig/arch/x86_64/kernel/tsc.c
+++ linux-2.6.19/arch/x86_64/kernel/tsc.c
@@ -218,7 +218,7 @@ static int tsc_update_callback(void)
 	/* check to see if we should switch to the safe clocksource: */
 	if (clocksource_tsc.rating != 50 && check_tsc_unstable()) {
 		clocksource_tsc.rating = 50;
-		clocksource_reselect();
+		clocksource_rating_change(&clocksource_tsc);
 		change = 1;
 	}
 	return change;
Index: linux-2.6.19/include/linux/clocksource.h
===================================================================
--- linux-2.6.19.orig/include/linux/clocksource.h
+++ linux-2.6.19/include/linux/clocksource.h
@@ -12,6 +12,9 @@
 #include <linux/timex.h>
 #include <linux/time.h>
 #include <linux/list.h>
+#include <linux/plist.h>
+#include <linux/sysdev.h>
+
 #include <asm/div64.h>
 #include <asm/io.h>
 
@@ -183,9 +186,9 @@ static inline void clocksource_calculate
 
 
 /* used to install a new clocksource */
-int clocksource_register(struct clocksource*);
-void clocksource_reselect(void);
-struct clocksource* clocksource_get_next(void);
+extern struct clocksource *clocksource_get_next(void);
+extern int clocksource_register(struct clocksource*);
+extern void clocksource_rating_change(struct clocksource*);
 
 #ifdef CONFIG_GENERIC_TIME_VSYSCALL
 extern void update_vsyscall(struct timespec *ts, struct clocksource *c);
Index: linux-2.6.19/kernel/time/clocksource.c
===================================================================
--- linux-2.6.19.orig/kernel/time/clocksource.c
+++ linux-2.6.19/kernel/time/clocksource.c
@@ -35,7 +35,7 @@
  * next_clocksource:
  *	pending next selected clocksource.
  * clocksource_list:
- *	linked list with the registered clocksources
+ *	rating sorted linked list with the registered clocksources
  * clocksource_lock:
  *	protects manipulations to curr_clocksource and next_clocksource
  *	and the clocksource_list
@@ -80,69 +80,105 @@ struct clocksource *clocksource_get_next
 }
 
 /**
- * select_clocksource - Finds the best registered clocksource.
+ * __is_registered - Returns a clocksource if it's registered
+ * @name:		name of the clocksource to return
  *
  * Private function. Must hold clocksource_lock when called.
  *
- * Looks through the list of registered clocksources, returning
- * the one with the highest rating value. If there is a clocksource
- * name that matches the override string, it returns that clocksource.
+ * Returns the clocksource if registered, zero otherwise.
+ * If no clocksources are registered the jiffies clock is
+ * returned.
  */
-static struct clocksource *select_clocksource(void)
+static struct clocksource * __is_registered(char * name)
 {
-	struct clocksource *best = NULL;
 	struct list_head *tmp;
 
 	list_for_each(tmp, &clocksource_list) {
 		struct clocksource *src;
 
 		src = list_entry(tmp, struct clocksource, list);
-		if (!best)
-			best = src;
-
-		/* check for override: */
-		if (strlen(src->name) == strlen(override_name) &&
-		    !strcmp(src->name, override_name)) {
-			best = src;
-			break;
-		}
-		/* pick the highest rating: */
-		if (src->rating > best->rating)
-		 	best = src;
+		if (!strcmp(src->name, name))
+			return src;
 	}
 
-	return best;
+	return 0;
 }
 
 /**
- * is_registered_source - Checks if clocksource is registered
- * @c:		pointer to a clocksource
+ * __get_clock - Finds a specific clocksource
+ * @name:		name of the clocksource to return
  *
- * Private helper function. Must hold clocksource_lock when called.
+ * Private function. Must hold clocksource_lock when called.
  *
- * Returns one if the clocksource is already registered, zero otherwise.
+ * Returns the clocksource if registered, zero otherwise.
+ * If the @name is null the highest rated clock is returned.
  */
-static int is_registered_source(struct clocksource *c)
+static inline struct clocksource * __get_clock(char * name)
+{
+
+	if (unlikely(list_empty(&clocksource_list)))
+		return &clocksource_jiffies;
+
+	if (!name)
+		return list_entry(clocksource_list.next,
+				  struct clocksource, list);
+
+	return __is_registered(name);
+}
+
+/**
+ * select_clocksource - Finds the best registered clocksource.
+ *
+ * Private function. Must hold clocksource_lock when called.
+ *
+ * Looks through the list of registered clocksources, returning
+ * the one with the highest rating value. If there is a clocksource
+ * name that matches the override string, it returns that clocksource.
+ */
+static struct clocksource *select_clocksource(void)
+{
+	if (!*override_name)
+		return list_entry(clocksource_list.next,
+				  struct clocksource, list);
+
+	return __get_clock(override_name);
+}
+
+/*
+ * __sorted_list_add - Sorted clocksource add
+ * @c:			clocksource to add
+ *
+ * Adds a clocksource to the clocksource_list in sorted order.
+ */
+static void __sorted_list_add(struct clocksource *c)
 {
-	int len = strlen(c->name);
 	struct list_head *tmp;
 
 	list_for_each(tmp, &clocksource_list) {
 		struct clocksource *src;
 
 		src = list_entry(tmp, struct clocksource, list);
-		if (strlen(src->name) == len &&	!strcmp(src->name, c->name))
-			return 1;
+
+		if (c->rating > src->rating) {
+			list_add_tail(&c->list, &src->list);
+			return;
+		}
 	}
 
-	return 0;
+	/*
+	 * If it's bigger/smaller than all the other entries put it
+	 * at the end.
+	 */
+	list_add_tail(&c->list, &clocksource_list);
 }
 
 /**
  * clocksource_register - Used to install new clocksources
  * @t:		clocksource to be registered
  *
- * Returns -EBUSY if registration fails, zero otherwise.
+ * Returns -EBUSY clock is already registered,
+ * Returns -EINVAL if clocksource is invalid,
+ * Return zero otherwise.
  */
 int clocksource_register(struct clocksource *c)
 {
@@ -150,14 +186,13 @@ int clocksource_register(struct clocksou
 	unsigned long flags;
 
 	spin_lock_irqsave(&clocksource_lock, flags);
-	/* check if clocksource is already registered */
-	if (is_registered_source(c)) {
-		printk("register_clocksource: Cannot register %s. "
+	if (unlikely(!list_empty(&c->list) && __is_registered(c->name))) {
+		printk("register_clocksource: Cannot register %s clocksource. "
 		       "Already registered!", c->name);
 		ret = -EBUSY;
 	} else {
-		/* register it */
- 		list_add(&c->list, &clocksource_list);
+		INIT_LIST_HEAD(&c->list);
+ 		__sorted_list_add(c);
 		/* scan the registered clocksources, and pick the best one */
 		next_clocksource = select_clocksource();
 	}
@@ -167,21 +202,31 @@ int clocksource_register(struct clocksou
 EXPORT_SYMBOL(clocksource_register);
 
 /**
- * clocksource_reselect - Rescan list for next clocksource
+ * clocksource_rating_change - Allows dynamic rating changes for register
+ *                           clocksources.
  *
- * A quick helper function to be used if a clocksource changes its
- * rating. Forces the clocksource list to be re-scanned for the best
- * clocksource.
+ * Signals that a clocksource is dynamically changing it's rating.
+ * This could happen if a clocksource becomes more/less stable.
  */
-void clocksource_reselect(void)
+void clocksource_rating_change(struct clocksource *c)
 {
 	unsigned long flags;
 
-	spin_lock_irqsave(&clocksource_lock, flags);
+	if (unlikely(list_empty(&c->list)))
+		return;
+
+  	spin_lock_irqsave(&clocksource_lock, flags);
+
+	/*
+	 * Re-register the clocksource with it's new rating.
+	 */
+	list_del_init(&c->list);
+	__sorted_list_add(c);
+
 	next_clocksource = select_clocksource();
 	spin_unlock_irqrestore(&clocksource_lock, flags);
 }
-EXPORT_SYMBOL(clocksource_reselect);
+EXPORT_SYMBOL(clocksource_rating_change);
 
 #ifdef CONFIG_SYSFS
 /**
@@ -247,7 +292,8 @@ static ssize_t sysfs_override_clocksourc
  * @dev:	unused
  * @buf:	char buffer to be filled with clocksource list
  *
- * Provides sysfs interface for listing registered clocksources
+ * Provides sysfs interface for listing registered clocksources.
+ * Output in priority sorted order.
  */
 static ssize_t
 sysfs_show_available_clocksources(struct sys_device *dev, char *buf)

-- 

[PATCH 03/12] clocksource: arm initialize list value

[Daniel Walker]

[-- Attachment #1: clocksource_arm_list_init.patch --]
[-- Type: text/plain, Size: 2019 bytes --]

Update arch/arm/ with list initialization.

Signed-Off-By: Daniel Walker <dwalker@mvista.com>

---
 arch/arm/mach-imx/time.c      |    1 +
 arch/arm/mach-ixp4xx/common.c |    1 +
 arch/arm/mach-netx/time.c     |    1 +
 arch/arm/mach-pxa/time.c      |    1 +
 4 files changed, 4 insertions(+)

Index: linux-2.6.19/arch/arm/mach-imx/time.c
===================================================================
--- linux-2.6.19.orig/arch/arm/mach-imx/time.c
+++ linux-2.6.19/arch/arm/mach-imx/time.c
@@ -87,6 +87,7 @@ static struct clocksource clocksource_im
 	.read		= imx_get_cycles,
 	.mask		= 0xFFFFFFFF,
 	.shift 		= 20,
+	.list		= LIST_HEAD_INIT(clocksource_imx.list),
 	.is_continuous 	= 1,
 };
 
Index: linux-2.6.19/arch/arm/mach-ixp4xx/common.c
===================================================================
--- linux-2.6.19.orig/arch/arm/mach-ixp4xx/common.c
+++ linux-2.6.19/arch/arm/mach-ixp4xx/common.c
@@ -396,6 +396,7 @@ static struct clocksource clocksource_ix
 	.mask		= CLOCKSOURCE_MASK(32),
 	.shift 		= 20,
 	.is_continuous 	= 1,
+	.list		= LIST_HEAD_INIT(clocksource_ixp4xx.list),
 };
 
 unsigned long ixp4xx_timer_freq = FREQ;
Index: linux-2.6.19/arch/arm/mach-netx/time.c
===================================================================
--- linux-2.6.19.orig/arch/arm/mach-netx/time.c
+++ linux-2.6.19/arch/arm/mach-netx/time.c
@@ -62,6 +62,7 @@ static struct clocksource clocksource_ne
 	.read		= netx_get_cycles,
 	.mask		= CLOCKSOURCE_MASK(32),
 	.shift 		= 20,
+	.list		= LIST_HEAD_INIT(clocksource_netx.list),
 	.is_continuous 	= 1,
 };
 
Index: linux-2.6.19/arch/arm/mach-pxa/time.c
===================================================================
--- linux-2.6.19.orig/arch/arm/mach-pxa/time.c
+++ linux-2.6.19/arch/arm/mach-pxa/time.c
@@ -112,6 +112,7 @@ static struct clocksource clocksource_px
 	.read           = pxa_get_cycles,
 	.mask           = CLOCKSOURCE_MASK(32),
 	.shift          = 20,
+	.list		= LIST_HEAD_INIT(clocksource_pxa.list),
 	.is_continuous  = 1,
 };
 

-- 

[PATCH 04/12] clocksource: avr32 initialize list value

[Daniel Walker]

[-- Attachment #1: clocksource_avr32_list_init.patch --]
[-- Type: text/plain, Size: 593 bytes --]

Update arch/avre32/ with list initialization.

Signed-Off-By: Daniel Walker <dwalker@mvista.com>

---
 arch/avr32/kernel/time.c |    1 +
 1 file changed, 1 insertion(+)

Index: linux-2.6.19/arch/avr32/kernel/time.c
===================================================================
--- linux-2.6.19.orig/arch/avr32/kernel/time.c
+++ linux-2.6.19/arch/avr32/kernel/time.c
@@ -37,6 +37,7 @@ static struct clocksource clocksource_av
 	.read		= read_cycle_count,
 	.mask		= CLOCKSOURCE_MASK(32),
 	.shift		= 16,
+	.list		= LIST_HEAD_INIT(clocksource_avr32.list),
 	.is_continuous	= 1,
 };
 

-- 

Re: [PATCH 04/12] clocksource: avr32 initialize list value

[Haavard Skinnemoen]

On 1/23/07, Daniel Walker <dwalker@mvista.com> wrote:
> Update arch/avre32/ with list initialization.
>
> Signed-Off-By: Daniel Walker <dwalker@mvista.com>

Looks good to me, although I suppose it wouldn't hurt to include
<linux/list.h> as well since there seems to be an ongoing effort to
reduce the number of files that are included implicitly through other
headers. Might save us from some trouble later.

I've closed your support request, so you shouldn't expect a reply from
avr32@atmel.com (I was going to reply through the support system, but
I couldn't figure out how to prevent it from top-quoting.) Probably
best if you send patches directly to me at hskinnemoen@atmel.com, but
the support address is a nice fallback if I'm unable to respond for
some reason.

Haavard

Re: [PATCH 04/12] clocksource: avr32 initialize list value

[Daniel Walker]

On Tue, 2007-01-23 at 13:42 +0100, Haavard Skinnemoen wrote:
> On 1/23/07, Daniel Walker <dwalker@mvista.com> wrote:
> > Update arch/avre32/ with list initialization.
> >
> > Signed-Off-By: Daniel Walker <dwalker@mvista.com>
> 
> Looks good to me, although I suppose it wouldn't hurt to include
> <linux/list.h> as well since there seems to be an ongoing effort to
> reduce the number of files that are included implicitly through other
> headers. Might save us from some trouble later.

It shouldn't be outright implicit since it's included in clocksource.h .
That has to be included before you can create a clocksource. I could see
adding list.h if it wasn't clear which include was bringing it in.

> I've closed your support request, so you shouldn't expect a reply from
> avr32@atmel.com (I was going to reply through the support system, but
> I couldn't figure out how to prevent it from top-quoting.) Probably
> best if you send patches directly to me at hskinnemoen@atmel.com, but
> the support address is a nice fallback if I'm unable to respond for
> some reason.

I got that address from the maintainers file,

AVR32 ARCHITECTURE
P:      Atmel AVR32 Support Team
M:      avr32@atmel.com
P:      Haavard Skinnemoen
M:      hskinnemoen@atmel.com
W:      http://www.atmel.com/products/AVR32/
W:      http://avr32linux.org/
W:      http://avrfreaks.net/
S:      Supported

I just picked the top address .. If it makes sense you might want to
consider dropping that address from the maintainers file. It seems a
little formal for submitting patches. In the future I'll send them to
you.

Daniel

Re: [PATCH 04/12] clocksource: avr32 initialize list value

[Haavard Skinnemoen]

On Tue, 23 Jan 2007 05:02:49 -0800
Daniel Walker <dwalker@mvista.com> wrote:

> On Tue, 2007-01-23 at 13:42 +0100, Haavard Skinnemoen wrote:
> > On 1/23/07, Daniel Walker <dwalker@mvista.com> wrote:
> > > Update arch/avre32/ with list initialization.
> > >
> > > Signed-Off-By: Daniel Walker <dwalker@mvista.com>
> > 
> > Looks good to me, although I suppose it wouldn't hurt to include
> > <linux/list.h> as well since there seems to be an ongoing effort to
> > reduce the number of files that are included implicitly through other
> > headers. Might save us from some trouble later.
> 
> It shouldn't be outright implicit since it's included in clocksource.h .
> That has to be included before you can create a clocksource. I could see
> adding list.h if it wasn't clear which include was bringing it in.

Ah, of course. Just disregard that suggestion, then.

> > I've closed your support request, so you shouldn't expect a reply from
> > avr32@atmel.com (I was going to reply through the support system, but
> > I couldn't figure out how to prevent it from top-quoting.) Probably
> > best if you send patches directly to me at hskinnemoen@atmel.com, but
> > the support address is a nice fallback if I'm unable to respond for
> > some reason.
> 
> I got that address from the maintainers file,
> 
> AVR32 ARCHITECTURE
> P:      Atmel AVR32 Support Team
> M:      avr32@atmel.com
> P:      Haavard Skinnemoen
> M:      hskinnemoen@atmel.com
> W:      http://www.atmel.com/products/AVR32/
> W:      http://avr32linux.org/
> W:      http://avrfreaks.net/
> S:      Supported
> 
> I just picked the top address .. If it makes sense you might want to
> consider dropping that address from the maintainers file. It seems a
> little formal for submitting patches. In the future I'll send them to
> you.

Right. I guess we should at least switch the order so that I'm the
first one on the list. I'll have a chat with the support guys, maybe we
should just drop that address from MAINTAINERS.

Håvard

[PATCH 05/12] clocksource: mips initialize list value

[Daniel Walker]

[-- Attachment #1: clocksource_mips_list_init.patch --]
[-- Type: text/plain, Size: 593 bytes --]

Update arch/mips/ with list initialization.

Signed-Off-By: Daniel Walker <dwalker@mvista.com>

---
 arch/mips/kernel/time.c |    1 +
 1 file changed, 1 insertion(+)

Index: linux-2.6.19/arch/mips/kernel/time.c
===================================================================
--- linux-2.6.19.orig/arch/mips/kernel/time.c
+++ linux-2.6.19/arch/mips/kernel/time.c
@@ -307,6 +307,7 @@ static unsigned int __init calibrate_hpt
 struct clocksource clocksource_mips = {
 	.name		= "MIPS",
 	.mask		= 0xffffffff,
+	.list		= LIST_HEAD_INIT(clocksource_mips.list),
 	.is_continuous	= 1,
 };
 

-- 

[PATCH 06/12] clocksource: i386 initialize list value

[Daniel Walker]

[-- Attachment #1: clocksource_i386_list_init.patch --]
[-- Type: text/plain, Size: 1532 bytes --]

Update arch/i386/ with list initialization.

Signed-Off-By: Daniel Walker <dwalker@mvista.com>

---
 arch/i386/kernel/hpet.c  |    1 +
 arch/i386/kernel/i8253.c |    1 +
 arch/i386/kernel/tsc.c   |    1 +
 3 files changed, 3 insertions(+)

Index: linux-2.6.19/arch/i386/kernel/hpet.c
===================================================================
--- linux-2.6.19.orig/arch/i386/kernel/hpet.c
+++ linux-2.6.19/arch/i386/kernel/hpet.c
@@ -282,6 +282,7 @@ static struct clocksource clocksource_hp
 	.mask		= HPET_MASK,
 	.shift		= HPET_SHIFT,
 	.is_continuous	= 1,
+	.list		= LIST_HEAD_INIT(clocksource_hpet.list),
 };
 
 static int __init init_hpet_clocksource(void)
Index: linux-2.6.19/arch/i386/kernel/i8253.c
===================================================================
--- linux-2.6.19.orig/arch/i386/kernel/i8253.c
+++ linux-2.6.19/arch/i386/kernel/i8253.c
@@ -177,6 +177,7 @@ static struct clocksource clocksource_pi
 	.mask	= CLOCKSOURCE_MASK(32),
 	.mult	= 0,
 	.shift	= 20,
+	.list	= LIST_HEAD_INIT(clocksource_pit.list),
 };
 
 static int __init init_pit_clocksource(void)
Index: linux-2.6.19/arch/i386/kernel/tsc.c
===================================================================
--- linux-2.6.19.orig/arch/i386/kernel/tsc.c
+++ linux-2.6.19/arch/i386/kernel/tsc.c
@@ -352,6 +352,7 @@ static struct clocksource clocksource_ts
 	.shift			= 22,
 	.update_callback	= tsc_update_callback,
 	.is_continuous		= 1,
+	.list			= LIST_HEAD_INIT(clocksource_tsc.list),
 };
 
 static int tsc_update_callback(void)

-- 

[PATCH 07/12] clocksource: x86_64 initialize list value

[Daniel Walker]

[-- Attachment #1: clocksource_x86_64_list_init.patch --]
[-- Type: text/plain, Size: 1098 bytes --]

Update arch/x86_64/ with list initialization.

Signed-Off-By: Daniel Walker <dwalker@mvista.com>

---
 arch/x86_64/kernel/hpet.c |    1 +
 arch/x86_64/kernel/tsc.c  |    1 +
 2 files changed, 2 insertions(+)

Index: linux-2.6.19/arch/x86_64/kernel/hpet.c
===================================================================
--- linux-2.6.19.orig/arch/x86_64/kernel/hpet.c
+++ linux-2.6.19/arch/x86_64/kernel/hpet.c
@@ -472,6 +472,7 @@ struct clocksource clocksource_hpet = {
 	.shift		= HPET_SHIFT,
 	.is_continuous	= 1,
 	.vread		= vread_hpet,
+	.list		= LIST_HEAD_INIT(clocksource_hpet.list),
 };
 
 static int __init init_hpet_clocksource(void)
Index: linux-2.6.19/arch/x86_64/kernel/tsc.c
===================================================================
--- linux-2.6.19.orig/arch/x86_64/kernel/tsc.c
+++ linux-2.6.19/arch/x86_64/kernel/tsc.c
@@ -209,6 +209,7 @@ static struct clocksource clocksource_ts
 	.update_callback	= tsc_update_callback,
 	.is_continuous		= 1,
 	.vread			= vread_tsc,
+	.list			= LIST_HEAD_INIT(clocksource_tsc.list),
 };
 
 static int tsc_update_callback(void)

-- 

[PATCH 08/12] clocksource: driver initialize list value

[Daniel Walker]

[-- Attachment #1: clocksource_drivers_list_init.patch --]
[-- Type: text/plain, Size: 1577 bytes --]

Update drivers/clocksource/ with list initialization.

Signed-Off-By: Daniel Walker <dwalker@mvista.com>

---
 drivers/clocksource/acpi_pm.c    |    1 +
 drivers/clocksource/cyclone.c    |    1 +
 drivers/clocksource/scx200_hrt.c |    1 +
 3 files changed, 3 insertions(+)

Index: linux-2.6.19/drivers/clocksource/acpi_pm.c
===================================================================
--- linux-2.6.19.orig/drivers/clocksource/acpi_pm.c
+++ linux-2.6.19/drivers/clocksource/acpi_pm.c
@@ -74,6 +74,7 @@ static struct clocksource clocksource_ac
 	.mult		= 0, /*to be caluclated*/
 	.shift		= 22,
 	.is_continuous	= 1,
+	.list		= LIST_HEAD_INIT(clocksource_acpi_pm.list),
 };
 
 
Index: linux-2.6.19/drivers/clocksource/cyclone.c
===================================================================
--- linux-2.6.19.orig/drivers/clocksource/cyclone.c
+++ linux-2.6.19/drivers/clocksource/cyclone.c
@@ -32,6 +32,7 @@ static struct clocksource clocksource_cy
 	.mult		= 10,
 	.shift		= 0,
 	.is_continuous	= 1,
+	.list		= LIST_HEAD_INIT(clocksource_cyclone.list),
 };
 
 static int __init init_cyclone_clocksource(void)
Index: linux-2.6.19/drivers/clocksource/scx200_hrt.c
===================================================================
--- linux-2.6.19.orig/drivers/clocksource/scx200_hrt.c
+++ linux-2.6.19/drivers/clocksource/scx200_hrt.c
@@ -58,6 +58,7 @@ static struct clocksource cs_hrt = {
 	.read		= read_hrt,
 	.mask		= CLOCKSOURCE_MASK(32),
 	.is_continuous	= 1,
+	.list		= LIST_HEAD_INIT(cs_hrt.list),
 	/* mult, shift are set based on mhz27 flag */
 };
 

-- 

[PATCH 09/12] clocksource: initialize list value

[Daniel Walker]

[-- Attachment #1: clocksource_list_init.patch --]
[-- Type: text/plain, Size: 1666 bytes --]

A change to clocksource initialization. If the list field is initialized 
it allows clocksource_register to complete faster since it doesn't have 
to scan the list of clocks doing strcmp on each looking for duplicates.

Signed-Off-By: Daniel Walker <dwalker@mvista.com>

---
 kernel/time/clocksource.c |    3 +--
 kernel/time/jiffies.c     |    1 +
 2 files changed, 2 insertions(+), 2 deletions(-)

Index: linux-2.6.19/kernel/time/clocksource.c
===================================================================
--- linux-2.6.19.orig/kernel/time/clocksource.c
+++ linux-2.6.19/kernel/time/clocksource.c
@@ -186,12 +186,11 @@ int clocksource_register(struct clocksou
 	unsigned long flags;
 
 	spin_lock_irqsave(&clocksource_lock, flags);
-	if (unlikely(!list_empty(&c->list) && __is_registered(c->name))) {
+	if (unlikely(!list_empty(&c->list))) {
 		printk("register_clocksource: Cannot register %s clocksource. "
 		       "Already registered!", c->name);
 		ret = -EBUSY;
 	} else {
-		INIT_LIST_HEAD(&c->list);
  		__sorted_list_add(c);
 		/* scan the registered clocksources, and pick the best one */
 		next_clocksource = select_clocksource();
Index: linux-2.6.19/kernel/time/jiffies.c
===================================================================
--- linux-2.6.19.orig/kernel/time/jiffies.c
+++ linux-2.6.19/kernel/time/jiffies.c
@@ -63,6 +63,7 @@ struct clocksource clocksource_jiffies =
 	.mult		= NSEC_PER_JIFFY << JIFFIES_SHIFT, /* details above */
 	.shift		= JIFFIES_SHIFT,
 	.is_continuous	= 0, /* tick based, not free running */
+	.list		= LIST_HEAD_INIT(clocksource_jiffies.list),
 };
 
 static int __init init_jiffies_clocksource(void)

-- 

[PATCH 10/12] clocksource: add block notifier

[Daniel Walker]

[-- Attachment #1: clocksource_add_block_notify_on_new_clock.patch --]
[-- Type: text/plain, Size: 3606 bytes --]

Adds a call back interface for register/rating change events. This is also used
later in this series to signal other interesting events.


Signed-Off-By: Daniel Walker <dwalker@mvista.com>

---
 include/linux/clocksource.h |   37 +++++++++++++++++++++++++++++++++++++
 include/linux/timekeeping.h |    3 +++
 kernel/time/clocksource.c   |   10 ++++++++++
 3 files changed, 50 insertions(+)

Index: linux-2.6.19/include/linux/clocksource.h
===================================================================
--- linux-2.6.19.orig/include/linux/clocksource.h
+++ linux-2.6.19/include/linux/clocksource.h
@@ -14,6 +14,7 @@
 #include <linux/list.h>
 #include <linux/plist.h>
 #include <linux/sysdev.h>
+#include <linux/notifier.h>
 
 #include <asm/div64.h>
 #include <asm/io.h>
@@ -24,6 +25,42 @@ typedef u64 cycle_t;
 /* XXX - Would like a better way for initializing curr_clocksource */
 extern struct clocksource clocksource_jiffies;
 
+/*
+ * Allows inlined calling for notifier routines.
+ */
+extern struct atomic_notifier_head clocksource_list_notifier;
+
+/*
+ * Block notifier flags.
+ */
+#define CLOCKSOURCE_NOTIFY_REGISTER	1
+#define CLOCKSOURCE_NOTIFY_RATING	2
+#define CLOCKSOURCE_NOTIFY_FREQ		4
+
+/**
+ * clocksource_notifier_register - Registers a list change notifier
+ * @nb:		pointer to a notifier block
+ *
+ * Returns zero always.
+ */
+static inline int clocksource_notifier_register(struct notifier_block *nb)
+{
+	return atomic_notifier_chain_register(&clocksource_list_notifier, nb);
+}
+
+/**
+ * clocksource_freq_change - Allows notification of dynamic frequency changes.
+ *
+ * Signals that a clocksource is dynamically changing it's frequency.
+ * This could happen if a clocksource becomes more/less stable.
+ */
+static inline void clocksource_freq_change(struct clocksource *c)
+{
+	atomic_notifier_call_chain(&clocksource_list_notifier,
+				   CLOCKSOURCE_NOTIFY_FREQ, c);
+}
+
+
 /**
  * struct clocksource - hardware abstraction for a free running counter
  *	Provides mostly state-free accessors to the underlying hardware.
Index: linux-2.6.19/include/linux/timekeeping.h
===================================================================
--- linux-2.6.19.orig/include/linux/timekeeping.h
+++ linux-2.6.19/include/linux/timekeeping.h
@@ -14,6 +14,9 @@ static inline int change_clocksource(voi
 {
 	return 0;
 }
+
+static inline void change_clocksource(void) { }
+
 #endif /* !CONFIG_GENERIC_TIME */
 
 #endif /* _LINUX_TIMEKEEPING_H */
Index: linux-2.6.19/kernel/time/clocksource.c
===================================================================
--- linux-2.6.19.orig/kernel/time/clocksource.c
+++ linux-2.6.19/kernel/time/clocksource.c
@@ -49,6 +49,8 @@ static DEFINE_SPINLOCK(clocksource_lock)
 static char override_name[32];
 static int finished_booting;
 
+ATOMIC_NOTIFIER_HEAD(clocksource_list_notifier);
+
 /* clocksource_done_booting - Called near the end of bootup
  *
  * Hack to avoid lots of clocksource churn at boot time
@@ -196,6 +198,10 @@ int clocksource_register(struct clocksou
 		next_clocksource = select_clocksource();
 	}
 	spin_unlock_irqrestore(&clocksource_lock, flags);
+
+	atomic_notifier_call_chain(&clocksource_list_notifier,
+ 				   CLOCKSOURCE_NOTIFY_REGISTER, c);
+
 	return ret;
 }
 EXPORT_SYMBOL(clocksource_register);
@@ -224,6 +230,10 @@ void clocksource_rating_change(struct cl
 
 	next_clocksource = select_clocksource();
 	spin_unlock_irqrestore(&clocksource_lock, flags);
+
+	atomic_notifier_call_chain(&clocksource_list_notifier,
+				   CLOCKSOURCE_NOTIFY_RATING, c);
+
 }
 EXPORT_SYMBOL(clocksource_rating_change);
 

-- 

[PATCH 11/12] clocksource: remove update_callback

[Daniel Walker]

[-- Attachment #1: clocksource_remove_update_callback.patch --]
[-- Type: text/plain, Size: 7213 bytes --]

Uses the block notifier to replace the functionality of update_callback().
update_callback() was a special case specifically for the tsc, but including
it in the clocksource structure duplicated it needlessly for other clocks.

Signed-Off-By: Daniel Walker <dwalker@mvista.com>

---
 arch/i386/kernel/tsc.c      |   35 ++++++++++-------------------------
 arch/x86_64/kernel/tsc.c    |   19 +++++--------------
 include/linux/clocksource.h |    2 --
 include/linux/timekeeping.h |    1 +
 kernel/time/timekeeping.c   |   32 ++++++++++++++++++++++++++++++--
 5 files changed, 46 insertions(+), 43 deletions(-)

Index: linux-2.6.19/arch/i386/kernel/tsc.c
===================================================================
--- linux-2.6.19.orig/arch/i386/kernel/tsc.c
+++ linux-2.6.19/arch/i386/kernel/tsc.c
@@ -51,8 +51,7 @@ static int __init tsc_setup(char *str)
 __setup("notsc", tsc_setup);
 
 /*
- * code to mark and check if the TSC is unstable
- * due to cpufreq or due to unsynced TSCs
+ * Flag that denotes an unstable tsc and check function.
  */
 static int tsc_unstable;
 
@@ -61,12 +60,6 @@ static inline int check_tsc_unstable(voi
 	return tsc_unstable;
 }
 
-void mark_tsc_unstable(void)
-{
-	tsc_unstable = 1;
-}
-EXPORT_SYMBOL_GPL(mark_tsc_unstable);
-
 /* Accellerators for sched_clock()
  * convert from cycles(64bits) => nanoseconds (64bits)
  *  basic equation:
@@ -180,6 +173,7 @@ int recalibrate_cpu_khz(void)
 	if (cpu_has_tsc) {
 		cpu_khz = calculate_cpu_khz();
 		tsc_khz = cpu_khz;
+		mark_tsc_unstable();
 		cpu_data[0].loops_per_jiffy =
 			cpufreq_scale(cpu_data[0].loops_per_jiffy,
 					cpu_khz_old, cpu_khz);
@@ -332,7 +326,6 @@ core_initcall(cpufreq_tsc);
 /* clock source code */
 
 static unsigned long current_tsc_khz = 0;
-static int tsc_update_callback(void);
 
 static cycle_t read_tsc(void)
 {
@@ -350,32 +343,24 @@ static struct clocksource clocksource_ts
 	.mask			= CLOCKSOURCE_MASK(64),
 	.mult			= 0, /* to be set */
 	.shift			= 22,
-	.update_callback	= tsc_update_callback,
 	.is_continuous		= 1,
 	.list			= LIST_HEAD_INIT(clocksource_tsc.list),
 };
 
-static int tsc_update_callback(void)
+/*
+ * code to mark if the TSC is unstable
+ * due to cpufreq or due to unsynced TSCs
+ */
+void mark_tsc_unstable(void)
 {
-	int change = 0;
-
 	/* check to see if we should switch to the safe clocksource: */
-	if (clocksource_tsc.rating != 0 && check_tsc_unstable()) {
+	if (unlikely(!tsc_unstable && clocksource_tsc.rating != 0)) {
 		clocksource_tsc.rating = 0;
 		clocksource_rating_change(&clocksource_tsc);
-		change = 1;
-	}
-
-	/* only update if tsc_khz has changed: */
-	if (current_tsc_khz != tsc_khz) {
-		current_tsc_khz = tsc_khz;
-		clocksource_tsc.mult = clocksource_khz2mult(current_tsc_khz,
-							clocksource_tsc.shift);
-		change = 1;
 	}
-
-	return change;
+	tsc_unstable = 1;
 }
+EXPORT_SYMBOL_GPL(mark_tsc_unstable);
 
 static int __init dmi_mark_tsc_unstable(struct dmi_system_id *d)
 {
Index: linux-2.6.19/arch/x86_64/kernel/tsc.c
===================================================================
--- linux-2.6.19.orig/arch/x86_64/kernel/tsc.c
+++ linux-2.6.19/arch/x86_64/kernel/tsc.c
@@ -47,11 +47,6 @@ static inline int check_tsc_unstable(voi
 	return tsc_unstable;
 }
 
-void mark_tsc_unstable(void)
-{
-	tsc_unstable = 1;
-}
-EXPORT_SYMBOL_GPL(mark_tsc_unstable);
 
 #ifdef CONFIG_CPU_FREQ
 
@@ -185,8 +180,6 @@ __setup("notsc", notsc_setup);
 
 /* clock source code: */
 
-static int tsc_update_callback(void);
-
 static cycle_t read_tsc(void)
 {
 	cycle_t ret = (cycle_t)get_cycles_sync();
@@ -206,24 +199,22 @@ static struct clocksource clocksource_ts
 	.mask			= (cycle_t)-1,
 	.mult			= 0, /* to be set */
 	.shift			= 22,
-	.update_callback	= tsc_update_callback,
 	.is_continuous		= 1,
 	.vread			= vread_tsc,
 	.list			= LIST_HEAD_INIT(clocksource_tsc.list),
 };
 
-static int tsc_update_callback(void)
+void mark_tsc_unstable(void)
 {
-	int change = 0;
-
 	/* check to see if we should switch to the safe clocksource: */
-	if (clocksource_tsc.rating != 50 && check_tsc_unstable()) {
+	if (unlikely(!tsc_unstable && clocksource_tsc.rating != 50)) {
 		clocksource_tsc.rating = 50;
 		clocksource_rating_change(&clocksource_tsc);
-		change = 1;
 	}
-	return change;
+
+	tsc_unstable = 1;
 }
+EXPORT_SYMBOL_GPL(mark_tsc_unstable);
 
 static int __init init_tsc_clocksource(void)
 {
Index: linux-2.6.19/include/linux/clocksource.h
===================================================================
--- linux-2.6.19.orig/include/linux/clocksource.h
+++ linux-2.6.19/include/linux/clocksource.h
@@ -87,7 +87,6 @@ static inline void clocksource_freq_chan
  *			subtraction of non 64 bit counters
  * @mult:		cycle to nanosecond multiplier
  * @shift:		cycle to nanosecond divisor (power of two)
- * @update_callback:	called when safe to alter clocksource values
  * @is_continuous:	defines if clocksource is free-running.
  * @vread:		vsyscall based read
  * @cycle_interval:	Used internally by timekeeping core, please ignore.
@@ -101,7 +100,6 @@ struct clocksource {
 	cycle_t mask;
 	u32 mult;
 	u32 shift;
-	int (*update_callback)(void);
 	int is_continuous;
 	cycle_t (*vread)(void);
 
Index: linux-2.6.19/include/linux/timekeeping.h
===================================================================
--- linux-2.6.19.orig/include/linux/timekeeping.h
+++ linux-2.6.19/include/linux/timekeeping.h
@@ -16,6 +16,7 @@ static inline int change_clocksource(voi
 }
 
 static inline void change_clocksource(void) { }
+static inline void timekeeping_init_notifier(void) { }
 
 #endif /* !CONFIG_GENERIC_TIME */
 
Index: linux-2.6.19/kernel/time/timekeeping.c
===================================================================
--- linux-2.6.19.orig/kernel/time/timekeeping.c
+++ linux-2.6.19/kernel/time/timekeeping.c
@@ -19,6 +19,7 @@ static unsigned long timekeeping_suspend
  * Clock used for timekeeping
  */
 struct clocksource *clock = &clocksource_jiffies;
+atomic_t clock_recalc_interval = ATOMIC_INIT(0);
 
 /*
  * The current time
@@ -169,8 +170,9 @@ static int change_clocksource(void)
 		printk(KERN_INFO "Time: %s clocksource has been installed.\n",
 		       clock->name);
 		return 1;
-	} else if (clock->update_callback) {
-		return clock->update_callback();
+	} else if (unlikely(atomic_read(&clock_recalc_interval))) {
+		atomic_set(&clock_recalc_interval, 0);
+		return 1;
 	}
 	return 0;
 }
@@ -192,6 +194,29 @@ int timekeeping_is_continuous(void)
 
 	return ret;
 }
+
+static int
+clocksource_callback(struct notifier_block *nb, unsigned long op, void *c)
+{
+	if (c == clock && op == CLOCKSOURCE_NOTIFY_FREQ &&
+	    !atomic_read(&clock_recalc_interval))
+		atomic_inc(&clock_recalc_interval);
+
+	return 0;
+}
+
+static struct notifier_block clocksource_nb = {
+	.notifier_call = clocksource_callback,
+};
+
+void __init timekeeping_init_notifier(void)
+{
+	/*
+	 * Needs to be done early, incase a clock is unstable
+	 * early.
+	 */
+	clocksource_notifier_register(&clocksource_nb);
+}
 #endif /* CONFIG_GENERIC_TIME */
 
 /**
@@ -230,6 +255,9 @@ void __init timekeeping_init(void)
 		-xtime.tv_sec, -xtime.tv_nsec);
 
 	write_sequnlock_irqrestore(&xtime_lock, flags);
+
+	timekeeping_init_notifier();
+
 }
 
 

-- 

[PATCH 12/12] clocksource: atomic signals

[Daniel Walker]

[-- Attachment #1: clocksource_add_atomic_singals.patch --]
[-- Type: text/plain, Size: 5815 bytes --]

Modifies the way clocks are switched to in the timekeeping code. The original
code would constantly monitor the clocksource list checking for newly added
clocksources. I modified this by using atomic types to signal when a new clock
is added. This allows the operation to be used only when it's needed.

The fast path is also reduced to checking a single atomic value.

Signed-Off-By: Daniel Walker <dwalker@mvista.com>

---
 include/linux/clocksource.h |    5 ++++
 include/linux/timekeeping.h |   10 --------
 kernel/time/clocksource.c   |    6 +++++
 kernel/time/timekeeping.c   |   51 +++++++++++++++++++++++++++++++-------------
 4 files changed, 49 insertions(+), 23 deletions(-)

Index: linux-2.6.19/include/linux/clocksource.h
===================================================================
--- linux-2.6.19.orig/include/linux/clocksource.h
+++ linux-2.6.19/include/linux/clocksource.h
@@ -26,6 +26,11 @@ typedef u64 cycle_t;
 extern struct clocksource clocksource_jiffies;
 
 /*
+ * Atomic signal that is specific to timekeeping.
+ */
+extern atomic_t clock_check;
+
+/*
  * Allows inlined calling for notifier routines.
  */
 extern struct atomic_notifier_head clocksource_list_notifier;
Index: linux-2.6.19/include/linux/timekeeping.h
===================================================================
--- linux-2.6.19.orig/include/linux/timekeeping.h
+++ linux-2.6.19/include/linux/timekeeping.h
@@ -5,15 +5,7 @@
 
 extern void update_wall_time(void);
 
-#ifdef CONFIG_GENERIC_TIME
-
-extern struct clocksource *clock;
-
-#else /* CONFIG_GENERIC_TIME */
-static inline int change_clocksource(void)
-{
-	return 0;
-}
+#ifndef CONFIG_GENERIC_TIME
 
 static inline void change_clocksource(void) { }
 static inline void timekeeping_init_notifier(void) { }
Index: linux-2.6.19/kernel/time/clocksource.c
===================================================================
--- linux-2.6.19.orig/kernel/time/clocksource.c
+++ linux-2.6.19/kernel/time/clocksource.c
@@ -50,6 +50,7 @@ static char override_name[32];
 static int finished_booting;
 
 ATOMIC_NOTIFIER_HEAD(clocksource_list_notifier);
+atomic_t clock_check = ATOMIC_INIT(0);
 
 /* clocksource_done_booting - Called near the end of bootup
  *
@@ -58,6 +59,8 @@ ATOMIC_NOTIFIER_HEAD(clocksource_list_no
 static int __init clocksource_done_booting(void)
 {
 	finished_booting = 1;
+	/* Check for a new clock now */
+	atomic_inc(&clock_check);
 	return 0;
 }
 
@@ -291,6 +294,9 @@ static ssize_t sysfs_override_clocksourc
 	/* try to select it: */
 	next_clocksource = select_clocksource();
 
+	/* Signal that there is a new clocksource */
+	atomic_inc(&clock_check);
+
 	spin_unlock_irq(&clocksource_lock);
 
 	return ret;
Index: linux-2.6.19/kernel/time/timekeeping.c
===================================================================
--- linux-2.6.19.orig/kernel/time/timekeeping.c
+++ linux-2.6.19/kernel/time/timekeeping.c
@@ -3,6 +3,7 @@
 #include <linux/module.h>
 #include <linux/timekeeping.h>
 #include <linux/time.h>
+#include <linux/timex.h>
 #include <linux/hrtimer.h>
 
 /*
@@ -19,7 +20,6 @@ static unsigned long timekeeping_suspend
  * Clock used for timekeeping
  */
 struct clocksource *clock = &clocksource_jiffies;
-atomic_t clock_recalc_interval = ATOMIC_INIT(0);
 
 /*
  * The current time
@@ -150,11 +150,12 @@ int do_settimeofday(struct timespec *tv)
 EXPORT_SYMBOL(do_settimeofday);
 
 /**
- * change_clocksource - Swaps clocksources if a new one is available
+ * timkeeping_change_clocksource - Swaps clocksources if a new one is available
  *
  * Accumulates current time interval and initializes new clocksource
+ * Needs to be called with the xtime_lock held.
  */
-static int change_clocksource(void)
+static int timekeeping_change_clocksource(void)
 {
 	struct clocksource *new;
 	cycle_t now;
@@ -169,9 +170,15 @@ static int change_clocksource(void)
 		clock->cycle_last = now;
 		printk(KERN_INFO "Time: %s clocksource has been installed.\n",
 		       clock->name);
+		hrtimer_clock_notify();
+		clock->error = 0;
+		clock->xtime_nsec = 0;
+		clocksource_calculate_interval(clock, tick_nsec);
 		return 1;
-	} else if (unlikely(atomic_read(&clock_recalc_interval))) {
-		atomic_set(&clock_recalc_interval, 0);
+	} else {
+		clock->error = 0;
+		clock->xtime_nsec = 0;
+		clocksource_calculate_interval(clock, tick_nsec);
 		return 1;
 	}
 	return 0;
@@ -198,9 +205,14 @@ int timekeeping_is_continuous(void)
 static int
 clocksource_callback(struct notifier_block *nb, unsigned long op, void *c)
 {
-	if (c == clock && op == CLOCKSOURCE_NOTIFY_FREQ &&
-	    !atomic_read(&clock_recalc_interval))
-		atomic_inc(&clock_recalc_interval);
+	if (likely(c != clock))
+		return 0;
+
+	switch (op) {
+		case CLOCKSOURCE_NOTIFY_FREQ:
+		case CLOCKSOURCE_NOTIFY_RATING:
+			atomic_inc(&clock_check);
+	}
 
 	return 0;
 }
@@ -325,6 +337,13 @@ static int __init timekeeping_init_devic
 	int error = sysdev_class_register(&timekeeping_sysclass);
 	if (!error)
 		error = sysdev_register(&device_timer);
+
+#ifdef CONFIG_GENERIC_TIME
+	/*
+	 * Now we signal to switch to a high-res clock.
+	 */
+	atomic_inc(&clock_check);
+#endif
 	return error;
 }
 
@@ -467,13 +486,17 @@ void update_wall_time(void)
 	xtime.tv_nsec = (s64)clock->xtime_nsec >> clock->shift;
 	clock->xtime_nsec -= (s64)xtime.tv_nsec << clock->shift;
 
-	/* check to see if there is a new clocksource to use */
-	if (change_clocksource()) {
-		clock->error = 0;
-		clock->xtime_nsec = 0;
-		hrtimer_clock_notify();
-		clocksource_calculate_interval(clock, tick_nsec);
+#ifdef CONFIG_GENERIC_TIME
+	/*
+	 * check to see if there is a new clocksource to use
+	 */
+	if (unlikely(atomic_read(&clock_check))) {
+
+		timekeeping_change_clocksource();
+
+		atomic_set(&clock_check, 0);
 	}
+#endif /* CONFIG_GENERIC_TIME */
 	update_vsyscall(&xtime, clock);
 }
 

--