From 5d33b883aed81c6fbcd09c6f7c3619eee850a7e2 Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Thu, 25 Apr 2013 20:31:43 +0000 Subject: clocksource: Always verify highres capability If a clocksource has a (wrong) high rating, but can't be used as a timebase for oneshot tick mode, it is unconditionally selected even when the system is already in oneshot tick mode. This causes full system failure. Verify the clocksource selection against the oneshot mode. Signed-off-by: Thomas Gleixner Acked-by: John Stultz Cc: Magnus Damm Link: http://lkml.kernel.org/r/20130425143435.635040849@linutronix.de Signed-off-by: Thomas Gleixner --- kernel/time/clocksource.c | 31 ++++++++++++++++++++++++++----- 1 file changed, 26 insertions(+), 5 deletions(-) (limited to 'kernel/time/clocksource.c') diff --git a/kernel/time/clocksource.c b/kernel/time/clocksource.c index c9583382141a..dda5c7130d93 100644 --- a/kernel/time/clocksource.c +++ b/kernel/time/clocksource.c @@ -553,6 +553,26 @@ static u64 clocksource_max_deferment(struct clocksource *cs) #ifndef CONFIG_ARCH_USES_GETTIMEOFFSET +static struct clocksource *clocksource_find_best(bool oneshot) +{ + struct clocksource *cs; + + if (!finished_booting || list_empty(&clocksource_list)) + return NULL; + + /* + * We pick the clocksource with the highest rating. If oneshot + * mode is active, we pick the highres valid clocksource with + * the best rating. + */ + list_for_each_entry(cs, &clocksource_list, list) { + if (oneshot && !(cs->flags & CLOCK_SOURCE_VALID_FOR_HRES)) + continue; + return cs; + } + return NULL; +} + /** * clocksource_select - Select the best clocksource available * @@ -563,12 +583,14 @@ static u64 clocksource_max_deferment(struct clocksource *cs) */ static void clocksource_select(void) { + bool oneshot = tick_oneshot_mode_active(); struct clocksource *best, *cs; - if (!finished_booting || list_empty(&clocksource_list)) + /* Find the best suitable clocksource */ + best = clocksource_find_best(oneshot); + if (!best) return; - /* First clocksource on the list has the best rating. */ - best = list_first_entry(&clocksource_list, struct clocksource, list); + /* Check for the override clocksource. */ list_for_each_entry(cs, &clocksource_list, list) { if (strcmp(cs->name, override_name) != 0) @@ -578,8 +600,7 @@ static void clocksource_select(void) * capable clocksource if the tick code is in oneshot * mode (highres or nohz) */ - if (!(cs->flags & CLOCK_SOURCE_VALID_FOR_HRES) && - tick_oneshot_mode_active()) { + if (!(cs->flags & CLOCK_SOURCE_VALID_FOR_HRES) && oneshot) { /* Override clocksource cannot be used. */ printk(KERN_WARNING "Override clocksource %s is not " "HRT compatible. Cannot switch while in " -- cgit v1.2.3 From ba919d1caa2e624eb8c6cae1f2ce0a253e697d45 Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Thu, 25 Apr 2013 20:31:44 +0000 Subject: clocksource: Let timekeeping_notify return success/error timekeeping_notify() can fail due cs->enable() failure. Though the caller does not notice and happily keeps the wrong clocksource as the current one. Let the caller know about failure, so the current clocksource will be shown correctly in sysfs. Signed-off-by: Thomas Gleixner Acked-by: John Stultz Cc: Magnus Damm Link: http://lkml.kernel.org/r/20130425143435.696321912@linutronix.de Signed-off-by: Thomas Gleixner --- include/linux/clocksource.h | 2 +- kernel/time/clocksource.c | 6 +++--- kernel/time/timekeeping.c | 5 +++-- 3 files changed, 7 insertions(+), 6 deletions(-) (limited to 'kernel/time/clocksource.c') diff --git a/include/linux/clocksource.h b/include/linux/clocksource.h index 7279b94c01da..aa6ba44e75d5 100644 --- a/include/linux/clocksource.h +++ b/include/linux/clocksource.h @@ -321,7 +321,7 @@ static inline void __clocksource_updatefreq_khz(struct clocksource *cs, u32 khz) } -extern void timekeeping_notify(struct clocksource *clock); +extern int timekeeping_notify(struct clocksource *clock); extern cycle_t clocksource_mmio_readl_up(struct clocksource *); extern cycle_t clocksource_mmio_readl_down(struct clocksource *); diff --git a/kernel/time/clocksource.c b/kernel/time/clocksource.c index dda5c7130d93..1923a340bd91 100644 --- a/kernel/time/clocksource.c +++ b/kernel/time/clocksource.c @@ -611,10 +611,10 @@ static void clocksource_select(void) best = cs; break; } - if (curr_clocksource != best) { - printk(KERN_INFO "Switching to clocksource %s\n", best->name); + + if (curr_clocksource != best && !timekeeping_notify(best)) { + pr_info("Switched to clocksource %s\n", best->name); curr_clocksource = best; - timekeeping_notify(curr_clocksource); } } diff --git a/kernel/time/timekeeping.c b/kernel/time/timekeeping.c index 98cd470bbe49..da6e10c7a378 100644 --- a/kernel/time/timekeeping.c +++ b/kernel/time/timekeeping.c @@ -648,14 +648,15 @@ static int change_clocksource(void *data) * 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) +int timekeeping_notify(struct clocksource *clock) { struct timekeeper *tk = &timekeeper; if (tk->clock == clock) - return; + return 0; stop_machine(change_clocksource, clock, NULL); tick_clock_notify(); + return tk->clock == clock ? 0 : -1; } /** -- cgit v1.2.3 From f5a2e34375a5e2b711aea488ac3ae50eeba6d57c Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Thu, 25 Apr 2013 20:31:45 +0000 Subject: clocksource: Allow clocksource select to skip current clocksource Preparatory patch for clocksource unbind support. Split out code from clocksource_select and modify it, so it skips the current clocksource on request and tries to find a fallback clocksource. Convert all existing users. No functional change. Signed-off-by: Thomas Gleixner Cc: John Stultz Cc: Magnus Damm Link: http://lkml.kernel.org/r/20130425143435.834965397@linutronix.de Signed-off-by: Thomas Gleixner --- kernel/time/clocksource.c | 31 ++++++++++++++++++++----------- 1 file changed, 20 insertions(+), 11 deletions(-) (limited to 'kernel/time/clocksource.c') diff --git a/kernel/time/clocksource.c b/kernel/time/clocksource.c index 1923a340bd91..9782997cb6cf 100644 --- a/kernel/time/clocksource.c +++ b/kernel/time/clocksource.c @@ -553,7 +553,7 @@ static u64 clocksource_max_deferment(struct clocksource *cs) #ifndef CONFIG_ARCH_USES_GETTIMEOFFSET -static struct clocksource *clocksource_find_best(bool oneshot) +static struct clocksource *clocksource_find_best(bool oneshot, bool skipcur) { struct clocksource *cs; @@ -566,6 +566,8 @@ static struct clocksource *clocksource_find_best(bool oneshot) * the best rating. */ list_for_each_entry(cs, &clocksource_list, list) { + if (skipcur && cs == curr_clocksource) + continue; if (oneshot && !(cs->flags & CLOCK_SOURCE_VALID_FOR_HRES)) continue; return cs; @@ -573,26 +575,20 @@ static struct clocksource *clocksource_find_best(bool oneshot) return NULL; } -/** - * clocksource_select - Select the best clocksource available - * - * Private function. Must hold clocksource_mutex when called. - * - * Select the clocksource with the best rating, or the clocksource, - * which is selected by userspace override. - */ -static void clocksource_select(void) +static void __clocksource_select(bool skipcur) { bool oneshot = tick_oneshot_mode_active(); struct clocksource *best, *cs; /* Find the best suitable clocksource */ - best = clocksource_find_best(oneshot); + best = clocksource_find_best(oneshot, skipcur); if (!best) return; /* Check for the override clocksource. */ list_for_each_entry(cs, &clocksource_list, list) { + if (skipcur && cs == curr_clocksource) + continue; if (strcmp(cs->name, override_name) != 0) continue; /* @@ -618,6 +614,19 @@ static void clocksource_select(void) } } +/** + * clocksource_select - Select the best clocksource available + * + * Private function. Must hold clocksource_mutex when called. + * + * Select the clocksource with the best rating, or the clocksource, + * which is selected by userspace override. + */ +static void clocksource_select(void) +{ + return __clocksource_select(false); +} + #else /* !CONFIG_ARCH_USES_GETTIMEOFFSET */ static inline void clocksource_select(void) { } -- cgit v1.2.3 From 29b5407819f59731c9423238fae03b756822708c Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Thu, 25 Apr 2013 20:31:45 +0000 Subject: clocksource: Split out user string input Split out the user string input for clocksource override. Preparatory patch for unbind. [ jstultz: Fix an off by one error ] Signed-off-by: Thomas Gleixner Cc: John Stultz Cc: Magnus Damm Link: http://lkml.kernel.org/r/20130425143435.895851338@linutronix.de Signed-off-by: Thomas Gleixner --- kernel/time/clocksource.c | 37 +++++++++++++++++++++++-------------- 1 file changed, 23 insertions(+), 14 deletions(-) (limited to 'kernel/time/clocksource.c') diff --git a/kernel/time/clocksource.c b/kernel/time/clocksource.c index 9782997cb6cf..d7f1a45c2fa5 100644 --- a/kernel/time/clocksource.c +++ b/kernel/time/clocksource.c @@ -174,7 +174,8 @@ clocks_calc_mult_shift(u32 *mult, u32 *shift, u32 from, u32 to, u32 maxsec) static struct clocksource *curr_clocksource; static LIST_HEAD(clocksource_list); static DEFINE_MUTEX(clocksource_mutex); -static char override_name[32]; +#define CS_NAME_LEN 32 +static char override_name[CS_NAME_LEN]; static int finished_booting; #ifdef CONFIG_CLOCKSOURCE_WATCHDOG @@ -838,6 +839,23 @@ sysfs_show_current_clocksources(struct device *dev, return count; } +static size_t clocksource_get_uname(const char *buf, char *dst, size_t cnt) +{ + size_t ret = cnt; + + /* strings from sysfs write are not 0 terminated! */ + if (!cnt || cnt >= CS_NAME_LEN) + return -EINVAL; + + /* strip of \n: */ + if (buf[cnt-1] == '\n') + cnt--; + if (cnt > 0) + memcpy(dst, buf, cnt); + dst[cnt] = 0; + return ret; +} + /** * sysfs_override_clocksource - interface for manually overriding clocksource * @dev: unused @@ -852,22 +870,13 @@ static ssize_t sysfs_override_clocksource(struct device *dev, struct device_attribute *attr, const char *buf, size_t count) { - size_t ret = count; - - /* strings from sysfs write are not 0 terminated! */ - if (count >= sizeof(override_name)) - return -EINVAL; - - /* strip of \n: */ - if (buf[count-1] == '\n') - count--; + size_t ret; mutex_lock(&clocksource_mutex); - if (count > 0) - memcpy(override_name, buf, count); - override_name[count] = 0; - clocksource_select(); + ret = clocksource_get_uname(buf, override_name, count); + if (ret >= 0) + clocksource_select(); mutex_unlock(&clocksource_mutex); -- cgit v1.2.3 From 7eaeb34305dee26634f7c98ae62646da5cebe91d Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Thu, 25 Apr 2013 20:31:46 +0000 Subject: clocksource: Provide unbind interface in sysfs With the module refcount held for the current clocksource there is no way to unload the module. Provide a sysfs interface which allows to unbind the clocksource. One could argue that the clocksource override could be (ab)used to do so, but the clocksource override cannot be used from the kernel itself, while an unbind function can be used to programmatically check whether a clocksource can be shutdown or not. The unbind functionality uses the new skip current feature of clocksource_select and verifies that a fallback clocksource has been installed. If the clocksource which should be unbound is the current clocksource and no fallback can be found, unbind returns -EBUSY. This does not support the unbinding of a clocksource which is used as the watchdog clocksource. No point in fostering crappy hardware. Signed-off-by: Thomas Gleixner Cc: John Stultz Cc: Magnus Damm Link: http://lkml.kernel.org/r/20130425143435.964218245@linutronix.de Signed-off-by: Thomas Gleixner --- kernel/time/clocksource.c | 73 +++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 73 insertions(+) (limited to 'kernel/time/clocksource.c') diff --git a/kernel/time/clocksource.c b/kernel/time/clocksource.c index d7f1a45c2fa5..791d1aeb17ac 100644 --- a/kernel/time/clocksource.c +++ b/kernel/time/clocksource.c @@ -440,6 +440,11 @@ static int clocksource_watchdog_kthread(void *data) return 0; } +static bool clocksource_is_watchdog(struct clocksource *cs) +{ + return cs == watchdog; +} + #else /* CONFIG_CLOCKSOURCE_WATCHDOG */ static void clocksource_enqueue_watchdog(struct clocksource *cs) @@ -451,6 +456,7 @@ static void clocksource_enqueue_watchdog(struct clocksource *cs) static inline void clocksource_dequeue_watchdog(struct clocksource *cs) { } static inline void clocksource_resume_watchdog(void) { } static inline int clocksource_watchdog_kthread(void *data) { return 0; } +static bool clocksource_is_watchdog(struct clocksource *cs) { return false; } #endif /* CONFIG_CLOCKSOURCE_WATCHDOG */ @@ -628,6 +634,11 @@ static void clocksource_select(void) return __clocksource_select(false); } +static void clocksource_select_fallback(void) +{ + return __clocksource_select(true); +} + #else /* !CONFIG_ARCH_USES_GETTIMEOFFSET */ static inline void clocksource_select(void) { } @@ -803,6 +814,29 @@ void clocksource_change_rating(struct clocksource *cs, int rating) } EXPORT_SYMBOL(clocksource_change_rating); +/* + * Unbind clocksource @cs. Called with clocksource_mutex held + */ +static int clocksource_unbind(struct clocksource *cs) +{ + /* + * I really can't convince myself to support this on hardware + * designed by lobotomized monkeys. + */ + if (clocksource_is_watchdog(cs)) + return -EBUSY; + + if (cs == curr_clocksource) { + /* Select and try to install a replacement clock source */ + clocksource_select_fallback(); + if (curr_clocksource == cs) + return -EBUSY; + } + clocksource_dequeue_watchdog(cs); + list_del_init(&cs->list); + return 0; +} + /** * clocksource_unregister - remove a registered clocksource * @cs: clocksource to be unregistered @@ -883,6 +917,40 @@ static ssize_t sysfs_override_clocksource(struct device *dev, return ret; } +/** + * sysfs_unbind_current_clocksource - interface for manually unbinding clocksource + * @dev: unused + * @attr: unused + * @buf: unused + * @count: length of buffer + * + * Takes input from sysfs interface for manually unbinding a clocksource. + */ +static ssize_t sysfs_unbind_clocksource(struct device *dev, + struct device_attribute *attr, + const char *buf, size_t count) +{ + struct clocksource *cs; + char name[CS_NAME_LEN]; + size_t ret; + + ret = clocksource_get_uname(buf, name, count); + if (ret < 0) + return ret; + + ret = -ENODEV; + mutex_lock(&clocksource_mutex); + list_for_each_entry(cs, &clocksource_list, list) { + if (strcmp(cs->name, name)) + continue; + ret = clocksource_unbind(cs); + break; + } + mutex_unlock(&clocksource_mutex); + + return ret ? ret : count; +} + /** * sysfs_show_available_clocksources - sysfs interface for listing clocksource * @dev: unused @@ -925,6 +993,8 @@ sysfs_show_available_clocksources(struct device *dev, static DEVICE_ATTR(current_clocksource, 0644, sysfs_show_current_clocksources, sysfs_override_clocksource); +static DEVICE_ATTR(unbind_clocksource, 0200, NULL, sysfs_unbind_clocksource); + static DEVICE_ATTR(available_clocksource, 0444, sysfs_show_available_clocksources, NULL); @@ -948,6 +1018,9 @@ static int __init init_clocksource_sysfs(void) error = device_create_file( &device_clocksource, &dev_attr_current_clocksource); + if (!error) + error = device_create_file(&device_clocksource, + &dev_attr_unbind_clocksource); if (!error) error = device_create_file( &device_clocksource, -- cgit v1.2.3 From a89c7edbe7d7aa80f507915f3dd801211b116b79 Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Thu, 25 Apr 2013 20:31:46 +0000 Subject: clocksource: Let clocksource_unregister() return success/error The unregister call can fail, if the clocksource is the current one and there is no replacement clocksource available. It can also fail, if the clocksource is the watchdog clocksource and I'm not going to provide support for this. Signed-off-by: Thomas Gleixner Cc: John Stultz Cc: Magnus Damm Link: http://lkml.kernel.org/r/20130425143436.029915527@linutronix.de Signed-off-by: Thomas Gleixner --- include/linux/clocksource.h | 2 +- kernel/time/clocksource.c | 33 ++++++++++++--------------------- 2 files changed, 13 insertions(+), 22 deletions(-) (limited to 'kernel/time/clocksource.c') diff --git a/include/linux/clocksource.h b/include/linux/clocksource.h index 32a895b114d8..2f39a4911668 100644 --- a/include/linux/clocksource.h +++ b/include/linux/clocksource.h @@ -282,7 +282,7 @@ static inline s64 clocksource_cyc2ns(cycle_t cycles, u32 mult, u32 shift) extern int clocksource_register(struct clocksource*); -extern void clocksource_unregister(struct clocksource*); +extern int clocksource_unregister(struct clocksource*); extern void clocksource_touch_watchdog(void); extern struct clocksource* clocksource_get_next(void); extern void clocksource_change_rating(struct clocksource *cs, int rating); diff --git a/kernel/time/clocksource.c b/kernel/time/clocksource.c index 791d1aeb17ac..31b90332f47b 100644 --- a/kernel/time/clocksource.c +++ b/kernel/time/clocksource.c @@ -389,28 +389,17 @@ static void clocksource_enqueue_watchdog(struct clocksource *cs) static void clocksource_dequeue_watchdog(struct clocksource *cs) { - struct clocksource *tmp; unsigned long flags; spin_lock_irqsave(&watchdog_lock, flags); - if (cs->flags & CLOCK_SOURCE_MUST_VERIFY) { - /* cs is a watched clocksource. */ - list_del_init(&cs->wd_list); - } else if (cs == watchdog) { - /* Reset watchdog cycles */ - clocksource_reset_watchdog(); - /* Current watchdog is removed. Find an alternative. */ - watchdog = NULL; - list_for_each_entry(tmp, &clocksource_list, list) { - if (tmp == cs || tmp->flags & CLOCK_SOURCE_MUST_VERIFY) - continue; - if (!watchdog || tmp->rating > watchdog->rating) - watchdog = tmp; + if (cs != watchdog) { + if (cs->flags & CLOCK_SOURCE_MUST_VERIFY) { + /* cs is a watched clocksource. */ + list_del_init(&cs->wd_list); + /* Check if the watchdog timer needs to be stopped. */ + clocksource_stop_watchdog(); } } - cs->flags &= ~CLOCK_SOURCE_WATCHDOG; - /* Check if the watchdog timer needs to be stopped. */ - clocksource_stop_watchdog(); spin_unlock_irqrestore(&watchdog_lock, flags); } @@ -841,13 +830,15 @@ static int clocksource_unbind(struct clocksource *cs) * clocksource_unregister - remove a registered clocksource * @cs: clocksource to be unregistered */ -void clocksource_unregister(struct clocksource *cs) +int clocksource_unregister(struct clocksource *cs) { + int ret = 0; + mutex_lock(&clocksource_mutex); - clocksource_dequeue_watchdog(cs); - list_del(&cs->list); - clocksource_select(); + if (!list_empty(&cs->list)) + ret = clocksource_unbind(cs); mutex_unlock(&clocksource_mutex); + return ret; } EXPORT_SYMBOL(clocksource_unregister); -- cgit v1.2.3 From 03e13cf5ee60584fe0c831682c67212effb7fca4 Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Thu, 25 Apr 2013 20:31:50 +0000 Subject: clockevents: Implement unbind functionality Provide a sysfs interface to allow unbinding of clockevent devices. The device is unbound if it is unused or if there is a replacement device available. Unbinding of broadcast devices is not supported as we don't want to foster that nonsense. If no replacement device is available the unbind returns -EBUSY. Unbind is available from the kernel and through sysfs, which is necessary to drop the module refcount. Signed-off-by: Thomas Gleixner Cc: John Stultz Cc: Magnus Damm Link: http://lkml.kernel.org/r/20130425143436.499216659@linutronix.de Signed-off-by: Thomas Gleixner --- include/linux/clockchips.h | 1 + kernel/time/clockevents.c | 125 ++++++++++++++++++++++++++++++++++++++++++++ kernel/time/clocksource.c | 9 ++-- kernel/time/tick-common.c | 24 +++++++++ kernel/time/tick-internal.h | 7 +++ 5 files changed, 162 insertions(+), 4 deletions(-) (limited to 'kernel/time/clocksource.c') diff --git a/include/linux/clockchips.h b/include/linux/clockchips.h index ae1193bcf074..0857922e8ad0 100644 --- a/include/linux/clockchips.h +++ b/include/linux/clockchips.h @@ -141,6 +141,7 @@ static inline unsigned long div_sc(unsigned long ticks, unsigned long nsec, extern u64 clockevent_delta2ns(unsigned long latch, struct clock_event_device *evt); extern void clockevents_register_device(struct clock_event_device *dev); +extern int clockevents_unbind_device(struct clock_event_device *ced, int cpu); extern void clockevents_config(struct clock_event_device *dev, u32 freq); extern void clockevents_config_and_register(struct clock_event_device *dev, diff --git a/kernel/time/clockevents.c b/kernel/time/clockevents.c index 0a23f4f29934..38959c866789 100644 --- a/kernel/time/clockevents.c +++ b/kernel/time/clockevents.c @@ -25,6 +25,13 @@ static LIST_HEAD(clockevent_devices); static LIST_HEAD(clockevents_released); /* Protection for the above */ static DEFINE_RAW_SPINLOCK(clockevents_lock); +/* Protection for unbind operations */ +static DEFINE_MUTEX(clockevents_mutex); + +struct ce_unbind { + struct clock_event_device *ce; + int res; +}; /** * clockevents_delta2ns - Convert a latch value (device ticks) to nanoseconds @@ -245,6 +252,90 @@ static void clockevents_notify_released(void) } } +/* + * Try to install a replacement clock event device + */ +static int clockevents_replace(struct clock_event_device *ced) +{ + struct clock_event_device *dev, *newdev = NULL; + + list_for_each_entry(dev, &clockevent_devices, list) { + if (dev == ced || dev->mode != CLOCK_EVT_MODE_UNUSED) + continue; + + if (!tick_check_replacement(newdev, dev)) + continue; + + if (!try_module_get(dev->owner)) + continue; + + if (newdev) + module_put(newdev->owner); + newdev = dev; + } + if (newdev) { + tick_install_replacement(newdev); + list_del_init(&ced->list); + } + return newdev ? 0 : -EBUSY; +} + +/* + * Called with clockevents_mutex and clockevents_lock held + */ +static int __clockevents_try_unbind(struct clock_event_device *ced, int cpu) +{ + /* Fast track. Device is unused */ + if (ced->mode == CLOCK_EVT_MODE_UNUSED) { + list_del_init(&ced->list); + return 0; + } + + return ced == per_cpu(tick_cpu_device, cpu).evtdev ? -EAGAIN : -EBUSY; +} + +/* + * SMP function call to unbind a device + */ +static void __clockevents_unbind(void *arg) +{ + struct ce_unbind *cu = arg; + int res; + + raw_spin_lock(&clockevents_lock); + res = __clockevents_try_unbind(cu->ce, smp_processor_id()); + if (res == -EAGAIN) + res = clockevents_replace(cu->ce); + cu->res = res; + raw_spin_unlock(&clockevents_lock); +} + +/* + * Issues smp function call to unbind a per cpu device. Called with + * clockevents_mutex held. + */ +static int clockevents_unbind(struct clock_event_device *ced, int cpu) +{ + struct ce_unbind cu = { .ce = ced, .res = -ENODEV }; + + smp_call_function_single(cpu, __clockevents_unbind, &cu, 1); + return cu.res; +} + +/* + * Unbind a clockevents device. + */ +int clockevents_unbind_device(struct clock_event_device *ced, int cpu) +{ + int ret; + + mutex_lock(&clockevents_mutex); + ret = clockevents_unbind(ced, cpu); + mutex_unlock(&clockevents_mutex); + return ret; +} +EXPORT_SYMBOL_GPL(clockevents_unbind); + /** * clockevents_register_device - register a clock event device * @dev: device to register @@ -487,6 +578,38 @@ static ssize_t sysfs_show_current_tick_dev(struct device *dev, } static DEVICE_ATTR(current_device, 0444, sysfs_show_current_tick_dev, NULL); +/* We don't support the abomination of removable broadcast devices */ +static ssize_t sysfs_unbind_tick_dev(struct device *dev, + struct device_attribute *attr, + const char *buf, size_t count) +{ + char name[CS_NAME_LEN]; + size_t ret = sysfs_get_uname(buf, name, count); + struct clock_event_device *ce; + + if (ret < 0) + return ret; + + ret = -ENODEV; + mutex_lock(&clockevents_mutex); + raw_spin_lock_irq(&clockevents_lock); + list_for_each_entry(ce, &clockevent_devices, list) { + if (!strcmp(ce->name, name)) { + ret = __clockevents_try_unbind(ce, dev->id); + break; + } + } + raw_spin_unlock_irq(&clockevents_lock); + /* + * We hold clockevents_mutex, so ce can't go away + */ + if (ret == -EAGAIN) + ret = clockevents_unbind(ce, dev->id); + mutex_unlock(&clockevents_mutex); + return ret ? ret : count; +} +static DEVICE_ATTR(unbind_device, 0200, NULL, sysfs_unbind_tick_dev); + #ifdef CONFIG_GENERIC_CLOCKEVENTS_BROADCAST static struct device tick_bc_dev = { .init_name = "broadcast", @@ -529,6 +652,8 @@ static int __init tick_init_sysfs(void) err = device_register(dev); if (!err) err = device_create_file(dev, &dev_attr_current_device); + if (!err) + err = device_create_file(dev, &dev_attr_unbind_device); if (err) return err; } diff --git a/kernel/time/clocksource.c b/kernel/time/clocksource.c index 31b90332f47b..6d05b00410cc 100644 --- a/kernel/time/clocksource.c +++ b/kernel/time/clocksource.c @@ -31,6 +31,8 @@ #include #include +#include "tick-internal.h" + void timecounter_init(struct timecounter *tc, const struct cyclecounter *cc, u64 start_tstamp) @@ -174,7 +176,6 @@ clocks_calc_mult_shift(u32 *mult, u32 *shift, u32 from, u32 to, u32 maxsec) static struct clocksource *curr_clocksource; static LIST_HEAD(clocksource_list); static DEFINE_MUTEX(clocksource_mutex); -#define CS_NAME_LEN 32 static char override_name[CS_NAME_LEN]; static int finished_booting; @@ -864,7 +865,7 @@ sysfs_show_current_clocksources(struct device *dev, return count; } -static size_t clocksource_get_uname(const char *buf, char *dst, size_t cnt) +size_t sysfs_get_uname(const char *buf, char *dst, size_t cnt) { size_t ret = cnt; @@ -899,7 +900,7 @@ static ssize_t sysfs_override_clocksource(struct device *dev, mutex_lock(&clocksource_mutex); - ret = clocksource_get_uname(buf, override_name, count); + ret = sysfs_get_uname(buf, override_name, count); if (ret >= 0) clocksource_select(); @@ -925,7 +926,7 @@ static ssize_t sysfs_unbind_clocksource(struct device *dev, char name[CS_NAME_LEN]; size_t ret; - ret = clocksource_get_uname(buf, name, count); + ret = sysfs_get_uname(buf, name, count); if (ret < 0) return ret; diff --git a/kernel/time/tick-common.c b/kernel/time/tick-common.c index c34021650348..5edfb4806032 100644 --- a/kernel/time/tick-common.c +++ b/kernel/time/tick-common.c @@ -205,6 +205,17 @@ static void tick_setup_device(struct tick_device *td, tick_setup_oneshot(newdev, handler, next_event); } +void tick_install_replacement(struct clock_event_device *newdev) +{ + struct tick_device *td = &__get_cpu_var(tick_cpu_device); + int cpu = smp_processor_id(); + + clockevents_exchange_device(td->evtdev, newdev); + tick_setup_device(td, newdev, cpu, cpumask_of(cpu)); + if (newdev->features & CLOCK_EVT_FEAT_ONESHOT) + tick_oneshot_notify(); +} + static bool tick_check_percpu(struct clock_event_device *curdev, struct clock_event_device *newdev, int cpu) { @@ -236,6 +247,19 @@ static bool tick_check_preferred(struct clock_event_device *curdev, return !curdev || newdev->rating > curdev->rating; } +/* + * Check whether the new device is a better fit than curdev. curdev + * can be NULL ! + */ +bool tick_check_replacement(struct clock_event_device *curdev, + struct clock_event_device *newdev) +{ + if (tick_check_percpu(curdev, newdev, smp_processor_id())) + return false; + + return tick_check_preferred(curdev, newdev); +} + /* * Check, if the new registered device should be used. Called with * clockevents_lock held and interrupts disabled. diff --git a/kernel/time/tick-internal.h b/kernel/time/tick-internal.h index 06bfc8802dfb..be1690eaecff 100644 --- a/kernel/time/tick-internal.h +++ b/kernel/time/tick-internal.h @@ -11,6 +11,8 @@ extern seqlock_t jiffies_lock; #define TICK_DO_TIMER_NONE -1 #define TICK_DO_TIMER_BOOT -2 +#define CS_NAME_LEN 32 + DECLARE_PER_CPU(struct tick_device, tick_cpu_device); extern ktime_t tick_next_period; extern ktime_t tick_period; @@ -23,9 +25,14 @@ extern void tick_handover_do_timer(int *cpup); extern void tick_shutdown(unsigned int *cpup); extern void tick_suspend(void); extern void tick_resume(void); +extern bool tick_check_replacement(struct clock_event_device *curdev, + struct clock_event_device *newdev); +extern void tick_install_replacement(struct clock_event_device *dev); extern void clockevents_shutdown(struct clock_event_device *dev); +extern size_t sysfs_get_uname(const char *buf, char *dst, size_t cnt); + /* * NO_HZ / high resolution timer shared code */ -- cgit v1.2.3 From 1eaff67266b6b6c97bbd33cf2c20577822836413 Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Tue, 28 May 2013 09:48:46 +0200 Subject: clocksource: Implement clocksource_select_fallback() for CONFIG_ARCH_USES_GETTIMEOFFSET=y commit 7eaeb34305 (clocksource: Provide unbind interface in sysfs) implemented clocksource_select_fallback() which is not defined for CONFIG_ARCH_USES_GETTIMEOFFSET=y. Add an empty inline function for that. Reported-by: Ingo Molnar Reported-by: fengguang.wu@intel.com Signed-off-by: Thomas Gleixner --- kernel/time/clocksource.c | 1 + 1 file changed, 1 insertion(+) (limited to 'kernel/time/clocksource.c') diff --git a/kernel/time/clocksource.c b/kernel/time/clocksource.c index 6d05b00410cc..e713ef7d19a7 100644 --- a/kernel/time/clocksource.c +++ b/kernel/time/clocksource.c @@ -632,6 +632,7 @@ static void clocksource_select_fallback(void) #else /* !CONFIG_ARCH_USES_GETTIMEOFFSET */ static inline void clocksource_select(void) { } +static inline void clocksource_select_fallback(void) { } #endif -- cgit v1.2.3 From 332962f2c88868ed3cdab466870baaa34dd58612 Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Thu, 4 Jul 2013 22:46:45 +0200 Subject: clocksource: Reselect clocksource when watchdog validated high-res capability Up to commit 5d33b883a (clocksource: Always verify highres capability) we had no sanity check when selecting a clocksource, which prevented that a non highres capable clocksource is used when the system already switched to highres/nohz mode. The new sanity check works as Alex and Tim found out. It prevents the TSC from being used. This happens because on x86 the boot process looks like this: tsc_start_freqency_validation(TSC); clocksource_register(HPET); clocksource_done_booting(); clocksource_select() Selects HPET which is valid for high-res switch_to_highres(); clocksource_register(TSC); TSC is not selected, because it is not yet flagged as VALID_HIGH_RES clocksource_watchdog() Validates TSC for highres, but that does not make TSC the current clocksource. Before the sanity check was added, we installed TSC unvalidated which worked most of the time. If the TSC was really detected as unstable, then the unstable logic removed it and installed HPET again. The sanity check is correct and needed. So the watchdog needs to kick a reselection of the clocksource, when it qualifies TSC as a valid high res clocksource. To solve this, we mark the clocksource which got the flag CLOCK_SOURCE_VALID_FOR_HRES set by the watchdog with an new flag CLOCK_SOURCE_RESELECT and trigger the watchdog thread. The watchdog thread evaluates the flag and invokes clocksource_select() when set. To avoid that the clocksource_done_booting() code, which is about to install the first real clocksource anyway, needs to go through clocksource_select and tick_oneshot_notify() pointlessly, split out the clocksource_watchdog_kthread() list walk code and invoke the select/notify only when called from clocksource_watchdog_kthread(). So clocksource_done_booting() can utilize the same splitout code without the select/notify invocation and the clocksource_mutex unlock/relock dance. Reported-and-tested-by: Alex Shi Cc: Hans Peter Anvin Cc: Tim Chen Cc: Andi Kleen Tested-by: Peter Zijlstra Cc: Ingo Molnar Cc: Davidlohr Bueso Cc: John Stultz Link: http://lkml.kernel.org/r/alpine.DEB.2.02.1307042239150.11637@ionos.tec.linutronix.de Signed-off-by: Thomas Gleixner --- include/linux/clocksource.h | 1 + kernel/time/clocksource.c | 57 +++++++++++++++++++++++++++++++++------------ 2 files changed, 43 insertions(+), 15 deletions(-) (limited to 'kernel/time/clocksource.c') diff --git a/include/linux/clocksource.h b/include/linux/clocksource.h index 2f39a4911668..dbbf8aa7731b 100644 --- a/include/linux/clocksource.h +++ b/include/linux/clocksource.h @@ -210,6 +210,7 @@ struct clocksource { #define CLOCK_SOURCE_VALID_FOR_HRES 0x20 #define CLOCK_SOURCE_UNSTABLE 0x40 #define CLOCK_SOURCE_SUSPEND_NONSTOP 0x80 +#define CLOCK_SOURCE_RESELECT 0x100 /* simplify initialization of mask field */ #define CLOCKSOURCE_MASK(bits) (cycle_t)((bits) < 64 ? ((1ULL<<(bits))-1) : -1) diff --git a/kernel/time/clocksource.c b/kernel/time/clocksource.c index e713ef7d19a7..50a8736757f3 100644 --- a/kernel/time/clocksource.c +++ b/kernel/time/clocksource.c @@ -181,6 +181,7 @@ static int finished_booting; #ifdef CONFIG_CLOCKSOURCE_WATCHDOG static void clocksource_watchdog_work(struct work_struct *work); +static void clocksource_select(void); static LIST_HEAD(watchdog_list); static struct clocksource *watchdog; @@ -301,13 +302,30 @@ static void clocksource_watchdog(unsigned long data) if (!(cs->flags & CLOCK_SOURCE_VALID_FOR_HRES) && (cs->flags & CLOCK_SOURCE_IS_CONTINUOUS) && (watchdog->flags & CLOCK_SOURCE_IS_CONTINUOUS)) { + /* Mark it valid for high-res. */ cs->flags |= CLOCK_SOURCE_VALID_FOR_HRES; + + /* + * clocksource_done_booting() will sort it if + * finished_booting is not set yet. + */ + if (!finished_booting) + continue; + /* - * We just marked the clocksource as highres-capable, - * notify the rest of the system as well so that we - * transition into high-res mode: + * If this is not the current clocksource let + * the watchdog thread reselect it. Due to the + * change to high res this clocksource might + * be preferred now. If it is the current + * clocksource let the tick code know about + * that change. */ - tick_clock_notify(); + if (cs != curr_clocksource) { + cs->flags |= CLOCK_SOURCE_RESELECT; + schedule_work(&watchdog_work); + } else { + tick_clock_notify(); + } } } @@ -404,19 +422,25 @@ static void clocksource_dequeue_watchdog(struct clocksource *cs) spin_unlock_irqrestore(&watchdog_lock, flags); } -static int clocksource_watchdog_kthread(void *data) +static int __clocksource_watchdog_kthread(void) { struct clocksource *cs, *tmp; unsigned long flags; LIST_HEAD(unstable); + int select = 0; - mutex_lock(&clocksource_mutex); spin_lock_irqsave(&watchdog_lock, flags); - list_for_each_entry_safe(cs, tmp, &watchdog_list, wd_list) + list_for_each_entry_safe(cs, tmp, &watchdog_list, wd_list) { if (cs->flags & CLOCK_SOURCE_UNSTABLE) { list_del_init(&cs->wd_list); list_add(&cs->wd_list, &unstable); + select = 1; + } + if (cs->flags & CLOCK_SOURCE_RESELECT) { + cs->flags &= ~CLOCK_SOURCE_RESELECT; + select = 1; } + } /* Check if the watchdog timer needs to be stopped. */ clocksource_stop_watchdog(); spin_unlock_irqrestore(&watchdog_lock, flags); @@ -426,6 +450,14 @@ static int clocksource_watchdog_kthread(void *data) list_del_init(&cs->wd_list); __clocksource_change_rating(cs, 0); } + return select; +} + +static int clocksource_watchdog_kthread(void *data) +{ + mutex_lock(&clocksource_mutex); + if (__clocksource_watchdog_kthread()) + clocksource_select(); mutex_unlock(&clocksource_mutex); return 0; } @@ -445,7 +477,7 @@ static void clocksource_enqueue_watchdog(struct clocksource *cs) static inline void clocksource_dequeue_watchdog(struct clocksource *cs) { } static inline void clocksource_resume_watchdog(void) { } -static inline int clocksource_watchdog_kthread(void *data) { return 0; } +static inline int __clocksource_watchdog_kthread(void) { return 0; } static bool clocksource_is_watchdog(struct clocksource *cs) { return false; } #endif /* CONFIG_CLOCKSOURCE_WATCHDOG */ @@ -647,16 +679,11 @@ static int __init clocksource_done_booting(void) { mutex_lock(&clocksource_mutex); curr_clocksource = clocksource_default_clock(); - mutex_unlock(&clocksource_mutex); - finished_booting = 1; - /* * Run the watchdog first to eliminate unstable clock sources */ - clocksource_watchdog_kthread(NULL); - - mutex_lock(&clocksource_mutex); + __clocksource_watchdog_kthread(); clocksource_select(); mutex_unlock(&clocksource_mutex); return 0; @@ -789,7 +816,6 @@ static void __clocksource_change_rating(struct clocksource *cs, int rating) list_del(&cs->list); cs->rating = rating; clocksource_enqueue(cs); - clocksource_select(); } /** @@ -801,6 +827,7 @@ void clocksource_change_rating(struct clocksource *cs, int rating) { mutex_lock(&clocksource_mutex); __clocksource_change_rating(cs, rating); + clocksource_select(); mutex_unlock(&clocksource_mutex); } EXPORT_SYMBOL(clocksource_change_rating); -- cgit v1.2.3