From: Martin Schwidefsky update_wall_time calls change_clocksource HZ times per second to check if a new clock source is available. In close to 100% of all calls there is no new clock. Replace the tick based check by an update done with stop_machine. Cc: Ingo Molnar Cc: Thomas Gleixner Cc: john stultz Cc: Daniel Walker Signed-off-by: Martin Schwidefsky --- include/linux/clocksource.h | 2 kernel/time/clocksource.c | 112 +++++++++++++++++--------------------------- kernel/time/timekeeping.c | 41 ++++++++++------ 3 files changed, 72 insertions(+), 83 deletions(-) Index: linux-2.6/include/linux/clocksource.h =================================================================== --- linux-2.6.orig/include/linux/clocksource.h +++ linux-2.6/include/linux/clocksource.h @@ -291,4 +291,6 @@ static inline void update_vsyscall_tz(vo } #endif +extern void timekeeping_notify(struct clocksource *clock); + #endif /* _LINUX_CLOCKSOURCE_H */ Index: linux-2.6/kernel/time/clocksource.c =================================================================== --- linux-2.6.orig/kernel/time/clocksource.c +++ linux-2.6/kernel/time/clocksource.c @@ -109,35 +109,17 @@ EXPORT_SYMBOL(timecounter_cyc2time); /*[Clocksource internal variables]--------- * curr_clocksource: * currently selected clocksource. - * next_clocksource: - * pending next selected clocksource. * clocksource_list: * linked list with the registered clocksources - * clocksource_lock: - * protects manipulations to curr_clocksource and next_clocksource - * and the clocksource_list + * clocksource_mutex: + * protects manipulations to curr_clocksource and the clocksource_list * override_name: * Name of the user-specified clocksource. */ static struct clocksource *curr_clocksource; -static struct clocksource *next_clocksource; static LIST_HEAD(clocksource_list); -static DEFINE_SPINLOCK(clocksource_lock); +static DEFINE_MUTEX(clocksource_mutex); static char override_name[32]; -static int finished_booting; - -/* clocksource_done_booting - Called near the end of core bootup - * - * Hack to avoid lots of clocksource churn at boot time. - * We use fs_initcall because we want this to start before - * device_initcall but after subsys_initcall. - */ -static int __init clocksource_done_booting(void) -{ - finished_booting = 1; - return 0; -} -fs_initcall(clocksource_done_booting); #ifdef CONFIG_CLOCKSOURCE_WATCHDOG static LIST_HEAD(watchdog_list); @@ -356,18 +338,16 @@ static inline void clocksource_resume_wa void clocksource_resume(void) { struct clocksource *cs; - unsigned long flags; - spin_lock_irqsave(&clocksource_lock, flags); + mutex_lock(&clocksource_mutex); - list_for_each_entry(cs, &clocksource_list, list) { + list_for_each_entry(cs, &clocksource_list, list) if (cs->resume) cs->resume(); - } clocksource_resume_watchdog(); - spin_unlock_irqrestore(&clocksource_lock, flags); + mutex_unlock(&clocksource_mutex); } /** @@ -383,28 +363,13 @@ void clocksource_touch_watchdog(void) } #ifdef CONFIG_GENERIC_TIME -/** - * clocksource_get_next - Returns the selected clocksource - * - */ -struct clocksource *clocksource_get_next(void) -{ - unsigned long flags; - spin_lock_irqsave(&clocksource_lock, flags); - if (next_clocksource && finished_booting) { - curr_clocksource = next_clocksource; - next_clocksource = NULL; - } - spin_unlock_irqrestore(&clocksource_lock, flags); - - return curr_clocksource; -} +static int finished_booting; /** * clocksource_select - Select the best clocksource available * - * Private function. Must hold clocksource_lock when called. + * Private function. Must hold clocksource_mutex when called. * * Select the clocksource with the best rating, or the clocksource, * which is selected by userspace override. @@ -413,7 +378,7 @@ static void clocksource_select(void) { struct clocksource *best, *cs; - if (list_empty(&clocksource_list)) + if (!finished_booting || list_empty(&clocksource_list)) return; /* First clocksource on the list has the best rating. */ best = list_first_entry(&clocksource_list, struct clocksource, list); @@ -438,13 +403,31 @@ static void clocksource_select(void) best = cs; break; } - if (curr_clocksource != best) - next_clocksource = best; + if (curr_clocksource != best) { + printk(KERN_INFO "Switching to clocksource %s\n", best->name); + curr_clocksource = best; + timekeeping_notify(curr_clocksource); + } } +/* + * clocksource_done_booting - Called near the end of core bootup + * + * Hack to avoid lots of clocksource churn at boot time. + * We use fs_initcall because we want this to start before + * device_initcall but after subsys_initcall. + */ +static int __init clocksource_done_booting(void) +{ + finished_booting = 1; + clocksource_select(); + return 0; +} +fs_initcall(clocksource_done_booting); + #else /* CONFIG_GENERIC_TIME */ -static void clocksource_select(void) { } +static inline void clocksource_select(void) { } #endif @@ -471,13 +454,11 @@ static void clocksource_enqueue(struct c */ int clocksource_register(struct clocksource *cs) { - unsigned long flags; - - spin_lock_irqsave(&clocksource_lock, flags); + mutex_lock(&clocksource_mutex); clocksource_enqueue(cs); clocksource_select(); - spin_unlock_irqrestore(&clocksource_lock, flags); clocksource_enqueue_watchdog(cs); + mutex_unlock(&clocksource_mutex); return 0; } EXPORT_SYMBOL(clocksource_register); @@ -487,14 +468,12 @@ EXPORT_SYMBOL(clocksource_register); */ void clocksource_change_rating(struct clocksource *cs, int rating) { - unsigned long flags; - - spin_lock_irqsave(&clocksource_lock, flags); + mutex_lock(&clocksource_mutex); list_del(&cs->list); cs->rating = rating; clocksource_enqueue(cs); clocksource_select(); - spin_unlock_irqrestore(&clocksource_lock, flags); + mutex_unlock(&clocksource_mutex); } EXPORT_SYMBOL(clocksource_change_rating); @@ -503,13 +482,11 @@ EXPORT_SYMBOL(clocksource_change_rating) */ void clocksource_unregister(struct clocksource *cs) { - unsigned long flags; - + mutex_lock(&clocksource_mutex); clocksource_dequeue_watchdog(cs); - spin_lock_irqsave(&clocksource_lock, flags); list_del(&cs->list); clocksource_select(); - spin_unlock_irqrestore(&clocksource_lock, flags); + mutex_unlock(&clocksource_mutex); } EXPORT_SYMBOL(clocksource_unregister); @@ -527,9 +504,9 @@ sysfs_show_current_clocksources(struct s { ssize_t count = 0; - spin_lock_irq(&clocksource_lock); + mutex_lock(&clocksource_mutex); count = snprintf(buf, PAGE_SIZE, "%s\n", curr_clocksource->name); - spin_unlock_irq(&clocksource_lock); + mutex_unlock(&clocksource_mutex); return count; } @@ -555,14 +532,14 @@ static ssize_t sysfs_override_clocksourc if (buf[count-1] == '\n') count--; - spin_lock_irq(&clocksource_lock); + mutex_lock(&clocksource_mutex); if (count > 0) memcpy(override_name, buf, count); override_name[count] = 0; clocksource_select(); - spin_unlock_irq(&clocksource_lock); + mutex_unlock(&clocksource_mutex); return count; } @@ -582,7 +559,7 @@ sysfs_show_available_clocksources(struct struct clocksource *src; ssize_t count = 0; - spin_lock_irq(&clocksource_lock); + mutex_lock(&clocksource_mutex); list_for_each_entry(src, &clocksource_list, list) { /* * Don't show non-HRES clocksource if the tick code is @@ -594,7 +571,7 @@ sysfs_show_available_clocksources(struct max((ssize_t)PAGE_SIZE - count, (ssize_t)0), "%s ", src->name); } - spin_unlock_irq(&clocksource_lock); + mutex_unlock(&clocksource_mutex); count += snprintf(buf + count, max((ssize_t)PAGE_SIZE - count, (ssize_t)0), "\n"); @@ -649,11 +626,10 @@ device_initcall(init_clocksource_sysfs); */ static int __init boot_override_clocksource(char* str) { - unsigned long flags; - spin_lock_irqsave(&clocksource_lock, flags); + mutex_lock(&clocksource_mutex); if (str) strlcpy(override_name, str, sizeof(override_name)); - spin_unlock_irqrestore(&clocksource_lock, flags); + mutex_unlock(&clocksource_mutex); return 1; } Index: linux-2.6/kernel/time/timekeeping.c =================================================================== --- linux-2.6.orig/kernel/time/timekeeping.c +++ linux-2.6/kernel/time/timekeeping.c @@ -18,6 +18,7 @@ #include #include #include +#include /* Structure holding internal timekeeping values. */ struct timekeeper { @@ -179,6 +180,7 @@ void timekeeping_leap_insert(int leapsec } #ifdef CONFIG_GENERIC_TIME + /** * timekeeping_forward_now - update clock to the current time * @@ -351,31 +353,40 @@ EXPORT_SYMBOL(do_settimeofday); * * Accumulates current time interval and initializes new clocksource */ -static void change_clocksource(void) +static int change_clocksource(void *data) { struct clocksource *new, *old; - new = clocksource_get_next(); - - if (!new || timekeeper.clock == new) - return; + new = (struct clocksource *) data; timekeeping_forward_now(); + if (!new->enable || new->enable(new) == 0) { + old = timekeeper.clock; + timekeeper_setup_internals(new); + if (old->disable) + old->disable(old); + } + return 0; +} - if (new->enable && ! new->enable(new)) +/** + * timekeeping_notify - Install a new clock source + * @clock: pointer to the clock source + * + * This function is called from clocksource.c after a new, better clock + * source has been registered. The caller holds the clocksource_mutex. + */ +void timekeeping_notify(struct clocksource *clock) +{ + if (timekeeper.clock == clock) return; - - old = timekeeper.clock; - timekeeper_setup_internals(new); - - if (old->disable) - old->disable(old); - + stop_machine(change_clocksource, clock, NULL); tick_clock_notify(); } + #else /* GENERIC_TIME */ + static inline void timekeeping_forward_now(void) { } -static inline void change_clocksource(void) { } /** * ktime_get - get the monotonic time in ktime_t format @@ -416,6 +427,7 @@ void ktime_get_ts(struct timespec *ts) ts->tv_nsec + tomono.tv_nsec); } EXPORT_SYMBOL_GPL(ktime_get_ts); + #endif /* !GENERIC_TIME */ /** @@ -772,7 +784,6 @@ void update_wall_time(void) update_xtime_cache(nsecs); /* check to see if there is a new clocksource to use */ - change_clocksource(); update_vsyscall(&xtime, timekeeper.clock); } -- blue skies, Martin. "Reality continues to ruin my life." - Calvin. -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/