diff options
Diffstat (limited to 'kernel/time')
41 files changed, 8704 insertions, 3633 deletions
diff --git a/kernel/time/Kconfig b/kernel/time/Kconfig index fcc42353f125..b0b97a60aaa6 100644 --- a/kernel/time/Kconfig +++ b/kernel/time/Kconfig @@ -17,22 +17,13 @@ config ARCH_CLOCKSOURCE_DATA config ARCH_CLOCKSOURCE_INIT bool -# Clocksources require validation of the clocksource against the last -# cycle update - x86/TSC misfeature -config CLOCKSOURCE_VALIDATE_LAST_CYCLE - bool - # Timekeeping vsyscall support config GENERIC_TIME_VSYSCALL bool -# Old style timekeeping -config ARCH_USES_GETTIMEOFFSET - bool - # The generic clock events infrastructure config GENERIC_CLOCKEVENTS - bool + def_bool !LEGACY_TIMER_TICK # Architecture can handle broadcast in a driver-agnostic way config ARCH_HAS_TICK_BROADCAST @@ -43,6 +34,11 @@ config GENERIC_CLOCKEVENTS_BROADCAST bool depends on GENERIC_CLOCKEVENTS +# Handle broadcast in default_idle_call() +config GENERIC_CLOCKEVENTS_BROADCAST_IDLE + bool + depends on GENERIC_CLOCKEVENTS_BROADCAST + # Automatically adjust the min. reprogramming time for # clock event device config GENERIC_CLOCKEVENTS_MIN_ADJUST @@ -52,6 +48,40 @@ config GENERIC_CLOCKEVENTS_MIN_ADJUST config GENERIC_CMOS_UPDATE bool +# Select to handle posix CPU timers from task_work +# and not from the timer interrupt context +config HAVE_POSIX_CPU_TIMERS_TASK_WORK + bool + +config POSIX_CPU_TIMERS_TASK_WORK + bool + default y if POSIX_TIMERS && HAVE_POSIX_CPU_TIMERS_TASK_WORK + +config LEGACY_TIMER_TICK + bool + help + The legacy timer tick helper is used by platforms that + lack support for the generic clockevent framework. + New platforms should use generic clockevents instead. + +config TIME_KUNIT_TEST + tristate "KUnit test for kernel/time functions" if !KUNIT_ALL_TESTS + depends on KUNIT + default KUNIT_ALL_TESTS + help + Enable this option to test RTC library functions. + + If unsure, say N. + +config CONTEXT_TRACKING + bool + +config CONTEXT_TRACKING_IDLE + bool + select CONTEXT_TRACKING + help + Tracks idle state on behalf of RCU. + if GENERIC_CLOCKEVENTS menu "Timers subsystem" @@ -63,7 +93,6 @@ config TICK_ONESHOT config NO_HZ_COMMON bool - depends on !ARCH_USES_GETTIMEOFFSET && GENERIC_CLOCKEVENTS select TICK_ONESHOT choice @@ -78,7 +107,6 @@ config HZ_PERIODIC config NO_HZ_IDLE bool "Idle dynticks system (tickless idle)" - depends on !ARCH_USES_GETTIMEOFFSET && GENERIC_CLOCKEVENTS select NO_HZ_COMMON help This option enables a tickless idle system: timer interrupts @@ -90,10 +118,9 @@ config NO_HZ_IDLE config NO_HZ_FULL bool "Full dynticks system (tickless)" # NO_HZ_COMMON dependency - depends on !ARCH_USES_GETTIMEOFFSET && GENERIC_CLOCKEVENTS # We need at least one periodic CPU for timekeeping depends on SMP - depends on HAVE_CONTEXT_TRACKING + depends on HAVE_CONTEXT_TRACKING_USER # VIRT_CPU_ACCOUNTING_GEN dependency depends on HAVE_VIRT_CPU_ACCOUNTING_GEN select NO_HZ_COMMON @@ -108,48 +135,54 @@ config NO_HZ_FULL the task mostly runs in userspace and has few kernel activity. You need to fill up the nohz_full boot parameter with the - desired range of dynticks CPUs. + desired range of dynticks CPUs to use it. This is implemented at + the expense of some overhead in user <-> kernel transitions: + syscalls, exceptions and interrupts. - This is implemented at the expense of some overhead in user <-> kernel - transitions: syscalls, exceptions and interrupts. Even when it's - dynamically off. + By default, without passing the nohz_full parameter, this behaves just + like NO_HZ_IDLE. - Say N. + If you're a distro say Y. endchoice -config CONTEXT_TRACKING - bool +config CONTEXT_TRACKING_USER + bool + depends on HAVE_CONTEXT_TRACKING_USER + select CONTEXT_TRACKING + help + Track transitions between kernel and user on behalf of RCU and + tickless cputime accounting. The former case relies on context + tracking to enter/exit RCU extended quiescent states. -config CONTEXT_TRACKING_FORCE - bool "Force context tracking" - depends on CONTEXT_TRACKING +config CONTEXT_TRACKING_USER_FORCE + bool "Force user context tracking" + depends on CONTEXT_TRACKING_USER default y if !NO_HZ_FULL help The major pre-requirement for full dynticks to work is to - support the context tracking subsystem. But there are also + support the user context tracking subsystem. But there are also other dependencies to provide in order to make the full dynticks working. This option stands for testing when an arch implements the - context tracking backend but doesn't yet fullfill all the + user context tracking backend but doesn't yet fulfill all the requirements to make the full dynticks feature working. Without the full dynticks, there is no way to test the support - for context tracking and the subsystems that rely on it: RCU + for user context tracking and the subsystems that rely on it: RCU userspace extended quiescent state and tickless cputime accounting. This option copes with the absence of the full - dynticks subsystem by forcing the context tracking on all + dynticks subsystem by forcing the user context tracking on all CPUs in the system. Say Y only if you're working on the development of an - architecture backend for the context tracking. + architecture backend for the user context tracking. Say N otherwise, this option brings an overhead that you don't want in production. config NO_HZ bool "Old Idle dynticks config" - depends on !ARCH_USES_GETTIMEOFFSET && GENERIC_CLOCKEVENTS help This is the old config entry that enables dynticks idle. We keep it around for a little while to enforce backward @@ -157,12 +190,24 @@ config NO_HZ config HIGH_RES_TIMERS bool "High Resolution Timer Support" - depends on !ARCH_USES_GETTIMEOFFSET && GENERIC_CLOCKEVENTS select TICK_ONESHOT help This option enables high resolution timer support. If your hardware is not capable then this option only increases the size of the kernel image. +config CLOCKSOURCE_WATCHDOG_MAX_SKEW_US + int "Clocksource watchdog maximum allowable skew (in microseconds)" + depends on CLOCKSOURCE_WATCHDOG + range 50 1000 + default 125 + help + Specify the maximum amount of allowable watchdog skew in + microseconds before reporting the clocksource to be unstable. + The default is based on a half-second clocksource watchdog + interval and NTP's maximum frequency drift of 500 parts + per million. If the clocksource is good enough for NTP, + it is good enough for the clocksource watchdog! + endmenu endif diff --git a/kernel/time/Makefile b/kernel/time/Makefile index c8f00168afe8..fe0ae82124fe 100644 --- a/kernel/time/Makefile +++ b/kernel/time/Makefile @@ -1,5 +1,5 @@ # SPDX-License-Identifier: GPL-2.0 -obj-y += time.o timer.o hrtimer.o +obj-y += time.o timer.o hrtimer.o sleep_timeout.o obj-y += timekeeping.o ntp.o clocksource.o jiffies.o timer_list.o obj-y += timeconv.o timecounter.o alarmtimer.o @@ -16,7 +16,13 @@ ifeq ($(CONFIG_GENERIC_CLOCKEVENTS_BROADCAST),y) endif obj-$(CONFIG_GENERIC_SCHED_CLOCK) += sched_clock.o obj-$(CONFIG_TICK_ONESHOT) += tick-oneshot.o tick-sched.o +obj-$(CONFIG_LEGACY_TIMER_TICK) += tick-legacy.o +ifeq ($(CONFIG_SMP),y) + obj-$(CONFIG_NO_HZ_COMMON) += timer_migration.o +endif obj-$(CONFIG_HAVE_GENERIC_VDSO) += vsyscall.o obj-$(CONFIG_DEBUG_FS) += timekeeping_debug.o obj-$(CONFIG_TEST_UDELAY) += test_udelay.o obj-$(CONFIG_TIME_NS) += namespace.o +obj-$(CONFIG_TEST_CLOCKSOURCE_WATCHDOG) += clocksource-wdtest.o +obj-$(CONFIG_TIME_KUNIT_TEST) += time_test.o diff --git a/kernel/time/alarmtimer.c b/kernel/time/alarmtimer.c index 2ffb466af77e..0ddccdff119a 100644 --- a/kernel/time/alarmtimer.c +++ b/kernel/time/alarmtimer.c @@ -2,13 +2,13 @@ /* * Alarmtimer interface * - * This interface provides a timer which is similarto hrtimers, + * This interface provides a timer which is similar to hrtimers, * but triggers a RTC alarm if the box is suspend. * * This interface is influenced by the Android RTC Alarm timer * interface. * - * Copyright (C) 2010 IBM Corperation + * Copyright (C) 2010 IBM Corporation * * Author: John Stultz <john.stultz@linaro.org> */ @@ -81,8 +81,7 @@ struct rtc_device *alarmtimer_get_rtcdev(void) } EXPORT_SYMBOL_GPL(alarmtimer_get_rtcdev); -static int alarmtimer_rtc_add_device(struct device *dev, - struct class_interface *class_intf) +static int alarmtimer_rtc_add_device(struct device *dev) { unsigned long flags; struct rtc_device *rtc = to_rtc_device(dev); @@ -92,7 +91,7 @@ static int alarmtimer_rtc_add_device(struct device *dev, if (rtcdev) return -EBUSY; - if (!rtc->ops->set_alarm) + if (!test_bit(RTC_FEATURE_ALARM, rtc->features)) return -1; if (!device_may_wakeup(rtc->dev.parent)) return -1; @@ -135,7 +134,7 @@ static struct class_interface alarmtimer_rtc_interface = { static int alarmtimer_rtc_interface_setup(void) { - alarmtimer_rtc_interface.class = rtc_class; + alarmtimer_rtc_interface.class = &rtc_class; return class_interface_register(&alarmtimer_rtc_interface); } static void alarmtimer_rtc_interface_remove(void) @@ -192,34 +191,21 @@ static void alarmtimer_dequeue(struct alarm_base *base, struct alarm *alarm) * When a alarm timer fires, this runs through the timerqueue to * see which alarms expired, and runs those. If there are more alarm * timers queued for the future, we set the hrtimer to fire when - * when the next future alarm timer expires. + * the next future alarm timer expires. */ static enum hrtimer_restart alarmtimer_fired(struct hrtimer *timer) { struct alarm *alarm = container_of(timer, struct alarm, timer); struct alarm_base *base = &alarm_bases[alarm->type]; - unsigned long flags; - int ret = HRTIMER_NORESTART; - int restart = ALARMTIMER_NORESTART; - spin_lock_irqsave(&base->lock, flags); - alarmtimer_dequeue(base, alarm); - spin_unlock_irqrestore(&base->lock, flags); + scoped_guard (spinlock_irqsave, &base->lock) + alarmtimer_dequeue(base, alarm); if (alarm->function) - restart = alarm->function(alarm, base->get_ktime()); - - spin_lock_irqsave(&base->lock, flags); - if (restart != ALARMTIMER_NORESTART) { - hrtimer_set_expires(&alarm->timer, alarm->node.expires); - alarmtimer_enqueue(base, alarm); - ret = HRTIMER_RESTART; - } - spin_unlock_irqrestore(&base->lock, flags); + alarm->function(alarm, base->get_ktime()); trace_alarmtimer_fired(alarm, base->get_ktime()); - return ret; - + return HRTIMER_NORESTART; } ktime_t alarm_expires_remaining(const struct alarm *alarm) @@ -291,6 +277,17 @@ static int alarmtimer_suspend(struct device *dev) rtc_timer_cancel(rtc, &rtctimer); rtc_read_time(rtc, &tm); now = rtc_tm_to_ktime(tm); + + /* + * If the RTC alarm timer only supports a limited time offset, set the + * alarm time to the maximum supported value. + * The system may wake up earlier (possibly much earlier) than expected + * when the alarmtimer runs. This is the best the kernel can do if + * the alarmtimer exceeds the time that the rtc device can be programmed + * for. + */ + min = rtc_bound_alarmtime(rtc, min); + now = ktime_add(now, min); /* Set alarm, if in the past reject suspend briefly to handle */ @@ -324,10 +321,9 @@ static int alarmtimer_resume(struct device *dev) static void __alarm_init(struct alarm *alarm, enum alarmtimer_type type, - enum alarmtimer_restart (*function)(struct alarm *, ktime_t)) + void (*function)(struct alarm *, ktime_t)) { timerqueue_init(&alarm->node); - alarm->timer.function = alarmtimer_fired; alarm->function = function; alarm->type = type; alarm->state = ALARMTIMER_STATE_INACTIVE; @@ -340,10 +336,10 @@ __alarm_init(struct alarm *alarm, enum alarmtimer_type type, * @function: callback that is run when the alarm fires */ void alarm_init(struct alarm *alarm, enum alarmtimer_type type, - enum alarmtimer_restart (*function)(struct alarm *, ktime_t)) + void (*function)(struct alarm *, ktime_t)) { - hrtimer_init(&alarm->timer, alarm_bases[type].base_clockid, - HRTIMER_MODE_ABS); + hrtimer_setup(&alarm->timer, alarmtimer_fired, alarm_bases[type].base_clockid, + HRTIMER_MODE_ABS); __alarm_init(alarm, type, function); } EXPORT_SYMBOL_GPL(alarm_init); @@ -527,37 +523,18 @@ static enum alarmtimer_type clock2alarm(clockid_t clockid) /** * alarm_handle_timer - Callback for posix timers * @alarm: alarm that fired + * @now: time at the timer expiration * * Posix timer callback for expired alarm timers. + * + * Return: whether the timer is to be restarted */ -static enum alarmtimer_restart alarm_handle_timer(struct alarm *alarm, - ktime_t now) +static void alarm_handle_timer(struct alarm *alarm, ktime_t now) { - struct k_itimer *ptr = container_of(alarm, struct k_itimer, - it.alarm.alarmtimer); - enum alarmtimer_restart result = ALARMTIMER_NORESTART; - unsigned long flags; - int si_private = 0; + struct k_itimer *ptr = container_of(alarm, struct k_itimer, it.alarm.alarmtimer); - spin_lock_irqsave(&ptr->it_lock, flags); - - ptr->it_active = 0; - if (ptr->it_interval) - si_private = ++ptr->it_requeue_pending; - - if (posix_timer_event(ptr, si_private) && ptr->it_interval) { - /* - * Handle ignored signals and rearm the timer. This will go - * away once we handle ignored signals proper. - */ - ptr->it_overrun += alarm_forward_now(alarm, ptr->it_interval); - ++ptr->it_requeue_pending; - ptr->it_active = 1; - result = ALARMTIMER_RESTART; - } - spin_unlock_irqrestore(&ptr->it_lock, flags); - - return result; + guard(spinlock_irqsave)(&ptr->it_lock); + posix_timer_queue_signal(ptr); } /** @@ -715,24 +692,24 @@ static int alarm_timer_create(struct k_itimer *new_timer) /** * alarmtimer_nsleep_wakeup - Wakeup function for alarm_timer_nsleep * @alarm: ptr to alarm that fired + * @now: time at the timer expiration * * Wakes up the task that set the alarmtimer */ -static enum alarmtimer_restart alarmtimer_nsleep_wakeup(struct alarm *alarm, - ktime_t now) +static void alarmtimer_nsleep_wakeup(struct alarm *alarm, ktime_t now) { - struct task_struct *task = (struct task_struct *)alarm->data; + struct task_struct *task = alarm->data; alarm->data = NULL; if (task) wake_up_process(task); - return ALARMTIMER_NORESTART; } /** * alarmtimer_do_nsleep - Internal alarmtimer nsleep implementation * @alarm: ptr to alarmtimer * @absexp: absolute expiration time + * @type: alarm type (BOOTTIME/REALTIME). * * Sets the alarm timer and sleeps until it is fired or interrupted. */ @@ -777,10 +754,10 @@ static int alarmtimer_do_nsleep(struct alarm *alarm, ktime_t absexp, static void alarm_init_on_stack(struct alarm *alarm, enum alarmtimer_type type, - enum alarmtimer_restart (*function)(struct alarm *, ktime_t)) + void (*function)(struct alarm *, ktime_t)) { - hrtimer_init_on_stack(&alarm->timer, alarm_bases[type].base_clockid, - HRTIMER_MODE_ABS); + hrtimer_setup_on_stack(&alarm->timer, alarmtimer_fired, alarm_bases[type].base_clockid, + HRTIMER_MODE_ABS); __alarm_init(alarm, type, function); } @@ -804,9 +781,8 @@ static long __sched alarm_timer_nsleep_restart(struct restart_block *restart) /** * alarm_timer_nsleep - alarmtimer nanosleep * @which_clock: clockid - * @flags: determins abstime or relative + * @flags: determines abstime or relative * @tsreq: requested sleep time (abs or rel) - * @rmtp: remaining sleep time saved * * Handles clock_nanosleep calls against _ALARM clockids */ @@ -817,7 +793,7 @@ static int alarm_timer_nsleep(const clockid_t which_clock, int flags, struct restart_block *restart = ¤t->restart_block; struct alarm alarm; ktime_t exp; - int ret = 0; + int ret; if (!alarmtimer_get_rtcdev()) return -EOPNOTSUPP; @@ -848,9 +824,9 @@ static int alarm_timer_nsleep(const clockid_t which_clock, int flags, if (flags == TIMER_ABSTIME) return -ERESTARTNOHAND; - restart->fn = alarm_timer_nsleep_restart; restart->nanosleep.clockid = type; restart->nanosleep.expires = exp; + set_restart_fn(restart, alarm_timer_nsleep_restart); return ret; } @@ -908,7 +884,7 @@ static int __init alarmtimer_init(void) /* Initialize alarm bases */ alarm_bases[ALARM_REALTIME].base_clockid = CLOCK_REALTIME; alarm_bases[ALARM_REALTIME].get_ktime = &ktime_get_real; - alarm_bases[ALARM_REALTIME].get_timespec = ktime_get_real_ts64, + alarm_bases[ALARM_REALTIME].get_timespec = ktime_get_real_ts64; alarm_bases[ALARM_BOOTTIME].base_clockid = CLOCK_BOOTTIME; alarm_bases[ALARM_BOOTTIME].get_ktime = &ktime_get_boottime; alarm_bases[ALARM_BOOTTIME].get_timespec = get_boottime_timespec; diff --git a/kernel/time/clockevents.c b/kernel/time/clockevents.c index f5490222e134..f3e831f62906 100644 --- a/kernel/time/clockevents.c +++ b/kernel/time/clockevents.c @@ -76,7 +76,7 @@ static u64 cev_delta2ns(unsigned long latch, struct clock_event_device *evt, } /** - * clockevents_delta2ns - Convert a latch value (device ticks) to nanoseconds + * clockevent_delta2ns - Convert a latch value (device ticks) to nanoseconds * @latch: value to convert * @evt: pointer to clock event device descriptor * @@ -190,7 +190,7 @@ int clockevents_tick_resume(struct clock_event_device *dev) #ifdef CONFIG_GENERIC_CLOCKEVENTS_MIN_ADJUST -/* Limit min_delta to a jiffie */ +/* Limit min_delta to a jiffy */ #define MIN_DELTA_LIMIT (NSEC_PER_SEC / HZ) /** @@ -337,18 +337,25 @@ int clockevents_program_event(struct clock_event_device *dev, ktime_t expires, } /* - * Called after a notify add to make devices available which were - * released from the notifier call. + * Called after a clockevent has been added which might + * have replaced a current regular or broadcast device. A + * released normal device might be a suitable replacement + * for the current broadcast device. Similarly a released + * broadcast device might be a suitable replacement for a + * normal device. */ static void clockevents_notify_released(void) { struct clock_event_device *dev; + /* + * Keep iterating as long as tick_check_new_device() + * replaces a device. + */ while (!list_empty(&clockevents_released)) { dev = list_entry(clockevents_released.next, struct clock_event_device, list); - list_del(&dev->list); - list_add(&dev->list, &clockevent_devices); + list_move(&dev->list, &clockevent_devices); tick_check_new_device(dev); } } @@ -576,8 +583,7 @@ void clockevents_exchange_device(struct clock_event_device *old, if (old) { module_put(old->owner); clockevents_switch_state(old, CLOCK_EVT_STATE_DETACHED); - list_del(&old->list); - list_add(&old->list, &clockevents_released); + list_move(&old->list, &clockevents_released); } if (new) { @@ -612,38 +618,30 @@ void clockevents_resume(void) #ifdef CONFIG_HOTPLUG_CPU -# ifdef CONFIG_GENERIC_CLOCKEVENTS_BROADCAST /** - * tick_offline_cpu - Take CPU out of the broadcast mechanism + * tick_offline_cpu - Shutdown all clock events related + * to this CPU and take it out of the + * broadcast mechanism. * @cpu: The outgoing CPU * - * Called on the outgoing CPU after it took itself offline. + * Called by the dying CPU during teardown. */ void tick_offline_cpu(unsigned int cpu) { - raw_spin_lock(&clockevents_lock); - tick_broadcast_offline(cpu); - raw_spin_unlock(&clockevents_lock); -} -# endif - -/** - * tick_cleanup_dead_cpu - Cleanup the tick and clockevents of a dead cpu - */ -void tick_cleanup_dead_cpu(int cpu) -{ struct clock_event_device *dev, *tmp; - unsigned long flags; - raw_spin_lock_irqsave(&clockevents_lock, flags); + raw_spin_lock(&clockevents_lock); + tick_broadcast_offline(cpu); tick_shutdown(cpu); + /* * Unregister the clock event devices which were - * released from the users in the notify chain. + * released above. */ list_for_each_entry_safe(dev, tmp, &clockevents_released, list) list_del(&dev->list); + /* * Now check whether the CPU has left unused per cpu devices */ @@ -655,12 +653,13 @@ void tick_cleanup_dead_cpu(int cpu) list_del(&dev->list); } } - raw_spin_unlock_irqrestore(&clockevents_lock, flags); + + raw_spin_unlock(&clockevents_lock); } #endif #ifdef CONFIG_SYSFS -static struct bus_type clockevents_subsys = { +static const struct bus_type clockevents_subsys = { .name = "clockevents", .dev_name = "clockevent", }; @@ -668,9 +667,9 @@ static struct bus_type clockevents_subsys = { static DEFINE_PER_CPU(struct device, tick_percpu_dev); static struct tick_device *tick_get_tick_dev(struct device *dev); -static ssize_t sysfs_show_current_tick_dev(struct device *dev, - struct device_attribute *attr, - char *buf) +static ssize_t current_device_show(struct device *dev, + struct device_attribute *attr, + char *buf) { struct tick_device *td; ssize_t count = 0; @@ -678,20 +677,20 @@ static ssize_t sysfs_show_current_tick_dev(struct device *dev, raw_spin_lock_irq(&clockevents_lock); td = tick_get_tick_dev(dev); if (td && td->evtdev) - count = snprintf(buf, PAGE_SIZE, "%s\n", td->evtdev->name); + count = sysfs_emit(buf, "%s\n", td->evtdev->name); raw_spin_unlock_irq(&clockevents_lock); return count; } -static DEVICE_ATTR(current_device, 0444, sysfs_show_current_tick_dev, NULL); +static DEVICE_ATTR_RO(current_device); /* 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) +static ssize_t unbind_device_store(struct device *dev, + struct device_attribute *attr, + const char *buf, size_t count) { char name[CS_NAME_LEN]; ssize_t ret = sysfs_get_uname(buf, name, count); - struct clock_event_device *ce; + struct clock_event_device *ce = NULL, *iter; if (ret < 0) return ret; @@ -699,9 +698,10 @@ static ssize_t sysfs_unbind_tick_dev(struct device *dev, 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); + list_for_each_entry(iter, &clockevent_devices, list) { + if (!strcmp(iter->name, name)) { + ret = __clockevents_try_unbind(iter, dev->id); + ce = iter; break; } } @@ -714,7 +714,7 @@ static ssize_t sysfs_unbind_tick_dev(struct device *dev, mutex_unlock(&clockevents_mutex); return ret ? ret : count; } -static DEVICE_ATTR(unbind_device, 0200, NULL, sysfs_unbind_tick_dev); +static DEVICE_ATTR_WO(unbind_device); #ifdef CONFIG_GENERIC_CLOCKEVENTS_BROADCAST static struct device tick_bc_dev = { diff --git a/kernel/time/clocksource-wdtest.c b/kernel/time/clocksource-wdtest.c new file mode 100644 index 000000000000..38dae590b29f --- /dev/null +++ b/kernel/time/clocksource-wdtest.c @@ -0,0 +1,204 @@ +// SPDX-License-Identifier: GPL-2.0+ +/* + * Unit test for the clocksource watchdog. + * + * Copyright (C) 2021 Facebook, Inc. + * + * Author: Paul E. McKenney <paulmck@kernel.org> + */ +#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt + +#include <linux/device.h> +#include <linux/clocksource.h> +#include <linux/init.h> +#include <linux/module.h> +#include <linux/sched.h> /* for spin_unlock_irq() using preempt_count() m68k */ +#include <linux/tick.h> +#include <linux/kthread.h> +#include <linux/delay.h> +#include <linux/prandom.h> +#include <linux/cpu.h> + +#include "tick-internal.h" + +MODULE_LICENSE("GPL"); +MODULE_DESCRIPTION("Clocksource watchdog unit test"); +MODULE_AUTHOR("Paul E. McKenney <paulmck@kernel.org>"); + +static int holdoff = IS_BUILTIN(CONFIG_TEST_CLOCKSOURCE_WATCHDOG) ? 10 : 0; +module_param(holdoff, int, 0444); +MODULE_PARM_DESC(holdoff, "Time to wait to start test (s)."); + +/* Watchdog kthread's task_struct pointer for debug purposes. */ +static struct task_struct *wdtest_task; + +static u64 wdtest_jiffies_read(struct clocksource *cs) +{ + return (u64)jiffies; +} + +static struct clocksource clocksource_wdtest_jiffies = { + .name = "wdtest-jiffies", + .rating = 1, /* lowest valid rating*/ + .uncertainty_margin = TICK_NSEC, + .read = wdtest_jiffies_read, + .mask = CLOCKSOURCE_MASK(32), + .flags = CLOCK_SOURCE_MUST_VERIFY, + .mult = TICK_NSEC << JIFFIES_SHIFT, /* details above */ + .shift = JIFFIES_SHIFT, + .max_cycles = 10, +}; + +static int wdtest_ktime_read_ndelays; +static bool wdtest_ktime_read_fuzz; + +static u64 wdtest_ktime_read(struct clocksource *cs) +{ + int wkrn = READ_ONCE(wdtest_ktime_read_ndelays); + static int sign = 1; + u64 ret; + + if (wkrn) { + udelay(cs->uncertainty_margin / 250); + WRITE_ONCE(wdtest_ktime_read_ndelays, wkrn - 1); + } + ret = ktime_get_real_fast_ns(); + if (READ_ONCE(wdtest_ktime_read_fuzz)) { + sign = -sign; + ret = ret + sign * 100 * NSEC_PER_MSEC; + } + return ret; +} + +static void wdtest_ktime_cs_mark_unstable(struct clocksource *cs) +{ + pr_info("--- Marking %s unstable due to clocksource watchdog.\n", cs->name); +} + +#define KTIME_FLAGS (CLOCK_SOURCE_IS_CONTINUOUS | \ + CLOCK_SOURCE_VALID_FOR_HRES | \ + CLOCK_SOURCE_MUST_VERIFY | \ + CLOCK_SOURCE_VERIFY_PERCPU) + +static struct clocksource clocksource_wdtest_ktime = { + .name = "wdtest-ktime", + .rating = 300, + .read = wdtest_ktime_read, + .mask = CLOCKSOURCE_MASK(64), + .flags = KTIME_FLAGS, + .mark_unstable = wdtest_ktime_cs_mark_unstable, + .list = LIST_HEAD_INIT(clocksource_wdtest_ktime.list), +}; + +/* Reset the clocksource if needed. */ +static void wdtest_ktime_clocksource_reset(void) +{ + if (clocksource_wdtest_ktime.flags & CLOCK_SOURCE_UNSTABLE) { + clocksource_unregister(&clocksource_wdtest_ktime); + clocksource_wdtest_ktime.flags = KTIME_FLAGS; + schedule_timeout_uninterruptible(HZ / 10); + clocksource_register_khz(&clocksource_wdtest_ktime, 1000 * 1000); + } +} + +/* Run the specified series of watchdog tests. */ +static int wdtest_func(void *arg) +{ + unsigned long j1, j2; + int i, max_retries; + char *s; + + schedule_timeout_uninterruptible(holdoff * HZ); + + /* + * Verify that jiffies-like clocksources get the manually + * specified uncertainty margin. + */ + pr_info("--- Verify jiffies-like uncertainty margin.\n"); + __clocksource_register(&clocksource_wdtest_jiffies); + WARN_ON_ONCE(clocksource_wdtest_jiffies.uncertainty_margin != TICK_NSEC); + + j1 = clocksource_wdtest_jiffies.read(&clocksource_wdtest_jiffies); + schedule_timeout_uninterruptible(HZ); + j2 = clocksource_wdtest_jiffies.read(&clocksource_wdtest_jiffies); + WARN_ON_ONCE(j1 == j2); + + clocksource_unregister(&clocksource_wdtest_jiffies); + + /* + * Verify that tsc-like clocksources are assigned a reasonable + * uncertainty margin. + */ + pr_info("--- Verify tsc-like uncertainty margin.\n"); + clocksource_register_khz(&clocksource_wdtest_ktime, 1000 * 1000); + WARN_ON_ONCE(clocksource_wdtest_ktime.uncertainty_margin < NSEC_PER_USEC); + + j1 = clocksource_wdtest_ktime.read(&clocksource_wdtest_ktime); + udelay(1); + j2 = clocksource_wdtest_ktime.read(&clocksource_wdtest_ktime); + pr_info("--- tsc-like times: %lu - %lu = %lu.\n", j2, j1, j2 - j1); + WARN_ONCE(time_before(j2, j1 + NSEC_PER_USEC), + "Expected at least 1000ns, got %lu.\n", j2 - j1); + + /* Verify tsc-like stability with various numbers of errors injected. */ + max_retries = clocksource_get_max_watchdog_retry(); + for (i = 0; i <= max_retries + 1; i++) { + if (i <= 1 && i < max_retries) + s = ""; + else if (i <= max_retries) + s = ", expect message"; + else + s = ", expect clock skew"; + pr_info("--- Watchdog with %dx error injection, %d retries%s.\n", i, max_retries, s); + WRITE_ONCE(wdtest_ktime_read_ndelays, i); + schedule_timeout_uninterruptible(2 * HZ); + WARN_ON_ONCE(READ_ONCE(wdtest_ktime_read_ndelays)); + WARN_ON_ONCE((i <= max_retries) != + !(clocksource_wdtest_ktime.flags & CLOCK_SOURCE_UNSTABLE)); + wdtest_ktime_clocksource_reset(); + } + + /* Verify tsc-like stability with clock-value-fuzz error injection. */ + pr_info("--- Watchdog clock-value-fuzz error injection, expect clock skew and per-CPU mismatches.\n"); + WRITE_ONCE(wdtest_ktime_read_fuzz, true); + schedule_timeout_uninterruptible(2 * HZ); + WARN_ON_ONCE(!(clocksource_wdtest_ktime.flags & CLOCK_SOURCE_UNSTABLE)); + clocksource_verify_percpu(&clocksource_wdtest_ktime); + WRITE_ONCE(wdtest_ktime_read_fuzz, false); + + clocksource_unregister(&clocksource_wdtest_ktime); + + pr_info("--- Done with test.\n"); + return 0; +} + +static void wdtest_print_module_parms(void) +{ + pr_alert("--- holdoff=%d\n", holdoff); +} + +/* Cleanup function. */ +static void clocksource_wdtest_cleanup(void) +{ +} + +static int __init clocksource_wdtest_init(void) +{ + int ret = 0; + + wdtest_print_module_parms(); + + /* Create watchdog-test task. */ + wdtest_task = kthread_run(wdtest_func, NULL, "wdtest"); + if (IS_ERR(wdtest_task)) { + ret = PTR_ERR(wdtest_task); + pr_warn("%s: Failed to create wdtest kthread.\n", __func__); + wdtest_task = NULL; + return ret; + } + + return 0; +} + +module_init(clocksource_wdtest_init); +module_exit(clocksource_wdtest_cleanup); diff --git a/kernel/time/clocksource.c b/kernel/time/clocksource.c index 02441ead3c3b..2a7802ec480c 100644 --- a/kernel/time/clocksource.c +++ b/kernel/time/clocksource.c @@ -14,10 +14,24 @@ #include <linux/sched.h> /* for spin_unlock_irq() using preempt_count() m68k */ #include <linux/tick.h> #include <linux/kthread.h> +#include <linux/prandom.h> +#include <linux/cpu.h> #include "tick-internal.h" #include "timekeeping_internal.h" +static void clocksource_enqueue(struct clocksource *cs); + +static noinline u64 cycles_to_nsec_safe(struct clocksource *cs, u64 start, u64 end) +{ + u64 delta = clocksource_delta(end, start, cs->mask, cs->max_raw_delta); + + if (likely(delta < cs->max_cycles)) + return clocksource_cyc2ns(delta, cs->mult, cs->shift); + + return mul_u64_u32_shr(delta, cs->mult, cs->shift); +} + /** * clocks_calc_mult_shift - calculate mult/shift factors for scaled math of clocks * @mult: pointer to mult variable @@ -38,7 +52,7 @@ * calculated mult and shift factors. This guarantees that no 64bit * overflow happens when the input value of the conversion is * multiplied with the calculated mult factor. Larger ranges may - * reduce the conversion accuracy by chosing smaller mult and shift + * reduce the conversion accuracy by choosing smaller mult and shift * factors. */ void @@ -93,6 +107,48 @@ static char override_name[CS_NAME_LEN]; static int finished_booting; static u64 suspend_start; +/* + * Interval: 0.5sec. + */ +#define WATCHDOG_INTERVAL (HZ >> 1) +#define WATCHDOG_INTERVAL_MAX_NS ((2 * WATCHDOG_INTERVAL) * (NSEC_PER_SEC / HZ)) + +/* + * Threshold: 0.0312s, when doubled: 0.0625s. + */ +#define WATCHDOG_THRESHOLD (NSEC_PER_SEC >> 5) + +/* + * Maximum permissible delay between two readouts of the watchdog + * clocksource surrounding a read of the clocksource being validated. + * This delay could be due to SMIs, NMIs, or to VCPU preemptions. Used as + * a lower bound for cs->uncertainty_margin values when registering clocks. + * + * The default of 500 parts per million is based on NTP's limits. + * If a clocksource is good enough for NTP, it is good enough for us! + * + * In other words, by default, even if a clocksource is extremely + * precise (for example, with a sub-nanosecond period), the maximum + * permissible skew between the clocksource watchdog and the clocksource + * under test is not permitted to go below the 500ppm minimum defined + * by MAX_SKEW_USEC. This 500ppm minimum may be overridden using the + * CLOCKSOURCE_WATCHDOG_MAX_SKEW_US Kconfig option. + */ +#ifdef CONFIG_CLOCKSOURCE_WATCHDOG_MAX_SKEW_US +#define MAX_SKEW_USEC CONFIG_CLOCKSOURCE_WATCHDOG_MAX_SKEW_US +#else +#define MAX_SKEW_USEC (125 * WATCHDOG_INTERVAL / HZ) +#endif + +/* + * Default for maximum permissible skew when cs->uncertainty_margin is + * not specified, and the lower bound even when cs->uncertainty_margin + * is specified. This is also the default that is used when registering + * clocks with unspecifed cs->uncertainty_margin, so this macro is used + * even in CONFIG_CLOCKSOURCE_WATCHDOG=n kernels. + */ +#define WATCHDOG_MAX_SKEW (MAX_SKEW_USEC * NSEC_PER_USEC) + #ifdef CONFIG_CLOCKSOURCE_WATCHDOG static void clocksource_watchdog_work(struct work_struct *work); static void clocksource_select(void); @@ -104,6 +160,7 @@ static DECLARE_WORK(watchdog_work, clocksource_watchdog_work); static DEFINE_SPINLOCK(watchdog_lock); static int watchdog_running; static atomic_t watchdog_reset_pending; +static int64_t watchdog_max_interval; static inline void clocksource_watchdog_lock(unsigned long *flags) { @@ -116,13 +173,6 @@ static inline void clocksource_watchdog_unlock(unsigned long *flags) } static int clocksource_watchdog_kthread(void *data); -static void __clocksource_change_rating(struct clocksource *cs, int rating); - -/* - * Interval: 0.5sec Threshold: 0.0625s - */ -#define WATCHDOG_INTERVAL (HZ >> 1) -#define WATCHDOG_THRESHOLD (NSEC_PER_SEC >> 4) static void clocksource_watchdog_work(struct work_struct *work) { @@ -142,6 +192,13 @@ static void clocksource_watchdog_work(struct work_struct *work) kthread_run(clocksource_watchdog_kthread, NULL, "kwatchdog"); } +static void clocksource_change_rating(struct clocksource *cs, int rating) +{ + list_del(&cs->list); + cs->rating = rating; + clocksource_enqueue(cs); +} + static void __clocksource_unstable(struct clocksource *cs) { cs->flags &= ~(CLOCK_SOURCE_VALID_FOR_HRES | CLOCK_SOURCE_WATCHDOG); @@ -184,12 +241,201 @@ void clocksource_mark_unstable(struct clocksource *cs) spin_unlock_irqrestore(&watchdog_lock, flags); } -static void clocksource_watchdog(struct timer_list *unused) +static int verify_n_cpus = 8; +module_param(verify_n_cpus, int, 0644); + +enum wd_read_status { + WD_READ_SUCCESS, + WD_READ_UNSTABLE, + WD_READ_SKIP +}; + +static enum wd_read_status cs_watchdog_read(struct clocksource *cs, u64 *csnow, u64 *wdnow) +{ + int64_t md = 2 * watchdog->uncertainty_margin; + unsigned int nretries, max_retries; + int64_t wd_delay, wd_seq_delay; + u64 wd_end, wd_end2; + + max_retries = clocksource_get_max_watchdog_retry(); + for (nretries = 0; nretries <= max_retries; nretries++) { + local_irq_disable(); + *wdnow = watchdog->read(watchdog); + *csnow = cs->read(cs); + wd_end = watchdog->read(watchdog); + wd_end2 = watchdog->read(watchdog); + local_irq_enable(); + + wd_delay = cycles_to_nsec_safe(watchdog, *wdnow, wd_end); + if (wd_delay <= md + cs->uncertainty_margin) { + if (nretries > 1 && nretries >= max_retries) { + pr_warn("timekeeping watchdog on CPU%d: %s retried %d times before success\n", + smp_processor_id(), watchdog->name, nretries); + } + return WD_READ_SUCCESS; + } + + /* + * Now compute delay in consecutive watchdog read to see if + * there is too much external interferences that cause + * significant delay in reading both clocksource and watchdog. + * + * If consecutive WD read-back delay > md, report + * system busy, reinit the watchdog and skip the current + * watchdog test. + */ + wd_seq_delay = cycles_to_nsec_safe(watchdog, wd_end, wd_end2); + if (wd_seq_delay > md) + goto skip_test; + } + + pr_warn("timekeeping watchdog on CPU%d: wd-%s-wd excessive read-back delay of %lldns vs. limit of %ldns, wd-wd read-back delay only %lldns, attempt %d, marking %s unstable\n", + smp_processor_id(), cs->name, wd_delay, WATCHDOG_MAX_SKEW, wd_seq_delay, nretries, cs->name); + return WD_READ_UNSTABLE; + +skip_test: + pr_info("timekeeping watchdog on CPU%d: %s wd-wd read-back delay of %lldns\n", + smp_processor_id(), watchdog->name, wd_seq_delay); + pr_info("wd-%s-wd read-back delay of %lldns, clock-skew test skipped!\n", + cs->name, wd_delay); + return WD_READ_SKIP; +} + +static u64 csnow_mid; +static cpumask_t cpus_ahead; +static cpumask_t cpus_behind; +static cpumask_t cpus_chosen; + +static void clocksource_verify_choose_cpus(void) +{ + int cpu, i, n = verify_n_cpus; + + if (n < 0) { + /* Check all of the CPUs. */ + cpumask_copy(&cpus_chosen, cpu_online_mask); + cpumask_clear_cpu(smp_processor_id(), &cpus_chosen); + return; + } + + /* If no checking desired, or no other CPU to check, leave. */ + cpumask_clear(&cpus_chosen); + if (n == 0 || num_online_cpus() <= 1) + return; + + /* Make sure to select at least one CPU other than the current CPU. */ + cpu = cpumask_first(cpu_online_mask); + if (cpu == smp_processor_id()) + cpu = cpumask_next(cpu, cpu_online_mask); + if (WARN_ON_ONCE(cpu >= nr_cpu_ids)) + return; + cpumask_set_cpu(cpu, &cpus_chosen); + + /* Force a sane value for the boot parameter. */ + if (n > nr_cpu_ids) + n = nr_cpu_ids; + + /* + * Randomly select the specified number of CPUs. If the same + * CPU is selected multiple times, that CPU is checked only once, + * and no replacement CPU is selected. This gracefully handles + * situations where verify_n_cpus is greater than the number of + * CPUs that are currently online. + */ + for (i = 1; i < n; i++) { + cpu = get_random_u32_below(nr_cpu_ids); + cpu = cpumask_next(cpu - 1, cpu_online_mask); + if (cpu >= nr_cpu_ids) + cpu = cpumask_first(cpu_online_mask); + if (!WARN_ON_ONCE(cpu >= nr_cpu_ids)) + cpumask_set_cpu(cpu, &cpus_chosen); + } + + /* Don't verify ourselves. */ + cpumask_clear_cpu(smp_processor_id(), &cpus_chosen); +} + +static void clocksource_verify_one_cpu(void *csin) +{ + struct clocksource *cs = (struct clocksource *)csin; + + csnow_mid = cs->read(cs); +} + +void clocksource_verify_percpu(struct clocksource *cs) +{ + int64_t cs_nsec, cs_nsec_max = 0, cs_nsec_min = LLONG_MAX; + u64 csnow_begin, csnow_end; + int cpu, testcpu; + s64 delta; + + if (verify_n_cpus == 0) + return; + cpumask_clear(&cpus_ahead); + cpumask_clear(&cpus_behind); + cpus_read_lock(); + migrate_disable(); + clocksource_verify_choose_cpus(); + if (cpumask_empty(&cpus_chosen)) { + migrate_enable(); + cpus_read_unlock(); + pr_warn("Not enough CPUs to check clocksource '%s'.\n", cs->name); + return; + } + testcpu = smp_processor_id(); + pr_info("Checking clocksource %s synchronization from CPU %d to CPUs %*pbl.\n", + cs->name, testcpu, cpumask_pr_args(&cpus_chosen)); + preempt_disable(); + for_each_cpu(cpu, &cpus_chosen) { + if (cpu == testcpu) + continue; + csnow_begin = cs->read(cs); + smp_call_function_single(cpu, clocksource_verify_one_cpu, cs, 1); + csnow_end = cs->read(cs); + delta = (s64)((csnow_mid - csnow_begin) & cs->mask); + if (delta < 0) + cpumask_set_cpu(cpu, &cpus_behind); + delta = (csnow_end - csnow_mid) & cs->mask; + if (delta < 0) + cpumask_set_cpu(cpu, &cpus_ahead); + cs_nsec = cycles_to_nsec_safe(cs, csnow_begin, csnow_end); + if (cs_nsec > cs_nsec_max) + cs_nsec_max = cs_nsec; + if (cs_nsec < cs_nsec_min) + cs_nsec_min = cs_nsec; + } + preempt_enable(); + migrate_enable(); + cpus_read_unlock(); + if (!cpumask_empty(&cpus_ahead)) + pr_warn(" CPUs %*pbl ahead of CPU %d for clocksource %s.\n", + cpumask_pr_args(&cpus_ahead), testcpu, cs->name); + if (!cpumask_empty(&cpus_behind)) + pr_warn(" CPUs %*pbl behind CPU %d for clocksource %s.\n", + cpumask_pr_args(&cpus_behind), testcpu, cs->name); + if (!cpumask_empty(&cpus_ahead) || !cpumask_empty(&cpus_behind)) + pr_warn(" CPU %d check durations %lldns - %lldns for clocksource %s.\n", + testcpu, cs_nsec_min, cs_nsec_max, cs->name); +} +EXPORT_SYMBOL_GPL(clocksource_verify_percpu); + +static inline void clocksource_reset_watchdog(void) { struct clocksource *cs; - u64 csnow, wdnow, cslast, wdlast, delta; - int64_t wd_nsec, cs_nsec; + + list_for_each_entry(cs, &watchdog_list, wd_list) + cs->flags &= ~CLOCK_SOURCE_WATCHDOG; +} + + +static void clocksource_watchdog(struct timer_list *unused) +{ + int64_t wd_nsec, cs_nsec, interval; + u64 csnow, wdnow, cslast, wdlast; int next_cpu, reset_pending; + struct clocksource *cs; + enum wd_read_status read_ret; + unsigned long extra_wait = 0; + u32 md; spin_lock(&watchdog_lock); if (!watchdog_running) @@ -206,10 +452,31 @@ static void clocksource_watchdog(struct timer_list *unused) continue; } - local_irq_disable(); - csnow = cs->read(cs); - wdnow = watchdog->read(watchdog); - local_irq_enable(); + read_ret = cs_watchdog_read(cs, &csnow, &wdnow); + + if (read_ret == WD_READ_UNSTABLE) { + /* Clock readout unreliable, so give it up. */ + __clocksource_unstable(cs); + continue; + } + + /* + * When WD_READ_SKIP is returned, it means the system is likely + * under very heavy load, where the latency of reading + * watchdog/clocksource is very big, and affect the accuracy of + * watchdog check. So give system some space and suspend the + * watchdog check for 5 minutes. + */ + if (read_ret == WD_READ_SKIP) { + /* + * As the watchdog timer will be suspended, and + * cs->last could keep unchanged for 5 minutes, reset + * the counters. + */ + clocksource_reset_watchdog(); + extra_wait = HZ * 300; + break; + } /* Clocksource initialized ? */ if (!(cs->flags & CLOCK_SOURCE_WATCHDOG) || @@ -220,12 +487,8 @@ static void clocksource_watchdog(struct timer_list *unused) continue; } - delta = clocksource_delta(wdnow, cs->wd_last, watchdog->mask); - wd_nsec = clocksource_cyc2ns(delta, watchdog->mult, - watchdog->shift); - - delta = clocksource_delta(csnow, cs->cs_last, cs->mask); - cs_nsec = clocksource_cyc2ns(delta, cs->mult, cs->shift); + wd_nsec = cycles_to_nsec_safe(watchdog, cs->wd_last, wdnow); + cs_nsec = cycles_to_nsec_safe(cs, cs->cs_last, csnow); wdlast = cs->wd_last; /* save these in case we print them */ cslast = cs->cs_last; cs->cs_last = csnow; @@ -234,14 +497,50 @@ static void clocksource_watchdog(struct timer_list *unused) if (atomic_read(&watchdog_reset_pending)) continue; + /* + * The processing of timer softirqs can get delayed (usually + * on account of ksoftirqd not getting to run in a timely + * manner), which causes the watchdog interval to stretch. + * Skew detection may fail for longer watchdog intervals + * on account of fixed margins being used. + * Some clocksources, e.g. acpi_pm, cannot tolerate + * watchdog intervals longer than a few seconds. + */ + interval = max(cs_nsec, wd_nsec); + if (unlikely(interval > WATCHDOG_INTERVAL_MAX_NS)) { + if (system_state > SYSTEM_SCHEDULING && + interval > 2 * watchdog_max_interval) { + watchdog_max_interval = interval; + pr_warn("Long readout interval, skipping watchdog check: cs_nsec: %lld wd_nsec: %lld\n", + cs_nsec, wd_nsec); + } + watchdog_timer.expires = jiffies; + continue; + } + /* Check the deviation from the watchdog clocksource. */ - if (abs(cs_nsec - wd_nsec) > WATCHDOG_THRESHOLD) { + md = cs->uncertainty_margin + watchdog->uncertainty_margin; + if (abs(cs_nsec - wd_nsec) > md) { + s64 cs_wd_msec; + s64 wd_msec; + u32 wd_rem; + pr_warn("timekeeping watchdog on CPU%d: Marking clocksource '%s' as unstable because the skew is too large:\n", smp_processor_id(), cs->name); - pr_warn(" '%s' wd_now: %llx wd_last: %llx mask: %llx\n", - watchdog->name, wdnow, wdlast, watchdog->mask); - pr_warn(" '%s' cs_now: %llx cs_last: %llx mask: %llx\n", - cs->name, csnow, cslast, cs->mask); + pr_warn(" '%s' wd_nsec: %lld wd_now: %llx wd_last: %llx mask: %llx\n", + watchdog->name, wd_nsec, wdnow, wdlast, watchdog->mask); + pr_warn(" '%s' cs_nsec: %lld cs_now: %llx cs_last: %llx mask: %llx\n", + cs->name, cs_nsec, csnow, cslast, cs->mask); + cs_wd_msec = div_s64_rem(cs_nsec - wd_nsec, 1000 * 1000, &wd_rem); + wd_msec = div_s64_rem(wd_nsec, 1000 * 1000, &wd_rem); + pr_warn(" Clocksource '%s' skewed %lld ns (%lld ms) over watchdog '%s' interval of %lld ns (%lld ms)\n", + cs->name, cs_nsec - wd_nsec, cs_wd_msec, watchdog->name, wd_nsec, wd_msec); + if (curr_clocksource == cs) + pr_warn(" '%s' is current clocksource.\n", cs->name); + else if (curr_clocksource) + pr_warn(" '%s' (not '%s') is current clocksource.\n", curr_clocksource->name, cs->name); + else + pr_warn(" No current clocksource.\n"); __clocksource_unstable(cs); continue; } @@ -299,7 +598,7 @@ static void clocksource_watchdog(struct timer_list *unused) * pair clocksource_stop_watchdog() clocksource_start_watchdog(). */ if (!timer_pending(&watchdog_timer)) { - watchdog_timer.expires += WATCHDOG_INTERVAL; + watchdog_timer.expires += WATCHDOG_INTERVAL + extra_wait; add_timer_on(&watchdog_timer, next_cpu); } out: @@ -324,14 +623,6 @@ static inline void clocksource_stop_watchdog(void) watchdog_running = 0; } -static inline void clocksource_reset_watchdog(void) -{ - struct clocksource *cs; - - list_for_each_entry(cs, &watchdog_list, wd_list) - cs->flags &= ~CLOCK_SOURCE_WATCHDOG; -} - static void clocksource_resume_watchdog(void) { atomic_inc(&watchdog_reset_pending); @@ -407,11 +698,17 @@ static int __clocksource_watchdog_kthread(void) unsigned long flags; int select = 0; + /* Do any required per-CPU skew verification. */ + if (curr_clocksource && + curr_clocksource->flags & CLOCK_SOURCE_UNSTABLE && + curr_clocksource->flags & CLOCK_SOURCE_VERIFY_PERCPU) + clocksource_verify_percpu(curr_clocksource); + spin_lock_irqsave(&watchdog_lock, flags); list_for_each_entry_safe(cs, tmp, &watchdog_list, wd_list) { if (cs->flags & CLOCK_SOURCE_UNSTABLE) { list_del_init(&cs->wd_list); - __clocksource_change_rating(cs, 0); + clocksource_change_rating(cs, 0); select = 1; } if (cs->flags & CLOCK_SOURCE_RESELECT) { @@ -518,7 +815,7 @@ static void clocksource_suspend_select(bool fallback) * the suspend time when resuming system. * * This function is called late in the suspend process from timekeeping_suspend(), - * that means processes are freezed, non-boot cpus and interrupts are disabled + * that means processes are frozen, non-boot cpus and interrupts are disabled * now. It is therefore possible to start the suspend timer without taking the * clocksource mutex. */ @@ -562,7 +859,7 @@ void clocksource_start_suspend_timing(struct clocksource *cs, u64 start_cycles) */ u64 clocksource_stop_suspend_timing(struct clocksource *cs, u64 cycle_now) { - u64 now, delta, nsec = 0; + u64 now, nsec = 0; if (!suspend_clocksource) return 0; @@ -577,12 +874,8 @@ u64 clocksource_stop_suspend_timing(struct clocksource *cs, u64 cycle_now) else now = suspend_clocksource->read(suspend_clocksource); - if (now > suspend_start) { - delta = clocksource_delta(now, suspend_start, - suspend_clocksource->mask); - nsec = mul_u64_u32_shr(delta, suspend_clocksource->mult, - suspend_clocksource->shift); - } + if (now > suspend_start) + nsec = cycles_to_nsec_safe(suspend_clocksource, suspend_start, now); /* * Disable the suspend timer to save power if current clocksource is @@ -703,9 +996,16 @@ static inline void clocksource_update_max_deferment(struct clocksource *cs) cs->max_idle_ns = clocks_calc_max_nsecs(cs->mult, cs->shift, cs->maxadj, cs->mask, &cs->max_cycles); -} -#ifndef CONFIG_ARCH_USES_GETTIMEOFFSET + /* + * Threshold for detecting negative motion in clocksource_delta(). + * + * Allow for 0.875 of the counter width so that overly long idle + * sleeps, which go slightly over mask/2, do not trigger the + * negative motion detection. + */ + cs->max_raw_delta = (cs->mask >> 1) + (cs->mask >> 2) + (cs->mask >> 3); +} static struct clocksource *clocksource_find_best(bool oneshot, bool skipcur) { @@ -798,12 +1098,6 @@ static void clocksource_select_fallback(void) __clocksource_select(true); } -#else /* !CONFIG_ARCH_USES_GETTIMEOFFSET */ -static inline void clocksource_select(void) { } -static inline void clocksource_select_fallback(void) { } - -#endif - /* * clocksource_done_booting - Called near the end of core bootup * @@ -884,6 +1178,31 @@ void __clocksource_update_freq_scale(struct clocksource *cs, u32 scale, u32 freq clocks_calc_mult_shift(&cs->mult, &cs->shift, freq, NSEC_PER_SEC / scale, sec * scale); } + + /* + * If the uncertainty margin is not specified, calculate it. If + * both scale and freq are non-zero, calculate the clock period, but + * bound below at 2*WATCHDOG_MAX_SKEW, that is, 500ppm by default. + * However, if either of scale or freq is zero, be very conservative + * and take the tens-of-milliseconds WATCHDOG_THRESHOLD value + * for the uncertainty margin. Allow stupidly small uncertainty + * margins to be specified by the caller for testing purposes, + * but warn to discourage production use of this capability. + * + * Bottom line: The sum of the uncertainty margins of the + * watchdog clocksource and the clocksource under test will be at + * least 500ppm by default. For more information, please see the + * comment preceding CONFIG_CLOCKSOURCE_WATCHDOG_MAX_SKEW_US above. + */ + if (scale && freq && !cs->uncertainty_margin) { + cs->uncertainty_margin = NSEC_PER_SEC / (scale * freq); + if (cs->uncertainty_margin < 2 * WATCHDOG_MAX_SKEW) + cs->uncertainty_margin = 2 * WATCHDOG_MAX_SKEW; + } else if (!cs->uncertainty_margin) { + cs->uncertainty_margin = WATCHDOG_THRESHOLD; + } + WARN_ON_ONCE(cs->uncertainty_margin < 2 * WATCHDOG_MAX_SKEW); + /* * Ensure clocksources that have large 'mult' values don't overflow * when adjusted. @@ -928,6 +1247,8 @@ int __clocksource_register_scale(struct clocksource *cs, u32 scale, u32 freq) clocksource_arch_init(cs); + if (WARN_ON_ONCE((unsigned int)cs->id >= CSID_MAX)) + cs->id = CSID_GENERIC; if (cs->vdso_clock_mode < 0 || cs->vdso_clock_mode >= VDSO_CLOCKMODE_MAX) { pr_warn("clocksource %s registered with invalid VDSO mode %d. Disabling VDSO support.\n", @@ -954,34 +1275,6 @@ int __clocksource_register_scale(struct clocksource *cs, u32 scale, u32 freq) } EXPORT_SYMBOL_GPL(__clocksource_register_scale); -static void __clocksource_change_rating(struct clocksource *cs, int rating) -{ - list_del(&cs->list); - cs->rating = rating; - clocksource_enqueue(cs); -} - -/** - * clocksource_change_rating - Change the rating of a registered clocksource - * @cs: clocksource to be changed - * @rating: new rating - */ -void clocksource_change_rating(struct clocksource *cs, int rating) -{ - unsigned long flags; - - mutex_lock(&clocksource_mutex); - clocksource_watchdog_lock(&flags); - __clocksource_change_rating(cs, rating); - clocksource_watchdog_unlock(&flags); - - clocksource_select(); - clocksource_select_watchdog(false); - clocksource_suspend_select(false); - mutex_unlock(&clocksource_mutex); -} -EXPORT_SYMBOL(clocksource_change_rating); - /* * Unbind clocksource @cs. Called with clocksource_mutex held */ @@ -1052,7 +1345,7 @@ static ssize_t current_clocksource_show(struct device *dev, ssize_t count = 0; mutex_lock(&clocksource_mutex); - count = snprintf(buf, PAGE_SIZE, "%s\n", curr_clocksource->name); + count = sysfs_emit(buf, "%s\n", curr_clocksource->name); mutex_unlock(&clocksource_mutex); return count; @@ -1182,7 +1475,7 @@ static struct attribute *clocksource_attrs[] = { }; ATTRIBUTE_GROUPS(clocksource); -static struct bus_type clocksource_subsys = { +static const struct bus_type clocksource_subsys = { .name = "clocksource", .dev_name = "clocksource", }; @@ -1217,7 +1510,7 @@ static int __init boot_override_clocksource(char* str) { mutex_lock(&clocksource_mutex); if (str) - strlcpy(override_name, str, sizeof(override_name)); + strscpy(override_name, str, sizeof(override_name)); mutex_unlock(&clocksource_mutex); return 1; } diff --git a/kernel/time/hrtimer.c b/kernel/time/hrtimer.c index d89da1c7e005..deb1aa32814e 100644 --- a/kernel/time/hrtimer.c +++ b/kernel/time/hrtimer.c @@ -38,6 +38,7 @@ #include <linux/sched/deadline.h> #include <linux/sched/nohz.h> #include <linux/sched/debug.h> +#include <linux/sched/isolation.h> #include <linux/timer.h> #include <linux/freezer.h> #include <linux/compat.h> @@ -57,6 +58,8 @@ #define HRTIMER_ACTIVE_SOFT (HRTIMER_ACTIVE_HARD << MASK_SHIFT) #define HRTIMER_ACTIVE_ALL (HRTIMER_ACTIVE_SOFT | HRTIMER_ACTIVE_HARD) +static void retrigger_next_event(void *arg); + /* * The timer bases: * @@ -110,7 +113,8 @@ DEFINE_PER_CPU(struct hrtimer_cpu_base, hrtimer_bases) = .clockid = CLOCK_TAI, .get_time = &ktime_get_clocktai, }, - } + }, + .csd = CSD_INIT(retrigger_next_event, NULL) }; static const int hrtimer_clock_to_base_table[MAX_CLOCKS] = { @@ -123,6 +127,14 @@ static const int hrtimer_clock_to_base_table[MAX_CLOCKS] = { [CLOCK_TAI] = HRTIMER_BASE_TAI, }; +static inline bool hrtimer_base_is_online(struct hrtimer_cpu_base *base) +{ + if (!IS_ENABLED(CONFIG_HOTPLUG_CPU)) + return true; + else + return likely(base->online); +} + /* * Functions and macros which are different for UP/SMP systems are kept in a * single place @@ -135,16 +147,15 @@ static const int hrtimer_clock_to_base_table[MAX_CLOCKS] = { * timer->base->cpu_base */ static struct hrtimer_cpu_base migration_cpu_base = { - .clock_base = { { .cpu_base = &migration_cpu_base, }, }, + .clock_base = { { + .cpu_base = &migration_cpu_base, + .seq = SEQCNT_RAW_SPINLOCK_ZERO(migration_cpu_base.seq, + &migration_cpu_base.lock), + }, }, }; #define migration_base migration_cpu_base.clock_base[0] -static inline bool is_migration_base(struct hrtimer_clock_base *base) -{ - return base == &migration_base; -} - /* * We are using hashed locking: holding per_cpu(hrtimer_bases)[n].lock * means that all timers which are tied to this base via timer->base are @@ -160,6 +171,7 @@ static inline bool is_migration_base(struct hrtimer_clock_base *base) static struct hrtimer_clock_base *lock_hrtimer_base(const struct hrtimer *timer, unsigned long *flags) + __acquires(&timer->base->lock) { struct hrtimer_clock_base *base; @@ -177,27 +189,54 @@ struct hrtimer_clock_base *lock_hrtimer_base(const struct hrtimer *timer, } /* - * We do not migrate the timer when it is expiring before the next - * event on the target cpu. When high resolution is enabled, we cannot - * reprogram the target cpu hardware and we would cause it to fire - * late. To keep it simple, we handle the high resolution enabled and - * disabled case similar. + * Check if the elected target is suitable considering its next + * event and the hotplug state of the current CPU. + * + * If the elected target is remote and its next event is after the timer + * to queue, then a remote reprogram is necessary. However there is no + * guarantee the IPI handling the operation would arrive in time to meet + * the high resolution deadline. In this case the local CPU becomes a + * preferred target, unless it is offline. + * + * High and low resolution modes are handled the same way for simplicity. * * Called with cpu_base->lock of target cpu held. */ -static int -hrtimer_check_target(struct hrtimer *timer, struct hrtimer_clock_base *new_base) +static bool hrtimer_suitable_target(struct hrtimer *timer, struct hrtimer_clock_base *new_base, + struct hrtimer_cpu_base *new_cpu_base, + struct hrtimer_cpu_base *this_cpu_base) { ktime_t expires; + /* + * The local CPU clockevent can be reprogrammed. Also get_target_base() + * guarantees it is online. + */ + if (new_cpu_base == this_cpu_base) + return true; + + /* + * The offline local CPU can't be the default target if the + * next remote target event is after this timer. Keep the + * elected new base. An IPI will we issued to reprogram + * it as a last resort. + */ + if (!hrtimer_base_is_online(this_cpu_base)) + return true; + expires = ktime_sub(hrtimer_get_expires(timer), new_base->offset); - return expires < new_base->cpu_base->expires_next; + + return expires >= new_base->cpu_base->expires_next; } -static inline -struct hrtimer_cpu_base *get_target_base(struct hrtimer_cpu_base *base, - int pinned) +static inline struct hrtimer_cpu_base *get_target_base(struct hrtimer_cpu_base *base, int pinned) { + if (!hrtimer_base_is_online(base)) { + int cpu = cpumask_any_and(cpu_online_mask, housekeeping_cpumask(HK_TYPE_TIMER)); + + return &per_cpu(hrtimer_bases, cpu); + } + #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON) if (static_branch_likely(&timers_migration_enabled) && !pinned) return &per_cpu(hrtimer_bases, get_nohz_timer_target()); @@ -248,8 +287,8 @@ again: raw_spin_unlock(&base->cpu_base->lock); raw_spin_lock(&new_base->cpu_base->lock); - if (new_cpu_base != this_cpu_base && - hrtimer_check_target(timer, new_base)) { + if (!hrtimer_suitable_target(timer, new_base, new_cpu_base, + this_cpu_base)) { raw_spin_unlock(&new_base->cpu_base->lock); raw_spin_lock(&base->cpu_base->lock); new_cpu_base = this_cpu_base; @@ -258,8 +297,7 @@ again: } WRITE_ONCE(timer->base, new_base); } else { - if (new_cpu_base != this_cpu_base && - hrtimer_check_target(timer, new_base)) { + if (!hrtimer_suitable_target(timer, new_base, new_cpu_base, this_cpu_base)) { new_cpu_base = this_cpu_base; goto again; } @@ -269,13 +307,9 @@ again: #else /* CONFIG_SMP */ -static inline bool is_migration_base(struct hrtimer_clock_base *base) -{ - return false; -} - static inline struct hrtimer_clock_base * lock_hrtimer_base(const struct hrtimer *timer, unsigned long *flags) + __acquires(&timer->base->cpu_base->lock) { struct hrtimer_clock_base *base = timer->base; @@ -338,7 +372,7 @@ EXPORT_SYMBOL_GPL(ktime_add_safe); #ifdef CONFIG_DEBUG_OBJECTS_TIMERS -static struct debug_obj_descr hrtimer_debug_descr; +static const struct debug_obj_descr hrtimer_debug_descr; static void *hrtimer_debug_hint(void *addr) { @@ -373,7 +407,7 @@ static bool hrtimer_fixup_activate(void *addr, enum debug_obj_state state) switch (state) { case ODEBUG_STATE_ACTIVE: WARN_ON(1); - /* fall through */ + fallthrough; default: return false; } @@ -397,7 +431,7 @@ static bool hrtimer_fixup_free(void *addr, enum debug_obj_state state) } } -static struct debug_obj_descr hrtimer_debug_descr = { +static const struct debug_obj_descr hrtimer_debug_descr = { .name = "hrtimer", .debug_hint = hrtimer_debug_hint, .fixup_init = hrtimer_fixup_init, @@ -410,6 +444,11 @@ static inline void debug_hrtimer_init(struct hrtimer *timer) debug_object_init(timer, &hrtimer_debug_descr); } +static inline void debug_hrtimer_init_on_stack(struct hrtimer *timer) +{ + debug_object_init_on_stack(timer, &hrtimer_debug_descr); +} + static inline void debug_hrtimer_activate(struct hrtimer *timer, enum hrtimer_mode mode) { @@ -421,33 +460,6 @@ static inline void debug_hrtimer_deactivate(struct hrtimer *timer) debug_object_deactivate(timer, &hrtimer_debug_descr); } -static inline void debug_hrtimer_free(struct hrtimer *timer) -{ - debug_object_free(timer, &hrtimer_debug_descr); -} - -static void __hrtimer_init(struct hrtimer *timer, clockid_t clock_id, - enum hrtimer_mode mode); - -void hrtimer_init_on_stack(struct hrtimer *timer, clockid_t clock_id, - enum hrtimer_mode mode) -{ - debug_object_init_on_stack(timer, &hrtimer_debug_descr); - __hrtimer_init(timer, clock_id, mode); -} -EXPORT_SYMBOL_GPL(hrtimer_init_on_stack); - -static void __hrtimer_init_sleeper(struct hrtimer_sleeper *sl, - clockid_t clock_id, enum hrtimer_mode mode); - -void hrtimer_init_sleeper_on_stack(struct hrtimer_sleeper *sl, - clockid_t clock_id, enum hrtimer_mode mode) -{ - debug_object_init_on_stack(&sl->timer, &hrtimer_debug_descr); - __hrtimer_init_sleeper(sl, clock_id, mode); -} -EXPORT_SYMBOL_GPL(hrtimer_init_sleeper_on_stack); - void destroy_hrtimer_on_stack(struct hrtimer *timer) { debug_object_free(timer, &hrtimer_debug_descr); @@ -457,6 +469,7 @@ EXPORT_SYMBOL_GPL(destroy_hrtimer_on_stack); #else static inline void debug_hrtimer_init(struct hrtimer *timer) { } +static inline void debug_hrtimer_init_on_stack(struct hrtimer *timer) { } static inline void debug_hrtimer_activate(struct hrtimer *timer, enum hrtimer_mode mode) { } static inline void debug_hrtimer_deactivate(struct hrtimer *timer) { } @@ -470,6 +483,13 @@ debug_init(struct hrtimer *timer, clockid_t clockid, trace_hrtimer_init(timer, clockid, mode); } +static inline void debug_init_on_stack(struct hrtimer *timer, clockid_t clockid, + enum hrtimer_mode mode) +{ + debug_hrtimer_init_on_stack(timer); + trace_hrtimer_init(timer, clockid, mode); +} + static inline void debug_activate(struct hrtimer *timer, enum hrtimer_mode mode) { @@ -547,8 +567,11 @@ static ktime_t __hrtimer_next_event_base(struct hrtimer_cpu_base *cpu_base, } /* - * Recomputes cpu_base::*next_timer and returns the earliest expires_next but - * does not set cpu_base::*expires_next, that is done by hrtimer_reprogram. + * Recomputes cpu_base::*next_timer and returns the earliest expires_next + * but does not set cpu_base::*expires_next, that is done by + * hrtimer[_force]_reprogram and hrtimer_interrupt only. When updating + * cpu_base::*expires_next right away, reprogramming logic would no longer + * work. * * When a softirq is pending, we can ignore the HRTIMER_ACTIVE_SOFT bases, * those timers will get run whenever the softirq gets handled, at the end of @@ -589,6 +612,37 @@ __hrtimer_get_next_event(struct hrtimer_cpu_base *cpu_base, unsigned int active_ return expires_next; } +static ktime_t hrtimer_update_next_event(struct hrtimer_cpu_base *cpu_base) +{ + ktime_t expires_next, soft = KTIME_MAX; + + /* + * If the soft interrupt has already been activated, ignore the + * soft bases. They will be handled in the already raised soft + * interrupt. + */ + if (!cpu_base->softirq_activated) { + soft = __hrtimer_get_next_event(cpu_base, HRTIMER_ACTIVE_SOFT); + /* + * Update the soft expiry time. clock_settime() might have + * affected it. + */ + cpu_base->softirq_expires_next = soft; + } + + expires_next = __hrtimer_get_next_event(cpu_base, HRTIMER_ACTIVE_HARD); + /* + * If a softirq timer is expiring first, update cpu_base->next_timer + * and program the hardware with the soft expiry time. + */ + if (expires_next > soft) { + cpu_base->next_timer = cpu_base->softirq_next_timer; + expires_next = soft; + } + + return expires_next; +} + static inline ktime_t hrtimer_update_base(struct hrtimer_cpu_base *base) { ktime_t *offs_real = &base->clock_base[HRTIMER_BASE_REALTIME].offset; @@ -608,48 +662,16 @@ static inline ktime_t hrtimer_update_base(struct hrtimer_cpu_base *base) /* * Is the high resolution mode active ? */ -static inline int __hrtimer_hres_active(struct hrtimer_cpu_base *cpu_base) +static inline int hrtimer_hres_active(struct hrtimer_cpu_base *cpu_base) { return IS_ENABLED(CONFIG_HIGH_RES_TIMERS) ? cpu_base->hres_active : 0; } -static inline int hrtimer_hres_active(void) -{ - return __hrtimer_hres_active(this_cpu_ptr(&hrtimer_bases)); -} - -/* - * Reprogram the event source with checking both queues for the - * next event - * Called with interrupts disabled and base->lock held - */ -static void -hrtimer_force_reprogram(struct hrtimer_cpu_base *cpu_base, int skip_equal) +static void __hrtimer_reprogram(struct hrtimer_cpu_base *cpu_base, + struct hrtimer *next_timer, + ktime_t expires_next) { - ktime_t expires_next; - - /* - * Find the current next expiration time. - */ - expires_next = __hrtimer_get_next_event(cpu_base, HRTIMER_ACTIVE_ALL); - - if (cpu_base->next_timer && cpu_base->next_timer->is_soft) { - /* - * When the softirq is activated, hrtimer has to be - * programmed with the first hard hrtimer because soft - * timer interrupt could occur too late. - */ - if (cpu_base->softirq_activated) - expires_next = __hrtimer_get_next_event(cpu_base, - HRTIMER_ACTIVE_HARD); - else - cpu_base->softirq_expires_next = expires_next; - } - - if (skip_equal && expires_next == cpu_base->expires_next) - return; - cpu_base->expires_next = expires_next; /* @@ -666,13 +688,31 @@ hrtimer_force_reprogram(struct hrtimer_cpu_base *cpu_base, int skip_equal) * T1 is removed, so this code is called and would reprogram * the hardware to 5s from now. Any hrtimer_start after that * will not reprogram the hardware due to hang_detected being - * set. So we'd effectivly block all timers until the T2 event + * set. So we'd effectively block all timers until the T2 event * fires. */ - if (!__hrtimer_hres_active(cpu_base) || cpu_base->hang_detected) + if (!hrtimer_hres_active(cpu_base) || cpu_base->hang_detected) return; - tick_program_event(cpu_base->expires_next, 1); + tick_program_event(expires_next, 1); +} + +/* + * Reprogram the event source with checking both queues for the + * next event + * Called with interrupts disabled and base->lock held + */ +static void +hrtimer_force_reprogram(struct hrtimer_cpu_base *cpu_base, int skip_equal) +{ + ktime_t expires_next; + + expires_next = hrtimer_update_next_event(cpu_base); + + if (skip_equal && expires_next == cpu_base->expires_next) + return; + + __hrtimer_reprogram(cpu_base, cpu_base->next_timer, expires_next); } /* High resolution timer related functions */ @@ -704,24 +744,6 @@ static inline int hrtimer_is_hres_enabled(void) } /* - * Retrigger next event is called after clock was set - * - * Called with interrupts disabled via on_each_cpu() - */ -static void retrigger_next_event(void *arg) -{ - struct hrtimer_cpu_base *base = this_cpu_ptr(&hrtimer_bases); - - if (!__hrtimer_hres_active(base)) - return; - - raw_spin_lock(&base->lock); - hrtimer_update_base(base); - hrtimer_force_reprogram(base, 0); - raw_spin_unlock(&base->lock); -} - -/* * Switch to high resolution mode */ static void hrtimer_switch_to_hres(void) @@ -736,34 +758,59 @@ static void hrtimer_switch_to_hres(void) base->hres_active = 1; hrtimer_resolution = HIGH_RES_NSEC; - tick_setup_sched_timer(); + tick_setup_sched_timer(true); /* "Retrigger" the interrupt to get things going */ retrigger_next_event(NULL); } -static void clock_was_set_work(struct work_struct *work) -{ - clock_was_set(); -} +#else -static DECLARE_WORK(hrtimer_work, clock_was_set_work); +static inline int hrtimer_is_hres_enabled(void) { return 0; } +static inline void hrtimer_switch_to_hres(void) { } +#endif /* CONFIG_HIGH_RES_TIMERS */ /* - * Called from timekeeping and resume code to reprogram the hrtimer - * interrupt device on all cpus. + * Retrigger next event is called after clock was set with interrupts + * disabled through an SMP function call or directly from low level + * resume code. + * + * This is only invoked when: + * - CONFIG_HIGH_RES_TIMERS is enabled. + * - CONFIG_NOHZ_COMMON is enabled + * + * For the other cases this function is empty and because the call sites + * are optimized out it vanishes as well, i.e. no need for lots of + * #ifdeffery. */ -void clock_was_set_delayed(void) +static void retrigger_next_event(void *arg) { - schedule_work(&hrtimer_work); -} - -#else + struct hrtimer_cpu_base *base = this_cpu_ptr(&hrtimer_bases); -static inline int hrtimer_is_hres_enabled(void) { return 0; } -static inline void hrtimer_switch_to_hres(void) { } -static inline void retrigger_next_event(void *arg) { } + /* + * When high resolution mode or nohz is active, then the offsets of + * CLOCK_REALTIME/TAI/BOOTTIME have to be updated. Otherwise the + * next tick will take care of that. + * + * If high resolution mode is active then the next expiring timer + * must be reevaluated and the clock event device reprogrammed if + * necessary. + * + * In the NOHZ case the update of the offset and the reevaluation + * of the next expiring timer is enough. The return from the SMP + * function call will take care of the reprogramming in case the + * CPU was in a NOHZ idle sleep. + */ + if (!hrtimer_hres_active(base) && !tick_nohz_active) + return; -#endif /* CONFIG_HIGH_RES_TIMERS */ + raw_spin_lock(&base->lock); + hrtimer_update_base(base); + if (hrtimer_hres_active(base)) + hrtimer_force_reprogram(base, 0); + else + hrtimer_update_next_event(base); + raw_spin_unlock(&base->lock); +} /* * When a timer is enqueued and expires earlier than the already enqueued @@ -818,75 +865,161 @@ static void hrtimer_reprogram(struct hrtimer *timer, bool reprogram) if (base->cpu_base != cpu_base) return; + if (expires >= cpu_base->expires_next) + return; + /* - * If the hrtimer interrupt is running, then it will - * reevaluate the clock bases and reprogram the clock event - * device. The callbacks are always executed in hard interrupt - * context so we don't need an extra check for a running - * callback. + * If the hrtimer interrupt is running, then it will reevaluate the + * clock bases and reprogram the clock event device. */ if (cpu_base->in_hrtirq) return; - if (expires >= cpu_base->expires_next) - return; - - /* Update the pointer to the next expiring timer */ cpu_base->next_timer = timer; - cpu_base->expires_next = expires; + + __hrtimer_reprogram(cpu_base, timer, expires); +} + +static bool update_needs_ipi(struct hrtimer_cpu_base *cpu_base, + unsigned int active) +{ + struct hrtimer_clock_base *base; + unsigned int seq; + ktime_t expires; /* - * If hres is not active, hardware does not have to be - * programmed yet. + * Update the base offsets unconditionally so the following + * checks whether the SMP function call is required works. * - * If a hang was detected in the last timer interrupt then we - * do not schedule a timer which is earlier than the expiry - * which we enforced in the hang detection. We want the system - * to make progress. + * The update is safe even when the remote CPU is in the hrtimer + * interrupt or the hrtimer soft interrupt and expiring affected + * bases. Either it will see the update before handling a base or + * it will see it when it finishes the processing and reevaluates + * the next expiring timer. */ - if (!__hrtimer_hres_active(cpu_base) || cpu_base->hang_detected) - return; + seq = cpu_base->clock_was_set_seq; + hrtimer_update_base(cpu_base); /* - * Program the timer hardware. We enforce the expiry for - * events which are already in the past. + * If the sequence did not change over the update then the + * remote CPU already handled it. */ - tick_program_event(expires, 1); + if (seq == cpu_base->clock_was_set_seq) + return false; + + /* + * If the remote CPU is currently handling an hrtimer interrupt, it + * will reevaluate the first expiring timer of all clock bases + * before reprogramming. Nothing to do here. + */ + if (cpu_base->in_hrtirq) + return false; + + /* + * Walk the affected clock bases and check whether the first expiring + * timer in a clock base is moving ahead of the first expiring timer of + * @cpu_base. If so, the IPI must be invoked because per CPU clock + * event devices cannot be remotely reprogrammed. + */ + active &= cpu_base->active_bases; + + for_each_active_base(base, cpu_base, active) { + struct timerqueue_node *next; + + next = timerqueue_getnext(&base->active); + expires = ktime_sub(next->expires, base->offset); + if (expires < cpu_base->expires_next) + return true; + + /* Extra check for softirq clock bases */ + if (base->clockid < HRTIMER_BASE_MONOTONIC_SOFT) + continue; + if (cpu_base->softirq_activated) + continue; + if (expires < cpu_base->softirq_expires_next) + return true; + } + return false; } /* - * Clock realtime was set + * Clock was set. This might affect CLOCK_REALTIME, CLOCK_TAI and + * CLOCK_BOOTTIME (for late sleep time injection). * - * Change the offset of the realtime clock vs. the monotonic - * clock. - * - * We might have to reprogram the high resolution timer interrupt. On - * SMP we call the architecture specific code to retrigger _all_ high - * resolution timer interrupts. On UP we just disable interrupts and - * call the high resolution interrupt code. + * This requires to update the offsets for these clocks + * vs. CLOCK_MONOTONIC. When high resolution timers are enabled, then this + * also requires to eventually reprogram the per CPU clock event devices + * when the change moves an affected timer ahead of the first expiring + * timer on that CPU. Obviously remote per CPU clock event devices cannot + * be reprogrammed. The other reason why an IPI has to be sent is when the + * system is in !HIGH_RES and NOHZ mode. The NOHZ mode updates the offsets + * in the tick, which obviously might be stopped, so this has to bring out + * the remote CPU which might sleep in idle to get this sorted. */ -void clock_was_set(void) +void clock_was_set(unsigned int bases) { -#ifdef CONFIG_HIGH_RES_TIMERS - /* Retrigger the CPU local events everywhere */ - on_each_cpu(retrigger_next_event, NULL, 1); -#endif + struct hrtimer_cpu_base *cpu_base = raw_cpu_ptr(&hrtimer_bases); + cpumask_var_t mask; + int cpu; + + if (!hrtimer_hres_active(cpu_base) && !tick_nohz_active) + goto out_timerfd; + + if (!zalloc_cpumask_var(&mask, GFP_KERNEL)) { + on_each_cpu(retrigger_next_event, NULL, 1); + goto out_timerfd; + } + + /* Avoid interrupting CPUs if possible */ + cpus_read_lock(); + for_each_online_cpu(cpu) { + unsigned long flags; + + cpu_base = &per_cpu(hrtimer_bases, cpu); + raw_spin_lock_irqsave(&cpu_base->lock, flags); + + if (update_needs_ipi(cpu_base, bases)) + cpumask_set_cpu(cpu, mask); + + raw_spin_unlock_irqrestore(&cpu_base->lock, flags); + } + + preempt_disable(); + smp_call_function_many(mask, retrigger_next_event, NULL, 1); + preempt_enable(); + cpus_read_unlock(); + free_cpumask_var(mask); + +out_timerfd: timerfd_clock_was_set(); } +static void clock_was_set_work(struct work_struct *work) +{ + clock_was_set(CLOCK_SET_WALL); +} + +static DECLARE_WORK(hrtimer_work, clock_was_set_work); + +/* + * Called from timekeeping code to reprogram the hrtimer interrupt device + * on all cpus and to notify timerfd. + */ +void clock_was_set_delayed(void) +{ + schedule_work(&hrtimer_work); +} + /* - * During resume we might have to reprogram the high resolution timer - * interrupt on all online CPUs. However, all other CPUs will be - * stopped with IRQs interrupts disabled so the clock_was_set() call - * must be deferred. + * Called during resume either directly from via timekeeping_resume() + * or in the case of s2idle from tick_unfreeze() to ensure that the + * hrtimers are up to date. */ -void hrtimers_resume(void) +void hrtimers_resume_local(void) { lockdep_assert_irqs_disabled(); /* Retrigger on the local CPU */ retrigger_next_event(NULL); - /* And schedule a retrigger for all others */ - clock_was_set_delayed(); } /* @@ -894,26 +1027,29 @@ void hrtimers_resume(void) */ static inline void unlock_hrtimer_base(const struct hrtimer *timer, unsigned long *flags) + __releases(&timer->base->cpu_base->lock) { raw_spin_unlock_irqrestore(&timer->base->cpu_base->lock, *flags); } /** - * hrtimer_forward - forward the timer expiry + * hrtimer_forward() - forward the timer expiry * @timer: hrtimer to forward * @now: forward past this time * @interval: the interval to forward * * Forward the timer expiry so it will expire in the future. - * Returns the number of overruns. * - * Can be safely called from the callback function of @timer. If - * called from other contexts @timer must neither be enqueued nor - * running the callback and the caller needs to take care of - * serialization. + * .. note:: + * This only updates the timer expiry value and does not requeue the timer. + * + * There is also a variant of the function hrtimer_forward_now(). * - * Note: This only updates the timer expiry value and does not requeue - * the timer. + * Context: Can be safely called from the callback function of @timer. If called + * from other contexts @timer must neither be enqueued nor running the + * callback and the caller needs to take care of serialization. + * + * Return: The number of overruns are returned. */ u64 hrtimer_forward(struct hrtimer *timer, ktime_t now, ktime_t interval) { @@ -956,13 +1092,13 @@ EXPORT_SYMBOL_GPL(hrtimer_forward); * The timer is inserted in expiry order. Insertion into the * red black tree is O(log(n)). Must hold the base lock. * - * Returns 1 when the new timer is the leftmost timer in the tree. + * Returns true when the new timer is the leftmost timer in the tree. */ -static int enqueue_hrtimer(struct hrtimer *timer, - struct hrtimer_clock_base *base, - enum hrtimer_mode mode) +static bool enqueue_hrtimer(struct hrtimer *timer, struct hrtimer_clock_base *base, + enum hrtimer_mode mode) { debug_activate(timer, mode); + WARN_ON_ONCE(!base->cpu_base->online); base->cpu_base->active_bases |= 1 << base->index; @@ -1002,7 +1138,7 @@ static void __remove_hrtimer(struct hrtimer *timer, * cpu_base->next_timer. This happens when we remove the first * timer on a remote cpu. No harm as we never dereference * cpu_base->next_timer. So the worst thing what can happen is - * an superflous call to hrtimer_force_reprogram() on the + * an superfluous call to hrtimer_force_reprogram() on the * remote cpu later on if the same timer gets enqueued again. */ if (reprogram && timer == cpu_base->next_timer) @@ -1013,12 +1149,13 @@ static void __remove_hrtimer(struct hrtimer *timer, * remove hrtimer, called with base lock held */ static inline int -remove_hrtimer(struct hrtimer *timer, struct hrtimer_clock_base *base, bool restart) +remove_hrtimer(struct hrtimer *timer, struct hrtimer_clock_base *base, + bool restart, bool keep_local) { u8 state = timer->state; if (state & HRTIMER_STATE_ENQUEUED) { - int reprogram; + bool reprogram; /* * Remove the timer and force reprogramming when high @@ -1031,8 +1168,16 @@ remove_hrtimer(struct hrtimer *timer, struct hrtimer_clock_base *base, bool rest debug_deactivate(timer); reprogram = base->cpu_base == this_cpu_ptr(&hrtimer_bases); + /* + * If the timer is not restarted then reprogramming is + * required if the timer is local. If it is local and about + * to be restarted, avoid programming it twice (on removal + * and a moment later when it's requeued). + */ if (!restart) state = HRTIMER_STATE_INACTIVE; + else + reprogram &= !keep_local; __remove_hrtimer(timer, base, state, reprogram); return 1; @@ -1047,7 +1192,7 @@ static inline ktime_t hrtimer_update_lowres(struct hrtimer *timer, ktime_t tim, /* * CONFIG_TIME_LOW_RES indicates that the system has no way to return * granular time values. For relative timers we add hrtimer_resolution - * (i.e. one jiffie) to prevent short timeouts. + * (i.e. one jiffy) to prevent short timeouts. */ timer->is_rel = mode & HRTIMER_MODE_REL; if (timer->is_rel) @@ -1085,10 +1230,39 @@ static int __hrtimer_start_range_ns(struct hrtimer *timer, ktime_t tim, u64 delta_ns, const enum hrtimer_mode mode, struct hrtimer_clock_base *base) { + struct hrtimer_cpu_base *this_cpu_base = this_cpu_ptr(&hrtimer_bases); struct hrtimer_clock_base *new_base; + bool force_local, first; - /* Remove an active timer from the queue: */ - remove_hrtimer(timer, base, true); + /* + * If the timer is on the local cpu base and is the first expiring + * timer then this might end up reprogramming the hardware twice + * (on removal and on enqueue). To avoid that by prevent the + * reprogram on removal, keep the timer local to the current CPU + * and enforce reprogramming after it is queued no matter whether + * it is the new first expiring timer again or not. + */ + force_local = base->cpu_base == this_cpu_base; + force_local &= base->cpu_base->next_timer == timer; + + /* + * Don't force local queuing if this enqueue happens on a unplugged + * CPU after hrtimer_cpu_dying() has been invoked. + */ + force_local &= this_cpu_base->online; + + /* + * Remove an active timer from the queue. In case it is not queued + * on the current CPU, make sure that remove_hrtimer() updates the + * remote data correctly. + * + * If it's on the current CPU and the first expiring timer, then + * skip reprogramming, keep the timer local and enforce + * reprogramming later if it was the first expiring timer. This + * avoids programming the underlying clock event twice (once at + * removal and once after enqueue). + */ + remove_hrtimer(timer, base, true, force_local); if (mode & HRTIMER_MODE_REL) tim = ktime_add_safe(tim, base->get_time()); @@ -1098,9 +1272,43 @@ static int __hrtimer_start_range_ns(struct hrtimer *timer, ktime_t tim, hrtimer_set_expires_range_ns(timer, tim, delta_ns); /* Switch the timer base, if necessary: */ - new_base = switch_hrtimer_base(timer, base, mode & HRTIMER_MODE_PINNED); + if (!force_local) { + new_base = switch_hrtimer_base(timer, base, + mode & HRTIMER_MODE_PINNED); + } else { + new_base = base; + } - return enqueue_hrtimer(timer, new_base, mode); + first = enqueue_hrtimer(timer, new_base, mode); + if (!force_local) { + /* + * If the current CPU base is online, then the timer is + * never queued on a remote CPU if it would be the first + * expiring timer there. + */ + if (hrtimer_base_is_online(this_cpu_base)) + return first; + + /* + * Timer was enqueued remote because the current base is + * already offline. If the timer is the first to expire, + * kick the remote CPU to reprogram the clock event. + */ + if (first) { + struct hrtimer_cpu_base *new_cpu_base = new_base->cpu_base; + + smp_call_function_single_async(new_cpu_base->cpu, &new_cpu_base->csd); + } + return 0; + } + + /* + * Timer was forced to stay on the current CPU to avoid + * reprogramming on removal and enqueue. Force reprogram the + * hardware by evaluating the new first expiring timer. + */ + hrtimer_force_reprogram(new_base->cpu_base, 1); + return 0; } /** @@ -1118,6 +1326,8 @@ void hrtimer_start_range_ns(struct hrtimer *timer, ktime_t tim, struct hrtimer_clock_base *base; unsigned long flags; + if (WARN_ON_ONCE(!timer->function)) + return; /* * Check whether the HRTIMER_MODE_SOFT bit and hrtimer.is_soft * match on CONFIG_PREEMPT_RT = n. With PREEMPT_RT check the hard @@ -1166,7 +1376,7 @@ int hrtimer_try_to_cancel(struct hrtimer *timer) base = lock_hrtimer_base(timer, &flags); if (!hrtimer_callback_running(timer)) - ret = remove_hrtimer(timer, base, false); + ret = remove_hrtimer(timer, base, false, false); unlock_hrtimer_base(timer, &flags); @@ -1182,11 +1392,13 @@ static void hrtimer_cpu_base_init_expiry_lock(struct hrtimer_cpu_base *base) } static void hrtimer_cpu_base_lock_expiry(struct hrtimer_cpu_base *base) + __acquires(&base->softirq_expiry_lock) { spin_lock(&base->softirq_expiry_lock); } static void hrtimer_cpu_base_unlock_expiry(struct hrtimer_cpu_base *base) + __releases(&base->softirq_expiry_lock) { spin_unlock(&base->softirq_expiry_lock); } @@ -1195,7 +1407,7 @@ static void hrtimer_cpu_base_unlock_expiry(struct hrtimer_cpu_base *base) * The counterpart to hrtimer_cancel_wait_running(). * * If there is a waiter for cpu_base->expiry_lock, then it was waiting for - * the timer callback to finish. Drop expiry_lock and reaquire it. That + * the timer callback to finish. Drop expiry_lock and reacquire it. That * allows the waiter to acquire the lock and make progress. */ static void hrtimer_sync_wait_running(struct hrtimer_cpu_base *cpu_base, @@ -1209,6 +1421,18 @@ static void hrtimer_sync_wait_running(struct hrtimer_cpu_base *cpu_base, } } +#ifdef CONFIG_SMP +static __always_inline bool is_migration_base(struct hrtimer_clock_base *base) +{ + return base == &migration_base; +} +#else +static __always_inline bool is_migration_base(struct hrtimer_clock_base *base) +{ + return false; +} +#endif + /* * This function is called on PREEMPT_RT kernels when the fast path * deletion of a timer failed because the timer callback function was @@ -1285,7 +1509,7 @@ int hrtimer_cancel(struct hrtimer *timer) EXPORT_SYMBOL_GPL(hrtimer_cancel); /** - * hrtimer_get_remaining - get remaining time for the timer + * __hrtimer_get_remaining - get remaining time for the timer * @timer: the timer to read * @adjust: adjust relative timers when CONFIG_TIME_LOW_RES=y */ @@ -1319,7 +1543,7 @@ u64 hrtimer_get_next_event(void) raw_spin_lock_irqsave(&cpu_base->lock, flags); - if (!__hrtimer_hres_active(cpu_base)) + if (!hrtimer_hres_active(cpu_base)) expires = __hrtimer_get_next_event(cpu_base, HRTIMER_ACTIVE_ALL); raw_spin_unlock_irqrestore(&cpu_base->lock, flags); @@ -1342,7 +1566,7 @@ u64 hrtimer_next_event_without(const struct hrtimer *exclude) raw_spin_lock_irqsave(&cpu_base->lock, flags); - if (__hrtimer_hres_active(cpu_base)) { + if (hrtimer_hres_active(cpu_base)) { unsigned int active; if (!cpu_base->softirq_activated) { @@ -1373,6 +1597,11 @@ static inline int hrtimer_clockid_to_base(clockid_t clock_id) return HRTIMER_BASE_MONOTONIC; } +static enum hrtimer_restart hrtimer_dummy_timeout(struct hrtimer *unused) +{ + return HRTIMER_NORESTART; +} + static void __hrtimer_init(struct hrtimer *timer, clockid_t clock_id, enum hrtimer_mode mode) { @@ -1381,7 +1610,7 @@ static void __hrtimer_init(struct hrtimer *timer, clockid_t clock_id, int base; /* - * On PREEMPT_RT enabled kernels hrtimers which are not explicitely + * On PREEMPT_RT enabled kernels hrtimers which are not explicitly * marked for hard interrupt expiry mode are moved into soft * interrupt context for latency reasons and because the callbacks * can invoke functions which might sleep on RT, e.g. spin_lock(). @@ -1409,11 +1638,23 @@ static void __hrtimer_init(struct hrtimer *timer, clockid_t clock_id, timerqueue_init(&timer->node); } +static void __hrtimer_setup(struct hrtimer *timer, + enum hrtimer_restart (*function)(struct hrtimer *), + clockid_t clock_id, enum hrtimer_mode mode) +{ + __hrtimer_init(timer, clock_id, mode); + + if (WARN_ON_ONCE(!function)) + timer->function = hrtimer_dummy_timeout; + else + timer->function = function; +} + /** * hrtimer_init - initialize a timer to the given clock * @timer: the timer to be initialized * @clock_id: the clock to be used - * @mode: The modes which are relevant for intitialization: + * @mode: The modes which are relevant for initialization: * HRTIMER_MODE_ABS, HRTIMER_MODE_REL, HRTIMER_MODE_ABS_SOFT, * HRTIMER_MODE_REL_SOFT * @@ -1429,6 +1670,46 @@ void hrtimer_init(struct hrtimer *timer, clockid_t clock_id, } EXPORT_SYMBOL_GPL(hrtimer_init); +/** + * hrtimer_setup - initialize a timer to the given clock + * @timer: the timer to be initialized + * @function: the callback function + * @clock_id: the clock to be used + * @mode: The modes which are relevant for initialization: + * HRTIMER_MODE_ABS, HRTIMER_MODE_REL, HRTIMER_MODE_ABS_SOFT, + * HRTIMER_MODE_REL_SOFT + * + * The PINNED variants of the above can be handed in, + * but the PINNED bit is ignored as pinning happens + * when the hrtimer is started + */ +void hrtimer_setup(struct hrtimer *timer, enum hrtimer_restart (*function)(struct hrtimer *), + clockid_t clock_id, enum hrtimer_mode mode) +{ + debug_init(timer, clock_id, mode); + __hrtimer_setup(timer, function, clock_id, mode); +} +EXPORT_SYMBOL_GPL(hrtimer_setup); + +/** + * hrtimer_setup_on_stack - initialize a timer on stack memory + * @timer: The timer to be initialized + * @function: the callback function + * @clock_id: The clock to be used + * @mode: The timer mode + * + * Similar to hrtimer_setup(), except that this one must be used if struct hrtimer is in stack + * memory. + */ +void hrtimer_setup_on_stack(struct hrtimer *timer, + enum hrtimer_restart (*function)(struct hrtimer *), + clockid_t clock_id, enum hrtimer_mode mode) +{ + debug_init_on_stack(timer, clock_id, mode); + __hrtimer_setup(timer, function, clock_id, mode); +} +EXPORT_SYMBOL_GPL(hrtimer_setup_on_stack); + /* * A timer is active, when it is enqueued into the rbtree or the * callback function is running or it's in the state of being migrated @@ -1470,7 +1751,7 @@ EXPORT_SYMBOL_GPL(hrtimer_active); * insufficient for that. * * The sequence numbers are required because otherwise we could still observe - * a false negative if the read side got smeared over multiple consequtive + * a false negative if the read side got smeared over multiple consecutive * __run_hrtimer() invocations. */ @@ -1571,7 +1852,7 @@ static void __hrtimer_run_queues(struct hrtimer_cpu_base *cpu_base, ktime_t now, * minimizing wakeups, not running timers at the * earliest interrupt after their soft expiration. * This allows us to avoid using a Priority Search - * Tree, which can answer a stabbing querry for + * Tree, which can answer a stabbing query for * overlapping intervals and instead use the simple * BST we already have. * We don't add extra wakeups by delaying timers that @@ -1588,7 +1869,7 @@ static void __hrtimer_run_queues(struct hrtimer_cpu_base *cpu_base, ktime_t now, } } -static __latent_entropy void hrtimer_run_softirq(struct softirq_action *h) +static __latent_entropy void hrtimer_run_softirq(void) { struct hrtimer_cpu_base *cpu_base = this_cpu_ptr(&hrtimer_bases); unsigned long flags; @@ -1640,13 +1921,13 @@ retry: if (!ktime_before(now, cpu_base->softirq_expires_next)) { cpu_base->softirq_expires_next = KTIME_MAX; cpu_base->softirq_activated = 1; - raise_softirq_irqoff(HRTIMER_SOFTIRQ); + raise_timer_softirq(HRTIMER_SOFTIRQ); } __hrtimer_run_queues(cpu_base, now, flags, HRTIMER_ACTIVE_HARD); - /* Reevaluate the clock bases for the next expiry */ - expires_next = __hrtimer_get_next_event(cpu_base, HRTIMER_ACTIVE_ALL); + /* Reevaluate the clock bases for the [soft] next expiry */ + expires_next = hrtimer_update_next_event(cpu_base); /* * Store the new expiry value so the migration code can verify * against it. @@ -1703,25 +1984,7 @@ retry: tick_program_event(expires_next, 1); pr_warn_once("hrtimer: interrupt took %llu ns\n", ktime_to_ns(delta)); } - -/* called with interrupts disabled */ -static inline void __hrtimer_peek_ahead_timers(void) -{ - struct tick_device *td; - - if (!hrtimer_hres_active()) - return; - - td = this_cpu_ptr(&tick_cpu_device); - if (td && td->evtdev) - hrtimer_interrupt(td->evtdev); -} - -#else /* CONFIG_HIGH_RES_TIMERS */ - -static inline void __hrtimer_peek_ahead_timers(void) { } - -#endif /* !CONFIG_HIGH_RES_TIMERS */ +#endif /* !CONFIG_HIGH_RES_TIMERS */ /* * Called from run_local_timers in hardirq context every jiffy @@ -1732,7 +1995,7 @@ void hrtimer_run_queues(void) unsigned long flags; ktime_t now; - if (__hrtimer_hres_active(cpu_base)) + if (hrtimer_hres_active(cpu_base)) return; /* @@ -1753,7 +2016,7 @@ void hrtimer_run_queues(void) if (!ktime_before(now, cpu_base->softirq_expires_next)) { cpu_base->softirq_expires_next = KTIME_MAX; cpu_base->softirq_activated = 1; - raise_softirq_irqoff(HRTIMER_SOFTIRQ); + raise_timer_softirq(HRTIMER_SOFTIRQ); } __hrtimer_run_queues(cpu_base, now, flags, HRTIMER_ACTIVE_HARD); @@ -1791,7 +2054,7 @@ void hrtimer_sleeper_start_expires(struct hrtimer_sleeper *sl, * Make the enqueue delivery mode check work on RT. If the sleeper * was initialized for hard interrupt delivery, force the mode bit. * This is a special case for hrtimer_sleepers because - * hrtimer_init_sleeper() determines the delivery mode on RT so the + * __hrtimer_init_sleeper() determines the delivery mode on RT so the * fiddling with this decision is avoided at the call sites. */ if (IS_ENABLED(CONFIG_PREEMPT_RT) && sl->timer.is_hard) @@ -1805,7 +2068,7 @@ static void __hrtimer_init_sleeper(struct hrtimer_sleeper *sl, clockid_t clock_id, enum hrtimer_mode mode) { /* - * On PREEMPT_RT enabled kernels hrtimers which are not explicitely + * On PREEMPT_RT enabled kernels hrtimers which are not explicitly * marked for hard interrupt expiry mode are moved into soft * interrupt context either for latency reasons or because the * hrtimer callback takes regular spinlocks or invokes other @@ -1818,13 +2081,13 @@ static void __hrtimer_init_sleeper(struct hrtimer_sleeper *sl, * the same CPU. That causes a latency spike due to the wakeup of * a gazillion threads. * - * OTOH, priviledged real-time user space applications rely on the + * OTOH, privileged real-time user space applications rely on the * low latency of hard interrupt wakeups. If the current task is in * a real-time scheduling class, mark the mode for hard interrupt * expiry. */ if (IS_ENABLED(CONFIG_PREEMPT_RT)) { - if (task_is_realtime(current) && !(mode & HRTIMER_MODE_SOFT)) + if (rt_or_dl_task_policy(current) && !(mode & HRTIMER_MODE_SOFT)) mode |= HRTIMER_MODE_HARD; } @@ -1834,19 +2097,18 @@ static void __hrtimer_init_sleeper(struct hrtimer_sleeper *sl, } /** - * hrtimer_init_sleeper - initialize sleeper to the given clock + * hrtimer_setup_sleeper_on_stack - initialize a sleeper in stack memory * @sl: sleeper to be initialized * @clock_id: the clock to be used * @mode: timer mode abs/rel */ -void hrtimer_init_sleeper(struct hrtimer_sleeper *sl, clockid_t clock_id, - enum hrtimer_mode mode) +void hrtimer_setup_sleeper_on_stack(struct hrtimer_sleeper *sl, + clockid_t clock_id, enum hrtimer_mode mode) { - debug_init(&sl->timer, clock_id, mode); + debug_init_on_stack(&sl->timer, clock_id, mode); __hrtimer_init_sleeper(sl, clock_id, mode); - } -EXPORT_SYMBOL_GPL(hrtimer_init_sleeper); +EXPORT_SYMBOL_GPL(hrtimer_setup_sleeper_on_stack); int nanosleep_copyout(struct restart_block *restart, struct timespec64 *ts) { @@ -1872,11 +2134,11 @@ static int __sched do_nanosleep(struct hrtimer_sleeper *t, enum hrtimer_mode mod struct restart_block *restart; do { - set_current_state(TASK_INTERRUPTIBLE); + set_current_state(TASK_INTERRUPTIBLE|TASK_FREEZABLE); hrtimer_sleeper_start_expires(t, mode); if (likely(t->task)) - freezable_schedule(); + schedule(); hrtimer_cancel(&t->timer); mode = HRTIMER_MODE_ABS; @@ -1907,8 +2169,7 @@ static long __sched hrtimer_nanosleep_restart(struct restart_block *restart) struct hrtimer_sleeper t; int ret; - hrtimer_init_sleeper_on_stack(&t, restart->nanosleep.clockid, - HRTIMER_MODE_ABS); + hrtimer_setup_sleeper_on_stack(&t, restart->nanosleep.clockid, HRTIMER_MODE_ABS); hrtimer_set_expires_tv64(&t.timer, restart->nanosleep.expires); ret = do_nanosleep(&t, HRTIMER_MODE_ABS); destroy_hrtimer_on_stack(&t.timer); @@ -1921,14 +2182,9 @@ long hrtimer_nanosleep(ktime_t rqtp, const enum hrtimer_mode mode, struct restart_block *restart; struct hrtimer_sleeper t; int ret = 0; - u64 slack; - - slack = current->timer_slack_ns; - if (dl_task(current) || rt_task(current)) - slack = 0; - hrtimer_init_sleeper_on_stack(&t, clockid, mode); - hrtimer_set_expires_range_ns(&t.timer, rqtp, slack); + hrtimer_setup_sleeper_on_stack(&t, clockid, mode); + hrtimer_set_expires_range_ns(&t.timer, rqtp, current->timer_slack_ns); ret = do_nanosleep(&t, mode); if (ret != -ERESTART_RESTARTBLOCK) goto out; @@ -1940,9 +2196,9 @@ long hrtimer_nanosleep(ktime_t rqtp, const enum hrtimer_mode mode, } restart = ¤t->restart_block; - restart->fn = hrtimer_nanosleep_restart; restart->nanosleep.clockid = t.timer.base->clockid; restart->nanosleep.expires = hrtimer_get_expires_tv64(&t.timer); + set_restart_fn(restart, hrtimer_nanosleep_restart); out: destroy_hrtimer_on_stack(&t.timer); return ret; @@ -1961,6 +2217,7 @@ SYSCALL_DEFINE2(nanosleep, struct __kernel_timespec __user *, rqtp, if (!timespec64_valid(&tu)) return -EINVAL; + current->restart_block.fn = do_no_restart_syscall; current->restart_block.nanosleep.type = rmtp ? TT_NATIVE : TT_NONE; current->restart_block.nanosleep.rmtp = rmtp; return hrtimer_nanosleep(timespec64_to_ktime(tu), HRTIMER_MODE_REL, @@ -1982,6 +2239,7 @@ SYSCALL_DEFINE2(nanosleep_time32, struct old_timespec32 __user *, rqtp, if (!timespec64_valid(&tu)) return -EINVAL; + current->restart_block.fn = do_no_restart_syscall; current->restart_block.nanosleep.type = rmtp ? TT_COMPAT : TT_NONE; current->restart_block.nanosleep.compat_rmtp = rmtp; return hrtimer_nanosleep(timespec64_to_ktime(tu), HRTIMER_MODE_REL, @@ -1998,11 +2256,23 @@ int hrtimers_prepare_cpu(unsigned int cpu) int i; for (i = 0; i < HRTIMER_MAX_CLOCK_BASES; i++) { - cpu_base->clock_base[i].cpu_base = cpu_base; - timerqueue_init_head(&cpu_base->clock_base[i].active); + struct hrtimer_clock_base *clock_b = &cpu_base->clock_base[i]; + + clock_b->cpu_base = cpu_base; + seqcount_raw_spinlock_init(&clock_b->seq, &cpu_base->lock); + timerqueue_init_head(&clock_b->active); } cpu_base->cpu = cpu; + hrtimer_cpu_base_init_expiry_lock(cpu_base); + return 0; +} + +int hrtimers_cpu_starting(unsigned int cpu) +{ + struct hrtimer_cpu_base *cpu_base = this_cpu_ptr(&hrtimer_bases); + + /* Clear out any left over state from a CPU down operation */ cpu_base->active_bases = 0; cpu_base->hres_active = 0; cpu_base->hang_detected = 0; @@ -2010,7 +2280,7 @@ int hrtimers_prepare_cpu(unsigned int cpu) cpu_base->softirq_next_timer = NULL; cpu_base->expires_next = KTIME_MAX; cpu_base->softirq_expires_next = KTIME_MAX; - hrtimer_cpu_base_init_expiry_lock(cpu_base); + cpu_base->online = 1; return 0; } @@ -2046,29 +2316,20 @@ static void migrate_hrtimer_list(struct hrtimer_clock_base *old_base, } } -int hrtimers_dead_cpu(unsigned int scpu) +int hrtimers_cpu_dying(unsigned int dying_cpu) { + int i, ncpu = cpumask_any_and(cpu_active_mask, housekeeping_cpumask(HK_TYPE_TIMER)); struct hrtimer_cpu_base *old_base, *new_base; - int i; - BUG_ON(cpu_online(scpu)); - tick_cancel_sched_timer(scpu); + old_base = this_cpu_ptr(&hrtimer_bases); + new_base = &per_cpu(hrtimer_bases, ncpu); /* - * this BH disable ensures that raise_softirq_irqoff() does - * not wakeup ksoftirqd (and acquire the pi-lock) while - * holding the cpu_base lock - */ - local_bh_disable(); - local_irq_disable(); - old_base = &per_cpu(hrtimer_bases, scpu); - new_base = this_cpu_ptr(&hrtimer_bases); - /* * The caller is globally serialized and nobody else * takes two locks at once, deadlock is not possible. */ - raw_spin_lock(&new_base->lock); - raw_spin_lock_nested(&old_base->lock, SINGLE_DEPTH_NESTING); + raw_spin_lock(&old_base->lock); + raw_spin_lock_nested(&new_base->lock, SINGLE_DEPTH_NESTING); for (i = 0; i < HRTIMER_MAX_CLOCK_BASES; i++) { migrate_hrtimer_list(&old_base->clock_base[i], @@ -2079,15 +2340,14 @@ int hrtimers_dead_cpu(unsigned int scpu) * The migration might have changed the first expiring softirq * timer on this CPU. Update it. */ - hrtimer_update_softirq_timer(new_base, false); + __hrtimer_get_next_event(new_base, HRTIMER_ACTIVE_SOFT); + /* Tell the other CPU to retrigger the next event */ + smp_call_function_single(ncpu, retrigger_next_event, NULL, 0); - raw_spin_unlock(&old_base->lock); raw_spin_unlock(&new_base->lock); + old_base->online = 0; + raw_spin_unlock(&old_base->lock); - /* Check, if we got expired work to do */ - __hrtimer_peek_ahead_timers(); - local_irq_enable(); - local_bh_enable(); return 0; } @@ -2096,123 +2356,6 @@ int hrtimers_dead_cpu(unsigned int scpu) void __init hrtimers_init(void) { hrtimers_prepare_cpu(smp_processor_id()); + hrtimers_cpu_starting(smp_processor_id()); open_softirq(HRTIMER_SOFTIRQ, hrtimer_run_softirq); } - -/** - * schedule_hrtimeout_range_clock - sleep until timeout - * @expires: timeout value (ktime_t) - * @delta: slack in expires timeout (ktime_t) - * @mode: timer mode - * @clock_id: timer clock to be used - */ -int __sched -schedule_hrtimeout_range_clock(ktime_t *expires, u64 delta, - const enum hrtimer_mode mode, clockid_t clock_id) -{ - struct hrtimer_sleeper t; - - /* - * Optimize when a zero timeout value is given. It does not - * matter whether this is an absolute or a relative time. - */ - if (expires && *expires == 0) { - __set_current_state(TASK_RUNNING); - return 0; - } - - /* - * A NULL parameter means "infinite" - */ - if (!expires) { - schedule(); - return -EINTR; - } - - hrtimer_init_sleeper_on_stack(&t, clock_id, mode); - hrtimer_set_expires_range_ns(&t.timer, *expires, delta); - hrtimer_sleeper_start_expires(&t, mode); - - if (likely(t.task)) - schedule(); - - hrtimer_cancel(&t.timer); - destroy_hrtimer_on_stack(&t.timer); - - __set_current_state(TASK_RUNNING); - - return !t.task ? 0 : -EINTR; -} - -/** - * schedule_hrtimeout_range - sleep until timeout - * @expires: timeout value (ktime_t) - * @delta: slack in expires timeout (ktime_t) - * @mode: timer mode - * - * Make the current task sleep until the given expiry time has - * elapsed. The routine will return immediately unless - * the current task state has been set (see set_current_state()). - * - * The @delta argument gives the kernel the freedom to schedule the - * actual wakeup to a time that is both power and performance friendly. - * The kernel give the normal best effort behavior for "@expires+@delta", - * but may decide to fire the timer earlier, but no earlier than @expires. - * - * You can set the task state as follows - - * - * %TASK_UNINTERRUPTIBLE - at least @timeout time is guaranteed to - * pass before the routine returns unless the current task is explicitly - * woken up, (e.g. by wake_up_process()). - * - * %TASK_INTERRUPTIBLE - the routine may return early if a signal is - * delivered to the current task or the current task is explicitly woken - * up. - * - * The current task state is guaranteed to be TASK_RUNNING when this - * routine returns. - * - * Returns 0 when the timer has expired. If the task was woken before the - * timer expired by a signal (only possible in state TASK_INTERRUPTIBLE) or - * by an explicit wakeup, it returns -EINTR. - */ -int __sched schedule_hrtimeout_range(ktime_t *expires, u64 delta, - const enum hrtimer_mode mode) -{ - return schedule_hrtimeout_range_clock(expires, delta, mode, - CLOCK_MONOTONIC); -} -EXPORT_SYMBOL_GPL(schedule_hrtimeout_range); - -/** - * schedule_hrtimeout - sleep until timeout - * @expires: timeout value (ktime_t) - * @mode: timer mode - * - * Make the current task sleep until the given expiry time has - * elapsed. The routine will return immediately unless - * the current task state has been set (see set_current_state()). - * - * You can set the task state as follows - - * - * %TASK_UNINTERRUPTIBLE - at least @timeout time is guaranteed to - * pass before the routine returns unless the current task is explicitly - * woken up, (e.g. by wake_up_process()). - * - * %TASK_INTERRUPTIBLE - the routine may return early if a signal is - * delivered to the current task or the current task is explicitly woken - * up. - * - * The current task state is guaranteed to be TASK_RUNNING when this - * routine returns. - * - * Returns 0 when the timer has expired. If the task was woken before the - * timer expired by a signal (only possible in state TASK_INTERRUPTIBLE) or - * by an explicit wakeup, it returns -EINTR. - */ -int __sched schedule_hrtimeout(ktime_t *expires, - const enum hrtimer_mode mode) -{ - return schedule_hrtimeout_range(expires, 0, mode); -} -EXPORT_SYMBOL_GPL(schedule_hrtimeout); diff --git a/kernel/time/itimer.c b/kernel/time/itimer.c index ca4e6d57d68b..876d389b2e21 100644 --- a/kernel/time/itimer.c +++ b/kernel/time/itimer.c @@ -151,7 +151,27 @@ COMPAT_SYSCALL_DEFINE2(getitimer, int, which, #endif /* - * The timer is automagically restarted, when interval != 0 + * Invoked from dequeue_signal() when SIG_ALRM is delivered. + * + * Restart the ITIMER_REAL timer if it is armed as periodic timer. Doing + * this in the signal delivery path instead of self rearming prevents a DoS + * with small increments in the high reolution timer case and reduces timer + * noise in general. + */ +void posixtimer_rearm_itimer(struct task_struct *tsk) +{ + struct hrtimer *tmr = &tsk->signal->real_timer; + + if (!hrtimer_is_queued(tmr) && tsk->signal->it_real_incr != 0) { + hrtimer_forward(tmr, tmr->base->get_time(), + tsk->signal->it_real_incr); + hrtimer_restart(tmr); + } +} + +/* + * Interval timers are restarted in the signal delivery path. See + * posixtimer_rearm_itimer(). */ enum hrtimer_restart it_real_fn(struct hrtimer *timer) { @@ -172,10 +192,6 @@ static void set_cpu_itimer(struct task_struct *tsk, unsigned int clock_id, u64 oval, nval, ointerval, ninterval; struct cpu_itimer *it = &tsk->signal->it[clock_id]; - /* - * Use the to_ktime conversion because that clamps the maximum - * value to KTIME_MAX and avoid multiplication overflows. - */ nval = timespec64_to_ns(&value->it_value); ninterval = timespec64_to_ns(&value->it_interval); diff --git a/kernel/time/jiffies.c b/kernel/time/jiffies.c index eddcf4970444..bc4db9e5ab70 100644 --- a/kernel/time/jiffies.c +++ b/kernel/time/jiffies.c @@ -10,28 +10,9 @@ #include <linux/init.h> #include "timekeeping.h" +#include "tick-internal.h" -/* Since jiffies uses a simple TICK_NSEC multiplier - * conversion, the .shift value could be zero. However - * this would make NTP adjustments impossible as they are - * in units of 1/2^.shift. Thus we use JIFFIES_SHIFT to - * shift both the nominator and denominator the same - * amount, and give ntp adjustments in units of 1/2^8 - * - * The value 8 is somewhat carefully chosen, as anything - * larger can result in overflows. TICK_NSEC grows as HZ - * shrinks, so values greater than 8 overflow 32bits when - * HZ=100. - */ -#if HZ < 34 -#define JIFFIES_SHIFT 6 -#elif HZ < 67 -#define JIFFIES_SHIFT 7 -#else -#define JIFFIES_SHIFT 8 -#endif - static u64 jiffies_read(struct clocksource *cs) { return (u64) jiffies; @@ -44,22 +25,24 @@ static u64 jiffies_read(struct clocksource *cs) * the timer interrupt frequency HZ and it suffers * inaccuracies caused by missed or lost timer * interrupts and the inability for the timer - * interrupt hardware to accuratly tick at the + * interrupt hardware to accurately tick at the * requested HZ value. It is also not recommended * for "tick-less" systems. */ static struct clocksource clocksource_jiffies = { - .name = "jiffies", - .rating = 1, /* lowest valid rating*/ - .read = jiffies_read, - .mask = CLOCKSOURCE_MASK(32), - .mult = TICK_NSEC << JIFFIES_SHIFT, /* details above */ - .shift = JIFFIES_SHIFT, - .max_cycles = 10, + .name = "jiffies", + .rating = 1, /* lowest valid rating*/ + .uncertainty_margin = 32 * NSEC_PER_MSEC, + .read = jiffies_read, + .mask = CLOCKSOURCE_MASK(32), + .mult = TICK_NSEC << JIFFIES_SHIFT, /* details above */ + .shift = JIFFIES_SHIFT, + .max_cycles = 10, }; __cacheline_aligned_in_smp DEFINE_RAW_SPINLOCK(jiffies_lock); -__cacheline_aligned_in_smp seqcount_t jiffies_seq; +__cacheline_aligned_in_smp seqcount_raw_spinlock_t jiffies_seq = + SEQCNT_RAW_SPINLOCK_ZERO(jiffies_seq, &jiffies_lock); #if (BITS_PER_LONG < 64) u64 get_jiffies_64(void) diff --git a/kernel/time/namespace.c b/kernel/time/namespace.c index 5d9fc22d836a..0775b9ec952a 100644 --- a/kernel/time/namespace.c +++ b/kernel/time/namespace.c @@ -88,13 +88,13 @@ static struct time_namespace *clone_time_ns(struct user_namespace *user_ns, goto fail; err = -ENOMEM; - ns = kmalloc(sizeof(*ns), GFP_KERNEL); + ns = kmalloc(sizeof(*ns), GFP_KERNEL_ACCOUNT); if (!ns) goto fail_dec; - kref_init(&ns->kref); + refcount_set(&ns->ns.count, 1); - ns->vvar_page = alloc_page(GFP_KERNEL | __GFP_ZERO); + ns->vvar_page = alloc_page(GFP_KERNEL_ACCOUNT | __GFP_ZERO); if (!ns->vvar_page) goto fail_free; @@ -192,6 +192,24 @@ static void timens_setup_vdso_data(struct vdso_data *vdata, offset[CLOCK_BOOTTIME_ALARM] = boottime; } +struct page *find_timens_vvar_page(struct vm_area_struct *vma) +{ + if (likely(vma->vm_mm == current->mm)) + return current->nsproxy->time_ns->vvar_page; + + /* + * VM_PFNMAP | VM_IO protect .fault() handler from being called + * through interfaces like /proc/$pid/mem or + * process_vm_{readv,writev}() as long as there's no .access() + * in special_mapping_vmops(). + * For more details check_vma_flags() and __access_remote_vm() + */ + + WARN(1, "vvar_page accessed remotely"); + + return NULL; +} + /* * Protects possibly multiple offsets writers racing each other * and tasks entering the namespace. @@ -226,11 +244,8 @@ out: mutex_unlock(&offset_lock); } -void free_time_ns(struct kref *kref) +void free_time_ns(struct time_namespace *ns) { - struct time_namespace *ns; - - ns = container_of(kref, struct time_namespace, kref); dec_time_namespaces(ns->ucounts); put_user_ns(ns->user_ns); ns_free_inum(&ns->ns); @@ -280,11 +295,16 @@ static void timens_put(struct ns_common *ns) put_time_ns(to_time_ns(ns)); } +void timens_commit(struct task_struct *tsk, struct time_namespace *ns) +{ + timens_set_vvar_page(tsk, ns); + vdso_join_timens(tsk, ns); +} + static int timens_install(struct nsset *nsset, struct ns_common *new) { struct nsproxy *nsproxy = nsset->nsproxy; struct time_namespace *ns = to_time_ns(new); - int err; if (!current_is_single_threaded()) return -EUSERS; @@ -293,12 +313,6 @@ static int timens_install(struct nsset *nsset, struct ns_common *new) !ns_capable(nsset->cred->user_ns, CAP_SYS_ADMIN)) return -EPERM; - timens_set_vvar_page(current, ns); - - err = vdso_join_timens(current, ns); - if (err) - return err; - get_time_ns(ns); put_time_ns(nsproxy->time_ns); nsproxy->time_ns = ns; @@ -309,27 +323,20 @@ static int timens_install(struct nsset *nsset, struct ns_common *new) return 0; } -int timens_on_fork(struct nsproxy *nsproxy, struct task_struct *tsk) +void timens_on_fork(struct nsproxy *nsproxy, struct task_struct *tsk) { struct ns_common *nsc = &nsproxy->time_ns_for_children->ns; struct time_namespace *ns = to_time_ns(nsc); - int err; /* create_new_namespaces() already incremented the ref counter */ if (nsproxy->time_ns == nsproxy->time_ns_for_children) - return 0; - - timens_set_vvar_page(tsk, ns); - - err = vdso_join_timens(tsk, ns); - if (err) - return err; + return; get_time_ns(ns); put_time_ns(nsproxy->time_ns); nsproxy->time_ns = ns; - return 0; + timens_commit(tsk, ns); } static struct user_namespace *timens_owner(struct ns_common *ns) @@ -470,15 +477,9 @@ const struct proc_ns_operations timens_for_children_operations = { }; struct time_namespace init_time_ns = { - .kref = KREF_INIT(3), + .ns.count = REFCOUNT_INIT(3), .user_ns = &init_user_ns, .ns.inum = PROC_TIME_INIT_INO, .ns.ops = &timens_operations, .frozen_offsets = true, }; - -static int __init time_ns_init(void) -{ - return 0; -} -subsys_initcall(time_ns_init); diff --git a/kernel/time/ntp.c b/kernel/time/ntp.c index 069ca78fb0bf..163e7a2033b6 100644 --- a/kernel/time/ntp.c +++ b/kernel/time/ntp.c @@ -22,22 +22,79 @@ #include "ntp_internal.h" #include "timekeeping_internal.h" - -/* - * NTP timekeeping variables: +/** + * struct ntp_data - Structure holding all NTP related state + * @tick_usec: USER_HZ period in microseconds + * @tick_length: Adjusted tick length + * @tick_length_base: Base value for @tick_length + * @time_state: State of the clock synchronization + * @time_status: Clock status bits + * @time_offset: Time adjustment in nanoseconds + * @time_constant: PLL time constant + * @time_maxerror: Maximum error in microseconds holding the NTP sync distance + * (NTP dispersion + delay / 2) + * @time_esterror: Estimated error in microseconds holding NTP dispersion + * @time_freq: Frequency offset scaled nsecs/secs + * @time_reftime: Time at last adjustment in seconds + * @time_adjust: Adjustment value + * @ntp_tick_adj: Constant boot-param configurable NTP tick adjustment (upscaled) + * @ntp_next_leap_sec: Second value of the next pending leapsecond, or TIME64_MAX if no leap * - * Note: All of the NTP state is protected by the timekeeping locks. + * @pps_valid: PPS signal watchdog counter + * @pps_tf: PPS phase median filter + * @pps_jitter: PPS current jitter in nanoseconds + * @pps_fbase: PPS beginning of the last freq interval + * @pps_shift: PPS current interval duration in seconds (shift value) + * @pps_intcnt: PPS interval counter + * @pps_freq: PPS frequency offset in scaled ns/s + * @pps_stabil: PPS current stability in scaled ns/s + * @pps_calcnt: PPS monitor: calibration intervals + * @pps_jitcnt: PPS monitor: jitter limit exceeded + * @pps_stbcnt: PPS monitor: stability limit exceeded + * @pps_errcnt: PPS monitor: calibration errors + * + * Protected by the timekeeping locks. */ +struct ntp_data { + unsigned long tick_usec; + u64 tick_length; + u64 tick_length_base; + int time_state; + int time_status; + s64 time_offset; + long time_constant; + long time_maxerror; + long time_esterror; + s64 time_freq; + time64_t time_reftime; + long time_adjust; + s64 ntp_tick_adj; + time64_t ntp_next_leap_sec; +#ifdef CONFIG_NTP_PPS + int pps_valid; + long pps_tf[3]; + long pps_jitter; + struct timespec64 pps_fbase; + int pps_shift; + int pps_intcnt; + s64 pps_freq; + long pps_stabil; + long pps_calcnt; + long pps_jitcnt; + long pps_stbcnt; + long pps_errcnt; +#endif +}; - -/* USER_HZ period (usecs): */ -unsigned long tick_usec = USER_TICK_USEC; - -/* SHIFTED_HZ period (nsecs): */ -unsigned long tick_nsec; - -static u64 tick_length; -static u64 tick_length_base; +static struct ntp_data tk_ntp_data = { + .tick_usec = USER_TICK_USEC, + .time_state = TIME_OK, + .time_status = STA_UNSYNC, + .time_constant = 2, + .time_maxerror = NTP_PHASE_LIMIT, + .time_esterror = NTP_PHASE_LIMIT, + .ntp_next_leap_sec = TIME64_MAX, +}; #define SECS_PER_DAY 86400 #define MAX_TICKADJ 500LL /* usecs */ @@ -45,46 +102,6 @@ static u64 tick_length_base; (((MAX_TICKADJ * NSEC_PER_USEC) << NTP_SCALE_SHIFT) / NTP_INTERVAL_FREQ) #define MAX_TAI_OFFSET 100000 -/* - * phase-lock loop variables - */ - -/* - * clock synchronization status - * - * (TIME_ERROR prevents overwriting the CMOS clock) - */ -static int time_state = TIME_OK; - -/* clock status bits: */ -static int time_status = STA_UNSYNC; - -/* time adjustment (nsecs): */ -static s64 time_offset; - -/* pll time constant: */ -static long time_constant = 2; - -/* maximum error (usecs): */ -static long time_maxerror = NTP_PHASE_LIMIT; - -/* estimated error (usecs): */ -static long time_esterror = NTP_PHASE_LIMIT; - -/* frequency offset (scaled nsecs/secs): */ -static s64 time_freq; - -/* time at last adjustment (secs): */ -static time64_t time_reftime; - -static long time_adjust; - -/* constant (boot-param configurable) NTP tick adjustment (upscaled) */ -static s64 ntp_tick_adj; - -/* second value of the next pending leapsecond, or TIME64_MAX if no leap */ -static time64_t ntp_next_leap_sec = TIME64_MAX; - #ifdef CONFIG_NTP_PPS /* @@ -101,128 +118,115 @@ static time64_t ntp_next_leap_sec = TIME64_MAX; intervals to decrease it */ #define PPS_MAXWANDER 100000 /* max PPS freq wander (ns/s) */ -static int pps_valid; /* signal watchdog counter */ -static long pps_tf[3]; /* phase median filter */ -static long pps_jitter; /* current jitter (ns) */ -static struct timespec64 pps_fbase; /* beginning of the last freq interval */ -static int pps_shift; /* current interval duration (s) (shift) */ -static int pps_intcnt; /* interval counter */ -static s64 pps_freq; /* frequency offset (scaled ns/s) */ -static long pps_stabil; /* current stability (scaled ns/s) */ - /* - * PPS signal quality monitors - */ -static long pps_calcnt; /* calibration intervals */ -static long pps_jitcnt; /* jitter limit exceeded */ -static long pps_stbcnt; /* stability limit exceeded */ -static long pps_errcnt; /* calibration errors */ - - -/* PPS kernel consumer compensates the whole phase error immediately. + * PPS kernel consumer compensates the whole phase error immediately. * Otherwise, reduce the offset by a fixed factor times the time constant. */ -static inline s64 ntp_offset_chunk(s64 offset) +static inline s64 ntp_offset_chunk(struct ntp_data *ntpdata, s64 offset) { - if (time_status & STA_PPSTIME && time_status & STA_PPSSIGNAL) + if (ntpdata->time_status & STA_PPSTIME && ntpdata->time_status & STA_PPSSIGNAL) return offset; else - return shift_right(offset, SHIFT_PLL + time_constant); + return shift_right(offset, SHIFT_PLL + ntpdata->time_constant); } -static inline void pps_reset_freq_interval(void) +static inline void pps_reset_freq_interval(struct ntp_data *ntpdata) { - /* the PPS calibration interval may end - surprisingly early */ - pps_shift = PPS_INTMIN; - pps_intcnt = 0; + /* The PPS calibration interval may end surprisingly early */ + ntpdata->pps_shift = PPS_INTMIN; + ntpdata->pps_intcnt = 0; } /** * pps_clear - Clears the PPS state variables + * @ntpdata: Pointer to ntp data */ -static inline void pps_clear(void) +static inline void pps_clear(struct ntp_data *ntpdata) { - pps_reset_freq_interval(); - pps_tf[0] = 0; - pps_tf[1] = 0; - pps_tf[2] = 0; - pps_fbase.tv_sec = pps_fbase.tv_nsec = 0; - pps_freq = 0; + pps_reset_freq_interval(ntpdata); + ntpdata->pps_tf[0] = 0; + ntpdata->pps_tf[1] = 0; + ntpdata->pps_tf[2] = 0; + ntpdata->pps_fbase.tv_sec = ntpdata->pps_fbase.tv_nsec = 0; + ntpdata->pps_freq = 0; } -/* Decrease pps_valid to indicate that another second has passed since - * the last PPS signal. When it reaches 0, indicate that PPS signal is - * missing. +/* + * Decrease pps_valid to indicate that another second has passed since the + * last PPS signal. When it reaches 0, indicate that PPS signal is missing. */ -static inline void pps_dec_valid(void) +static inline void pps_dec_valid(struct ntp_data *ntpdata) { - if (pps_valid > 0) - pps_valid--; - else { - time_status &= ~(STA_PPSSIGNAL | STA_PPSJITTER | - STA_PPSWANDER | STA_PPSERROR); - pps_clear(); + if (ntpdata->pps_valid > 0) { + ntpdata->pps_valid--; + } else { + ntpdata->time_status &= ~(STA_PPSSIGNAL | STA_PPSJITTER | + STA_PPSWANDER | STA_PPSERROR); + pps_clear(ntpdata); } } -static inline void pps_set_freq(s64 freq) +static inline void pps_set_freq(struct ntp_data *ntpdata) { - pps_freq = freq; + ntpdata->pps_freq = ntpdata->time_freq; } -static inline int is_error_status(int status) +static inline bool is_error_status(int status) { return (status & (STA_UNSYNC|STA_CLOCKERR)) - /* PPS signal lost when either PPS time or - * PPS frequency synchronization requested + /* + * PPS signal lost when either PPS time or PPS frequency + * synchronization requested */ || ((status & (STA_PPSFREQ|STA_PPSTIME)) && !(status & STA_PPSSIGNAL)) - /* PPS jitter exceeded when - * PPS time synchronization requested */ + /* + * PPS jitter exceeded when PPS time synchronization + * requested + */ || ((status & (STA_PPSTIME|STA_PPSJITTER)) == (STA_PPSTIME|STA_PPSJITTER)) - /* PPS wander exceeded or calibration error when - * PPS frequency synchronization requested + /* + * PPS wander exceeded or calibration error when PPS + * frequency synchronization requested */ || ((status & STA_PPSFREQ) && (status & (STA_PPSWANDER|STA_PPSERROR))); } -static inline void pps_fill_timex(struct __kernel_timex *txc) +static inline void pps_fill_timex(struct ntp_data *ntpdata, struct __kernel_timex *txc) { - txc->ppsfreq = shift_right((pps_freq >> PPM_SCALE_INV_SHIFT) * + txc->ppsfreq = shift_right((ntpdata->pps_freq >> PPM_SCALE_INV_SHIFT) * PPM_SCALE_INV, NTP_SCALE_SHIFT); - txc->jitter = pps_jitter; - if (!(time_status & STA_NANO)) - txc->jitter = pps_jitter / NSEC_PER_USEC; - txc->shift = pps_shift; - txc->stabil = pps_stabil; - txc->jitcnt = pps_jitcnt; - txc->calcnt = pps_calcnt; - txc->errcnt = pps_errcnt; - txc->stbcnt = pps_stbcnt; + txc->jitter = ntpdata->pps_jitter; + if (!(ntpdata->time_status & STA_NANO)) + txc->jitter = ntpdata->pps_jitter / NSEC_PER_USEC; + txc->shift = ntpdata->pps_shift; + txc->stabil = ntpdata->pps_stabil; + txc->jitcnt = ntpdata->pps_jitcnt; + txc->calcnt = ntpdata->pps_calcnt; + txc->errcnt = ntpdata->pps_errcnt; + txc->stbcnt = ntpdata->pps_stbcnt; } #else /* !CONFIG_NTP_PPS */ -static inline s64 ntp_offset_chunk(s64 offset) +static inline s64 ntp_offset_chunk(struct ntp_data *ntpdata, s64 offset) { - return shift_right(offset, SHIFT_PLL + time_constant); + return shift_right(offset, SHIFT_PLL + ntpdata->time_constant); } -static inline void pps_reset_freq_interval(void) {} -static inline void pps_clear(void) {} -static inline void pps_dec_valid(void) {} -static inline void pps_set_freq(s64 freq) {} +static inline void pps_reset_freq_interval(struct ntp_data *ntpdata) {} +static inline void pps_clear(struct ntp_data *ntpdata) {} +static inline void pps_dec_valid(struct ntp_data *ntpdata) {} +static inline void pps_set_freq(struct ntp_data *ntpdata) {} -static inline int is_error_status(int status) +static inline bool is_error_status(int status) { return status & (STA_UNSYNC|STA_CLOCKERR); } -static inline void pps_fill_timex(struct __kernel_timex *txc) +static inline void pps_fill_timex(struct ntp_data *ntpdata, struct __kernel_timex *txc) { /* PPS is not implemented, so these are zero */ txc->ppsfreq = 0; @@ -237,138 +241,123 @@ static inline void pps_fill_timex(struct __kernel_timex *txc) #endif /* CONFIG_NTP_PPS */ - -/** - * ntp_synced - Returns 1 if the NTP status is not UNSYNC - * - */ -static inline int ntp_synced(void) -{ - return !(time_status & STA_UNSYNC); -} - - /* - * NTP methods: + * Update tick_length and tick_length_base, based on tick_usec, ntp_tick_adj and + * time_freq: */ - -/* - * Update (tick_length, tick_length_base, tick_nsec), based - * on (tick_usec, ntp_tick_adj, time_freq): - */ -static void ntp_update_frequency(void) +static void ntp_update_frequency(struct ntp_data *ntpdata) { - u64 second_length; - u64 new_base; + u64 second_length, new_base, tick_usec = (u64)ntpdata->tick_usec; - second_length = (u64)(tick_usec * NSEC_PER_USEC * USER_HZ) - << NTP_SCALE_SHIFT; + second_length = (u64)(tick_usec * NSEC_PER_USEC * USER_HZ) << NTP_SCALE_SHIFT; - second_length += ntp_tick_adj; - second_length += time_freq; + second_length += ntpdata->ntp_tick_adj; + second_length += ntpdata->time_freq; - tick_nsec = div_u64(second_length, HZ) >> NTP_SCALE_SHIFT; new_base = div_u64(second_length, NTP_INTERVAL_FREQ); /* - * Don't wait for the next second_overflow, apply - * the change to the tick length immediately: + * Don't wait for the next second_overflow, apply the change to the + * tick length immediately: */ - tick_length += new_base - tick_length_base; - tick_length_base = new_base; + ntpdata->tick_length += new_base - ntpdata->tick_length_base; + ntpdata->tick_length_base = new_base; } -static inline s64 ntp_update_offset_fll(s64 offset64, long secs) +static inline s64 ntp_update_offset_fll(struct ntp_data *ntpdata, s64 offset64, long secs) { - time_status &= ~STA_MODE; + ntpdata->time_status &= ~STA_MODE; if (secs < MINSEC) return 0; - if (!(time_status & STA_FLL) && (secs <= MAXSEC)) + if (!(ntpdata->time_status & STA_FLL) && (secs <= MAXSEC)) return 0; - time_status |= STA_MODE; + ntpdata->time_status |= STA_MODE; return div64_long(offset64 << (NTP_SCALE_SHIFT - SHIFT_FLL), secs); } -static void ntp_update_offset(long offset) +static void ntp_update_offset(struct ntp_data *ntpdata, long offset) { - s64 freq_adj; - s64 offset64; - long secs; + s64 freq_adj, offset64; + long secs, real_secs; - if (!(time_status & STA_PLL)) + if (!(ntpdata->time_status & STA_PLL)) return; - if (!(time_status & STA_NANO)) { + if (!(ntpdata->time_status & STA_NANO)) { /* Make sure the multiplication below won't overflow */ offset = clamp(offset, -USEC_PER_SEC, USEC_PER_SEC); offset *= NSEC_PER_USEC; } - /* - * Scale the phase adjustment and - * clamp to the operating range. - */ + /* Scale the phase adjustment and clamp to the operating range. */ offset = clamp(offset, -MAXPHASE, MAXPHASE); /* * Select how the frequency is to be controlled * and in which mode (PLL or FLL). */ - secs = (long)(__ktime_get_real_seconds() - time_reftime); - if (unlikely(time_status & STA_FREQHOLD)) + real_secs = __ktime_get_real_seconds(); + secs = (long)(real_secs - ntpdata->time_reftime); + if (unlikely(ntpdata->time_status & STA_FREQHOLD)) secs = 0; - time_reftime = __ktime_get_real_seconds(); + ntpdata->time_reftime = real_secs; offset64 = offset; - freq_adj = ntp_update_offset_fll(offset64, secs); + freq_adj = ntp_update_offset_fll(ntpdata, offset64, secs); /* * Clamp update interval to reduce PLL gain with low * sampling rate (e.g. intermittent network connection) * to avoid instability. */ - if (unlikely(secs > 1 << (SHIFT_PLL + 1 + time_constant))) - secs = 1 << (SHIFT_PLL + 1 + time_constant); + if (unlikely(secs > 1 << (SHIFT_PLL + 1 + ntpdata->time_constant))) + secs = 1 << (SHIFT_PLL + 1 + ntpdata->time_constant); freq_adj += (offset64 * secs) << - (NTP_SCALE_SHIFT - 2 * (SHIFT_PLL + 2 + time_constant)); + (NTP_SCALE_SHIFT - 2 * (SHIFT_PLL + 2 + ntpdata->time_constant)); - freq_adj = min(freq_adj + time_freq, MAXFREQ_SCALED); + freq_adj = min(freq_adj + ntpdata->time_freq, MAXFREQ_SCALED); - time_freq = max(freq_adj, -MAXFREQ_SCALED); + ntpdata->time_freq = max(freq_adj, -MAXFREQ_SCALED); - time_offset = div_s64(offset64 << NTP_SCALE_SHIFT, NTP_INTERVAL_FREQ); + ntpdata->time_offset = div_s64(offset64 << NTP_SCALE_SHIFT, NTP_INTERVAL_FREQ); } -/** - * ntp_clear - Clears the NTP state variables - */ -void ntp_clear(void) +static void __ntp_clear(struct ntp_data *ntpdata) { - time_adjust = 0; /* stop active adjtime() */ - time_status |= STA_UNSYNC; - time_maxerror = NTP_PHASE_LIMIT; - time_esterror = NTP_PHASE_LIMIT; + /* Stop active adjtime() */ + ntpdata->time_adjust = 0; + ntpdata->time_status |= STA_UNSYNC; + ntpdata->time_maxerror = NTP_PHASE_LIMIT; + ntpdata->time_esterror = NTP_PHASE_LIMIT; - ntp_update_frequency(); + ntp_update_frequency(ntpdata); - tick_length = tick_length_base; - time_offset = 0; + ntpdata->tick_length = ntpdata->tick_length_base; + ntpdata->time_offset = 0; - ntp_next_leap_sec = TIME64_MAX; + ntpdata->ntp_next_leap_sec = TIME64_MAX; /* Clear PPS state variables */ - pps_clear(); + pps_clear(ntpdata); +} + +/** + * ntp_clear - Clears the NTP state variables + */ +void ntp_clear(void) +{ + __ntp_clear(&tk_ntp_data); } u64 ntp_tick_length(void) { - return tick_length; + return tk_ntp_data.tick_length; } /** @@ -379,16 +368,17 @@ u64 ntp_tick_length(void) */ ktime_t ntp_get_next_leap(void) { + struct ntp_data *ntpdata = &tk_ntp_data; ktime_t ret; - if ((time_state == TIME_INS) && (time_status & STA_INS)) - return ktime_set(ntp_next_leap_sec, 0); + if ((ntpdata->time_state == TIME_INS) && (ntpdata->time_status & STA_INS)) + return ktime_set(ntpdata->ntp_next_leap_sec, 0); ret = KTIME_MAX; return ret; } /* - * this routine handles the overflow of the microsecond field + * This routine handles the overflow of the microsecond field * * The tricky bits of code to handle the accurate clock support * were provided by Dave Mills (Mills@UDEL.EDU) of NTP fame. @@ -399,6 +389,7 @@ ktime_t ntp_get_next_leap(void) */ int second_overflow(time64_t secs) { + struct ntp_data *ntpdata = &tk_ntp_data; s64 delta; int leap = 0; s32 rem; @@ -408,151 +399,157 @@ int second_overflow(time64_t secs) * day, the system clock is set back one second; if in leap-delete * state, the system clock is set ahead one second. */ - switch (time_state) { + switch (ntpdata->time_state) { case TIME_OK: - if (time_status & STA_INS) { - time_state = TIME_INS; + if (ntpdata->time_status & STA_INS) { + ntpdata->time_state = TIME_INS; div_s64_rem(secs, SECS_PER_DAY, &rem); - ntp_next_leap_sec = secs + SECS_PER_DAY - rem; - } else if (time_status & STA_DEL) { - time_state = TIME_DEL; + ntpdata->ntp_next_leap_sec = secs + SECS_PER_DAY - rem; + } else if (ntpdata->time_status & STA_DEL) { + ntpdata->time_state = TIME_DEL; div_s64_rem(secs + 1, SECS_PER_DAY, &rem); - ntp_next_leap_sec = secs + SECS_PER_DAY - rem; + ntpdata->ntp_next_leap_sec = secs + SECS_PER_DAY - rem; } break; case TIME_INS: - if (!(time_status & STA_INS)) { - ntp_next_leap_sec = TIME64_MAX; - time_state = TIME_OK; - } else if (secs == ntp_next_leap_sec) { + if (!(ntpdata->time_status & STA_INS)) { + ntpdata->ntp_next_leap_sec = TIME64_MAX; + ntpdata->time_state = TIME_OK; + } else if (secs == ntpdata->ntp_next_leap_sec) { leap = -1; - time_state = TIME_OOP; - printk(KERN_NOTICE - "Clock: inserting leap second 23:59:60 UTC\n"); + ntpdata->time_state = TIME_OOP; + pr_notice("Clock: inserting leap second 23:59:60 UTC\n"); } break; case TIME_DEL: - if (!(time_status & STA_DEL)) { - ntp_next_leap_sec = TIME64_MAX; - time_state = TIME_OK; - } else if (secs == ntp_next_leap_sec) { + if (!(ntpdata->time_status & STA_DEL)) { + ntpdata->ntp_next_leap_sec = TIME64_MAX; + ntpdata->time_state = TIME_OK; + } else if (secs == ntpdata->ntp_next_leap_sec) { leap = 1; - ntp_next_leap_sec = TIME64_MAX; - time_state = TIME_WAIT; - printk(KERN_NOTICE - "Clock: deleting leap second 23:59:59 UTC\n"); + ntpdata->ntp_next_leap_sec = TIME64_MAX; + ntpdata->time_state = TIME_WAIT; + pr_notice("Clock: deleting leap second 23:59:59 UTC\n"); } break; case TIME_OOP: - ntp_next_leap_sec = TIME64_MAX; - time_state = TIME_WAIT; + ntpdata->ntp_next_leap_sec = TIME64_MAX; + ntpdata->time_state = TIME_WAIT; break; case TIME_WAIT: - if (!(time_status & (STA_INS | STA_DEL))) - time_state = TIME_OK; + if (!(ntpdata->time_status & (STA_INS | STA_DEL))) + ntpdata->time_state = TIME_OK; break; } - /* Bump the maxerror field */ - time_maxerror += MAXFREQ / NSEC_PER_USEC; - if (time_maxerror > NTP_PHASE_LIMIT) { - time_maxerror = NTP_PHASE_LIMIT; - time_status |= STA_UNSYNC; + ntpdata->time_maxerror += MAXFREQ / NSEC_PER_USEC; + if (ntpdata->time_maxerror > NTP_PHASE_LIMIT) { + ntpdata->time_maxerror = NTP_PHASE_LIMIT; + ntpdata->time_status |= STA_UNSYNC; } /* Compute the phase adjustment for the next second */ - tick_length = tick_length_base; + ntpdata->tick_length = ntpdata->tick_length_base; - delta = ntp_offset_chunk(time_offset); - time_offset -= delta; - tick_length += delta; + delta = ntp_offset_chunk(ntpdata, ntpdata->time_offset); + ntpdata->time_offset -= delta; + ntpdata->tick_length += delta; /* Check PPS signal */ - pps_dec_valid(); + pps_dec_valid(ntpdata); - if (!time_adjust) + if (!ntpdata->time_adjust) goto out; - if (time_adjust > MAX_TICKADJ) { - time_adjust -= MAX_TICKADJ; - tick_length += MAX_TICKADJ_SCALED; + if (ntpdata->time_adjust > MAX_TICKADJ) { + ntpdata->time_adjust -= MAX_TICKADJ; + ntpdata->tick_length += MAX_TICKADJ_SCALED; goto out; } - if (time_adjust < -MAX_TICKADJ) { - time_adjust += MAX_TICKADJ; - tick_length -= MAX_TICKADJ_SCALED; + if (ntpdata->time_adjust < -MAX_TICKADJ) { + ntpdata->time_adjust += MAX_TICKADJ; + ntpdata->tick_length -= MAX_TICKADJ_SCALED; goto out; } - tick_length += (s64)(time_adjust * NSEC_PER_USEC / NTP_INTERVAL_FREQ) - << NTP_SCALE_SHIFT; - time_adjust = 0; + ntpdata->tick_length += (s64)(ntpdata->time_adjust * NSEC_PER_USEC / NTP_INTERVAL_FREQ) + << NTP_SCALE_SHIFT; + ntpdata->time_adjust = 0; out: return leap; } +#if defined(CONFIG_GENERIC_CMOS_UPDATE) || defined(CONFIG_RTC_SYSTOHC) static void sync_hw_clock(struct work_struct *work); -static DECLARE_DELAYED_WORK(sync_work, sync_hw_clock); - -static void sched_sync_hw_clock(struct timespec64 now, - unsigned long target_nsec, bool fail) +static DECLARE_WORK(sync_work, sync_hw_clock); +static struct hrtimer sync_hrtimer; +#define SYNC_PERIOD_NS (11ULL * 60 * NSEC_PER_SEC) +static enum hrtimer_restart sync_timer_callback(struct hrtimer *timer) { - struct timespec64 next; - - ktime_get_real_ts64(&next); - if (!fail) - next.tv_sec = 659; - else { - /* - * Try again as soon as possible. Delaying long periods - * decreases the accuracy of the work queue timer. Due to this - * the algorithm is very likely to require a short-sleep retry - * after the above long sleep to synchronize ts_nsec. - */ - next.tv_sec = 0; - } + queue_work(system_freezable_power_efficient_wq, &sync_work); - /* Compute the needed delay that will get to tv_nsec == target_nsec */ - next.tv_nsec = target_nsec - next.tv_nsec; - if (next.tv_nsec <= 0) - next.tv_nsec += NSEC_PER_SEC; - if (next.tv_nsec >= NSEC_PER_SEC) { - next.tv_sec++; - next.tv_nsec -= NSEC_PER_SEC; - } - - queue_delayed_work(system_power_efficient_wq, &sync_work, - timespec64_to_jiffies(&next)); + return HRTIMER_NORESTART; } -static void sync_rtc_clock(void) +static void sched_sync_hw_clock(unsigned long offset_nsec, bool retry) { - unsigned long target_nsec; - struct timespec64 adjust, now; - int rc; + ktime_t exp = ktime_set(ktime_get_real_seconds(), 0); - if (!IS_ENABLED(CONFIG_RTC_SYSTOHC)) - return; + if (retry) + exp = ktime_add_ns(exp, 2ULL * NSEC_PER_SEC - offset_nsec); + else + exp = ktime_add_ns(exp, SYNC_PERIOD_NS - offset_nsec); - ktime_get_real_ts64(&now); + hrtimer_start(&sync_hrtimer, exp, HRTIMER_MODE_ABS); +} - adjust = now; - if (persistent_clock_is_local) - adjust.tv_sec -= (sys_tz.tz_minuteswest * 60); +/* + * Check whether @now is correct versus the required time to update the RTC + * and calculate the value which needs to be written to the RTC so that the + * next seconds increment of the RTC after the write is aligned with the next + * seconds increment of clock REALTIME. + * + * tsched t1 write(t2.tv_sec - 1sec)) t2 RTC increments seconds + * + * t2.tv_nsec == 0 + * tsched = t2 - set_offset_nsec + * newval = t2 - NSEC_PER_SEC + * + * ==> neval = tsched + set_offset_nsec - NSEC_PER_SEC + * + * As the execution of this code is not guaranteed to happen exactly at + * tsched this allows it to happen within a fuzzy region: + * + * abs(now - tsched) < FUZZ + * + * If @now is not inside the allowed window the function returns false. + */ +static inline bool rtc_tv_nsec_ok(unsigned long set_offset_nsec, + struct timespec64 *to_set, + const struct timespec64 *now) +{ + /* Allowed error in tv_nsec, arbitrarily set to 5 jiffies in ns. */ + const unsigned long TIME_SET_NSEC_FUZZ = TICK_NSEC * 5; + struct timespec64 delay = {.tv_sec = -1, + .tv_nsec = set_offset_nsec}; - /* - * The current RTC in use will provide the target_nsec it wants to be - * called at, and does rtc_tv_nsec_ok internally. - */ - rc = rtc_set_ntp_time(adjust, &target_nsec); - if (rc == -ENODEV) - return; + *to_set = timespec64_add(*now, delay); - sched_sync_hw_clock(now, target_nsec, rc); + if (to_set->tv_nsec < TIME_SET_NSEC_FUZZ) { + to_set->tv_nsec = 0; + return true; + } + + if (to_set->tv_nsec > NSEC_PER_SEC - TIME_SET_NSEC_FUZZ) { + to_set->tv_sec++; + to_set->tv_nsec = 0; + return true; + } + return false; } #ifdef CONFIG_GENERIC_CMOS_UPDATE @@ -560,47 +557,55 @@ int __weak update_persistent_clock64(struct timespec64 now64) { return -ENODEV; } +#else +static inline int update_persistent_clock64(struct timespec64 now64) +{ + return -ENODEV; +} #endif -static bool sync_cmos_clock(void) +#ifdef CONFIG_RTC_SYSTOHC +/* Save NTP synchronized time to the RTC */ +static int update_rtc(struct timespec64 *to_set, unsigned long *offset_nsec) { - static bool no_cmos; - struct timespec64 now; - struct timespec64 adjust; - int rc = -EPROTO; - long target_nsec = NSEC_PER_SEC / 2; + struct rtc_device *rtc; + struct rtc_time tm; + int err = -ENODEV; - if (!IS_ENABLED(CONFIG_GENERIC_CMOS_UPDATE)) - return false; + rtc = rtc_class_open(CONFIG_RTC_SYSTOHC_DEVICE); + if (!rtc) + return -ENODEV; - if (no_cmos) - return false; + if (!rtc->ops || !rtc->ops->set_time) + goto out_close; - /* - * Historically update_persistent_clock64() has followed x86 - * semantics, which match the MC146818A/etc RTC. This RTC will store - * 'adjust' and then in .5s it will advance once second. - * - * Architectures are strongly encouraged to use rtclib and not - * implement this legacy API. - */ - ktime_get_real_ts64(&now); - if (rtc_tv_nsec_ok(-1 * target_nsec, &adjust, &now)) { - if (persistent_clock_is_local) - adjust.tv_sec -= (sys_tz.tz_minuteswest * 60); - rc = update_persistent_clock64(adjust); - /* - * The machine does not support update_persistent_clock64 even - * though it defines CONFIG_GENERIC_CMOS_UPDATE. - */ - if (rc == -ENODEV) { - no_cmos = true; - return false; - } + /* First call might not have the correct offset */ + if (*offset_nsec == rtc->set_offset_nsec) { + rtc_time64_to_tm(to_set->tv_sec, &tm); + err = rtc_set_time(rtc, &tm); + } else { + /* Store the update offset and let the caller try again */ + *offset_nsec = rtc->set_offset_nsec; + err = -EAGAIN; } +out_close: + rtc_class_close(rtc); + return err; +} +#else +static inline int update_rtc(struct timespec64 *to_set, unsigned long *offset_nsec) +{ + return -ENODEV; +} +#endif - sched_sync_hw_clock(now, target_nsec, rc); - return true; +/** + * ntp_synced - Tells whether the NTP status is not UNSYNC + * Returns: true if not UNSYNC, false otherwise + */ +static inline bool ntp_synced(void) +{ + return !(tk_ntp_data.time_status & STA_UNSYNC); } /* @@ -613,185 +618,226 @@ static bool sync_cmos_clock(void) */ static void sync_hw_clock(struct work_struct *work) { - if (!ntp_synced()) - return; + /* + * The default synchronization offset is 500ms for the deprecated + * update_persistent_clock64() under the assumption that it uses + * the infamous CMOS clock (MC146818). + */ + static unsigned long offset_nsec = NSEC_PER_SEC / 2; + struct timespec64 now, to_set; + int res = -EAGAIN; - if (sync_cmos_clock()) + /* + * Don't update if STA_UNSYNC is set and if ntp_notify_cmos_timer() + * managed to schedule the work between the timer firing and the + * work being able to rearm the timer. Wait for the timer to expire. + */ + if (!ntp_synced() || hrtimer_is_queued(&sync_hrtimer)) return; - sync_rtc_clock(); + ktime_get_real_ts64(&now); + /* If @now is not in the allowed window, try again */ + if (!rtc_tv_nsec_ok(offset_nsec, &to_set, &now)) + goto rearm; + + /* Take timezone adjusted RTCs into account */ + if (persistent_clock_is_local) + to_set.tv_sec -= (sys_tz.tz_minuteswest * 60); + + /* Try the legacy RTC first. */ + res = update_persistent_clock64(to_set); + if (res != -ENODEV) + goto rearm; + + /* Try the RTC class */ + res = update_rtc(&to_set, &offset_nsec); + if (res == -ENODEV) + return; +rearm: + sched_sync_hw_clock(offset_nsec, res != 0); } -void ntp_notify_cmos_timer(void) +void ntp_notify_cmos_timer(bool offset_set) { - if (!ntp_synced()) - return; + /* + * If the time jumped (using ADJ_SETOFFSET) cancels sync timer, + * which may have been running if the time was synchronized + * prior to the ADJ_SETOFFSET call. + */ + if (offset_set) + hrtimer_cancel(&sync_hrtimer); - if (IS_ENABLED(CONFIG_GENERIC_CMOS_UPDATE) || - IS_ENABLED(CONFIG_RTC_SYSTOHC)) - queue_delayed_work(system_power_efficient_wq, &sync_work, 0); + /* + * When the work is currently executed but has not yet the timer + * rearmed this queues the work immediately again. No big issue, + * just a pointless work scheduled. + */ + if (ntp_synced() && !hrtimer_is_queued(&sync_hrtimer)) + queue_work(system_freezable_power_efficient_wq, &sync_work); +} + +static void __init ntp_init_cmos_sync(void) +{ + hrtimer_init(&sync_hrtimer, CLOCK_REALTIME, HRTIMER_MODE_ABS); + sync_hrtimer.function = sync_timer_callback; } +#else /* CONFIG_GENERIC_CMOS_UPDATE) || defined(CONFIG_RTC_SYSTOHC) */ +static inline void __init ntp_init_cmos_sync(void) { } +#endif /* !CONFIG_GENERIC_CMOS_UPDATE) || defined(CONFIG_RTC_SYSTOHC) */ /* * Propagate a new txc->status value into the NTP state: */ -static inline void process_adj_status(const struct __kernel_timex *txc) +static inline void process_adj_status(struct ntp_data *ntpdata, const struct __kernel_timex *txc) { - if ((time_status & STA_PLL) && !(txc->status & STA_PLL)) { - time_state = TIME_OK; - time_status = STA_UNSYNC; - ntp_next_leap_sec = TIME64_MAX; - /* restart PPS frequency calibration */ - pps_reset_freq_interval(); + if ((ntpdata->time_status & STA_PLL) && !(txc->status & STA_PLL)) { + ntpdata->time_state = TIME_OK; + ntpdata->time_status = STA_UNSYNC; + ntpdata->ntp_next_leap_sec = TIME64_MAX; + /* Restart PPS frequency calibration */ + pps_reset_freq_interval(ntpdata); } /* * If we turn on PLL adjustments then reset the * reference time to current time. */ - if (!(time_status & STA_PLL) && (txc->status & STA_PLL)) - time_reftime = __ktime_get_real_seconds(); + if (!(ntpdata->time_status & STA_PLL) && (txc->status & STA_PLL)) + ntpdata->time_reftime = __ktime_get_real_seconds(); /* only set allowed bits */ - time_status &= STA_RONLY; - time_status |= txc->status & ~STA_RONLY; + ntpdata->time_status &= STA_RONLY; + ntpdata->time_status |= txc->status & ~STA_RONLY; } - -static inline void process_adjtimex_modes(const struct __kernel_timex *txc, +static inline void process_adjtimex_modes(struct ntp_data *ntpdata, const struct __kernel_timex *txc, s32 *time_tai) { if (txc->modes & ADJ_STATUS) - process_adj_status(txc); + process_adj_status(ntpdata, txc); if (txc->modes & ADJ_NANO) - time_status |= STA_NANO; + ntpdata->time_status |= STA_NANO; if (txc->modes & ADJ_MICRO) - time_status &= ~STA_NANO; + ntpdata->time_status &= ~STA_NANO; if (txc->modes & ADJ_FREQUENCY) { - time_freq = txc->freq * PPM_SCALE; - time_freq = min(time_freq, MAXFREQ_SCALED); - time_freq = max(time_freq, -MAXFREQ_SCALED); - /* update pps_freq */ - pps_set_freq(time_freq); + ntpdata->time_freq = txc->freq * PPM_SCALE; + ntpdata->time_freq = min(ntpdata->time_freq, MAXFREQ_SCALED); + ntpdata->time_freq = max(ntpdata->time_freq, -MAXFREQ_SCALED); + /* Update pps_freq */ + pps_set_freq(ntpdata); } if (txc->modes & ADJ_MAXERROR) - time_maxerror = txc->maxerror; + ntpdata->time_maxerror = clamp(txc->maxerror, 0, NTP_PHASE_LIMIT); if (txc->modes & ADJ_ESTERROR) - time_esterror = txc->esterror; + ntpdata->time_esterror = clamp(txc->esterror, 0, NTP_PHASE_LIMIT); if (txc->modes & ADJ_TIMECONST) { - time_constant = txc->constant; - if (!(time_status & STA_NANO)) - time_constant += 4; - time_constant = min(time_constant, (long)MAXTC); - time_constant = max(time_constant, 0l); + ntpdata->time_constant = clamp(txc->constant, 0, MAXTC); + if (!(ntpdata->time_status & STA_NANO)) + ntpdata->time_constant += 4; + ntpdata->time_constant = clamp(ntpdata->time_constant, 0, MAXTC); } - if (txc->modes & ADJ_TAI && - txc->constant >= 0 && txc->constant <= MAX_TAI_OFFSET) + if (txc->modes & ADJ_TAI && txc->constant >= 0 && txc->constant <= MAX_TAI_OFFSET) *time_tai = txc->constant; if (txc->modes & ADJ_OFFSET) - ntp_update_offset(txc->offset); + ntp_update_offset(ntpdata, txc->offset); if (txc->modes & ADJ_TICK) - tick_usec = txc->tick; + ntpdata->tick_usec = txc->tick; if (txc->modes & (ADJ_TICK|ADJ_FREQUENCY|ADJ_OFFSET)) - ntp_update_frequency(); + ntp_update_frequency(ntpdata); } - /* - * adjtimex mainly allows reading (and writing, if superuser) of + * adjtimex() mainly allows reading (and writing, if superuser) of * kernel time-keeping variables. used by xntpd. */ int __do_adjtimex(struct __kernel_timex *txc, const struct timespec64 *ts, s32 *time_tai, struct audit_ntp_data *ad) { + struct ntp_data *ntpdata = &tk_ntp_data; int result; if (txc->modes & ADJ_ADJTIME) { - long save_adjust = time_adjust; + long save_adjust = ntpdata->time_adjust; if (!(txc->modes & ADJ_OFFSET_READONLY)) { /* adjtime() is independent from ntp_adjtime() */ - time_adjust = txc->offset; - ntp_update_frequency(); + ntpdata->time_adjust = txc->offset; + ntp_update_frequency(ntpdata); audit_ntp_set_old(ad, AUDIT_NTP_ADJUST, save_adjust); - audit_ntp_set_new(ad, AUDIT_NTP_ADJUST, time_adjust); + audit_ntp_set_new(ad, AUDIT_NTP_ADJUST, ntpdata->time_adjust); } txc->offset = save_adjust; } else { /* If there are input parameters, then process them: */ if (txc->modes) { - audit_ntp_set_old(ad, AUDIT_NTP_OFFSET, time_offset); - audit_ntp_set_old(ad, AUDIT_NTP_FREQ, time_freq); - audit_ntp_set_old(ad, AUDIT_NTP_STATUS, time_status); + audit_ntp_set_old(ad, AUDIT_NTP_OFFSET, ntpdata->time_offset); + audit_ntp_set_old(ad, AUDIT_NTP_FREQ, ntpdata->time_freq); + audit_ntp_set_old(ad, AUDIT_NTP_STATUS, ntpdata->time_status); audit_ntp_set_old(ad, AUDIT_NTP_TAI, *time_tai); - audit_ntp_set_old(ad, AUDIT_NTP_TICK, tick_usec); + audit_ntp_set_old(ad, AUDIT_NTP_TICK, ntpdata->tick_usec); - process_adjtimex_modes(txc, time_tai); + process_adjtimex_modes(ntpdata, txc, time_tai); - audit_ntp_set_new(ad, AUDIT_NTP_OFFSET, time_offset); - audit_ntp_set_new(ad, AUDIT_NTP_FREQ, time_freq); - audit_ntp_set_new(ad, AUDIT_NTP_STATUS, time_status); + audit_ntp_set_new(ad, AUDIT_NTP_OFFSET, ntpdata->time_offset); + audit_ntp_set_new(ad, AUDIT_NTP_FREQ, ntpdata->time_freq); + audit_ntp_set_new(ad, AUDIT_NTP_STATUS, ntpdata->time_status); audit_ntp_set_new(ad, AUDIT_NTP_TAI, *time_tai); - audit_ntp_set_new(ad, AUDIT_NTP_TICK, tick_usec); + audit_ntp_set_new(ad, AUDIT_NTP_TICK, ntpdata->tick_usec); } - txc->offset = shift_right(time_offset * NTP_INTERVAL_FREQ, - NTP_SCALE_SHIFT); - if (!(time_status & STA_NANO)) - txc->offset = (u32)txc->offset / NSEC_PER_USEC; + txc->offset = shift_right(ntpdata->time_offset * NTP_INTERVAL_FREQ, NTP_SCALE_SHIFT); + if (!(ntpdata->time_status & STA_NANO)) + txc->offset = div_s64(txc->offset, NSEC_PER_USEC); } - result = time_state; /* mostly `TIME_OK' */ - /* check for errors */ - if (is_error_status(time_status)) + result = ntpdata->time_state; + if (is_error_status(ntpdata->time_status)) result = TIME_ERROR; - txc->freq = shift_right((time_freq >> PPM_SCALE_INV_SHIFT) * + txc->freq = shift_right((ntpdata->time_freq >> PPM_SCALE_INV_SHIFT) * PPM_SCALE_INV, NTP_SCALE_SHIFT); - txc->maxerror = time_maxerror; - txc->esterror = time_esterror; - txc->status = time_status; - txc->constant = time_constant; + txc->maxerror = ntpdata->time_maxerror; + txc->esterror = ntpdata->time_esterror; + txc->status = ntpdata->time_status; + txc->constant = ntpdata->time_constant; txc->precision = 1; txc->tolerance = MAXFREQ_SCALED / PPM_SCALE; - txc->tick = tick_usec; + txc->tick = ntpdata->tick_usec; txc->tai = *time_tai; - /* fill PPS status fields */ - pps_fill_timex(txc); + /* Fill PPS status fields */ + pps_fill_timex(ntpdata, txc); txc->time.tv_sec = ts->tv_sec; txc->time.tv_usec = ts->tv_nsec; - if (!(time_status & STA_NANO)) + if (!(ntpdata->time_status & STA_NANO)) txc->time.tv_usec = ts->tv_nsec / NSEC_PER_USEC; /* Handle leapsec adjustments */ - if (unlikely(ts->tv_sec >= ntp_next_leap_sec)) { - if ((time_state == TIME_INS) && (time_status & STA_INS)) { + if (unlikely(ts->tv_sec >= ntpdata->ntp_next_leap_sec)) { + if ((ntpdata->time_state == TIME_INS) && (ntpdata->time_status & STA_INS)) { result = TIME_OOP; txc->tai++; txc->time.tv_sec--; } - if ((time_state == TIME_DEL) && (time_status & STA_DEL)) { + if ((ntpdata->time_state == TIME_DEL) && (ntpdata->time_status & STA_DEL)) { result = TIME_WAIT; txc->tai--; txc->time.tv_sec++; } - if ((time_state == TIME_OOP) && - (ts->tv_sec == ntp_next_leap_sec)) { + if ((ntpdata->time_state == TIME_OOP) && (ts->tv_sec == ntpdata->ntp_next_leap_sec)) result = TIME_WAIT; - } } return result; @@ -799,17 +845,21 @@ int __do_adjtimex(struct __kernel_timex *txc, const struct timespec64 *ts, #ifdef CONFIG_NTP_PPS -/* actually struct pps_normtime is good old struct timespec, but it is +/* + * struct pps_normtime is basically a struct timespec, but it is * semantically different (and it is the reason why it was invented): * pps_normtime.nsec has a range of ( -NSEC_PER_SEC / 2, NSEC_PER_SEC / 2 ] - * while timespec.tv_nsec has a range of [0, NSEC_PER_SEC) */ + * while timespec.tv_nsec has a range of [0, NSEC_PER_SEC) + */ struct pps_normtime { s64 sec; /* seconds */ long nsec; /* nanoseconds */ }; -/* normalize the timestamp so that nsec is in the - ( -NSEC_PER_SEC / 2, NSEC_PER_SEC / 2 ] interval */ +/* + * Normalize the timestamp so that nsec is in the + * [ -NSEC_PER_SEC / 2, NSEC_PER_SEC / 2 ] interval + */ static inline struct pps_normtime pps_normalize_ts(struct timespec64 ts) { struct pps_normtime norm = { @@ -825,54 +875,57 @@ static inline struct pps_normtime pps_normalize_ts(struct timespec64 ts) return norm; } -/* get current phase correction and jitter */ -static inline long pps_phase_filter_get(long *jitter) +/* Get current phase correction and jitter */ +static inline long pps_phase_filter_get(struct ntp_data *ntpdata, long *jitter) { - *jitter = pps_tf[0] - pps_tf[1]; + *jitter = ntpdata->pps_tf[0] - ntpdata->pps_tf[1]; if (*jitter < 0) *jitter = -*jitter; /* TODO: test various filters */ - return pps_tf[0]; + return ntpdata->pps_tf[0]; } -/* add the sample to the phase filter */ -static inline void pps_phase_filter_add(long err) +/* Add the sample to the phase filter */ +static inline void pps_phase_filter_add(struct ntp_data *ntpdata, long err) { - pps_tf[2] = pps_tf[1]; - pps_tf[1] = pps_tf[0]; - pps_tf[0] = err; + ntpdata->pps_tf[2] = ntpdata->pps_tf[1]; + ntpdata->pps_tf[1] = ntpdata->pps_tf[0]; + ntpdata->pps_tf[0] = err; } -/* decrease frequency calibration interval length. - * It is halved after four consecutive unstable intervals. +/* + * Decrease frequency calibration interval length. It is halved after four + * consecutive unstable intervals. */ -static inline void pps_dec_freq_interval(void) +static inline void pps_dec_freq_interval(struct ntp_data *ntpdata) { - if (--pps_intcnt <= -PPS_INTCOUNT) { - pps_intcnt = -PPS_INTCOUNT; - if (pps_shift > PPS_INTMIN) { - pps_shift--; - pps_intcnt = 0; + if (--ntpdata->pps_intcnt <= -PPS_INTCOUNT) { + ntpdata->pps_intcnt = -PPS_INTCOUNT; + if (ntpdata->pps_shift > PPS_INTMIN) { + ntpdata->pps_shift--; + ntpdata->pps_intcnt = 0; } } } -/* increase frequency calibration interval length. - * It is doubled after four consecutive stable intervals. +/* + * Increase frequency calibration interval length. It is doubled after + * four consecutive stable intervals. */ -static inline void pps_inc_freq_interval(void) +static inline void pps_inc_freq_interval(struct ntp_data *ntpdata) { - if (++pps_intcnt >= PPS_INTCOUNT) { - pps_intcnt = PPS_INTCOUNT; - if (pps_shift < PPS_INTMAX) { - pps_shift++; - pps_intcnt = 0; + if (++ntpdata->pps_intcnt >= PPS_INTCOUNT) { + ntpdata->pps_intcnt = PPS_INTCOUNT; + if (ntpdata->pps_shift < PPS_INTMAX) { + ntpdata->pps_shift++; + ntpdata->pps_intcnt = 0; } } } -/* update clock frequency based on MONOTONIC_RAW clock PPS signal +/* + * Update clock frequency based on MONOTONIC_RAW clock PPS signal * timestamps * * At the end of the calibration interval the difference between the @@ -881,90 +934,88 @@ static inline void pps_inc_freq_interval(void) * too long, the data are discarded. * Returns the difference between old and new frequency values. */ -static long hardpps_update_freq(struct pps_normtime freq_norm) +static long hardpps_update_freq(struct ntp_data *ntpdata, struct pps_normtime freq_norm) { long delta, delta_mod; s64 ftemp; - /* check if the frequency interval was too long */ - if (freq_norm.sec > (2 << pps_shift)) { - time_status |= STA_PPSERROR; - pps_errcnt++; - pps_dec_freq_interval(); - printk_deferred(KERN_ERR - "hardpps: PPSERROR: interval too long - %lld s\n", - freq_norm.sec); + /* Check if the frequency interval was too long */ + if (freq_norm.sec > (2 << ntpdata->pps_shift)) { + ntpdata->time_status |= STA_PPSERROR; + ntpdata->pps_errcnt++; + pps_dec_freq_interval(ntpdata); + printk_deferred(KERN_ERR "hardpps: PPSERROR: interval too long - %lld s\n", + freq_norm.sec); return 0; } - /* here the raw frequency offset and wander (stability) is - * calculated. If the wander is less than the wander threshold - * the interval is increased; otherwise it is decreased. + /* + * Here the raw frequency offset and wander (stability) is + * calculated. If the wander is less than the wander threshold the + * interval is increased; otherwise it is decreased. */ ftemp = div_s64(((s64)(-freq_norm.nsec)) << NTP_SCALE_SHIFT, freq_norm.sec); - delta = shift_right(ftemp - pps_freq, NTP_SCALE_SHIFT); - pps_freq = ftemp; + delta = shift_right(ftemp - ntpdata->pps_freq, NTP_SCALE_SHIFT); + ntpdata->pps_freq = ftemp; if (delta > PPS_MAXWANDER || delta < -PPS_MAXWANDER) { - printk_deferred(KERN_WARNING - "hardpps: PPSWANDER: change=%ld\n", delta); - time_status |= STA_PPSWANDER; - pps_stbcnt++; - pps_dec_freq_interval(); - } else { /* good sample */ - pps_inc_freq_interval(); + printk_deferred(KERN_WARNING "hardpps: PPSWANDER: change=%ld\n", delta); + ntpdata->time_status |= STA_PPSWANDER; + ntpdata->pps_stbcnt++; + pps_dec_freq_interval(ntpdata); + } else { + /* Good sample */ + pps_inc_freq_interval(ntpdata); } - /* the stability metric is calculated as the average of recent - * frequency changes, but is used only for performance - * monitoring + /* + * The stability metric is calculated as the average of recent + * frequency changes, but is used only for performance monitoring */ delta_mod = delta; if (delta_mod < 0) delta_mod = -delta_mod; - pps_stabil += (div_s64(((s64)delta_mod) << - (NTP_SCALE_SHIFT - SHIFT_USEC), - NSEC_PER_USEC) - pps_stabil) >> PPS_INTMIN; - - /* if enabled, the system clock frequency is updated */ - if ((time_status & STA_PPSFREQ) != 0 && - (time_status & STA_FREQHOLD) == 0) { - time_freq = pps_freq; - ntp_update_frequency(); + ntpdata->pps_stabil += (div_s64(((s64)delta_mod) << (NTP_SCALE_SHIFT - SHIFT_USEC), + NSEC_PER_USEC) - ntpdata->pps_stabil) >> PPS_INTMIN; + + /* If enabled, the system clock frequency is updated */ + if ((ntpdata->time_status & STA_PPSFREQ) && !(ntpdata->time_status & STA_FREQHOLD)) { + ntpdata->time_freq = ntpdata->pps_freq; + ntp_update_frequency(ntpdata); } return delta; } -/* correct REALTIME clock phase error against PPS signal */ -static void hardpps_update_phase(long error) +/* Correct REALTIME clock phase error against PPS signal */ +static void hardpps_update_phase(struct ntp_data *ntpdata, long error) { long correction = -error; long jitter; - /* add the sample to the median filter */ - pps_phase_filter_add(correction); - correction = pps_phase_filter_get(&jitter); + /* Add the sample to the median filter */ + pps_phase_filter_add(ntpdata, correction); + correction = pps_phase_filter_get(ntpdata, &jitter); - /* Nominal jitter is due to PPS signal noise. If it exceeds the + /* + * Nominal jitter is due to PPS signal noise. If it exceeds the * threshold, the sample is discarded; otherwise, if so enabled, * the time offset is updated. */ - if (jitter > (pps_jitter << PPS_POPCORN)) { - printk_deferred(KERN_WARNING - "hardpps: PPSJITTER: jitter=%ld, limit=%ld\n", - jitter, (pps_jitter << PPS_POPCORN)); - time_status |= STA_PPSJITTER; - pps_jitcnt++; - } else if (time_status & STA_PPSTIME) { - /* correct the time using the phase offset */ - time_offset = div_s64(((s64)correction) << NTP_SCALE_SHIFT, - NTP_INTERVAL_FREQ); - /* cancel running adjtime() */ - time_adjust = 0; + if (jitter > (ntpdata->pps_jitter << PPS_POPCORN)) { + printk_deferred(KERN_WARNING "hardpps: PPSJITTER: jitter=%ld, limit=%ld\n", + jitter, (ntpdata->pps_jitter << PPS_POPCORN)); + ntpdata->time_status |= STA_PPSJITTER; + ntpdata->pps_jitcnt++; + } else if (ntpdata->time_status & STA_PPSTIME) { + /* Correct the time using the phase offset */ + ntpdata->time_offset = div_s64(((s64)correction) << NTP_SCALE_SHIFT, + NTP_INTERVAL_FREQ); + /* Cancel running adjtime() */ + ntpdata->time_adjust = 0; } - /* update jitter */ - pps_jitter += (jitter - pps_jitter) >> PPS_INTMIN; + /* Update jitter */ + ntpdata->pps_jitter += (jitter - ntpdata->pps_jitter) >> PPS_INTMIN; } /* @@ -982,60 +1033,62 @@ static void hardpps_update_phase(long error) void __hardpps(const struct timespec64 *phase_ts, const struct timespec64 *raw_ts) { struct pps_normtime pts_norm, freq_norm; + struct ntp_data *ntpdata = &tk_ntp_data; pts_norm = pps_normalize_ts(*phase_ts); - /* clear the error bits, they will be set again if needed */ - time_status &= ~(STA_PPSJITTER | STA_PPSWANDER | STA_PPSERROR); + /* Clear the error bits, they will be set again if needed */ + ntpdata->time_status &= ~(STA_PPSJITTER | STA_PPSWANDER | STA_PPSERROR); /* indicate signal presence */ - time_status |= STA_PPSSIGNAL; - pps_valid = PPS_VALID; + ntpdata->time_status |= STA_PPSSIGNAL; + ntpdata->pps_valid = PPS_VALID; - /* when called for the first time, - * just start the frequency interval */ - if (unlikely(pps_fbase.tv_sec == 0)) { - pps_fbase = *raw_ts; + /* + * When called for the first time, just start the frequency + * interval + */ + if (unlikely(ntpdata->pps_fbase.tv_sec == 0)) { + ntpdata->pps_fbase = *raw_ts; return; } - /* ok, now we have a base for frequency calculation */ - freq_norm = pps_normalize_ts(timespec64_sub(*raw_ts, pps_fbase)); - - /* check that the signal is in the range - * [1s - MAXFREQ us, 1s + MAXFREQ us], otherwise reject it */ - if ((freq_norm.sec == 0) || - (freq_norm.nsec > MAXFREQ * freq_norm.sec) || - (freq_norm.nsec < -MAXFREQ * freq_norm.sec)) { - time_status |= STA_PPSJITTER; - /* restart the frequency calibration interval */ - pps_fbase = *raw_ts; + /* Ok, now we have a base for frequency calculation */ + freq_norm = pps_normalize_ts(timespec64_sub(*raw_ts, ntpdata->pps_fbase)); + + /* + * Check that the signal is in the range + * [1s - MAXFREQ us, 1s + MAXFREQ us], otherwise reject it + */ + if ((freq_norm.sec == 0) || (freq_norm.nsec > MAXFREQ * freq_norm.sec) || + (freq_norm.nsec < -MAXFREQ * freq_norm.sec)) { + ntpdata->time_status |= STA_PPSJITTER; + /* Restart the frequency calibration interval */ + ntpdata->pps_fbase = *raw_ts; printk_deferred(KERN_ERR "hardpps: PPSJITTER: bad pulse\n"); return; } - /* signal is ok */ - - /* check if the current frequency interval is finished */ - if (freq_norm.sec >= (1 << pps_shift)) { - pps_calcnt++; - /* restart the frequency calibration interval */ - pps_fbase = *raw_ts; - hardpps_update_freq(freq_norm); + /* Signal is ok. Check if the current frequency interval is finished */ + if (freq_norm.sec >= (1 << ntpdata->pps_shift)) { + ntpdata->pps_calcnt++; + /* Restart the frequency calibration interval */ + ntpdata->pps_fbase = *raw_ts; + hardpps_update_freq(ntpdata, freq_norm); } - hardpps_update_phase(pts_norm.nsec); + hardpps_update_phase(ntpdata, pts_norm.nsec); } #endif /* CONFIG_NTP_PPS */ static int __init ntp_tick_adj_setup(char *str) { - int rc = kstrtos64(str, 0, &ntp_tick_adj); + int rc = kstrtos64(str, 0, &tk_ntp_data.ntp_tick_adj); if (rc) return rc; - ntp_tick_adj <<= NTP_SCALE_SHIFT; + tk_ntp_data.ntp_tick_adj <<= NTP_SCALE_SHIFT; return 1; } @@ -1044,4 +1097,5 @@ __setup("ntp_tick_adj=", ntp_tick_adj_setup); void __init ntp_init(void) { ntp_clear(); + ntp_init_cmos_sync(); } diff --git a/kernel/time/ntp_internal.h b/kernel/time/ntp_internal.h index 908ecaa65fc3..5a633dce9057 100644 --- a/kernel/time/ntp_internal.h +++ b/kernel/time/ntp_internal.h @@ -12,4 +12,11 @@ extern int __do_adjtimex(struct __kernel_timex *txc, const struct timespec64 *ts, s32 *time_tai, struct audit_ntp_data *ad); extern void __hardpps(const struct timespec64 *phase_ts, const struct timespec64 *raw_ts); + +#if defined(CONFIG_GENERIC_CMOS_UPDATE) || defined(CONFIG_RTC_SYSTOHC) +extern void ntp_notify_cmos_timer(bool offset_set); +#else +static inline void ntp_notify_cmos_timer(bool offset_set) { } +#endif + #endif /* _LINUX_NTP_INTERNAL_H */ diff --git a/kernel/time/posix-clock.c b/kernel/time/posix-clock.c index 77c0c2370b6d..1af0bb2cc45c 100644 --- a/kernel/time/posix-clock.c +++ b/kernel/time/posix-clock.c @@ -19,7 +19,8 @@ */ static struct posix_clock *get_posix_clock(struct file *fp) { - struct posix_clock *clk = fp->private_data; + struct posix_clock_context *pccontext = fp->private_data; + struct posix_clock *clk = pccontext->clk; down_read(&clk->rwsem); @@ -39,6 +40,7 @@ static void put_posix_clock(struct posix_clock *clk) static ssize_t posix_clock_read(struct file *fp, char __user *buf, size_t count, loff_t *ppos) { + struct posix_clock_context *pccontext = fp->private_data; struct posix_clock *clk = get_posix_clock(fp); int err = -EINVAL; @@ -46,7 +48,7 @@ static ssize_t posix_clock_read(struct file *fp, char __user *buf, return -ENODEV; if (clk->ops.read) - err = clk->ops.read(clk, fp->f_flags, buf, count); + err = clk->ops.read(pccontext, fp->f_flags, buf, count); put_posix_clock(clk); @@ -55,6 +57,7 @@ static ssize_t posix_clock_read(struct file *fp, char __user *buf, static __poll_t posix_clock_poll(struct file *fp, poll_table *wait) { + struct posix_clock_context *pccontext = fp->private_data; struct posix_clock *clk = get_posix_clock(fp); __poll_t result = 0; @@ -62,7 +65,7 @@ static __poll_t posix_clock_poll(struct file *fp, poll_table *wait) return EPOLLERR; if (clk->ops.poll) - result = clk->ops.poll(clk, fp, wait); + result = clk->ops.poll(pccontext, fp, wait); put_posix_clock(clk); @@ -72,6 +75,7 @@ static __poll_t posix_clock_poll(struct file *fp, poll_table *wait) static long posix_clock_ioctl(struct file *fp, unsigned int cmd, unsigned long arg) { + struct posix_clock_context *pccontext = fp->private_data; struct posix_clock *clk = get_posix_clock(fp); int err = -ENOTTY; @@ -79,7 +83,7 @@ static long posix_clock_ioctl(struct file *fp, return -ENODEV; if (clk->ops.ioctl) - err = clk->ops.ioctl(clk, cmd, arg); + err = clk->ops.ioctl(pccontext, cmd, arg); put_posix_clock(clk); @@ -90,6 +94,7 @@ static long posix_clock_ioctl(struct file *fp, static long posix_clock_compat_ioctl(struct file *fp, unsigned int cmd, unsigned long arg) { + struct posix_clock_context *pccontext = fp->private_data; struct posix_clock *clk = get_posix_clock(fp); int err = -ENOTTY; @@ -97,7 +102,7 @@ static long posix_clock_compat_ioctl(struct file *fp, return -ENODEV; if (clk->ops.ioctl) - err = clk->ops.ioctl(clk, cmd, arg); + err = clk->ops.ioctl(pccontext, cmd, arg); put_posix_clock(clk); @@ -110,6 +115,7 @@ static int posix_clock_open(struct inode *inode, struct file *fp) int err; struct posix_clock *clk = container_of(inode->i_cdev, struct posix_clock, cdev); + struct posix_clock_context *pccontext; down_read(&clk->rwsem); @@ -117,15 +123,23 @@ static int posix_clock_open(struct inode *inode, struct file *fp) err = -ENODEV; goto out; } - if (clk->ops.open) - err = clk->ops.open(clk, fp->f_mode); - else - err = 0; - - if (!err) { - get_device(clk->dev); - fp->private_data = clk; + pccontext = kzalloc(sizeof(*pccontext), GFP_KERNEL); + if (!pccontext) { + err = -ENOMEM; + goto out; } + pccontext->clk = clk; + if (clk->ops.open) { + err = clk->ops.open(pccontext, fp->f_mode); + if (err) { + kfree(pccontext); + goto out; + } + } + + fp->private_data = pccontext; + get_device(clk->dev); + err = 0; out: up_read(&clk->rwsem); return err; @@ -133,14 +147,20 @@ out: static int posix_clock_release(struct inode *inode, struct file *fp) { - struct posix_clock *clk = fp->private_data; + struct posix_clock_context *pccontext = fp->private_data; + struct posix_clock *clk; int err = 0; + if (!pccontext) + return -ENODEV; + clk = pccontext->clk; + if (clk->ops.release) - err = clk->ops.release(clk); + err = clk->ops.release(pccontext); put_device(clk->dev); + kfree(pccontext); fp->private_data = NULL; return err; @@ -148,7 +168,6 @@ static int posix_clock_release(struct inode *inode, struct file *fp) static const struct file_operations posix_clock_file_operations = { .owner = THIS_MODULE, - .llseek = no_llseek, .read = posix_clock_read, .poll = posix_clock_poll, .unlocked_ioctl = posix_clock_ioctl, @@ -290,6 +309,9 @@ static int pc_clock_settime(clockid_t id, const struct timespec64 *ts) struct posix_clock_desc cd; int err; + if (!timespec64_valid_strict(ts)) + return -EINVAL; + err = get_clock_desc(id, &cd); if (err) return err; diff --git a/kernel/time/posix-cpu-timers.c b/kernel/time/posix-cpu-timers.c index 165117996ea0..50e8d04ab661 100644 --- a/kernel/time/posix-cpu-timers.c +++ b/kernel/time/posix-cpu-timers.c @@ -15,6 +15,7 @@ #include <linux/workqueue.h> #include <linux/compat.h> #include <linux/sched/deadline.h> +#include <linux/task_work.h> #include "posix-timers.h" @@ -34,14 +35,20 @@ void posix_cputimers_group_init(struct posix_cputimers *pct, u64 cpu_limit) * tsk->signal->posix_cputimers.bases[clock].nextevt expiration cache if * necessary. Needs siglock protection since other code may update the * expiration cache as well. + * + * Returns 0 on success, -ESRCH on failure. Can fail if the task is exiting and + * we cannot lock_task_sighand. Cannot fail if task is current. */ -void update_rlimit_cpu(struct task_struct *task, unsigned long rlim_new) +int update_rlimit_cpu(struct task_struct *task, unsigned long rlim_new) { u64 nsecs = rlim_new * NSEC_PER_SEC; + unsigned long irq_fl; - spin_lock_irq(&task->sighand->siglock); + if (!lock_task_sighand(task, &irq_fl)) + return -ESRCH; set_process_cpu_timer(task, CPUCLOCK_PROF, &nsecs, NULL); - spin_unlock_irq(&task->sighand->siglock); + unlock_task_sighand(task, &irq_fl); + return 0; } /* @@ -236,13 +243,12 @@ static void proc_sample_cputime_atomic(struct task_cputime_atomic *at, */ static inline void __update_gt_cputime(atomic64_t *cputime, u64 sum_cputime) { - u64 curr_cputime; -retry: - curr_cputime = atomic64_read(cputime); - if (sum_cputime > curr_cputime) { - if (atomic64_cmpxchg(cputime, curr_cputime, sum_cputime) != curr_cputime) - goto retry; - } + u64 curr_cputime = atomic64_read(cputime); + + do { + if (sum_cputime <= curr_cputime) + return; + } while (!atomic64_try_cmpxchg(cputime, &curr_cputime, sum_cputime)); } static void update_gt_cputime(struct task_cputime_atomic *cputime_atomic, @@ -279,7 +285,7 @@ void thread_group_sample_cputime(struct task_struct *tsk, u64 *samples) * @tsk: Task for which cputime needs to be started * @samples: Storage for time samples * - * The thread group cputime accouting is avoided when there are no posix + * The thread group cputime accounting is avoided when there are no posix * CPU timers armed. Before starting a timer it's required to check whether * the time accounting is active. If not, a full update of the atomic * accounting store needs to be done and the accounting enabled. @@ -291,6 +297,8 @@ static void thread_group_start_cputime(struct task_struct *tsk, u64 *samples) struct thread_group_cputimer *cputimer = &tsk->signal->cputimer; struct posix_cputimers *pct = &tsk->signal->posix_cputimers; + lockdep_assert_task_sighand_held(tsk); + /* Check if cputimer isn't running. This is accessed without locking. */ if (!READ_ONCE(pct->timers_active)) { struct task_cputime sum; @@ -377,6 +385,7 @@ static int posix_cpu_clock_get(const clockid_t clock, struct timespec64 *tp) */ static int posix_cpu_timer_create(struct k_itimer *new_timer) { + static struct lock_class_key posix_cpu_timers_key; struct pid *pid; rcu_read_lock(); @@ -386,6 +395,17 @@ static int posix_cpu_timer_create(struct k_itimer *new_timer) return -EINVAL; } + /* + * If posix timer expiry is handled in task work context then + * timer::it_lock can be taken without disabling interrupts as all + * other locking happens in task context. This requires a separate + * lock class key otherwise regular posix timer expiry would record + * the lock class being taken in interrupt context and generate a + * false positive warning. + */ + if (IS_ENABLED(CONFIG_POSIX_CPU_TIMERS_TASK_WORK)) + lockdep_set_class(&new_timer->it_lock, &posix_cpu_timers_key); + new_timer->kclock = &clock_posix_cpu; timerqueue_init(&new_timer->it.cpu.node); new_timer->it.cpu.pid = get_pid(pid); @@ -393,6 +413,55 @@ static int posix_cpu_timer_create(struct k_itimer *new_timer) return 0; } +static struct posix_cputimer_base *timer_base(struct k_itimer *timer, + struct task_struct *tsk) +{ + int clkidx = CPUCLOCK_WHICH(timer->it_clock); + + if (CPUCLOCK_PERTHREAD(timer->it_clock)) + return tsk->posix_cputimers.bases + clkidx; + else + return tsk->signal->posix_cputimers.bases + clkidx; +} + +/* + * Force recalculating the base earliest expiration on the next tick. + * This will also re-evaluate the need to keep around the process wide + * cputime counter and tick dependency and eventually shut these down + * if necessary. + */ +static void trigger_base_recalc_expires(struct k_itimer *timer, + struct task_struct *tsk) +{ + struct posix_cputimer_base *base = timer_base(timer, tsk); + + base->nextevt = 0; +} + +/* + * Dequeue the timer and reset the base if it was its earliest expiration. + * It makes sure the next tick recalculates the base next expiration so we + * don't keep the costly process wide cputime counter around for a random + * amount of time, along with the tick dependency. + * + * If another timer gets queued between this and the next tick, its + * expiration will update the base next event if necessary on the next + * tick. + */ +static void disarm_timer(struct k_itimer *timer, struct task_struct *p) +{ + struct cpu_timer *ctmr = &timer->it.cpu; + struct posix_cputimer_base *base; + + if (!cpu_timer_dequeue(ctmr)) + return; + + base = timer_base(timer, p); + if (cpu_timer_getexpires(ctmr) == base->nextevt) + trigger_base_recalc_expires(timer, p); +} + + /* * Clean up a CPU-clock timer that is about to be destroyed. * This is called from timer deletion with the timer already locked. @@ -424,19 +493,28 @@ static int posix_cpu_timer_del(struct k_itimer *timer) */ WARN_ON_ONCE(ctmr->head || timerqueue_node_queued(&ctmr->node)); } else { - if (timer->it.cpu.firing) + if (timer->it.cpu.firing) { + /* + * Prevent signal delivery. The timer cannot be dequeued + * because it is on the firing list which is not protected + * by sighand->lock. The delivery path is waiting for + * the timer lock. So go back, unlock and retry. + */ + timer->it.cpu.firing = false; ret = TIMER_RETRY; - else - cpu_timer_dequeue(ctmr); - + } else { + disarm_timer(timer, p); + } unlock_task_sighand(p, &flags); } out: rcu_read_unlock(); - if (!ret) - put_pid(ctmr->pid); + if (!ret) { + put_pid(ctmr->pid); + timer->it_status = POSIX_TIMER_DISARMED; + } return ret; } @@ -486,16 +564,11 @@ void posix_cpu_timers_exit_group(struct task_struct *tsk) */ static void arm_timer(struct k_itimer *timer, struct task_struct *p) { - int clkidx = CPUCLOCK_WHICH(timer->it_clock); + struct posix_cputimer_base *base = timer_base(timer, p); struct cpu_timer *ctmr = &timer->it.cpu; u64 newexp = cpu_timer_getexpires(ctmr); - struct posix_cputimer_base *base; - - if (CPUCLOCK_PERTHREAD(timer->it_clock)) - base = p->posix_cputimers.bases + clkidx; - else - base = p->signal->posix_cputimers.bases + clkidx; + timer->it_status = POSIX_TIMER_ARMED; if (!cpu_timer_enqueue(&base->tqhead, ctmr)) return; @@ -511,7 +584,7 @@ static void arm_timer(struct k_itimer *timer, struct task_struct *p) if (CPUCLOCK_PERTHREAD(timer->it_clock)) tick_dep_set_task(p, TICK_DEP_BIT_POSIX_TIMER); else - tick_dep_set_signal(p->signal, TICK_DEP_BIT_POSIX_TIMER); + tick_dep_set_signal(p, TICK_DEP_BIT_POSIX_TIMER); } /* @@ -521,36 +594,25 @@ static void cpu_timer_fire(struct k_itimer *timer) { struct cpu_timer *ctmr = &timer->it.cpu; - if ((timer->it_sigev_notify & ~SIGEV_THREAD_ID) == SIGEV_NONE) { - /* - * User don't want any signal. - */ - cpu_timer_setexpires(ctmr, 0); - } else if (unlikely(timer->sigq == NULL)) { + timer->it_status = POSIX_TIMER_DISARMED; + + if (unlikely(ctmr->nanosleep)) { /* * This a special case for clock_nanosleep, * not a normal timer from sys_timer_create. */ wake_up_process(timer->it_process); cpu_timer_setexpires(ctmr, 0); - } else if (!timer->it_interval) { - /* - * One-shot timer. Clear it as soon as it's fired. - */ - posix_timer_event(timer, 0); - cpu_timer_setexpires(ctmr, 0); - } else if (posix_timer_event(timer, ++timer->it_requeue_pending)) { - /* - * The signal did not get queued because the signal - * was ignored, so we won't get any callback to - * reload the timer. But we need to keep it - * ticking in case the signal is deliverable next time. - */ - posix_cpu_timer_rearm(timer); - ++timer->it_requeue_pending; + } else { + posix_timer_queue_signal(timer); + /* Disable oneshot timers */ + if (!timer->it_interval) + cpu_timer_setexpires(ctmr, 0); } } +static void __posix_cpu_timer_get(struct k_itimer *timer, struct itimerspec64 *itp, u64 now); + /* * Guts of sys_timer_settime for CPU timers. * This is called with the timer locked and interrupts disabled. @@ -560,9 +622,10 @@ static void cpu_timer_fire(struct k_itimer *timer) static int posix_cpu_timer_set(struct k_itimer *timer, int timer_flags, struct itimerspec64 *new, struct itimerspec64 *old) { + bool sigev_none = timer->it_sigev_notify == SIGEV_NONE; clockid_t clkid = CPUCLOCK_WHICH(timer->it_clock); - u64 old_expires, new_expires, old_incr, val; struct cpu_timer *ctmr = &timer->it.cpu; + u64 old_expires, new_expires, now; struct sighand_struct *sighand; struct task_struct *p; unsigned long flags; @@ -599,155 +662,136 @@ static int posix_cpu_timer_set(struct k_itimer *timer, int timer_flags, return -ESRCH; } - /* - * Disarm any old timer after extracting its expiry time. - */ - old_incr = timer->it_interval; + /* Retrieve the current expiry time before disarming the timer */ old_expires = cpu_timer_getexpires(ctmr); if (unlikely(timer->it.cpu.firing)) { - timer->it.cpu.firing = -1; + /* + * Prevent signal delivery. The timer cannot be dequeued + * because it is on the firing list which is not protected + * by sighand->lock. The delivery path is waiting for + * the timer lock. So go back, unlock and retry. + */ + timer->it.cpu.firing = false; ret = TIMER_RETRY; } else { cpu_timer_dequeue(ctmr); + timer->it_status = POSIX_TIMER_DISARMED; } /* - * We need to sample the current value to convert the new - * value from to relative and absolute, and to convert the - * old value from absolute to relative. To set a process - * timer, we need a sample to balance the thread expiry - * times (in arm_timer). With an absolute time, we must - * check if it's already passed. In short, we need a sample. + * Sample the current clock for saving the previous setting + * and for rearming the timer. */ if (CPUCLOCK_PERTHREAD(timer->it_clock)) - val = cpu_clock_sample(clkid, p); + now = cpu_clock_sample(clkid, p); else - val = cpu_clock_sample_group(clkid, p, true); + now = cpu_clock_sample_group(clkid, p, !sigev_none); + /* Retrieve the previous expiry value if requested. */ if (old) { - if (old_expires == 0) { - old->it_value.tv_sec = 0; - old->it_value.tv_nsec = 0; - } else { - /* - * Update the timer in case it has overrun already. - * If it has, we'll report it as having overrun and - * with the next reloaded timer already ticking, - * though we are swallowing that pending - * notification here to install the new setting. - */ - u64 exp = bump_cpu_timer(timer, val); - - if (val < exp) { - old_expires = exp - val; - old->it_value = ns_to_timespec64(old_expires); - } else { - old->it_value.tv_nsec = 1; - old->it_value.tv_sec = 0; - } - } + old->it_value = (struct timespec64){ }; + if (old_expires) + __posix_cpu_timer_get(timer, old, now); } + /* Retry if the timer expiry is running concurrently */ if (unlikely(ret)) { - /* - * We are colliding with the timer actually firing. - * Punt after filling in the timer's old value, and - * disable this firing since we are already reporting - * it as an overrun (thanks to bump_cpu_timer above). - */ unlock_task_sighand(p, &flags); goto out; } - if (new_expires != 0 && !(timer_flags & TIMER_ABSTIME)) { - new_expires += val; - } + /* Convert relative expiry time to absolute */ + if (new_expires && !(timer_flags & TIMER_ABSTIME)) + new_expires += now; + + /* Set the new expiry time (might be 0) */ + cpu_timer_setexpires(ctmr, new_expires); /* - * Install the new expiry time (or zero). - * For a timer with no notification action, we don't actually - * arm the timer (we'll just fake it for timer_gettime). + * Arm the timer if it is not disabled, the new expiry value has + * not yet expired and the timer requires signal delivery. + * SIGEV_NONE timers are never armed. In case the timer is not + * armed, enforce the reevaluation of the timer base so that the + * process wide cputime counter can be disabled eventually. */ - cpu_timer_setexpires(ctmr, new_expires); - if (new_expires != 0 && val < new_expires) { - arm_timer(timer, p); + if (likely(!sigev_none)) { + if (new_expires && now < new_expires) + arm_timer(timer, p); + else + trigger_base_recalc_expires(timer, p); } unlock_task_sighand(p, &flags); - /* - * Install the new reload setting, and - * set up the signal and overrun bookkeeping. - */ - timer->it_interval = timespec64_to_ktime(new->it_interval); + + posix_timer_set_common(timer, new); /* - * This acts as a modification timestamp for the timer, - * so any automatic reload attempt will punt on seeing - * that we have reset the timer manually. + * If the new expiry time was already in the past the timer was not + * queued. Fire it immediately even if the thread never runs to + * accumulate more time on this clock. */ - timer->it_requeue_pending = (timer->it_requeue_pending + 2) & - ~REQUEUE_PENDING; - timer->it_overrun_last = 0; - timer->it_overrun = -1; - - if (new_expires != 0 && !(val < new_expires)) { - /* - * The designated time already passed, so we notify - * immediately, even if the thread never runs to - * accumulate more time on this clock. - */ + if (!sigev_none && new_expires && now >= new_expires) cpu_timer_fire(timer); - } - - ret = 0; - out: +out: rcu_read_unlock(); - if (old) - old->it_interval = ns_to_timespec64(old_incr); - return ret; } -static void posix_cpu_timer_get(struct k_itimer *timer, struct itimerspec64 *itp) +static void __posix_cpu_timer_get(struct k_itimer *timer, struct itimerspec64 *itp, u64 now) { - clockid_t clkid = CPUCLOCK_WHICH(timer->it_clock); - struct cpu_timer *ctmr = &timer->it.cpu; - u64 now, expires = cpu_timer_getexpires(ctmr); - struct task_struct *p; - - rcu_read_lock(); - p = cpu_timer_task_rcu(timer); - if (!p) - goto out; + bool sigev_none = timer->it_sigev_notify == SIGEV_NONE; + u64 expires, iv = timer->it_interval; /* - * Easy part: convert the reload time. + * Make sure that interval timers are moved forward for the + * following cases: + * - SIGEV_NONE timers which are never armed + * - Timers which expired, but the signal has not yet been + * delivered */ - itp->it_interval = ktime_to_timespec64(timer->it_interval); - - if (!expires) - goto out; + if (iv && timer->it_status != POSIX_TIMER_ARMED) + expires = bump_cpu_timer(timer, now); + else + expires = cpu_timer_getexpires(&timer->it.cpu); /* - * Sample the clock to take the difference with the expiry time. + * Expired interval timers cannot have a remaining time <= 0. + * The kernel has to move them forward so that the next + * timer expiry is > @now. */ - if (CPUCLOCK_PERTHREAD(timer->it_clock)) - now = cpu_clock_sample(clkid, p); - else - now = cpu_clock_sample_group(clkid, p, false); - if (now < expires) { itp->it_value = ns_to_timespec64(expires - now); } else { /* - * The timer should have expired already, but the firing - * hasn't taken place yet. Say it's just about to expire. + * A single shot SIGEV_NONE timer must return 0, when it is + * expired! Timers which have a real signal delivery mode + * must return a remaining time greater than 0 because the + * signal has not yet been delivered. */ - itp->it_value.tv_nsec = 1; - itp->it_value.tv_sec = 0; + if (!sigev_none) + itp->it_value.tv_nsec = 1; + } +} + +static void posix_cpu_timer_get(struct k_itimer *timer, struct itimerspec64 *itp) +{ + clockid_t clkid = CPUCLOCK_WHICH(timer->it_clock); + struct task_struct *p; + u64 now; + + rcu_read_lock(); + p = cpu_timer_task_rcu(timer); + if (p && cpu_timer_getexpires(&timer->it.cpu)) { + itp->it_interval = ktime_to_timespec64(timer->it_interval); + + if (CPUCLOCK_PERTHREAD(timer->it_clock)) + now = cpu_clock_sample(clkid, p); + else + now = cpu_clock_sample_group(clkid, p, false); + + __posix_cpu_timer_get(timer, itp, now); } -out: rcu_read_unlock(); } @@ -769,7 +813,9 @@ static u64 collect_timerqueue(struct timerqueue_head *head, if (++i == MAX_COLLECTED || now < expires) return expires; - ctmr->firing = 1; + ctmr->firing = true; + /* See posix_cpu_timer_wait_running() */ + rcu_assign_pointer(ctmr->handling, current); cpu_timer_dequeue(ctmr); list_add_tail(&ctmr->elist, firing); } @@ -793,7 +839,7 @@ static inline void check_dl_overrun(struct task_struct *tsk) { if (tsk->dl.dl_overrun) { tsk->dl.dl_overrun = 0; - __group_send_sig_info(SIGXCPU, SEND_SIG_PRIV, tsk); + send_signal_locked(SIGXCPU, SEND_SIG_PRIV, tsk, PIDTYPE_TGID); } } @@ -807,7 +853,7 @@ static bool check_rlimit(u64 time, u64 limit, int signo, bool rt, bool hard) rt ? "RT" : "CPU", hard ? "hard" : "soft", current->comm, task_pid_nr(current)); } - __group_send_sig_info(signo, SEND_SIG_PRIV, current); + send_signal_locked(signo, SEND_SIG_PRIV, current, PIDTYPE_TGID); return true; } @@ -881,7 +927,7 @@ static void check_cpu_itimer(struct task_struct *tsk, struct cpu_itimer *it, trace_itimer_expire(signo == SIGPROF ? ITIMER_PROF : ITIMER_VIRTUAL, task_tgid(tsk), cur_time); - __group_send_sig_info(signo, SEND_SIG_PRIV, tsk); + send_signal_locked(signo, SEND_SIG_PRIV, tsk, PIDTYPE_TGID); } if (it->expires && it->expires < *expires) @@ -979,6 +1025,11 @@ static void posix_cpu_timer_rearm(struct k_itimer *timer) if (!p) goto out; + /* Protect timer list r/w in arm_timer() */ + sighand = lock_task_sighand(p, &flags); + if (unlikely(sighand == NULL)) + goto out; + /* * Fetch the current sample and update the timer's expiry time. */ @@ -989,11 +1040,6 @@ static void posix_cpu_timer_rearm(struct k_itimer *timer) bump_cpu_timer(timer, now); - /* Protect timer list r/w in arm_timer() */ - sighand = lock_task_sighand(p, &flags); - if (unlikely(sighand == NULL)) - goto out; - /* * Now re-arm for the new expiry time. */ @@ -1080,43 +1126,233 @@ static inline bool fastpath_timer_check(struct task_struct *tsk) return false; } +static void handle_posix_cpu_timers(struct task_struct *tsk); + +#ifdef CONFIG_POSIX_CPU_TIMERS_TASK_WORK +static void posix_cpu_timers_work(struct callback_head *work) +{ + struct posix_cputimers_work *cw = container_of(work, typeof(*cw), work); + + mutex_lock(&cw->mutex); + handle_posix_cpu_timers(current); + mutex_unlock(&cw->mutex); +} + /* - * This is called from the timer interrupt handler. The irq handler has - * already updated our counts. We need to check if any timers fire now. - * Interrupts are disabled. + * Invoked from the posix-timer core when a cancel operation failed because + * the timer is marked firing. The caller holds rcu_read_lock(), which + * protects the timer and the task which is expiring it from being freed. */ -void run_posix_cpu_timers(void) +static void posix_cpu_timer_wait_running(struct k_itimer *timr) { - struct task_struct *tsk = current; - struct k_itimer *timer, *next; - unsigned long flags; - LIST_HEAD(firing); + struct task_struct *tsk = rcu_dereference(timr->it.cpu.handling); - lockdep_assert_irqs_disabled(); + /* Has the handling task completed expiry already? */ + if (!tsk) + return; + /* Ensure that the task cannot go away */ + get_task_struct(tsk); + /* Now drop the RCU protection so the mutex can be locked */ + rcu_read_unlock(); + /* Wait on the expiry mutex */ + mutex_lock(&tsk->posix_cputimers_work.mutex); + /* Release it immediately again. */ + mutex_unlock(&tsk->posix_cputimers_work.mutex); + /* Drop the task reference. */ + put_task_struct(tsk); + /* Relock RCU so the callsite is balanced */ + rcu_read_lock(); +} + +static void posix_cpu_timer_wait_running_nsleep(struct k_itimer *timr) +{ + /* Ensure that timr->it.cpu.handling task cannot go away */ + rcu_read_lock(); + spin_unlock_irq(&timr->it_lock); + posix_cpu_timer_wait_running(timr); + rcu_read_unlock(); + /* @timr is on stack and is valid */ + spin_lock_irq(&timr->it_lock); +} + +/* + * Clear existing posix CPU timers task work. + */ +void clear_posix_cputimers_work(struct task_struct *p) +{ /* - * The fast path checks that there are no expired thread or thread - * group timers. If that's so, just return. + * A copied work entry from the old task is not meaningful, clear it. + * N.B. init_task_work will not do this. */ - if (!fastpath_timer_check(tsk)) - return; + memset(&p->posix_cputimers_work.work, 0, + sizeof(p->posix_cputimers_work.work)); + init_task_work(&p->posix_cputimers_work.work, + posix_cpu_timers_work); + mutex_init(&p->posix_cputimers_work.mutex); + p->posix_cputimers_work.scheduled = false; +} - lockdep_posixtimer_enter(); - if (!lock_task_sighand(tsk, &flags)) { - lockdep_posixtimer_exit(); +/* + * Initialize posix CPU timers task work in init task. Out of line to + * keep the callback static and to avoid header recursion hell. + */ +void __init posix_cputimers_init_work(void) +{ + clear_posix_cputimers_work(current); +} + +/* + * Note: All operations on tsk->posix_cputimer_work.scheduled happen either + * in hard interrupt context or in task context with interrupts + * disabled. Aside of that the writer/reader interaction is always in the + * context of the current task, which means they are strict per CPU. + */ +static inline bool posix_cpu_timers_work_scheduled(struct task_struct *tsk) +{ + return tsk->posix_cputimers_work.scheduled; +} + +static inline void __run_posix_cpu_timers(struct task_struct *tsk) +{ + if (WARN_ON_ONCE(tsk->posix_cputimers_work.scheduled)) return; + + /* Schedule task work to actually expire the timers */ + tsk->posix_cputimers_work.scheduled = true; + task_work_add(tsk, &tsk->posix_cputimers_work.work, TWA_RESUME); +} + +static inline bool posix_cpu_timers_enable_work(struct task_struct *tsk, + unsigned long start) +{ + bool ret = true; + + /* + * On !RT kernels interrupts are disabled while collecting expired + * timers, so no tick can happen and the fast path check can be + * reenabled without further checks. + */ + if (!IS_ENABLED(CONFIG_PREEMPT_RT)) { + tsk->posix_cputimers_work.scheduled = false; + return true; } + /* - * Here we take off tsk->signal->cpu_timers[N] and - * tsk->cpu_timers[N] all the timers that are firing, and - * put them on the firing list. + * On RT enabled kernels ticks can happen while the expired timers + * are collected under sighand lock. But any tick which observes + * the CPUTIMERS_WORK_SCHEDULED bit set, does not run the fastpath + * checks. So reenabling the tick work has do be done carefully: + * + * Disable interrupts and run the fast path check if jiffies have + * advanced since the collecting of expired timers started. If + * jiffies have not advanced or the fast path check did not find + * newly expired timers, reenable the fast path check in the timer + * interrupt. If there are newly expired timers, return false and + * let the collection loop repeat. */ - check_thread_timers(tsk, &firing); + local_irq_disable(); + if (start != jiffies && fastpath_timer_check(tsk)) + ret = false; + else + tsk->posix_cputimers_work.scheduled = false; + local_irq_enable(); + + return ret; +} +#else /* CONFIG_POSIX_CPU_TIMERS_TASK_WORK */ +static inline void __run_posix_cpu_timers(struct task_struct *tsk) +{ + lockdep_posixtimer_enter(); + handle_posix_cpu_timers(tsk); + lockdep_posixtimer_exit(); +} + +static void posix_cpu_timer_wait_running(struct k_itimer *timr) +{ + cpu_relax(); +} - check_process_timers(tsk, &firing); +static void posix_cpu_timer_wait_running_nsleep(struct k_itimer *timr) +{ + spin_unlock_irq(&timr->it_lock); + cpu_relax(); + spin_lock_irq(&timr->it_lock); +} + +static inline bool posix_cpu_timers_work_scheduled(struct task_struct *tsk) +{ + return false; +} + +static inline bool posix_cpu_timers_enable_work(struct task_struct *tsk, + unsigned long start) +{ + return true; +} +#endif /* CONFIG_POSIX_CPU_TIMERS_TASK_WORK */ + +static void handle_posix_cpu_timers(struct task_struct *tsk) +{ + struct k_itimer *timer, *next; + unsigned long flags, start; + LIST_HEAD(firing); + + if (!lock_task_sighand(tsk, &flags)) + return; + + do { + /* + * On RT locking sighand lock does not disable interrupts, + * so this needs to be careful vs. ticks. Store the current + * jiffies value. + */ + start = READ_ONCE(jiffies); + barrier(); + + /* + * Here we take off tsk->signal->cpu_timers[N] and + * tsk->cpu_timers[N] all the timers that are firing, and + * put them on the firing list. + */ + check_thread_timers(tsk, &firing); + + check_process_timers(tsk, &firing); + + /* + * The above timer checks have updated the expiry cache and + * because nothing can have queued or modified timers after + * sighand lock was taken above it is guaranteed to be + * consistent. So the next timer interrupt fastpath check + * will find valid data. + * + * If timer expiry runs in the timer interrupt context then + * the loop is not relevant as timers will be directly + * expired in interrupt context. The stub function below + * returns always true which allows the compiler to + * optimize the loop out. + * + * If timer expiry is deferred to task work context then + * the following rules apply: + * + * - On !RT kernels no tick can have happened on this CPU + * after sighand lock was acquired because interrupts are + * disabled. So reenabling task work before dropping + * sighand lock and reenabling interrupts is race free. + * + * - On RT kernels ticks might have happened but the tick + * work ignored posix CPU timer handling because the + * CPUTIMERS_WORK_SCHEDULED bit is set. Reenabling work + * must be done very carefully including a check whether + * ticks have happened since the start of the timer + * expiry checks. posix_cpu_timers_enable_work() takes + * care of that and eventually lets the expiry checks + * run again. + */ + } while (!posix_cpu_timers_enable_work(tsk, start)); /* - * We must release these locks before taking any timer's lock. + * We must release sighand lock before taking any timer's lock. * There is a potential race with timer deletion here, as the * siglock now protects our private firing list. We have set * the firing flag in each timer, so that a deletion attempt @@ -1132,22 +1368,58 @@ void run_posix_cpu_timers(void) * timer call will interfere. */ list_for_each_entry_safe(timer, next, &firing, it.cpu.elist) { - int cpu_firing; + bool cpu_firing; + /* + * spin_lock() is sufficient here even independent of the + * expiry context. If expiry happens in hard interrupt + * context it's obvious. For task work context it's safe + * because all other operations on timer::it_lock happen in + * task context (syscall or exit). + */ spin_lock(&timer->it_lock); list_del_init(&timer->it.cpu.elist); cpu_firing = timer->it.cpu.firing; - timer->it.cpu.firing = 0; + timer->it.cpu.firing = false; /* - * The firing flag is -1 if we collided with a reset - * of the timer, which already reported this - * almost-firing as an overrun. So don't generate an event. + * If the firing flag is cleared then this raced with a + * timer rearm/delete operation. So don't generate an + * event. */ - if (likely(cpu_firing >= 0)) + if (likely(cpu_firing)) cpu_timer_fire(timer); + /* See posix_cpu_timer_wait_running() */ + rcu_assign_pointer(timer->it.cpu.handling, NULL); spin_unlock(&timer->it_lock); } - lockdep_posixtimer_exit(); +} + +/* + * This is called from the timer interrupt handler. The irq handler has + * already updated our counts. We need to check if any timers fire now. + * Interrupts are disabled. + */ +void run_posix_cpu_timers(void) +{ + struct task_struct *tsk = current; + + lockdep_assert_irqs_disabled(); + + /* + * If the actual expiry is deferred to task work context and the + * work is already scheduled there is no point to do anything here. + */ + if (posix_cpu_timers_work_scheduled(tsk)) + return; + + /* + * The fast path checks that there are no expired thread or thread + * group timers. If that's so, just return. + */ + if (!fastpath_timer_check(tsk)) + return; + + __run_posix_cpu_timers(tsk); } /* @@ -1180,9 +1452,8 @@ void set_process_cpu_timer(struct task_struct *tsk, unsigned int clkid, } } - if (!*newval) - return; - *newval += now; + if (*newval) + *newval += now; } /* @@ -1192,7 +1463,7 @@ void set_process_cpu_timer(struct task_struct *tsk, unsigned int clkid, if (*newval < *nextevt) *nextevt = *newval; - tick_dep_set_signal(tsk->signal, TICK_DEP_BIT_POSIX_TIMER); + tick_dep_set_signal(tsk, TICK_DEP_BIT_POSIX_TIMER); } static int do_cpu_nanosleep(const clockid_t which_clock, int flags, @@ -1212,6 +1483,7 @@ static int do_cpu_nanosleep(const clockid_t which_clock, int flags, timer.it_overrun = -1; error = posix_cpu_timer_create(&timer); timer.it_process = current; + timer.it.cpu.nanosleep = true; if (!error) { static struct itimerspec64 zero_it; @@ -1253,23 +1525,16 @@ static int do_cpu_nanosleep(const clockid_t which_clock, int flags, expires = cpu_timer_getexpires(&timer.it.cpu); error = posix_cpu_timer_set(&timer, 0, &zero_it, &it); if (!error) { - /* - * Timer is now unarmed, deletion can not fail. - */ + /* Timer is now unarmed, deletion can not fail. */ posix_cpu_timer_del(&timer); + } else { + while (error == TIMER_RETRY) { + posix_cpu_timer_wait_running_nsleep(&timer); + error = posix_cpu_timer_del(&timer); + } } - spin_unlock_irq(&timer.it_lock); - while (error == TIMER_RETRY) { - /* - * We need to handle case when timer was or is in the - * middle of firing. In other cases we already freed - * resources. - */ - spin_lock_irq(&timer.it_lock); - error = posix_cpu_timer_del(&timer); - spin_unlock_irq(&timer.it_lock); - } + spin_unlock_irq(&timer.it_lock); if ((it.it_value.tv_sec | it.it_value.tv_nsec) == 0) { /* @@ -1314,8 +1579,8 @@ static int posix_cpu_nsleep(const clockid_t which_clock, int flags, if (flags & TIMER_ABSTIME) return -ERESTARTNOHAND; - restart_block->fn = posix_cpu_nsleep_restart; restart_block->nanosleep.clockid = which_clock; + set_restart_fn(restart_block, posix_cpu_nsleep_restart); } return error; } @@ -1379,6 +1644,7 @@ const struct k_clock clock_posix_cpu = { .timer_del = posix_cpu_timer_del, .timer_get = posix_cpu_timer_get, .timer_rearm = posix_cpu_timer_rearm, + .timer_wait_running = posix_cpu_timer_wait_running, }; const struct k_clock clock_process = { diff --git a/kernel/time/posix-stubs.c b/kernel/time/posix-stubs.c index fcb3b21d8bdc..9b6fcb8d85e7 100644 --- a/kernel/time/posix-stubs.c +++ b/kernel/time/posix-stubs.c @@ -17,40 +17,6 @@ #include <linux/time_namespace.h> #include <linux/compat.h> -#ifdef CONFIG_ARCH_HAS_SYSCALL_WRAPPER -/* Architectures may override SYS_NI and COMPAT_SYS_NI */ -#include <asm/syscall_wrapper.h> -#endif - -asmlinkage long sys_ni_posix_timers(void) -{ - pr_err_once("process %d (%s) attempted a POSIX timer syscall " - "while CONFIG_POSIX_TIMERS is not set\n", - current->pid, current->comm); - return -ENOSYS; -} - -#ifndef SYS_NI -#define SYS_NI(name) SYSCALL_ALIAS(sys_##name, sys_ni_posix_timers) -#endif - -#ifndef COMPAT_SYS_NI -#define COMPAT_SYS_NI(name) SYSCALL_ALIAS(compat_sys_##name, sys_ni_posix_timers) -#endif - -SYS_NI(timer_create); -SYS_NI(timer_gettime); -SYS_NI(timer_getoverrun); -SYS_NI(timer_settime); -SYS_NI(timer_delete); -SYS_NI(clock_adjtime); -SYS_NI(getitimer); -SYS_NI(setitimer); -SYS_NI(clock_adjtime32); -#ifdef __ARCH_WANT_SYS_ALARM -SYS_NI(alarm); -#endif - /* * We preserve minimal support for CLOCK_REALTIME and CLOCK_MONOTONIC * as it is easy to remain compatible with little code. CLOCK_BOOTTIME @@ -70,7 +36,7 @@ SYSCALL_DEFINE2(clock_settime, const clockid_t, which_clock, return do_sys_settimeofday64(&new_tp, NULL); } -int do_clock_gettime(clockid_t which_clock, struct timespec64 *tp) +static int do_clock_gettime(clockid_t which_clock, struct timespec64 *tp) { switch (which_clock) { case CLOCK_REALTIME: @@ -90,6 +56,7 @@ int do_clock_gettime(clockid_t which_clock, struct timespec64 *tp) return 0; } + SYSCALL_DEFINE2(clock_gettime, const clockid_t, which_clock, struct __kernel_timespec __user *, tp) { @@ -146,6 +113,7 @@ SYSCALL_DEFINE4(clock_nanosleep, const clockid_t, which_clock, int, flags, return -EINVAL; if (flags & TIMER_ABSTIME) rmtp = NULL; + current->restart_block.fn = do_no_restart_syscall; current->restart_block.nanosleep.type = rmtp ? TT_NATIVE : TT_NONE; current->restart_block.nanosleep.rmtp = rmtp; texp = timespec64_to_ktime(t); @@ -156,18 +124,7 @@ SYSCALL_DEFINE4(clock_nanosleep, const clockid_t, which_clock, int, flags, which_clock); } -#ifdef CONFIG_COMPAT -COMPAT_SYS_NI(timer_create); -#endif - -#if defined(CONFIG_COMPAT) || defined(CONFIG_ALPHA) -COMPAT_SYS_NI(getitimer); -COMPAT_SYS_NI(setitimer); -#endif - #ifdef CONFIG_COMPAT_32BIT_TIME -SYS_NI(timer_settime32); -SYS_NI(timer_gettime32); SYSCALL_DEFINE2(clock_settime32, const clockid_t, which_clock, struct old_timespec32 __user *, tp) @@ -239,6 +196,7 @@ SYSCALL_DEFINE4(clock_nanosleep_time32, clockid_t, which_clock, int, flags, return -EINVAL; if (flags & TIMER_ABSTIME) rmtp = NULL; + current->restart_block.fn = do_no_restart_syscall; current->restart_block.nanosleep.type = rmtp ? TT_COMPAT : TT_NONE; current->restart_block.nanosleep.compat_rmtp = rmtp; texp = timespec64_to_ktime(t); diff --git a/kernel/time/posix-timers.c b/kernel/time/posix-timers.c index 07709ac30439..1b675aee99a9 100644 --- a/kernel/time/posix-timers.c +++ b/kernel/time/posix-timers.c @@ -35,20 +35,17 @@ #include "timekeeping.h" #include "posix-timers.h" -/* - * Management arrays for POSIX timers. Timers are now kept in static hash table - * with 512 entries. - * Timer ids are allocated by local routine, which selects proper hash head by - * key, constructed from current->signal address and per signal struct counter. - * This keeps timer ids unique per process, but now they can intersect between - * processes. - */ +static struct kmem_cache *posix_timers_cache; /* - * Lets keep our timers in a slab cache :-) + * Timers are managed in a hash table for lockless lookup. The hash key is + * constructed from current::signal and the timer ID and the timer is + * matched against current::signal and the timer ID when walking the hash + * bucket list. + * + * This allows checkpoint/restore to reconstruct the exact timer IDs for + * a process. */ -static struct kmem_cache *posix_timers_cache; - static DEFINE_HASHTABLE(posix_timers_hashtable, 9); static DEFINE_SPINLOCK(hash_lock); @@ -56,52 +53,12 @@ static const struct k_clock * const posix_clocks[]; static const struct k_clock *clockid_to_kclock(const clockid_t id); static const struct k_clock clock_realtime, clock_monotonic; -/* - * we assume that the new SIGEV_THREAD_ID shares no bits with the other - * SIGEV values. Here we put out an error if this assumption fails. - */ +/* SIGEV_THREAD_ID cannot share a bit with the other SIGEV values. */ #if SIGEV_THREAD_ID != (SIGEV_THREAD_ID & \ - ~(SIGEV_SIGNAL | SIGEV_NONE | SIGEV_THREAD)) + ~(SIGEV_SIGNAL | SIGEV_NONE | SIGEV_THREAD)) #error "SIGEV_THREAD_ID must not share bit with other SIGEV values!" #endif -/* - * The timer ID is turned into a timer address by idr_find(). - * Verifying a valid ID consists of: - * - * a) checking that idr_find() returns other than -1. - * b) checking that the timer id matches the one in the timer itself. - * c) that the timer owner is in the callers thread group. - */ - -/* - * CLOCKs: The POSIX standard calls for a couple of clocks and allows us - * to implement others. This structure defines the various - * clocks. - * - * RESOLUTION: Clock resolution is used to round up timer and interval - * times, NOT to report clock times, which are reported with as - * much resolution as the system can muster. In some cases this - * resolution may depend on the underlying clock hardware and - * may not be quantifiable until run time, and only then is the - * necessary code is written. The standard says we should say - * something about this issue in the documentation... - * - * FUNCTIONS: The CLOCKs structure defines possible functions to - * handle various clock functions. - * - * The standard POSIX timer management code assumes the - * following: 1.) The k_itimer struct (sched.h) is used for - * the timer. 2.) The list, it_lock, it_clock, it_id and - * it_pid fields are not modified by timer code. - * - * Permissions: It is assumed that the clock_settime() function defined - * for each clock will take care of permission checks. Some - * clocks may be set able by any user (i.e. local process - * clocks) others not. Currently the only set able clock we - * have is CLOCK_REALTIME and its high res counter part, both of - * which we beg off on and pass to do_sys_settimeofday(). - */ static struct k_itimer *__lock_timer(timer_t timer_id, unsigned long *flags); #define lock_timer(tid, flags) \ @@ -121,9 +78,9 @@ static struct k_itimer *__posix_timers_find(struct hlist_head *head, { struct k_itimer *timer; - hlist_for_each_entry_rcu(timer, head, t_hash, - lockdep_is_held(&hash_lock)) { - if ((timer->it_signal == sig) && (timer->it_id == id)) + hlist_for_each_entry_rcu(timer, head, t_hash, lockdep_is_held(&hash_lock)) { + /* timer->it_signal can be set concurrently */ + if ((READ_ONCE(timer->it_signal) == sig) && (timer->it_id == id)) return timer; } return NULL; @@ -140,25 +97,30 @@ static struct k_itimer *posix_timer_by_id(timer_t id) static int posix_timer_add(struct k_itimer *timer) { struct signal_struct *sig = current->signal; - int first_free_id = sig->posix_timer_id; struct hlist_head *head; - int ret = -ENOENT; + unsigned int cnt, id; - do { + /* + * FIXME: Replace this by a per signal struct xarray once there is + * a plan to handle the resulting CRIU regression gracefully. + */ + for (cnt = 0; cnt <= INT_MAX; cnt++) { spin_lock(&hash_lock); - head = &posix_timers_hashtable[hash(sig, sig->posix_timer_id)]; - if (!__posix_timers_find(head, sig, sig->posix_timer_id)) { + id = sig->next_posix_timer_id; + + /* Write the next ID back. Clamp it to the positive space */ + sig->next_posix_timer_id = (id + 1) & INT_MAX; + + head = &posix_timers_hashtable[hash(sig, id)]; + if (!__posix_timers_find(head, sig, id)) { hlist_add_head_rcu(&timer->t_hash, head); - ret = sig->posix_timer_id; + spin_unlock(&hash_lock); + return id; } - if (++sig->posix_timer_id < 0) - sig->posix_timer_id = 0; - if ((sig->posix_timer_id == first_free_id) && (ret == -ENOENT)) - /* Loop over all possible ids completed */ - ret = -EAGAIN; spin_unlock(&hash_lock); - } while (ret == -ENOENT); - return ret; + } + /* POSIX return code when no timer ID could be allocated */ + return -EAGAIN; } static inline void unlock_timer(struct k_itimer *timr, unsigned long flags) @@ -166,7 +128,6 @@ static inline void unlock_timer(struct k_itimer *timr, unsigned long flags) spin_unlock_irqrestore(&timr->it_lock, flags); } -/* Get clock_realtime */ static int posix_get_realtime_timespec(clockid_t which_clock, struct timespec64 *tp) { ktime_get_real_ts64(tp); @@ -178,7 +139,6 @@ static ktime_t posix_get_realtime_ktime(clockid_t which_clock) return ktime_get_real(); } -/* Set clock_realtime */ static int posix_clock_realtime_set(const clockid_t which_clock, const struct timespec64 *tp) { @@ -191,9 +151,6 @@ static int posix_clock_realtime_adj(const clockid_t which_clock, return do_adjtimex(t); } -/* - * Get monotonic time for posix timers - */ static int posix_get_monotonic_timespec(clockid_t which_clock, struct timespec64 *tp) { ktime_get_ts64(tp); @@ -206,9 +163,6 @@ static ktime_t posix_get_monotonic_ktime(clockid_t which_clock) return ktime_get(); } -/* - * Get monotonic-raw time for posix timers - */ static int posix_get_monotonic_raw(clockid_t which_clock, struct timespec64 *tp) { ktime_get_raw_ts64(tp); @@ -216,7 +170,6 @@ static int posix_get_monotonic_raw(clockid_t which_clock, struct timespec64 *tp) return 0; } - static int posix_get_realtime_coarse(clockid_t which_clock, struct timespec64 *tp) { ktime_get_coarse_real_ts64(tp); @@ -267,14 +220,11 @@ static int posix_get_hrtimer_res(clockid_t which_clock, struct timespec64 *tp) return 0; } -/* - * Initialize everything, well, just everything in Posix clocks/timers ;) - */ static __init int init_posix_timers(void) { posix_timers_cache = kmem_cache_create("posix_timers_cache", - sizeof (struct k_itimer), 0, SLAB_PANIC, - NULL); + sizeof(struct k_itimer), 0, + SLAB_PANIC | SLAB_ACCOUNT, NULL); return 0; } __initcall(init_posix_timers); @@ -283,11 +233,12 @@ __initcall(init_posix_timers); * The siginfo si_overrun field and the return value of timer_getoverrun(2) * are of type int. Clamp the overrun value to INT_MAX */ -static inline int timer_overrun_to_int(struct k_itimer *timr, int baseval) +static inline int timer_overrun_to_int(struct k_itimer *timr) { - s64 sum = timr->it_overrun_last + (s64)baseval; + if (timr->it_overrun_last > (s64)INT_MAX) + return INT_MAX; - return sum > (s64)INT_MAX ? INT_MAX : (int)sum; + return (int)timr->it_overrun_last; } static void common_hrtimer_rearm(struct k_itimer *timr) @@ -299,133 +250,78 @@ static void common_hrtimer_rearm(struct k_itimer *timr) hrtimer_restart(timer); } +static bool __posixtimer_deliver_signal(struct kernel_siginfo *info, struct k_itimer *timr) +{ + guard(spinlock)(&timr->it_lock); + + /* + * Check if the timer is still alive or whether it got modified + * since the signal was queued. In either case, don't rearm and + * drop the signal. + */ + if (timr->it_signal_seq != timr->it_sigqueue_seq || WARN_ON_ONCE(!timr->it_signal)) + return false; + + if (!timr->it_interval || WARN_ON_ONCE(timr->it_status != POSIX_TIMER_REQUEUE_PENDING)) + return true; + + timr->kclock->timer_rearm(timr); + timr->it_status = POSIX_TIMER_ARMED; + timr->it_overrun_last = timr->it_overrun; + timr->it_overrun = -1LL; + ++timr->it_signal_seq; + info->si_overrun = timer_overrun_to_int(timr); + return true; +} + /* - * This function is exported for use by the signal deliver code. It is - * called just prior to the info block being released and passes that - * block to us. It's function is to update the overrun entry AND to - * restart the timer. It should only be called if the timer is to be - * restarted (i.e. we have flagged this in the sys_private entry of the - * info block). - * - * To protect against the timer going away while the interrupt is queued, - * we require that the it_requeue_pending flag be set. + * This function is called from the signal delivery code. It decides + * whether the signal should be dropped and rearms interval timers. The + * timer can be unconditionally accessed as there is a reference held on + * it. */ -void posixtimer_rearm(struct kernel_siginfo *info) +bool posixtimer_deliver_signal(struct kernel_siginfo *info, struct sigqueue *timer_sigq) { - struct k_itimer *timr; - unsigned long flags; + struct k_itimer *timr = container_of(timer_sigq, struct k_itimer, sigq); + bool ret; - timr = lock_timer(info->si_tid, &flags); - if (!timr) - return; + /* + * Release siglock to ensure proper locking order versus + * timr::it_lock. Keep interrupts disabled. + */ + spin_unlock(¤t->sighand->siglock); - if (timr->it_interval && timr->it_requeue_pending == info->si_sys_private) { - timr->kclock->timer_rearm(timr); + ret = __posixtimer_deliver_signal(info, timr); - timr->it_active = 1; - timr->it_overrun_last = timr->it_overrun; - timr->it_overrun = -1LL; - ++timr->it_requeue_pending; + /* Drop the reference which was acquired when the signal was queued */ + posixtimer_putref(timr); - info->si_overrun = timer_overrun_to_int(timr, info->si_overrun); - } - - unlock_timer(timr, flags); + spin_lock(¤t->sighand->siglock); + return ret; } -int posix_timer_event(struct k_itimer *timr, int si_private) +void posix_timer_queue_signal(struct k_itimer *timr) { - enum pid_type type; - int ret = -1; - /* - * FIXME: if ->sigq is queued we can race with - * dequeue_signal()->posixtimer_rearm(). - * - * If dequeue_signal() sees the "right" value of - * si_sys_private it calls posixtimer_rearm(). - * We re-queue ->sigq and drop ->it_lock(). - * posixtimer_rearm() locks the timer - * and re-schedules it while ->sigq is pending. - * Not really bad, but not that we want. - */ - timr->sigq->info.si_sys_private = si_private; + lockdep_assert_held(&timr->it_lock); - type = !(timr->it_sigev_notify & SIGEV_THREAD_ID) ? PIDTYPE_TGID : PIDTYPE_PID; - ret = send_sigqueue(timr->sigq, timr->it_pid, type); - /* If we failed to send the signal the timer stops. */ - return ret > 0; + timr->it_status = timr->it_interval ? POSIX_TIMER_REQUEUE_PENDING : POSIX_TIMER_DISARMED; + posixtimer_send_sigqueue(timr); } /* - * This function gets called when a POSIX.1b interval timer expires. It - * is used as a callback from the kernel internal timer. The - * run_timer_list code ALWAYS calls with interrupts on. - - * This code is for CLOCK_REALTIME* and CLOCK_MONOTONIC* timers. + * This function gets called when a POSIX.1b interval timer expires from + * the HRTIMER interrupt (soft interrupt on RT kernels). + * + * Handles CLOCK_REALTIME, CLOCK_MONOTONIC, CLOCK_BOOTTIME and CLOCK_TAI + * based timers. */ static enum hrtimer_restart posix_timer_fn(struct hrtimer *timer) { - struct k_itimer *timr; - unsigned long flags; - int si_private = 0; - enum hrtimer_restart ret = HRTIMER_NORESTART; - - timr = container_of(timer, struct k_itimer, it.real.timer); - spin_lock_irqsave(&timr->it_lock, flags); + struct k_itimer *timr = container_of(timer, struct k_itimer, it.real.timer); - timr->it_active = 0; - if (timr->it_interval != 0) - si_private = ++timr->it_requeue_pending; - - if (posix_timer_event(timr, si_private)) { - /* - * signal was not sent because of sig_ignor - * we will not get a call back to restart it AND - * it should be restarted. - */ - if (timr->it_interval != 0) { - ktime_t now = hrtimer_cb_get_time(timer); - - /* - * FIXME: What we really want, is to stop this - * timer completely and restart it in case the - * SIG_IGN is removed. This is a non trivial - * change which involves sighand locking - * (sigh !), which we don't want to do late in - * the release cycle. - * - * For now we just let timers with an interval - * less than a jiffie expire every jiffie to - * avoid softirq starvation in case of SIG_IGN - * and a very small interval, which would put - * the timer right back on the softirq pending - * list. By moving now ahead of time we trick - * hrtimer_forward() to expire the timer - * later, while we still maintain the overrun - * accuracy, but have some inconsistency in - * the timer_gettime() case. This is at least - * better than a starved softirq. A more - * complex fix which solves also another related - * inconsistency is already in the pipeline. - */ -#ifdef CONFIG_HIGH_RES_TIMERS - { - ktime_t kj = NSEC_PER_SEC / HZ; - - if (timr->it_interval < kj) - now = ktime_add(now, kj); - } -#endif - timr->it_overrun += hrtimer_forward(timer, now, - timr->it_interval); - ret = HRTIMER_RESTART; - ++timr->it_requeue_pending; - timr->it_active = 1; - } - } - - unlock_timer(timr, flags); - return ret; + guard(spinlock_irqsave)(&timr->it_lock); + posix_timer_queue_signal(timr); + return HRTIMER_NORESTART; } static struct pid *good_sigevent(sigevent_t * event) @@ -439,12 +335,12 @@ static struct pid *good_sigevent(sigevent_t * event) rtn = pid_task(pid, PIDTYPE_PID); if (!rtn || !same_thread_group(rtn, current)) return NULL; - /* FALLTHRU */ + fallthrough; case SIGEV_SIGNAL: case SIGEV_THREAD: if (event->sigev_signo <= 0 || event->sigev_signo > SIGRTMAX) return NULL; - /* FALLTHRU */ + fallthrough; case SIGEV_NONE: return pid; default: @@ -452,40 +348,35 @@ static struct pid *good_sigevent(sigevent_t * event) } } -static struct k_itimer * alloc_posix_timer(void) +static struct k_itimer *alloc_posix_timer(void) { - struct k_itimer *tmr; - tmr = kmem_cache_zalloc(posix_timers_cache, GFP_KERNEL); + struct k_itimer *tmr = kmem_cache_zalloc(posix_timers_cache, GFP_KERNEL); + if (!tmr) return tmr; - if (unlikely(!(tmr->sigq = sigqueue_alloc()))) { + + if (unlikely(!posixtimer_init_sigqueue(&tmr->sigq))) { kmem_cache_free(posix_timers_cache, tmr); return NULL; } - clear_siginfo(&tmr->sigq->info); + rcuref_init(&tmr->rcuref, 1); return tmr; } -static void k_itimer_rcu_free(struct rcu_head *head) +void posixtimer_free_timer(struct k_itimer *tmr) { - struct k_itimer *tmr = container_of(head, struct k_itimer, rcu); - - kmem_cache_free(posix_timers_cache, tmr); + put_pid(tmr->it_pid); + if (tmr->sigq.ucounts) + dec_rlimit_put_ucounts(tmr->sigq.ucounts, UCOUNT_RLIMIT_SIGPENDING); + kfree_rcu(tmr, rcu); } -#define IT_ID_SET 1 -#define IT_ID_NOT_SET 0 -static void release_posix_timer(struct k_itimer *tmr, int it_id_set) +static void posix_timer_unhash_and_free(struct k_itimer *tmr) { - if (it_id_set) { - unsigned long flags; - spin_lock_irqsave(&hash_lock, flags); - hlist_del_rcu(&tmr->t_hash); - spin_unlock_irqrestore(&hash_lock, flags); - } - put_pid(tmr->it_pid); - sigqueue_free(tmr->sigq); - call_rcu(&tmr->rcu, k_itimer_rcu_free); + spin_lock(&hash_lock); + hlist_del_rcu(&tmr->t_hash); + spin_unlock(&hash_lock); + posixtimer_putref(tmr); } static int common_timer_create(struct k_itimer *new_timer) @@ -501,7 +392,6 @@ static int do_timer_create(clockid_t which_clock, struct sigevent *event, const struct k_clock *kc = clockid_to_kclock(which_clock); struct k_itimer *new_timer; int error, new_timer_id; - int it_id_set = IT_ID_NOT_SET; if (!kc) return -EINVAL; @@ -513,13 +403,18 @@ static int do_timer_create(clockid_t which_clock, struct sigevent *event, return -EAGAIN; spin_lock_init(&new_timer->it_lock); + + /* + * Add the timer to the hash table. The timer is not yet valid + * because new_timer::it_signal is still NULL. The timer id is also + * not yet visible to user space. + */ new_timer_id = posix_timer_add(new_timer); if (new_timer_id < 0) { - error = new_timer_id; - goto out; + posixtimer_free_timer(new_timer); + return new_timer_id; } - it_id_set = IT_ID_SET; new_timer->it_id = (timer_t) new_timer_id; new_timer->it_clock = which_clock; new_timer->kclock = kc; @@ -534,43 +429,51 @@ static int do_timer_create(clockid_t which_clock, struct sigevent *event, goto out; } new_timer->it_sigev_notify = event->sigev_notify; - new_timer->sigq->info.si_signo = event->sigev_signo; - new_timer->sigq->info.si_value = event->sigev_value; + new_timer->sigq.info.si_signo = event->sigev_signo; + new_timer->sigq.info.si_value = event->sigev_value; } else { new_timer->it_sigev_notify = SIGEV_SIGNAL; - new_timer->sigq->info.si_signo = SIGALRM; - memset(&new_timer->sigq->info.si_value, 0, sizeof(sigval_t)); - new_timer->sigq->info.si_value.sival_int = new_timer->it_id; + new_timer->sigq.info.si_signo = SIGALRM; + memset(&new_timer->sigq.info.si_value, 0, sizeof(sigval_t)); + new_timer->sigq.info.si_value.sival_int = new_timer->it_id; new_timer->it_pid = get_pid(task_tgid(current)); } - new_timer->sigq->info.si_tid = new_timer->it_id; - new_timer->sigq->info.si_code = SI_TIMER; + if (new_timer->it_sigev_notify & SIGEV_THREAD_ID) + new_timer->it_pid_type = PIDTYPE_PID; + else + new_timer->it_pid_type = PIDTYPE_TGID; + + new_timer->sigq.info.si_tid = new_timer->it_id; + new_timer->sigq.info.si_code = SI_TIMER; - if (copy_to_user(created_timer_id, - &new_timer_id, sizeof (new_timer_id))) { + if (copy_to_user(created_timer_id, &new_timer_id, sizeof (new_timer_id))) { error = -EFAULT; goto out; } - + /* + * After succesful copy out, the timer ID is visible to user space + * now but not yet valid because new_timer::signal is still NULL. + * + * Complete the initialization with the clock specific create + * callback. + */ error = kc->timer_create(new_timer); if (error) goto out; spin_lock_irq(¤t->sighand->siglock); - new_timer->it_signal = current->signal; - list_add(&new_timer->list, ¤t->signal->posix_timers); + /* This makes the timer valid in the hash table */ + WRITE_ONCE(new_timer->it_signal, current->signal); + hlist_add_head(&new_timer->list, ¤t->signal->posix_timers); spin_unlock_irq(¤t->sighand->siglock); - - return 0; /* - * In the case of the timer belonging to another task, after - * the task is unlocked, the timer is owned by the other task - * and may cease to exist at any time. Don't use or modify - * new_timer after the unlock call. + * After unlocking sighand::siglock @new_timer is subject to + * concurrent removal and cannot be touched anymore */ + return 0; out: - release_posix_timer(new_timer, it_id_set); + posix_timer_unhash_and_free(new_timer); return error; } @@ -604,13 +507,6 @@ COMPAT_SYSCALL_DEFINE3(timer_create, clockid_t, which_clock, } #endif -/* - * Locking issues: We need to protect the result of the id look up until - * we get the timer locked down so it is not deleted under us. The - * removal is done under the idr spinlock so we use that here to bridge - * the find to the timer lock. To avoid a dead lock, the timer id MUST - * be release with out holding the timer lock. - */ static struct k_itimer *__lock_timer(timer_t timer_id, unsigned long *flags) { struct k_itimer *timr; @@ -622,10 +518,42 @@ static struct k_itimer *__lock_timer(timer_t timer_id, unsigned long *flags) if ((unsigned long long)timer_id > INT_MAX) return NULL; + /* + * The hash lookup and the timers are RCU protected. + * + * Timers are added to the hash in invalid state where + * timr::it_signal == NULL. timer::it_signal is only set after the + * rest of the initialization succeeded. + * + * Timer destruction happens in steps: + * 1) Set timr::it_signal to NULL with timr::it_lock held + * 2) Release timr::it_lock + * 3) Remove from the hash under hash_lock + * 4) Put the reference count. + * + * The reference count might not drop to zero if timr::sigq is + * queued. In that case the signal delivery or flush will put the + * last reference count. + * + * When the reference count reaches zero, the timer is scheduled + * for RCU removal after the grace period. + * + * Holding rcu_read_lock() across the lookup ensures that + * the timer cannot be freed. + * + * The lookup validates locklessly that timr::it_signal == + * current::it_signal and timr::it_id == @timer_id. timr::it_id + * can't change, but timr::it_signal becomes NULL during + * destruction. + */ rcu_read_lock(); timr = posix_timer_by_id(timer_id); if (timr) { spin_lock_irqsave(&timr->it_lock, *flags); + /* + * Validate under timr::it_lock that timr::it_signal is + * still valid. Pairs with #1 above. + */ if (timr->it_signal == current->signal) { rcu_read_unlock(); return timr; @@ -652,20 +580,16 @@ static s64 common_hrtimer_forward(struct k_itimer *timr, ktime_t now) } /* - * Get the time remaining on a POSIX.1b interval timer. This function - * is ALWAYS called with spin_lock_irq on the timer, thus it must not - * mess with irq. + * Get the time remaining on a POSIX.1b interval timer. + * + * Two issues to handle here: * - * We have a couple of messes to clean up here. First there is the case - * of a timer that has a requeue pending. These timers should appear to - * be in the timer list with an expiry as if we were to requeue them - * now. + * 1) The timer has a requeue pending. The return value must appear as + * if the timer has been requeued right now. * - * The second issue is the SIGEV_NONE timer which may be active but is - * not really ever put in the timer list (to save system resources). - * This timer may be expired, and if so, we will do it here. Otherwise - * it is the same as a requeue pending timer WRT to what we should - * report. + * 2) The timer is a SIGEV_NONE timer. These timers are never enqueued + * into the hrtimer queue and therefore never expired. Emulate expiry + * here taking #1 into account. */ void common_timer_get(struct k_itimer *timr, struct itimerspec64 *cur_setting) { @@ -679,10 +603,14 @@ void common_timer_get(struct k_itimer *timr, struct itimerspec64 *cur_setting) /* interval timer ? */ if (iv) { cur_setting->it_interval = ktime_to_timespec64(iv); - } else if (!timr->it_active) { + } else if (timr->it_status == POSIX_TIMER_DISARMED) { /* - * SIGEV_NONE oneshot timers are never queued. Check them - * below. + * SIGEV_NONE oneshot timers are never queued and therefore + * timr->it_status is always DISARMED. The check below + * vs. remaining time will handle this case. + * + * For all other timers there is nothing to update here, so + * return. */ if (!sig_none) return; @@ -691,18 +619,29 @@ void common_timer_get(struct k_itimer *timr, struct itimerspec64 *cur_setting) now = kc->clock_get_ktime(timr->it_clock); /* - * When a requeue is pending or this is a SIGEV_NONE timer move the - * expiry time forward by intervals, so expiry is > now. + * If this is an interval timer and either has requeue pending or + * is a SIGEV_NONE timer move the expiry time forward by intervals, + * so expiry is > now. */ - if (iv && (timr->it_requeue_pending & REQUEUE_PENDING || sig_none)) + if (iv && timr->it_status != POSIX_TIMER_ARMED) timr->it_overrun += kc->timer_forward(timr, now); remaining = kc->timer_remaining(timr, now); - /* Return 0 only, when the timer is expired and not pending */ + /* + * As @now is retrieved before a possible timer_forward() and + * cannot be reevaluated by the compiler @remaining is based on the + * same @now value. Therefore @remaining is consistent vs. @now. + * + * Consequently all interval timers, i.e. @iv > 0, cannot have a + * remaining time <= 0 because timer_forward() guarantees to move + * them forward so that the next timer expiry is > @now. + */ if (remaining <= 0) { /* - * A single shot SIGEV_NONE timer must return 0, when - * it is expired ! + * A single shot SIGEV_NONE timer must return 0, when it is + * expired! Timers which have a real signal delivery mode + * must return a remaining time greater than 0 because the + * signal has not yet been delivered. */ if (!sig_none) cur_setting->it_value.tv_nsec = 1; @@ -711,11 +650,10 @@ void common_timer_get(struct k_itimer *timr, struct itimerspec64 *cur_setting) } } -/* Get the time remaining on a POSIX.1b interval timer. */ static int do_timer_gettime(timer_t timer_id, struct itimerspec64 *setting) { - struct k_itimer *timr; const struct k_clock *kc; + struct k_itimer *timr; unsigned long flags; int ret = 0; @@ -765,26 +703,35 @@ SYSCALL_DEFINE2(timer_gettime32, timer_t, timer_id, #endif -/* - * Get the number of overruns of a POSIX.1b interval timer. This is to - * be the overrun of the timer last delivered. At the same time we are - * accumulating overruns on the next timer. The overrun is frozen when - * the signal is delivered, either at the notify time (if the info block - * is not queued) or at the actual delivery time (as we are informed by - * the call back to posixtimer_rearm(). So all we need to do is - * to pick up the frozen overrun. +/** + * sys_timer_getoverrun - Get the number of overruns of a POSIX.1b interval timer + * @timer_id: The timer ID which identifies the timer + * + * The "overrun count" of a timer is one plus the number of expiration + * intervals which have elapsed between the first expiry, which queues the + * signal and the actual signal delivery. On signal delivery the "overrun + * count" is calculated and cached, so it can be returned directly here. + * + * As this is relative to the last queued signal the returned overrun count + * is meaningless outside of the signal delivery path and even there it + * does not accurately reflect the current state when user space evaluates + * it. + * + * Returns: + * -EINVAL @timer_id is invalid + * 1..INT_MAX The number of overruns related to the last delivered signal */ SYSCALL_DEFINE1(timer_getoverrun, timer_t, timer_id) { struct k_itimer *timr; - int overrun; unsigned long flags; + int overrun; timr = lock_timer(timer_id, &flags); if (!timr) return -EINVAL; - overrun = timer_overrun_to_int(timr, 0); + overrun = timer_overrun_to_int(timr); unlock_timer(timr, flags); return overrun; @@ -831,10 +778,18 @@ static void common_timer_wait_running(struct k_itimer *timer) } /* - * On PREEMPT_RT this prevent priority inversion against softirq kthread in - * case it gets preempted while executing a timer callback. See comments in - * hrtimer_cancel_wait_running. For PREEMPT_RT=n this just results in a - * cpu_relax(). + * On PREEMPT_RT this prevents priority inversion and a potential livelock + * against the ksoftirqd thread in case that ksoftirqd gets preempted while + * executing a hrtimer callback. + * + * See the comments in hrtimer_cancel_wait_running(). For PREEMPT_RT=n this + * just results in a cpu_relax(). + * + * For POSIX CPU timers with CONFIG_POSIX_CPU_TIMERS_TASK_WORK=n this is + * just a cpu_relax(). With CONFIG_POSIX_CPU_TIMERS_TASK_WORK=y this + * prevents spinning on an eventually scheduled out task and a livelock + * when the task which tries to delete or disarm the timer has preempted + * the task which runs the expiry in task work context. */ static struct k_itimer *timer_wait_running(struct k_itimer *timer, unsigned long *flags) @@ -846,6 +801,10 @@ static struct k_itimer *timer_wait_running(struct k_itimer *timer, rcu_read_lock(); unlock_timer(timer, *flags); + /* + * kc->timer_wait_running() might drop RCU lock. So @timer + * cannot be touched anymore after the function returns! + */ if (!WARN_ON_ONCE(!kc->timer_wait_running)) kc->timer_wait_running(timer); @@ -854,6 +813,21 @@ static struct k_itimer *timer_wait_running(struct k_itimer *timer, return lock_timer(timer_id, flags); } +/* + * Set up the new interval and reset the signal delivery data + */ +void posix_timer_set_common(struct k_itimer *timer, struct itimerspec64 *new_setting) +{ + if (new_setting->it_value.tv_sec || new_setting->it_value.tv_nsec) + timer->it_interval = timespec64_to_ktime(new_setting->it_interval); + else + timer->it_interval = 0; + + /* Reset overrun accounting */ + timer->it_overrun_last = 0; + timer->it_overrun = -1LL; +} + /* Set a POSIX.1b interval timer. */ int common_timer_set(struct k_itimer *timr, int flags, struct itimerspec64 *new_setting, @@ -866,8 +840,6 @@ int common_timer_set(struct k_itimer *timr, int flags, if (old_setting) common_timer_get(timr, old_setting); - /* Prevent rearming by clearing the interval */ - timr->it_interval = 0; /* * Careful here. On SMP systems the timer expiry function could be * active and spinning on timr->it_lock. @@ -875,23 +847,21 @@ int common_timer_set(struct k_itimer *timr, int flags, if (kc->timer_try_to_cancel(timr) < 0) return TIMER_RETRY; - timr->it_active = 0; - timr->it_requeue_pending = (timr->it_requeue_pending + 2) & - ~REQUEUE_PENDING; - timr->it_overrun_last = 0; + timr->it_status = POSIX_TIMER_DISARMED; + posix_timer_set_common(timr, new_setting); - /* Switch off the timer when it_value is zero */ + /* Keep timer disarmed when it_value is zero */ if (!new_setting->it_value.tv_sec && !new_setting->it_value.tv_nsec) return 0; - timr->it_interval = timespec64_to_ktime(new_setting->it_interval); expires = timespec64_to_ktime(new_setting->it_value); if (flags & TIMER_ABSTIME) expires = timens_ktime_to_host(timr->it_clock, expires); sigev_none = timr->it_sigev_notify == SIGEV_NONE; kc->timer_arm(timr, expires, flags & TIMER_ABSTIME, sigev_none); - timr->it_active = !sigev_none; + if (!sigev_none) + timr->it_status = POSIX_TIMER_ARMED; return 0; } @@ -902,7 +872,7 @@ static int do_timer_settime(timer_t timer_id, int tmr_flags, const struct k_clock *kc; struct k_itimer *timr; unsigned long flags; - int error = 0; + int error; if (!timespec64_valid(&new_spec64->it_interval) || !timespec64_valid(&new_spec64->it_value)) @@ -916,6 +886,12 @@ retry: if (!timr) return -EINVAL; + if (old_spec64) + old_spec64->it_interval = ktime_to_timespec64(timr->it_interval); + + /* Prevent signal delivery and rearming. */ + timr->it_signal_seq++; + kc = timr->kclock; if (WARN_ON_ONCE(!kc || !kc->timer_set)) error = -EINVAL; @@ -939,8 +915,7 @@ SYSCALL_DEFINE4(timer_settime, timer_t, timer_id, int, flags, const struct __kernel_itimerspec __user *, new_setting, struct __kernel_itimerspec __user *, old_setting) { - struct itimerspec64 new_spec, old_spec; - struct itimerspec64 *rtn = old_setting ? &old_spec : NULL; + struct itimerspec64 new_spec, old_spec, *rtn; int error = 0; if (!new_setting) @@ -949,6 +924,7 @@ SYSCALL_DEFINE4(timer_settime, timer_t, timer_id, int, flags, if (get_itimerspec64(&new_spec, new_setting)) return -EFAULT; + rtn = old_setting ? &old_spec : NULL; error = do_timer_settime(timer_id, flags, &new_spec, rtn); if (!error && old_setting) { if (put_itimerspec64(&old_spec, old_setting)) @@ -984,17 +960,31 @@ int common_timer_del(struct k_itimer *timer) { const struct k_clock *kc = timer->kclock; - timer->it_interval = 0; if (kc->timer_try_to_cancel(timer) < 0) return TIMER_RETRY; - timer->it_active = 0; + timer->it_status = POSIX_TIMER_DISARMED; return 0; } +/* + * If the deleted timer is on the ignored list, remove it and + * drop the associated reference. + */ +static inline void posix_timer_cleanup_ignored(struct k_itimer *tmr) +{ + if (!hlist_unhashed(&tmr->ignored_list)) { + hlist_del_init(&tmr->ignored_list); + posixtimer_putref(tmr); + } +} + static inline int timer_delete_hook(struct k_itimer *timer) { const struct k_clock *kc = timer->kclock; + /* Prevent signal delivery and rearming. */ + timer->it_signal_seq++; + if (WARN_ON_ONCE(!kc || !kc->timer_del)) return -EINVAL; return kc->timer_del(timer); @@ -1019,48 +1009,110 @@ retry_delete: } spin_lock(¤t->sighand->siglock); - list_del(&timer->list); - spin_unlock(¤t->sighand->siglock); + hlist_del(&timer->list); + posix_timer_cleanup_ignored(timer); /* - * This keeps any tasks waiting on the spin lock from thinking - * they got something (see the lock code above). + * A concurrent lookup could check timer::it_signal lockless. It + * will reevaluate with timer::it_lock held and observe the NULL. + * + * It must be written with siglock held so that the signal code + * observes timer->it_signal == NULL in do_sigaction(SIG_IGN), + * which prevents it from moving a pending signal of a deleted + * timer to the ignore list. */ - timer->it_signal = NULL; + WRITE_ONCE(timer->it_signal, NULL); + spin_unlock(¤t->sighand->siglock); unlock_timer(timer, flags); - release_posix_timer(timer, IT_ID_SET); + posix_timer_unhash_and_free(timer); return 0; } /* - * return timer owned by the process, used by exit_itimers + * Delete a timer if it is armed, remove it from the hash and schedule it + * for RCU freeing. */ static void itimer_delete(struct k_itimer *timer) { -retry_delete: - spin_lock_irq(&timer->it_lock); + unsigned long flags; + /* + * irqsave is required to make timer_wait_running() work. + */ + spin_lock_irqsave(&timer->it_lock, flags); + +retry_delete: + /* + * Even if the timer is not longer accessible from other tasks + * it still might be armed and queued in the underlying timer + * mechanism. Worse, that timer mechanism might run the expiry + * function concurrently. + */ if (timer_delete_hook(timer) == TIMER_RETRY) { - spin_unlock_irq(&timer->it_lock); + /* + * Timer is expired concurrently, prevent livelocks + * and pointless spinning on RT. + * + * timer_wait_running() drops timer::it_lock, which opens + * the possibility for another task to delete the timer. + * + * That's not possible here because this is invoked from + * do_exit() only for the last thread of the thread group. + * So no other task can access and delete that timer. + */ + if (WARN_ON_ONCE(timer_wait_running(timer, &flags) != timer)) + return; + goto retry_delete; } - list_del(&timer->list); + hlist_del(&timer->list); + + posix_timer_cleanup_ignored(timer); + + /* + * Setting timer::it_signal to NULL is technically not required + * here as nothing can access the timer anymore legitimately via + * the hash table. Set it to NULL nevertheless so that all deletion + * paths are consistent. + */ + WRITE_ONCE(timer->it_signal, NULL); - spin_unlock_irq(&timer->it_lock); - release_posix_timer(timer, IT_ID_SET); + spin_unlock_irqrestore(&timer->it_lock, flags); + posix_timer_unhash_and_free(timer); } /* - * This is called by do_exit or de_thread, only when there are no more - * references to the shared signal_struct. + * Invoked from do_exit() when the last thread of a thread group exits. + * At that point no other task can access the timers of the dying + * task anymore. */ -void exit_itimers(struct signal_struct *sig) +void exit_itimers(struct task_struct *tsk) { - struct k_itimer *tmr; + struct hlist_head timers; + + if (hlist_empty(&tsk->signal->posix_timers)) + return; + + /* Protect against concurrent read via /proc/$PID/timers */ + spin_lock_irq(&tsk->sighand->siglock); + hlist_move_list(&tsk->signal->posix_timers, &timers); + spin_unlock_irq(&tsk->sighand->siglock); + + /* The timers are not longer accessible via tsk::signal */ + while (!hlist_empty(&timers)) + itimer_delete(hlist_entry(timers.first, struct k_itimer, list)); - while (!list_empty(&sig->posix_timers)) { - tmr = list_entry(sig->posix_timers.next, struct k_itimer, list); - itimer_delete(tmr); + /* + * There should be no timers on the ignored list. itimer_delete() has + * mopped them up. + */ + if (!WARN_ON_ONCE(!hlist_empty(&tsk->signal->ignored_posix_timers))) + return; + + hlist_move_list(&tsk->signal->ignored_posix_timers, &timers); + while (!hlist_empty(&timers)) { + posix_timer_cleanup_ignored(hlist_entry(timers.first, struct k_itimer, + ignored_list)); } } @@ -1076,6 +1128,10 @@ SYSCALL_DEFINE2(clock_settime, const clockid_t, which_clock, if (get_timespec64(&new_tp, tp)) return -EFAULT; + /* + * Permission checks have to be done inside the clock specific + * setter callback. + */ return kc->clock_set(which_clock, &new_tp); } @@ -1126,6 +1182,79 @@ SYSCALL_DEFINE2(clock_adjtime, const clockid_t, which_clock, return err; } +/** + * sys_clock_getres - Get the resolution of a clock + * @which_clock: The clock to get the resolution for + * @tp: Pointer to a a user space timespec64 for storage + * + * POSIX defines: + * + * "The clock_getres() function shall return the resolution of any + * clock. Clock resolutions are implementation-defined and cannot be set by + * a process. If the argument res is not NULL, the resolution of the + * specified clock shall be stored in the location pointed to by res. If + * res is NULL, the clock resolution is not returned. If the time argument + * of clock_settime() is not a multiple of res, then the value is truncated + * to a multiple of res." + * + * Due to the various hardware constraints the real resolution can vary + * wildly and even change during runtime when the underlying devices are + * replaced. The kernel also can use hardware devices with different + * resolutions for reading the time and for arming timers. + * + * The kernel therefore deviates from the POSIX spec in various aspects: + * + * 1) The resolution returned to user space + * + * For CLOCK_REALTIME, CLOCK_MONOTONIC, CLOCK_BOOTTIME, CLOCK_TAI, + * CLOCK_REALTIME_ALARM, CLOCK_BOOTTIME_ALAREM and CLOCK_MONOTONIC_RAW + * the kernel differentiates only two cases: + * + * I) Low resolution mode: + * + * When high resolution timers are disabled at compile or runtime + * the resolution returned is nanoseconds per tick, which represents + * the precision at which timers expire. + * + * II) High resolution mode: + * + * When high resolution timers are enabled the resolution returned + * is always one nanosecond independent of the actual resolution of + * the underlying hardware devices. + * + * For CLOCK_*_ALARM the actual resolution depends on system + * state. When system is running the resolution is the same as the + * resolution of the other clocks. During suspend the actual + * resolution is the resolution of the underlying RTC device which + * might be way less precise than the clockevent device used during + * running state. + * + * For CLOCK_REALTIME_COARSE and CLOCK_MONOTONIC_COARSE the resolution + * returned is always nanoseconds per tick. + * + * For CLOCK_PROCESS_CPUTIME and CLOCK_THREAD_CPUTIME the resolution + * returned is always one nanosecond under the assumption that the + * underlying scheduler clock has a better resolution than nanoseconds + * per tick. + * + * For dynamic POSIX clocks (PTP devices) the resolution returned is + * always one nanosecond. + * + * 2) Affect on sys_clock_settime() + * + * The kernel does not truncate the time which is handed in to + * sys_clock_settime(). The kernel internal timekeeping is always using + * nanoseconds precision independent of the clocksource device which is + * used to read the time from. The resolution of that device only + * affects the presicion of the time returned by sys_clock_gettime(). + * + * Returns: + * 0 Success. @tp contains the resolution + * -EINVAL @which_clock is not a valid clock ID + * -EFAULT Copying the resolution to @tp faulted + * -ENODEV Dynamic POSIX clock is not backed by a device + * -EOPNOTSUPP Dynamic POSIX clock does not support getres() + */ SYSCALL_DEFINE2(clock_getres, const clockid_t, which_clock, struct __kernel_timespec __user *, tp) { @@ -1191,8 +1320,8 @@ SYSCALL_DEFINE2(clock_adjtime32, clockid_t, which_clock, err = do_clock_adjtime(which_clock, &ktx); - if (err >= 0) - err = put_old_timex32(utp, &ktx); + if (err >= 0 && put_old_timex32(utp, &ktx)) + return -EFAULT; return err; } @@ -1217,7 +1346,7 @@ SYSCALL_DEFINE2(clock_getres_time32, clockid_t, which_clock, #endif /* - * nanosleep for monotonic and realtime clocks + * sys_clock_nanosleep() for CLOCK_REALTIME and CLOCK_TAI */ static int common_nsleep(const clockid_t which_clock, int flags, const struct timespec64 *rqtp) @@ -1229,8 +1358,13 @@ static int common_nsleep(const clockid_t which_clock, int flags, which_clock); } +/* + * sys_clock_nanosleep() for CLOCK_MONOTONIC and CLOCK_BOOTTIME + * + * Absolute nanosleeps for these clocks are time-namespace adjusted. + */ static int common_nsleep_timens(const clockid_t which_clock, int flags, - const struct timespec64 *rqtp) + const struct timespec64 *rqtp) { ktime_t texp = timespec64_to_ktime(*rqtp); @@ -1261,6 +1395,7 @@ SYSCALL_DEFINE4(clock_nanosleep, const clockid_t, which_clock, int, flags, return -EINVAL; if (flags & TIMER_ABSTIME) rmtp = NULL; + current->restart_block.fn = do_no_restart_syscall; current->restart_block.nanosleep.type = rmtp ? TT_NATIVE : TT_NONE; current->restart_block.nanosleep.rmtp = rmtp; @@ -1288,6 +1423,7 @@ SYSCALL_DEFINE4(clock_nanosleep_time32, clockid_t, which_clock, int, flags, return -EINVAL; if (flags & TIMER_ABSTIME) rmtp = NULL; + current->restart_block.fn = do_no_restart_syscall; current->restart_block.nanosleep.type = rmtp ? TT_COMPAT : TT_NONE; current->restart_block.nanosleep.compat_rmtp = rmtp; diff --git a/kernel/time/posix-timers.h b/kernel/time/posix-timers.h index f32a2ebba9b8..61906f0688c1 100644 --- a/kernel/time/posix-timers.h +++ b/kernel/time/posix-timers.h @@ -1,6 +1,12 @@ /* SPDX-License-Identifier: GPL-2.0 */ #define TIMER_RETRY 1 +enum posix_timer_state { + POSIX_TIMER_DISARMED, + POSIX_TIMER_ARMED, + POSIX_TIMER_REQUEUE_PENDING, +}; + struct k_clock { int (*clock_getres)(const clockid_t which_clock, struct timespec64 *tp); @@ -36,10 +42,11 @@ extern const struct k_clock clock_process; extern const struct k_clock clock_thread; extern const struct k_clock alarm_clock; -int posix_timer_event(struct k_itimer *timr, int si_private); +void posix_timer_queue_signal(struct k_itimer *timr); void common_timer_get(struct k_itimer *timr, struct itimerspec64 *cur_setting); int common_timer_set(struct k_itimer *timr, int flags, struct itimerspec64 *new_setting, struct itimerspec64 *old_setting); +void posix_timer_set_common(struct k_itimer *timer, struct itimerspec64 *new_setting); int common_timer_del(struct k_itimer *timer); diff --git a/kernel/time/sched_clock.c b/kernel/time/sched_clock.c index fa3f800d7d76..fcca4e72f1ef 100644 --- a/kernel/time/sched_clock.c +++ b/kernel/time/sched_clock.c @@ -8,6 +8,7 @@ #include <linux/jiffies.h> #include <linux/ktime.h> #include <linux/kernel.h> +#include <linux/math.h> #include <linux/moduleparam.h> #include <linux/sched.h> #include <linux/sched/clock.h> @@ -20,31 +21,6 @@ #include "timekeeping.h" /** - * struct clock_read_data - data required to read from sched_clock() - * - * @epoch_ns: sched_clock() value at last update - * @epoch_cyc: Clock cycle value at last update. - * @sched_clock_mask: Bitmask for two's complement subtraction of non 64bit - * clocks. - * @read_sched_clock: Current clock source (or dummy source when suspended). - * @mult: Multipler for scaled math conversion. - * @shift: Shift value for scaled math conversion. - * - * Care must be taken when updating this structure; it is read by - * some very hot code paths. It occupies <=40 bytes and, when combined - * with the seqcount used to synchronize access, comfortably fits into - * a 64 byte cache line. - */ -struct clock_read_data { - u64 epoch_ns; - u64 epoch_cyc; - u64 sched_clock_mask; - u64 (*read_sched_clock)(void); - u32 mult; - u32 shift; -}; - -/** * struct clock_data - all data needed for sched_clock() (including * registration of a new clock source) * @@ -60,7 +36,7 @@ struct clock_read_data { * into a single 64-byte cache line. */ struct clock_data { - seqcount_t seq; + seqcount_latch_t seq; struct clock_read_data read_data[2]; ktime_t wrap_kt; unsigned long rate; @@ -88,29 +64,61 @@ static struct clock_data cd ____cacheline_aligned = { .actual_read_sched_clock = jiffy_sched_clock_read, }; -static inline u64 notrace cyc_to_ns(u64 cyc, u32 mult, u32 shift) +static __always_inline u64 cyc_to_ns(u64 cyc, u32 mult, u32 shift) { return (cyc * mult) >> shift; } -unsigned long long notrace sched_clock(void) +notrace struct clock_read_data *sched_clock_read_begin(unsigned int *seq) +{ + *seq = read_seqcount_latch(&cd.seq); + return cd.read_data + (*seq & 1); +} + +notrace int sched_clock_read_retry(unsigned int seq) +{ + return read_seqcount_latch_retry(&cd.seq, seq); +} + +static __always_inline unsigned long long __sched_clock(void) { - u64 cyc, res; - unsigned int seq; struct clock_read_data *rd; + unsigned int seq; + u64 cyc, res; do { - seq = raw_read_seqcount(&cd.seq); + seq = raw_read_seqcount_latch(&cd.seq); rd = cd.read_data + (seq & 1); cyc = (rd->read_sched_clock() - rd->epoch_cyc) & rd->sched_clock_mask; res = rd->epoch_ns + cyc_to_ns(cyc, rd->mult, rd->shift); - } while (read_seqcount_retry(&cd.seq, seq)); + } while (raw_read_seqcount_latch_retry(&cd.seq, seq)); return res; } +unsigned long long noinstr sched_clock_noinstr(void) +{ + return __sched_clock(); +} + +unsigned long long notrace sched_clock(void) +{ + unsigned long long ns; + preempt_disable_notrace(); + /* + * All of __sched_clock() is a seqcount_latch reader critical section, + * but relies on the raw helpers which are uninstrumented. For KCSAN, + * mark all accesses in __sched_clock() as atomic. + */ + kcsan_nestable_atomic_begin(); + ns = __sched_clock(); + kcsan_nestable_atomic_end(); + preempt_enable_notrace(); + return ns; +} + /* * Updating the data required to read the clock. * @@ -123,17 +131,19 @@ unsigned long long notrace sched_clock(void) */ static void update_clock_read_data(struct clock_read_data *rd) { - /* update the backup (odd) copy with the new data */ - cd.read_data[1] = *rd; - /* steer readers towards the odd copy */ - raw_write_seqcount_latch(&cd.seq); + write_seqcount_latch_begin(&cd.seq); /* now its safe for us to update the normal (even) copy */ cd.read_data[0] = *rd; /* switch readers back to the even copy */ - raw_write_seqcount_latch(&cd.seq); + write_seqcount_latch(&cd.seq); + + /* update the backup (odd) copy with the new data */ + cd.read_data[1] = *rd; + + write_seqcount_latch_end(&cd.seq); } /* @@ -214,15 +224,13 @@ sched_clock_register(u64 (*read)(void), int bits, unsigned long rate) r = rate; if (r >= 4000000) { - r /= 1000000; + r = DIV_ROUND_CLOSEST(r, 1000000); r_unit = 'M'; + } else if (r >= 4000) { + r = DIV_ROUND_CLOSEST(r, 1000); + r_unit = 'k'; } else { - if (r >= 1000) { - r /= 1000; - r_unit = 'k'; - } else { - r_unit = ' '; - } + r_unit = ' '; } /* Calculate the ns resolution of this counter */ @@ -244,7 +252,7 @@ void __init generic_sched_clock_init(void) { /* * If no sched_clock() function has been provided at that point, - * make it the final one one. + * make it the final one. */ if (cd.actual_read_sched_clock == jiffy_sched_clock_read) sched_clock_register(jiffy_sched_clock_read, BITS_PER_LONG, HZ); @@ -273,7 +281,7 @@ void __init generic_sched_clock_init(void) */ static u64 notrace suspended_sched_clock_read(void) { - unsigned int seq = raw_read_seqcount(&cd.seq); + unsigned int seq = read_seqcount_latch(&cd.seq); return cd.read_data[seq & 1].epoch_cyc; } diff --git a/kernel/time/sleep_timeout.c b/kernel/time/sleep_timeout.c new file mode 100644 index 000000000000..dfe939f6e4ec --- /dev/null +++ b/kernel/time/sleep_timeout.c @@ -0,0 +1,377 @@ +// SPDX-License-Identifier: GPL-2.0 +/* + * Kernel internal schedule timeout and sleeping functions + */ + +#include <linux/delay.h> +#include <linux/jiffies.h> +#include <linux/timer.h> +#include <linux/sched/signal.h> +#include <linux/sched/debug.h> + +#include "tick-internal.h" + +/* + * Since schedule_timeout()'s timer is defined on the stack, it must store + * the target task on the stack as well. + */ +struct process_timer { + struct timer_list timer; + struct task_struct *task; +}; + +static void process_timeout(struct timer_list *t) +{ + struct process_timer *timeout = from_timer(timeout, t, timer); + + wake_up_process(timeout->task); +} + +/** + * schedule_timeout - sleep until timeout + * @timeout: timeout value in jiffies + * + * Make the current task sleep until @timeout jiffies have elapsed. + * The function behavior depends on the current task state + * (see also set_current_state() description): + * + * %TASK_RUNNING - the scheduler is called, but the task does not sleep + * at all. That happens because sched_submit_work() does nothing for + * tasks in %TASK_RUNNING state. + * + * %TASK_UNINTERRUPTIBLE - at least @timeout jiffies are guaranteed to + * pass before the routine returns unless the current task is explicitly + * woken up, (e.g. by wake_up_process()). + * + * %TASK_INTERRUPTIBLE - the routine may return early if a signal is + * delivered to the current task or the current task is explicitly woken + * up. + * + * The current task state is guaranteed to be %TASK_RUNNING when this + * routine returns. + * + * Specifying a @timeout value of %MAX_SCHEDULE_TIMEOUT will schedule + * the CPU away without a bound on the timeout. In this case the return + * value will be %MAX_SCHEDULE_TIMEOUT. + * + * Returns: 0 when the timer has expired otherwise the remaining time in + * jiffies will be returned. In all cases the return value is guaranteed + * to be non-negative. + */ +signed long __sched schedule_timeout(signed long timeout) +{ + struct process_timer timer; + unsigned long expire; + + switch (timeout) { + case MAX_SCHEDULE_TIMEOUT: + /* + * These two special cases are useful to be comfortable + * in the caller. Nothing more. We could take + * MAX_SCHEDULE_TIMEOUT from one of the negative value + * but I' d like to return a valid offset (>=0) to allow + * the caller to do everything it want with the retval. + */ + schedule(); + goto out; + default: + /* + * Another bit of PARANOID. Note that the retval will be + * 0 since no piece of kernel is supposed to do a check + * for a negative retval of schedule_timeout() (since it + * should never happens anyway). You just have the printk() + * that will tell you if something is gone wrong and where. + */ + if (timeout < 0) { + pr_err("%s: wrong timeout value %lx\n", __func__, timeout); + dump_stack(); + __set_current_state(TASK_RUNNING); + goto out; + } + } + + expire = timeout + jiffies; + + timer.task = current; + timer_setup_on_stack(&timer.timer, process_timeout, 0); + timer.timer.expires = expire; + add_timer(&timer.timer); + schedule(); + del_timer_sync(&timer.timer); + + /* Remove the timer from the object tracker */ + destroy_timer_on_stack(&timer.timer); + + timeout = expire - jiffies; + + out: + return timeout < 0 ? 0 : timeout; +} +EXPORT_SYMBOL(schedule_timeout); + +/* + * __set_current_state() can be used in schedule_timeout_*() functions, because + * schedule_timeout() calls schedule() unconditionally. + */ + +/** + * schedule_timeout_interruptible - sleep until timeout (interruptible) + * @timeout: timeout value in jiffies + * + * See schedule_timeout() for details. + * + * Task state is set to TASK_INTERRUPTIBLE before starting the timeout. + */ +signed long __sched schedule_timeout_interruptible(signed long timeout) +{ + __set_current_state(TASK_INTERRUPTIBLE); + return schedule_timeout(timeout); +} +EXPORT_SYMBOL(schedule_timeout_interruptible); + +/** + * schedule_timeout_killable - sleep until timeout (killable) + * @timeout: timeout value in jiffies + * + * See schedule_timeout() for details. + * + * Task state is set to TASK_KILLABLE before starting the timeout. + */ +signed long __sched schedule_timeout_killable(signed long timeout) +{ + __set_current_state(TASK_KILLABLE); + return schedule_timeout(timeout); +} +EXPORT_SYMBOL(schedule_timeout_killable); + +/** + * schedule_timeout_uninterruptible - sleep until timeout (uninterruptible) + * @timeout: timeout value in jiffies + * + * See schedule_timeout() for details. + * + * Task state is set to TASK_UNINTERRUPTIBLE before starting the timeout. + */ +signed long __sched schedule_timeout_uninterruptible(signed long timeout) +{ + __set_current_state(TASK_UNINTERRUPTIBLE); + return schedule_timeout(timeout); +} +EXPORT_SYMBOL(schedule_timeout_uninterruptible); + +/** + * schedule_timeout_idle - sleep until timeout (idle) + * @timeout: timeout value in jiffies + * + * See schedule_timeout() for details. + * + * Task state is set to TASK_IDLE before starting the timeout. It is similar to + * schedule_timeout_uninterruptible(), except this task will not contribute to + * load average. + */ +signed long __sched schedule_timeout_idle(signed long timeout) +{ + __set_current_state(TASK_IDLE); + return schedule_timeout(timeout); +} +EXPORT_SYMBOL(schedule_timeout_idle); + +/** + * schedule_hrtimeout_range_clock - sleep until timeout + * @expires: timeout value (ktime_t) + * @delta: slack in expires timeout (ktime_t) + * @mode: timer mode + * @clock_id: timer clock to be used + * + * Details are explained in schedule_hrtimeout_range() function description as + * this function is commonly used. + */ +int __sched schedule_hrtimeout_range_clock(ktime_t *expires, u64 delta, + const enum hrtimer_mode mode, clockid_t clock_id) +{ + struct hrtimer_sleeper t; + + /* + * Optimize when a zero timeout value is given. It does not + * matter whether this is an absolute or a relative time. + */ + if (expires && *expires == 0) { + __set_current_state(TASK_RUNNING); + return 0; + } + + /* + * A NULL parameter means "infinite" + */ + if (!expires) { + schedule(); + return -EINTR; + } + + hrtimer_setup_sleeper_on_stack(&t, clock_id, mode); + hrtimer_set_expires_range_ns(&t.timer, *expires, delta); + hrtimer_sleeper_start_expires(&t, mode); + + if (likely(t.task)) + schedule(); + + hrtimer_cancel(&t.timer); + destroy_hrtimer_on_stack(&t.timer); + + __set_current_state(TASK_RUNNING); + + return !t.task ? 0 : -EINTR; +} +EXPORT_SYMBOL_GPL(schedule_hrtimeout_range_clock); + +/** + * schedule_hrtimeout_range - sleep until timeout + * @expires: timeout value (ktime_t) + * @delta: slack in expires timeout (ktime_t) + * @mode: timer mode + * + * Make the current task sleep until the given expiry time has + * elapsed. The routine will return immediately unless + * the current task state has been set (see set_current_state()). + * + * The @delta argument gives the kernel the freedom to schedule the + * actual wakeup to a time that is both power and performance friendly + * for regular (non RT/DL) tasks. + * The kernel give the normal best effort behavior for "@expires+@delta", + * but may decide to fire the timer earlier, but no earlier than @expires. + * + * You can set the task state as follows - + * + * %TASK_UNINTERRUPTIBLE - at least @timeout time is guaranteed to + * pass before the routine returns unless the current task is explicitly + * woken up, (e.g. by wake_up_process()). + * + * %TASK_INTERRUPTIBLE - the routine may return early if a signal is + * delivered to the current task or the current task is explicitly woken + * up. + * + * The current task state is guaranteed to be TASK_RUNNING when this + * routine returns. + * + * Returns: 0 when the timer has expired. If the task was woken before the + * timer expired by a signal (only possible in state TASK_INTERRUPTIBLE) or + * by an explicit wakeup, it returns -EINTR. + */ +int __sched schedule_hrtimeout_range(ktime_t *expires, u64 delta, + const enum hrtimer_mode mode) +{ + return schedule_hrtimeout_range_clock(expires, delta, mode, + CLOCK_MONOTONIC); +} +EXPORT_SYMBOL_GPL(schedule_hrtimeout_range); + +/** + * schedule_hrtimeout - sleep until timeout + * @expires: timeout value (ktime_t) + * @mode: timer mode + * + * See schedule_hrtimeout_range() for details. @delta argument of + * schedule_hrtimeout_range() is set to 0 and has therefore no impact. + */ +int __sched schedule_hrtimeout(ktime_t *expires, const enum hrtimer_mode mode) +{ + return schedule_hrtimeout_range(expires, 0, mode); +} +EXPORT_SYMBOL_GPL(schedule_hrtimeout); + +/** + * msleep - sleep safely even with waitqueue interruptions + * @msecs: Requested sleep duration in milliseconds + * + * msleep() uses jiffy based timeouts for the sleep duration. Because of the + * design of the timer wheel, the maximum additional percentage delay (slack) is + * 12.5%. This is only valid for timers which will end up in level 1 or a higher + * level of the timer wheel. For explanation of those 12.5% please check the + * detailed description about the basics of the timer wheel. + * + * The slack of timers which will end up in level 0 depends on sleep duration + * (msecs) and HZ configuration and can be calculated in the following way (with + * the timer wheel design restriction that the slack is not less than 12.5%): + * + * ``slack = MSECS_PER_TICK / msecs`` + * + * When the allowed slack of the callsite is known, the calculation could be + * turned around to find the minimal allowed sleep duration to meet the + * constraints. For example: + * + * * ``HZ=1000`` with ``slack=25%``: ``MSECS_PER_TICK / slack = 1 / (1/4) = 4``: + * all sleep durations greater or equal 4ms will meet the constraints. + * * ``HZ=1000`` with ``slack=12.5%``: ``MSECS_PER_TICK / slack = 1 / (1/8) = 8``: + * all sleep durations greater or equal 8ms will meet the constraints. + * * ``HZ=250`` with ``slack=25%``: ``MSECS_PER_TICK / slack = 4 / (1/4) = 16``: + * all sleep durations greater or equal 16ms will meet the constraints. + * * ``HZ=250`` with ``slack=12.5%``: ``MSECS_PER_TICK / slack = 4 / (1/8) = 32``: + * all sleep durations greater or equal 32ms will meet the constraints. + * + * See also the signal aware variant msleep_interruptible(). + */ +void msleep(unsigned int msecs) +{ + unsigned long timeout = msecs_to_jiffies(msecs); + + while (timeout) + timeout = schedule_timeout_uninterruptible(timeout); +} +EXPORT_SYMBOL(msleep); + +/** + * msleep_interruptible - sleep waiting for signals + * @msecs: Requested sleep duration in milliseconds + * + * See msleep() for some basic information. + * + * The difference between msleep() and msleep_interruptible() is that the sleep + * could be interrupted by a signal delivery and then returns early. + * + * Returns: The remaining time of the sleep duration transformed to msecs (see + * schedule_timeout() for details). + */ +unsigned long msleep_interruptible(unsigned int msecs) +{ + unsigned long timeout = msecs_to_jiffies(msecs); + + while (timeout && !signal_pending(current)) + timeout = schedule_timeout_interruptible(timeout); + return jiffies_to_msecs(timeout); +} +EXPORT_SYMBOL(msleep_interruptible); + +/** + * usleep_range_state - Sleep for an approximate time in a given state + * @min: Minimum time in usecs to sleep + * @max: Maximum time in usecs to sleep + * @state: State of the current task that will be while sleeping + * + * usleep_range_state() sleeps at least for the minimum specified time but not + * longer than the maximum specified amount of time. The range might reduce + * power usage by allowing hrtimers to coalesce an already scheduled interrupt + * with this hrtimer. In the worst case, an interrupt is scheduled for the upper + * bound. + * + * The sleeping task is set to the specified state before starting the sleep. + * + * In non-atomic context where the exact wakeup time is flexible, use + * usleep_range() or its variants instead of udelay(). The sleep improves + * responsiveness by avoiding the CPU-hogging busy-wait of udelay(). + */ +void __sched usleep_range_state(unsigned long min, unsigned long max, unsigned int state) +{ + ktime_t exp = ktime_add_us(ktime_get(), min); + u64 delta = (u64)(max - min) * NSEC_PER_USEC; + + if (WARN_ON_ONCE(max < min)) + delta = 0; + + for (;;) { + __set_current_state(state); + /* Do not return before the requested sleep time has elapsed */ + if (!schedule_hrtimeout_range(&exp, delta, HRTIMER_MODE_ABS)) + break; + } +} +EXPORT_SYMBOL(usleep_range_state); diff --git a/kernel/time/test_udelay.c b/kernel/time/test_udelay.c index 77c63005dc4e..783f2297111b 100644 --- a/kernel/time/test_udelay.c +++ b/kernel/time/test_udelay.c @@ -21,7 +21,6 @@ #define DEBUGFS_FILENAME "udelay_test" static DEFINE_MUTEX(udelay_test_lock); -static struct dentry *udelay_test_debugfs_file; static int udelay_test_usecs; static int udelay_test_iterations = DEFAULT_ITERATIONS; @@ -138,8 +137,8 @@ static const struct file_operations udelay_test_debugfs_ops = { static int __init udelay_test_init(void) { mutex_lock(&udelay_test_lock); - udelay_test_debugfs_file = debugfs_create_file(DEBUGFS_FILENAME, - S_IRUSR, NULL, NULL, &udelay_test_debugfs_ops); + debugfs_create_file(DEBUGFS_FILENAME, S_IRUSR, NULL, NULL, + &udelay_test_debugfs_ops); mutex_unlock(&udelay_test_lock); return 0; @@ -150,11 +149,12 @@ module_init(udelay_test_init); static void __exit udelay_test_exit(void) { mutex_lock(&udelay_test_lock); - debugfs_remove(udelay_test_debugfs_file); + debugfs_lookup_and_remove(DEBUGFS_FILENAME, NULL); mutex_unlock(&udelay_test_lock); } module_exit(udelay_test_exit); +MODULE_DESCRIPTION("udelay test module"); MODULE_AUTHOR("David Riley <davidriley@chromium.org>"); MODULE_LICENSE("GPL"); diff --git a/kernel/time/tick-broadcast-hrtimer.c b/kernel/time/tick-broadcast-hrtimer.c index b5a65e212df2..e28f9210f8a1 100644 --- a/kernel/time/tick-broadcast-hrtimer.c +++ b/kernel/time/tick-broadcast-hrtimer.c @@ -53,28 +53,23 @@ static int bc_set_next(ktime_t expires, struct clock_event_device *bc) * reasons. * * Each caller tries to arm the hrtimer on its own CPU, but if the - * hrtimer callbback function is currently running, then + * hrtimer callback function is currently running, then * hrtimer_start() cannot move it and the timer stays on the CPU on * which it is assigned at the moment. + */ + hrtimer_start(&bctimer, expires, HRTIMER_MODE_ABS_PINNED_HARD); + /* + * The core tick broadcast mode expects bc->bound_on to be set + * correctly to prevent a CPU which has the broadcast hrtimer + * armed from going deep idle. * - * As this can be called from idle code, the hrtimer_start() - * invocation has to be wrapped with RCU_NONIDLE() as - * hrtimer_start() can call into tracing. + * As tick_broadcast_lock is held, nothing can change the cpu + * base which was just established in hrtimer_start() above. So + * the below access is safe even without holding the hrtimer + * base lock. */ - RCU_NONIDLE( { - hrtimer_start(&bctimer, expires, HRTIMER_MODE_ABS_PINNED_HARD); - /* - * The core tick broadcast mode expects bc->bound_on to be set - * correctly to prevent a CPU which has the broadcast hrtimer - * armed from going deep idle. - * - * As tick_broadcast_lock is held, nothing can change the cpu - * base which was just established in hrtimer_start() above. So - * the below access is safe even without holding the hrtimer - * base lock. - */ - bc->bound_on = bctimer.base->cpu_base->cpu; - } ); + bc->bound_on = bctimer.base->cpu_base->cpu; + return 0; } diff --git a/kernel/time/tick-broadcast.c b/kernel/time/tick-broadcast.c index e51778c312f1..0207868c8b4d 100644 --- a/kernel/time/tick-broadcast.c +++ b/kernel/time/tick-broadcast.c @@ -33,14 +33,17 @@ static int tick_broadcast_forced; static __cacheline_aligned_in_smp DEFINE_RAW_SPINLOCK(tick_broadcast_lock); #ifdef CONFIG_TICK_ONESHOT -static void tick_broadcast_setup_oneshot(struct clock_event_device *bc); +static DEFINE_PER_CPU(struct clock_event_device *, tick_oneshot_wakeup_device); + +static void tick_broadcast_setup_oneshot(struct clock_event_device *bc, bool from_periodic); static void tick_broadcast_clear_oneshot(int cpu); static void tick_resume_broadcast_oneshot(struct clock_event_device *bc); # ifdef CONFIG_HOTPLUG_CPU static void tick_broadcast_oneshot_offline(unsigned int cpu); # endif #else -static inline void tick_broadcast_setup_oneshot(struct clock_event_device *bc) { BUG(); } +static inline void +tick_broadcast_setup_oneshot(struct clock_event_device *bc, bool from_periodic) { BUG(); } static inline void tick_broadcast_clear_oneshot(int cpu) { } static inline void tick_resume_broadcast_oneshot(struct clock_event_device *bc) { } # ifdef CONFIG_HOTPLUG_CPU @@ -61,6 +64,13 @@ struct cpumask *tick_get_broadcast_mask(void) return tick_broadcast_mask; } +static struct clock_event_device *tick_get_oneshot_wakeup_device(int cpu); + +const struct clock_event_device *tick_get_wakeup_device(int cpu) +{ + return tick_get_oneshot_wakeup_device(cpu); +} + /* * Start the device in periodic mode */ @@ -88,13 +98,75 @@ static bool tick_check_broadcast_device(struct clock_event_device *curdev, return !curdev || newdev->rating > curdev->rating; } +#ifdef CONFIG_TICK_ONESHOT +static struct clock_event_device *tick_get_oneshot_wakeup_device(int cpu) +{ + return per_cpu(tick_oneshot_wakeup_device, cpu); +} + +static void tick_oneshot_wakeup_handler(struct clock_event_device *wd) +{ + /* + * If we woke up early and the tick was reprogrammed in the + * meantime then this may be spurious but harmless. + */ + tick_receive_broadcast(); +} + +static bool tick_set_oneshot_wakeup_device(struct clock_event_device *newdev, + int cpu) +{ + struct clock_event_device *curdev = tick_get_oneshot_wakeup_device(cpu); + + if (!newdev) + goto set_device; + + if ((newdev->features & CLOCK_EVT_FEAT_DUMMY) || + (newdev->features & CLOCK_EVT_FEAT_C3STOP)) + return false; + + if (!(newdev->features & CLOCK_EVT_FEAT_PERCPU) || + !(newdev->features & CLOCK_EVT_FEAT_ONESHOT)) + return false; + + if (!cpumask_equal(newdev->cpumask, cpumask_of(cpu))) + return false; + + if (curdev && newdev->rating <= curdev->rating) + return false; + + if (!try_module_get(newdev->owner)) + return false; + + newdev->event_handler = tick_oneshot_wakeup_handler; +set_device: + clockevents_exchange_device(curdev, newdev); + per_cpu(tick_oneshot_wakeup_device, cpu) = newdev; + return true; +} +#else +static struct clock_event_device *tick_get_oneshot_wakeup_device(int cpu) +{ + return NULL; +} + +static bool tick_set_oneshot_wakeup_device(struct clock_event_device *newdev, + int cpu) +{ + return false; +} +#endif + /* * Conditionally install/replace broadcast device */ -void tick_install_broadcast_device(struct clock_event_device *dev) +void tick_install_broadcast_device(struct clock_event_device *dev, int cpu) { struct clock_event_device *cur = tick_broadcast_device.evtdev; + if (tick_set_oneshot_wakeup_device(dev, cpu)) + return; + if (!tick_check_broadcast_device(cur, dev)) return; @@ -107,6 +179,19 @@ void tick_install_broadcast_device(struct clock_event_device *dev) tick_broadcast_device.evtdev = dev; if (!cpumask_empty(tick_broadcast_mask)) tick_broadcast_start_periodic(dev); + + if (!(dev->features & CLOCK_EVT_FEAT_ONESHOT)) + return; + + /* + * If the system already runs in oneshot mode, switch the newly + * registered broadcast device to oneshot mode explicitly. + */ + if (tick_broadcast_oneshot_active()) { + tick_broadcast_switch_to_oneshot(); + return; + } + /* * Inform all cpus about this. We might be in a situation * where we did not switch to oneshot mode because the per cpu @@ -115,8 +200,7 @@ void tick_install_broadcast_device(struct clock_event_device *dev) * notification the systems stays stuck in periodic mode * forever. */ - if (dev->features & CLOCK_EVT_FEAT_ONESHOT) - tick_clock_notify(); + tick_clock_notify(); } /* @@ -157,7 +241,7 @@ static void tick_device_setup_broadcast_func(struct clock_event_device *dev) } /* - * Check, if the device is disfunctional and a place holder, which + * Check, if the device is dysfunctional and a placeholder, which * needs to be handled by the broadcast device. */ int tick_device_uses_broadcast(struct clock_event_device *dev, int cpu) @@ -181,7 +265,7 @@ int tick_device_uses_broadcast(struct clock_event_device *dev, int cpu) if (tick_broadcast_device.mode == TICKDEV_MODE_PERIODIC) tick_broadcast_start_periodic(bc); else - tick_broadcast_setup_oneshot(bc); + tick_broadcast_setup_oneshot(bc, false); ret = 1; } else { /* @@ -241,7 +325,6 @@ int tick_device_uses_broadcast(struct clock_event_device *dev, int cpu) return ret; } -#ifdef CONFIG_GENERIC_CLOCKEVENTS_BROADCAST int tick_receive_broadcast(void) { struct tick_device *td = this_cpu_ptr(&tick_cpu_device); @@ -256,7 +339,6 @@ int tick_receive_broadcast(void) evt->event_handler(evt); return 0; } -#endif /* * Broadcast the event to the cpus, which are set in the mask (mangled). @@ -331,7 +413,7 @@ static void tick_handle_periodic_broadcast(struct clock_event_device *dev) bc_local = tick_do_periodic_broadcast(); if (clockevent_state_oneshot(dev)) { - ktime_t next = ktime_add(dev->next_event, tick_period); + ktime_t next = ktime_add_ns(dev->next_event, TICK_NSEC); clockevents_program_event(dev, next, true); } @@ -381,7 +463,7 @@ void tick_broadcast_control(enum tick_broadcast_mode mode) switch (mode) { case TICK_BROADCAST_FORCE: tick_broadcast_forced = 1; - /* fall through */ + fallthrough; case TICK_BROADCAST_ON: cpumask_set_cpu(cpu, tick_broadcast_on); if (!cpumask_test_and_set_cpu(cpu, tick_broadcast_mask)) { @@ -391,7 +473,7 @@ void tick_broadcast_control(enum tick_broadcast_mode mode) * - the broadcast device exists * - the broadcast device is not a hrtimer based one * - the broadcast device is in periodic mode to - * avoid a hickup during switch to oneshot mode + * avoid a hiccup during switch to oneshot mode */ if (bc && !(bc->features & CLOCK_EVT_FEAT_HRTIMER) && tick_broadcast_device.mode == TICKDEV_MODE_PERIODIC) @@ -419,7 +501,7 @@ void tick_broadcast_control(enum tick_broadcast_mode mode) if (tick_broadcast_device.mode == TICKDEV_MODE_PERIODIC) tick_broadcast_start_periodic(bc); else - tick_broadcast_setup_oneshot(bc); + tick_broadcast_setup_oneshot(bc, false); } } out: @@ -541,9 +623,13 @@ struct cpumask *tick_get_broadcast_oneshot_mask(void) * to avoid a deep idle transition as we are about to get the * broadcast IPI right away. */ -int tick_check_broadcast_expired(void) +noinstr int tick_check_broadcast_expired(void) { +#ifdef _ASM_GENERIC_BITOPS_INSTRUMENTED_NON_ATOMIC_H + return arch_test_bit(smp_processor_id(), cpumask_bits(tick_broadcast_force_mask)); +#else return cpumask_test_cpu(smp_processor_id(), tick_broadcast_force_mask); +#endif } /* @@ -707,24 +793,16 @@ static void broadcast_shutdown_local(struct clock_event_device *bc, clockevents_switch_state(dev, CLOCK_EVT_STATE_SHUTDOWN); } -int __tick_broadcast_oneshot_control(enum tick_broadcast_state state) +static int ___tick_broadcast_oneshot_control(enum tick_broadcast_state state, + struct tick_device *td, + int cpu) { - struct clock_event_device *bc, *dev; - int cpu, ret = 0; + struct clock_event_device *bc, *dev = td->evtdev; + int ret = 0; ktime_t now; - /* - * If there is no broadcast device, tell the caller not to go - * into deep idle. - */ - if (!tick_broadcast_device.evtdev) - return -EBUSY; - - dev = this_cpu_ptr(&tick_cpu_device)->evtdev; - raw_spin_lock(&tick_broadcast_lock); bc = tick_broadcast_device.evtdev; - cpu = smp_processor_id(); if (state == TICK_BROADCAST_ENTER) { /* @@ -853,6 +931,53 @@ out: return ret; } +static int tick_oneshot_wakeup_control(enum tick_broadcast_state state, + struct tick_device *td, + int cpu) +{ + struct clock_event_device *dev, *wd; + + dev = td->evtdev; + if (td->mode != TICKDEV_MODE_ONESHOT) + return -EINVAL; + + wd = tick_get_oneshot_wakeup_device(cpu); + if (!wd) + return -ENODEV; + + switch (state) { + case TICK_BROADCAST_ENTER: + clockevents_switch_state(dev, CLOCK_EVT_STATE_ONESHOT_STOPPED); + clockevents_switch_state(wd, CLOCK_EVT_STATE_ONESHOT); + clockevents_program_event(wd, dev->next_event, 1); + break; + case TICK_BROADCAST_EXIT: + /* We may have transitioned to oneshot mode while idle */ + if (clockevent_get_state(wd) != CLOCK_EVT_STATE_ONESHOT) + return -ENODEV; + } + + return 0; +} + +int __tick_broadcast_oneshot_control(enum tick_broadcast_state state) +{ + struct tick_device *td = this_cpu_ptr(&tick_cpu_device); + int cpu = smp_processor_id(); + + if (!tick_oneshot_wakeup_control(state, td, cpu)) + return 0; + + if (tick_broadcast_device.evtdev) + return ___tick_broadcast_oneshot_control(state, td, cpu); + + /* + * If there is no broadcast or wakeup device, tell the caller not + * to go into deep idle. + */ + return -EBUSY; +} + /* * Reset the one shot broadcast for a cpu * @@ -877,50 +1002,122 @@ static void tick_broadcast_init_next_event(struct cpumask *mask, } } +static inline ktime_t tick_get_next_period(void) +{ + ktime_t next; + + /* + * Protect against concurrent updates (store /load tearing on + * 32bit). It does not matter if the time is already in the + * past. The broadcast device which is about to be programmed will + * fire in any case. + */ + raw_spin_lock(&jiffies_lock); + next = tick_next_period; + raw_spin_unlock(&jiffies_lock); + return next; +} + /** * tick_broadcast_setup_oneshot - setup the broadcast device + * @bc: the broadcast device + * @from_periodic: true if called from periodic mode */ -static void tick_broadcast_setup_oneshot(struct clock_event_device *bc) +static void tick_broadcast_setup_oneshot(struct clock_event_device *bc, + bool from_periodic) { int cpu = smp_processor_id(); + ktime_t nexttick = 0; if (!bc) return; - /* Set it up only once ! */ - if (bc->event_handler != tick_handle_oneshot_broadcast) { - int was_periodic = clockevent_state_periodic(bc); - - bc->event_handler = tick_handle_oneshot_broadcast; - + /* + * When the broadcast device was switched to oneshot by the first + * CPU handling the NOHZ change, the other CPUs will reach this + * code via hrtimer_run_queues() -> tick_check_oneshot_change() + * too. Set up the broadcast device only once! + */ + if (bc->event_handler == tick_handle_oneshot_broadcast) { /* - * We must be careful here. There might be other CPUs - * waiting for periodic broadcast. We need to set the - * oneshot_mask bits for those and program the - * broadcast device to fire. + * The CPU which switched from periodic to oneshot mode + * set the broadcast oneshot bit for all other CPUs which + * are in the general (periodic) broadcast mask to ensure + * that CPUs which wait for the periodic broadcast are + * woken up. + * + * Clear the bit for the local CPU as the set bit would + * prevent the first tick_broadcast_enter() after this CPU + * switched to oneshot state to program the broadcast + * device. + * + * This code can also be reached via tick_broadcast_control(), + * but this cannot avoid the tick_broadcast_clear_oneshot() + * as that would break the periodic to oneshot transition of + * secondary CPUs. But that's harmless as the below only + * clears already cleared bits. */ + tick_broadcast_clear_oneshot(cpu); + return; + } + + + bc->event_handler = tick_handle_oneshot_broadcast; + bc->next_event = KTIME_MAX; + + /* + * When the tick mode is switched from periodic to oneshot it must + * be ensured that CPUs which are waiting for periodic broadcast + * get their wake-up at the next tick. This is achieved by ORing + * tick_broadcast_mask into tick_broadcast_oneshot_mask. + * + * For other callers, e.g. broadcast device replacement, + * tick_broadcast_oneshot_mask must not be touched as this would + * set bits for CPUs which are already NOHZ, but not idle. Their + * next tick_broadcast_enter() would observe the bit set and fail + * to update the expiry time and the broadcast event device. + */ + if (from_periodic) { cpumask_copy(tmpmask, tick_broadcast_mask); + /* Remove the local CPU as it is obviously not idle */ cpumask_clear_cpu(cpu, tmpmask); - cpumask_or(tick_broadcast_oneshot_mask, - tick_broadcast_oneshot_mask, tmpmask); - - if (was_periodic && !cpumask_empty(tmpmask)) { - clockevents_switch_state(bc, CLOCK_EVT_STATE_ONESHOT); - tick_broadcast_init_next_event(tmpmask, - tick_next_period); - tick_broadcast_set_event(bc, cpu, tick_next_period); - } else - bc->next_event = KTIME_MAX; - } else { + cpumask_or(tick_broadcast_oneshot_mask, tick_broadcast_oneshot_mask, tmpmask); + /* - * The first cpu which switches to oneshot mode sets - * the bit for all other cpus which are in the general - * (periodic) broadcast mask. So the bit is set and - * would prevent the first broadcast enter after this - * to program the bc device. + * Ensure that the oneshot broadcast handler will wake the + * CPUs which are still waiting for periodic broadcast. */ - tick_broadcast_clear_oneshot(cpu); + nexttick = tick_get_next_period(); + tick_broadcast_init_next_event(tmpmask, nexttick); + + /* + * If the underlying broadcast clock event device is + * already in oneshot state, then there is nothing to do. + * The device was already armed for the next tick + * in tick_handle_broadcast_periodic() + */ + if (clockevent_state_oneshot(bc)) + return; } + + /* + * When switching from periodic to oneshot mode arm the broadcast + * device for the next tick. + * + * If the broadcast device has been replaced in oneshot mode and + * the oneshot broadcast mask is not empty, then arm it to expire + * immediately in order to reevaluate the next expiring timer. + * @nexttick is 0 and therefore in the past which will cause the + * clockevent code to force an event. + * + * For both cases the programming can be avoided when the oneshot + * broadcast mask is empty. + * + * tick_broadcast_set_event() implicitly switches the broadcast + * device to oneshot state. + */ + if (!cpumask_empty(tick_broadcast_oneshot_mask)) + tick_broadcast_set_event(bc, cpu, nexttick); } /* @@ -929,14 +1126,16 @@ static void tick_broadcast_setup_oneshot(struct clock_event_device *bc) void tick_broadcast_switch_to_oneshot(void) { struct clock_event_device *bc; + enum tick_device_mode oldmode; unsigned long flags; raw_spin_lock_irqsave(&tick_broadcast_lock, flags); + oldmode = tick_broadcast_device.mode; tick_broadcast_device.mode = TICKDEV_MODE_ONESHOT; bc = tick_broadcast_device.evtdev; if (bc) - tick_broadcast_setup_oneshot(bc); + tick_broadcast_setup_oneshot(bc, oldmode == TICKDEV_MODE_PERIODIC); raw_spin_unlock_irqrestore(&tick_broadcast_lock, flags); } @@ -951,6 +1150,30 @@ void hotplug_cpu__broadcast_tick_pull(int deadcpu) bc = tick_broadcast_device.evtdev; if (bc && broadcast_needs_cpu(bc, deadcpu)) { + /* + * If the broadcast force bit of the current CPU is set, + * then the current CPU has not yet reprogrammed the local + * timer device to avoid a ping-pong race. See + * ___tick_broadcast_oneshot_control(). + * + * If the broadcast device is hrtimer based then + * programming the broadcast event below does not have any + * effect because the local clockevent device is not + * running and not programmed because the broadcast event + * is not earlier than the pending event of the local clock + * event device. As a consequence all CPUs waiting for a + * broadcast event are stuck forever. + * + * Detect this condition and reprogram the cpu local timer + * device to avoid the starvation. + */ + if (tick_check_broadcast_expired()) { + struct tick_device *td = this_cpu_ptr(&tick_cpu_device); + + cpumask_clear_cpu(smp_processor_id(), tick_broadcast_force_mask); + tick_program_event(td->evtdev->next_event, 1); + } + /* This moves the broadcast assignment to this CPU: */ clockevents_program_event(bc, bc->next_event, 1); } @@ -962,6 +1185,9 @@ void hotplug_cpu__broadcast_tick_pull(int deadcpu) */ static void tick_broadcast_oneshot_offline(unsigned int cpu) { + if (tick_get_oneshot_wakeup_device(cpu)) + tick_set_oneshot_wakeup_device(NULL, cpu); + /* * Clear the broadcast masks for the dead cpu, but do not stop * the broadcast device! diff --git a/kernel/time/tick-common.c b/kernel/time/tick-common.c index 6c9c342dd0e5..a47bcf71defc 100644 --- a/kernel/time/tick-common.c +++ b/kernel/time/tick-common.c @@ -7,6 +7,7 @@ * Copyright(C) 2005-2007, Red Hat, Inc., Ingo Molnar * Copyright(C) 2006-2007, Timesys Corp., Thomas Gleixner */ +#include <linux/compiler.h> #include <linux/cpu.h> #include <linux/err.h> #include <linux/hrtimer.h> @@ -27,10 +28,11 @@ */ DEFINE_PER_CPU(struct tick_device, tick_cpu_device); /* - * Tick next event: keeps track of the tick time + * Tick next event: keeps track of the tick time. It's updated by the + * CPU which handles the tick and protected by jiffies_lock. There is + * no requirement to write hold the jiffies seqcount for it. */ ktime_t tick_next_period; -ktime_t tick_period; /* * tick_do_timer_cpu is a timer core internal variable which holds the CPU NR @@ -83,12 +85,12 @@ int tick_is_oneshot_available(void) */ static void tick_periodic(int cpu) { - if (tick_do_timer_cpu == cpu) { + if (READ_ONCE(tick_do_timer_cpu) == cpu) { raw_spin_lock(&jiffies_lock); write_seqcount_begin(&jiffies_seq); /* Keep track of the next tick event */ - tick_next_period = ktime_add(tick_next_period, tick_period); + tick_next_period = ktime_add_ns(tick_next_period, TICK_NSEC); do_timer(1); write_seqcount_end(&jiffies_seq); @@ -110,15 +112,13 @@ void tick_handle_periodic(struct clock_event_device *dev) tick_periodic(cpu); -#if defined(CONFIG_HIGH_RES_TIMERS) || defined(CONFIG_NO_HZ_COMMON) /* * The cpu might have transitioned to HIGHRES or NOHZ mode via * update_process_times() -> run_local_timers() -> * hrtimer_run_queues(). */ - if (dev->event_handler != tick_handle_periodic) + if (IS_ENABLED(CONFIG_TICK_ONESHOT) && dev->event_handler != tick_handle_periodic) return; -#endif if (!clockevent_state_oneshot(dev)) return; @@ -127,7 +127,7 @@ void tick_handle_periodic(struct clock_event_device *dev) * Setup the next period for devices, which do not have * periodic mode: */ - next = ktime_add(next, tick_period); + next = ktime_add_ns(next, TICK_NSEC); if (!clockevents_program_event(dev, next, false)) return; @@ -173,31 +173,11 @@ void tick_setup_periodic(struct clock_event_device *dev, int broadcast) for (;;) { if (!clockevents_program_event(dev, next, false)) return; - next = ktime_add(next, tick_period); + next = ktime_add_ns(next, TICK_NSEC); } } } -#ifdef CONFIG_NO_HZ_FULL -static void giveup_do_timer(void *info) -{ - int cpu = *(unsigned int *)info; - - WARN_ON(tick_do_timer_cpu != smp_processor_id()); - - tick_do_timer_cpu = cpu; -} - -static void tick_take_do_timer_from_boot(void) -{ - int cpu = smp_processor_id(); - int from = tick_do_timer_boot_cpu; - - if (from >= 0 && from != cpu) - smp_call_function_single(from, giveup_do_timer, &cpu, 1); -} -#endif - /* * Setup the tick device */ @@ -216,26 +196,30 @@ static void tick_setup_device(struct tick_device *td, * If no cpu took the do_timer update, assign it to * this cpu: */ - if (tick_do_timer_cpu == TICK_DO_TIMER_BOOT) { - tick_do_timer_cpu = cpu; - + if (READ_ONCE(tick_do_timer_cpu) == TICK_DO_TIMER_BOOT) { + WRITE_ONCE(tick_do_timer_cpu, cpu); tick_next_period = ktime_get(); - tick_period = NSEC_PER_SEC / HZ; #ifdef CONFIG_NO_HZ_FULL /* - * The boot CPU may be nohz_full, in which case set - * tick_do_timer_boot_cpu so the first housekeeping - * secondary that comes up will take do_timer from - * us. + * The boot CPU may be nohz_full, in which case the + * first housekeeping secondary will take do_timer() + * from it. */ if (tick_nohz_full_cpu(cpu)) tick_do_timer_boot_cpu = cpu; - } else if (tick_do_timer_boot_cpu != -1 && - !tick_nohz_full_cpu(cpu)) { - tick_take_do_timer_from_boot(); + } else if (tick_do_timer_boot_cpu != -1 && !tick_nohz_full_cpu(cpu)) { tick_do_timer_boot_cpu = -1; - WARN_ON(tick_do_timer_cpu != cpu); + /* + * The boot CPU will stay in periodic (NOHZ disabled) + * mode until clocksource_done_booting() called after + * smp_init() selects a high resolution clocksource and + * timekeeping_notify() kicks the NOHZ stuff alive. + * + * So this WRITE_ONCE can only race with the READ_ONCE + * check in tick_periodic() but this race is harmless. + */ + WRITE_ONCE(tick_do_timer_cpu, cpu); #endif } @@ -348,12 +332,7 @@ void tick_check_new_device(struct clock_event_device *newdev) td = &per_cpu(tick_cpu_device, cpu); curdev = td->evtdev; - /* cpu local device ? */ - if (!tick_check_percpu(curdev, newdev, cpu)) - goto out_bc; - - /* Preference decision */ - if (!tick_check_preferred(curdev, newdev)) + if (!tick_check_replacement(curdev, newdev)) goto out_bc; if (!try_module_get(newdev->owner)) @@ -378,7 +357,7 @@ out_bc: /* * Can the new device be used as a broadcast device ? */ - tick_install_broadcast_device(newdev); + tick_install_broadcast_device(newdev, cpu); } /** @@ -404,20 +383,31 @@ int tick_broadcast_oneshot_control(enum tick_broadcast_state state) EXPORT_SYMBOL_GPL(tick_broadcast_oneshot_control); #ifdef CONFIG_HOTPLUG_CPU +void tick_assert_timekeeping_handover(void) +{ + WARN_ON_ONCE(tick_do_timer_cpu == smp_processor_id()); +} /* - * Transfer the do_timer job away from a dying cpu. - * - * Called with interrupts disabled. Not locking required. If - * tick_do_timer_cpu is owned by this cpu, nothing can change it. + * Stop the tick and transfer the timekeeping job away from a dying cpu. */ -void tick_handover_do_timer(void) +int tick_cpu_dying(unsigned int dying_cpu) { - if (tick_do_timer_cpu == smp_processor_id()) { - int cpu = cpumask_first(cpu_online_mask); + /* + * If the current CPU is the timekeeper, it's the only one that can + * safely hand over its duty. Also all online CPUs are in stop + * machine, guaranteed not to be idle, therefore there is no + * concurrency and it's safe to pick any online successor. + */ + if (tick_do_timer_cpu == dying_cpu) + tick_do_timer_cpu = cpumask_first(cpu_online_mask); - tick_do_timer_cpu = (cpu < nr_cpu_ids) ? cpu : - TICK_DO_TIMER_NONE; - } + /* Make sure the CPU won't try to retake the timekeeping duty */ + tick_sched_timer_dying(dying_cpu); + + /* Remove CPU from timer broadcasting */ + tick_offline_cpu(dying_cpu); + + return 0; } /* @@ -479,6 +469,13 @@ void tick_resume_local(void) else tick_resume_oneshot(); } + + /* + * Ensure that hrtimers are up to date and the clockevents device + * is reprogrammed correctly when high resolution timers are + * enabled. + */ + hrtimers_resume_local(); } /** diff --git a/kernel/time/tick-internal.h b/kernel/time/tick-internal.h index 7b2496136729..faac36de35b9 100644 --- a/kernel/time/tick-internal.h +++ b/kernel/time/tick-internal.h @@ -8,6 +8,11 @@ #include "timekeeping.h" #include "tick-sched.h" +struct timer_events { + u64 local; + u64 global; +}; + #ifdef CONFIG_GENERIC_CLOCKEVENTS # define TICK_DO_TIMER_NONE -1 @@ -15,12 +20,12 @@ DECLARE_PER_CPU(struct tick_device, tick_cpu_device); extern ktime_t tick_next_period; -extern ktime_t tick_period; extern int tick_do_timer_cpu __read_mostly; extern void tick_setup_periodic(struct clock_event_device *dev, int broadcast); extern void tick_handle_periodic(struct clock_event_device *dev); extern void tick_check_new_device(struct clock_event_device *dev); +extern void tick_offline_cpu(unsigned int cpu); extern void tick_shutdown(unsigned int cpu); extern void tick_suspend(void); extern void tick_resume(void); @@ -57,12 +62,11 @@ extern int clockevents_program_event(struct clock_event_device *dev, ktime_t expires, bool force); extern void clockevents_handle_noop(struct clock_event_device *dev); extern int __clockevents_update_freq(struct clock_event_device *dev, u32 freq); -extern ssize_t sysfs_get_uname(const char *buf, char *dst, size_t cnt); /* Broadcasting support */ # ifdef CONFIG_GENERIC_CLOCKEVENTS_BROADCAST extern int tick_device_uses_broadcast(struct clock_event_device *dev, int cpu); -extern void tick_install_broadcast_device(struct clock_event_device *dev); +extern void tick_install_broadcast_device(struct clock_event_device *dev, int cpu); extern int tick_is_broadcast_device(struct clock_event_device *dev); extern void tick_suspend_broadcast(void); extern void tick_resume_broadcast(void); @@ -72,8 +76,9 @@ extern void tick_set_periodic_handler(struct clock_event_device *dev, int broadc extern int tick_broadcast_update_freq(struct clock_event_device *dev, u32 freq); extern struct tick_device *tick_get_broadcast_device(void); extern struct cpumask *tick_get_broadcast_mask(void); +extern const struct clock_event_device *tick_get_wakeup_device(int cpu); # else /* !CONFIG_GENERIC_CLOCKEVENTS_BROADCAST: */ -static inline void tick_install_broadcast_device(struct clock_event_device *dev) { } +static inline void tick_install_broadcast_device(struct clock_event_device *dev, int cpu) { } static inline int tick_is_broadcast_device(struct clock_event_device *dev) { return 0; } static inline int tick_device_uses_broadcast(struct clock_event_device *dev, int cpu) { return 0; } static inline void tick_do_periodic_broadcast(struct clock_event_device *d) { } @@ -153,8 +158,16 @@ static inline void tick_nohz_init(void) { } #ifdef CONFIG_NO_HZ_COMMON extern unsigned long tick_nohz_active; extern void timers_update_nohz(void); +extern u64 get_jiffies_update(unsigned long *basej); # ifdef CONFIG_SMP extern struct static_key_false timers_migration_enabled; +extern void fetch_next_timer_interrupt_remote(unsigned long basej, u64 basem, + struct timer_events *tevt, + unsigned int cpu); +extern void timer_lock_remote_bases(unsigned int cpu); +extern void timer_unlock_remote_bases(unsigned int cpu); +extern bool timer_base_is_idle(void); +extern void timer_expire_remote(unsigned int cpu); # endif #else /* CONFIG_NO_HZ_COMMON */ static inline void timers_update_nohz(void) { } @@ -164,4 +177,39 @@ static inline void timers_update_nohz(void) { } DECLARE_PER_CPU(struct hrtimer_cpu_base, hrtimer_bases); extern u64 get_next_timer_interrupt(unsigned long basej, u64 basem); +u64 timer_base_try_to_set_idle(unsigned long basej, u64 basem, bool *idle); void timer_clear_idle(void); + +#define CLOCK_SET_WALL \ + (BIT(HRTIMER_BASE_REALTIME) | BIT(HRTIMER_BASE_REALTIME_SOFT) | \ + BIT(HRTIMER_BASE_TAI) | BIT(HRTIMER_BASE_TAI_SOFT)) + +#define CLOCK_SET_BOOT \ + (BIT(HRTIMER_BASE_BOOTTIME) | BIT(HRTIMER_BASE_BOOTTIME_SOFT)) + +void clock_was_set(unsigned int bases); +void clock_was_set_delayed(void); + +void hrtimers_resume_local(void); + +/* Since jiffies uses a simple TICK_NSEC multiplier + * conversion, the .shift value could be zero. However + * this would make NTP adjustments impossible as they are + * in units of 1/2^.shift. Thus we use JIFFIES_SHIFT to + * shift both the nominator and denominator the same + * amount, and give ntp adjustments in units of 1/2^8 + * + * The value 8 is somewhat carefully chosen, as anything + * larger can result in overflows. TICK_NSEC grows as HZ + * shrinks, so values greater than 8 overflow 32bits when + * HZ=100. + */ +#if HZ < 34 +#define JIFFIES_SHIFT 6 +#elif HZ < 67 +#define JIFFIES_SHIFT 7 +#else +#define JIFFIES_SHIFT 8 +#endif + +extern ssize_t sysfs_get_uname(const char *buf, char *dst, size_t cnt); diff --git a/kernel/time/tick-legacy.c b/kernel/time/tick-legacy.c new file mode 100644 index 000000000000..af225b32f5b3 --- /dev/null +++ b/kernel/time/tick-legacy.c @@ -0,0 +1,37 @@ +// SPDX-License-Identifier: GPL-2.0 +/* + * Timer tick function for architectures that lack generic clockevents, + * consolidated here from m68k/ia64/parisc/arm. + */ + +#include <linux/irq.h> +#include <linux/profile.h> +#include <linux/timekeeper_internal.h> + +#include "tick-internal.h" + +/** + * legacy_timer_tick() - advances the timekeeping infrastructure + * @ticks: number of ticks, that have elapsed since the last call. + * + * This is used by platforms that have not been converted to + * generic clockevents. + * + * If 'ticks' is zero, the CPU is not handling timekeeping, so + * only perform process accounting and profiling. + * + * Must be called with interrupts disabled. + */ +void legacy_timer_tick(unsigned long ticks) +{ + if (ticks) { + raw_spin_lock(&jiffies_lock); + write_seqcount_begin(&jiffies_seq); + do_timer(ticks); + write_seqcount_end(&jiffies_seq); + raw_spin_unlock(&jiffies_lock); + update_wall_time(); + } + update_process_times(user_mode(get_irq_regs())); + profile_tick(CPU_PROFILING); +} diff --git a/kernel/time/tick-oneshot.c b/kernel/time/tick-oneshot.c index f9745d47425a..5e2c2c26b3cc 100644 --- a/kernel/time/tick-oneshot.c +++ b/kernel/time/tick-oneshot.c @@ -18,7 +18,7 @@ #include "tick-internal.h" /** - * tick_program_event + * tick_program_event - program the CPU local timer device for the next event */ int tick_program_event(ktime_t expires, int force) { @@ -45,7 +45,7 @@ int tick_program_event(ktime_t expires, int force) } /** - * tick_resume_onshot - resume oneshot mode + * tick_resume_oneshot - resume oneshot mode */ void tick_resume_oneshot(void) { @@ -99,7 +99,7 @@ int tick_switch_to_oneshot(void (*handler)(struct clock_event_device *)) } /** - * tick_check_oneshot_mode - check whether the system is in oneshot mode + * tick_oneshot_mode_active - check whether the system is in oneshot mode * * returns 1 when either nohz or highres are enabled. otherwise 0. */ diff --git a/kernel/time/tick-sched.c b/kernel/time/tick-sched.c index 3e2dc9b8858c..fa058510af9c 100644 --- a/kernel/time/tick-sched.c +++ b/kernel/time/tick-sched.c @@ -4,10 +4,11 @@ * Copyright(C) 2005-2007, Red Hat, Inc., Ingo Molnar * Copyright(C) 2006-2007 Timesys Corp., Thomas Gleixner * - * No idle tick implementation for low and high resolution timers + * NOHZ implementation for low and high resolution timers * * Started by: Thomas Gleixner and Ingo Molnar */ +#include <linux/compiler.h> #include <linux/cpu.h> #include <linux/err.h> #include <linux/hrtimer.h> @@ -20,6 +21,7 @@ #include <linux/sched/clock.h> #include <linux/sched/stat.h> #include <linux/sched/nohz.h> +#include <linux/sched/loadavg.h> #include <linux/module.h> #include <linux/irq_work.h> #include <linux/posix-timers.h> @@ -42,9 +44,10 @@ struct tick_sched *tick_get_tick_sched(int cpu) return &per_cpu(tick_cpu_sched, cpu); } -#if defined(CONFIG_NO_HZ_COMMON) || defined(CONFIG_HIGH_RES_TIMERS) /* - * The time, when the last jiffy update happened. Protected by jiffies_lock. + * The time when the last jiffy update happened. Write access must hold + * jiffies_lock and jiffies_seq. tick_nohz_next_event() needs to get a + * consistent view of jiffies and last_jiffies_update. */ static ktime_t last_jiffies_update; @@ -53,50 +56,95 @@ static ktime_t last_jiffies_update; */ static void tick_do_update_jiffies64(ktime_t now) { - unsigned long ticks = 0; - ktime_t delta; + unsigned long ticks = 1; + ktime_t delta, nextp; /* - * Do a quick check without holding jiffies_lock: - * The READ_ONCE() pairs with two updates done later in this function. + * 64-bit can do a quick check without holding the jiffies lock and + * without looking at the sequence count. The smp_load_acquire() + * pairs with the update done later in this function. + * + * 32-bit cannot do that because the store of 'tick_next_period' + * consists of two 32-bit stores, and the first store could be + * moved by the CPU to a random point in the future. */ - delta = ktime_sub(now, READ_ONCE(last_jiffies_update)); - if (delta < tick_period) - return; + if (IS_ENABLED(CONFIG_64BIT)) { + if (ktime_before(now, smp_load_acquire(&tick_next_period))) + return; + } else { + unsigned int seq; + + /* + * Avoid contention on 'jiffies_lock' and protect the quick + * check with the sequence count. + */ + do { + seq = read_seqcount_begin(&jiffies_seq); + nextp = tick_next_period; + } while (read_seqcount_retry(&jiffies_seq, seq)); + + if (ktime_before(now, nextp)) + return; + } - /* Reevaluate with jiffies_lock held */ + /* Quick check failed, i.e. update is required. */ raw_spin_lock(&jiffies_lock); + /* + * Re-evaluate with the lock held. Another CPU might have done the + * update already. + */ + if (ktime_before(now, tick_next_period)) { + raw_spin_unlock(&jiffies_lock); + return; + } + write_seqcount_begin(&jiffies_seq); - delta = ktime_sub(now, last_jiffies_update); - if (delta >= tick_period) { + delta = ktime_sub(now, tick_next_period); + if (unlikely(delta >= TICK_NSEC)) { + /* Slow path for long idle sleep times */ + s64 incr = TICK_NSEC; - delta = ktime_sub(delta, tick_period); - /* Pairs with the lockless read in this function. */ - WRITE_ONCE(last_jiffies_update, - ktime_add(last_jiffies_update, tick_period)); + ticks += ktime_divns(delta, incr); - /* Slow path for long timeouts */ - if (unlikely(delta >= tick_period)) { - s64 incr = ktime_to_ns(tick_period); + last_jiffies_update = ktime_add_ns(last_jiffies_update, + incr * ticks); + } else { + last_jiffies_update = ktime_add_ns(last_jiffies_update, + TICK_NSEC); + } - ticks = ktime_divns(delta, incr); + /* Advance jiffies to complete the 'jiffies_seq' protected job */ + jiffies_64 += ticks; - /* Pairs with the lockless read in this function. */ - WRITE_ONCE(last_jiffies_update, - ktime_add_ns(last_jiffies_update, - incr * ticks)); - } - do_timer(++ticks); + /* Keep the tick_next_period variable up to date */ + nextp = ktime_add_ns(last_jiffies_update, TICK_NSEC); - /* Keep the tick_next_period variable up to date */ - tick_next_period = ktime_add(last_jiffies_update, tick_period); + if (IS_ENABLED(CONFIG_64BIT)) { + /* + * Pairs with smp_load_acquire() in the lockless quick + * check above, and ensures that the update to 'jiffies_64' is + * not reordered vs. the store to 'tick_next_period', neither + * by the compiler nor by the CPU. + */ + smp_store_release(&tick_next_period, nextp); } else { - write_seqcount_end(&jiffies_seq); - raw_spin_unlock(&jiffies_lock); - return; + /* + * A plain store is good enough on 32-bit, as the quick check + * above is protected by the sequence count. + */ + tick_next_period = nextp; } + + /* + * Release the sequence count. calc_global_load() below is not + * protected by it, but 'jiffies_lock' needs to be held to prevent + * concurrent invocations. + */ write_seqcount_end(&jiffies_seq); + + calc_global_load(); + raw_spin_unlock(&jiffies_lock); update_wall_time(); } @@ -110,58 +158,110 @@ static ktime_t tick_init_jiffy_update(void) raw_spin_lock(&jiffies_lock); write_seqcount_begin(&jiffies_seq); - /* Did we start the jiffies update yet ? */ - if (last_jiffies_update == 0) + + /* Have we started the jiffies update yet ? */ + if (last_jiffies_update == 0) { + u32 rem; + + /* + * Ensure that the tick is aligned to a multiple of + * TICK_NSEC. + */ + div_u64_rem(tick_next_period, TICK_NSEC, &rem); + if (rem) + tick_next_period += TICK_NSEC - rem; + last_jiffies_update = tick_next_period; + } period = last_jiffies_update; + write_seqcount_end(&jiffies_seq); raw_spin_unlock(&jiffies_lock); + return period; } +static inline int tick_sched_flag_test(struct tick_sched *ts, + unsigned long flag) +{ + return !!(ts->flags & flag); +} + +static inline void tick_sched_flag_set(struct tick_sched *ts, + unsigned long flag) +{ + lockdep_assert_irqs_disabled(); + ts->flags |= flag; +} + +static inline void tick_sched_flag_clear(struct tick_sched *ts, + unsigned long flag) +{ + lockdep_assert_irqs_disabled(); + ts->flags &= ~flag; +} + +#define MAX_STALLED_JIFFIES 5 + static void tick_sched_do_timer(struct tick_sched *ts, ktime_t now) { - int cpu = smp_processor_id(); + int tick_cpu, cpu = smp_processor_id(); -#ifdef CONFIG_NO_HZ_COMMON /* * Check if the do_timer duty was dropped. We don't care about * concurrency: This happens only when the CPU in charge went * into a long sleep. If two CPUs happen to assign themselves to * this duty, then the jiffies update is still serialized by - * jiffies_lock. + * 'jiffies_lock'. * * If nohz_full is enabled, this should not happen because the - * tick_do_timer_cpu never relinquishes. + * 'tick_do_timer_cpu' CPU never relinquishes. */ - if (unlikely(tick_do_timer_cpu == TICK_DO_TIMER_NONE)) { + tick_cpu = READ_ONCE(tick_do_timer_cpu); + + if (IS_ENABLED(CONFIG_NO_HZ_COMMON) && unlikely(tick_cpu == TICK_DO_TIMER_NONE)) { #ifdef CONFIG_NO_HZ_FULL - WARN_ON(tick_nohz_full_running); + WARN_ON_ONCE(tick_nohz_full_running); #endif - tick_do_timer_cpu = cpu; + WRITE_ONCE(tick_do_timer_cpu, cpu); + tick_cpu = cpu; } -#endif - /* Check, if the jiffies need an update */ - if (tick_do_timer_cpu == cpu) + /* Check if jiffies need an update */ + if (tick_cpu == cpu) tick_do_update_jiffies64(now); - if (ts->inidle) + /* + * If the jiffies update stalled for too long (timekeeper in stop_machine() + * or VMEXIT'ed for several msecs), force an update. + */ + if (ts->last_tick_jiffies != jiffies) { + ts->stalled_jiffies = 0; + ts->last_tick_jiffies = READ_ONCE(jiffies); + } else { + if (++ts->stalled_jiffies == MAX_STALLED_JIFFIES) { + tick_do_update_jiffies64(now); + ts->stalled_jiffies = 0; + ts->last_tick_jiffies = READ_ONCE(jiffies); + } + } + + if (tick_sched_flag_test(ts, TS_FLAG_INIDLE)) ts->got_idle_tick = 1; } static void tick_sched_handle(struct tick_sched *ts, struct pt_regs *regs) { -#ifdef CONFIG_NO_HZ_COMMON /* * When we are idle and the tick is stopped, we have to touch * the watchdog as we might not schedule for a really long - * time. This happens on complete idle SMP systems while + * time. This happens on completely idle SMP systems while * waiting on the login prompt. We also increment the "start of * idle" jiffy stamp so the idle accounting adjustment we do - * when we go busy again does not account too much ticks. + * when we go busy again does not account too many ticks. */ - if (ts->tick_stopped) { + if (IS_ENABLED(CONFIG_NO_HZ_COMMON) && + tick_sched_flag_test(ts, TS_FLAG_STOPPED)) { touch_softlockup_watchdog_sched(); if (is_idle_task(current)) ts->idle_jiffies++; @@ -172,14 +272,48 @@ static void tick_sched_handle(struct tick_sched *ts, struct pt_regs *regs) */ ts->next_tick = 0; } -#endif + update_process_times(user_mode(regs)); profile_tick(CPU_PROFILING); } -#endif + +/* + * We rearm the timer until we get disabled by the idle code. + * Called with interrupts disabled. + */ +static enum hrtimer_restart tick_nohz_handler(struct hrtimer *timer) +{ + struct tick_sched *ts = container_of(timer, struct tick_sched, sched_timer); + struct pt_regs *regs = get_irq_regs(); + ktime_t now = ktime_get(); + + tick_sched_do_timer(ts, now); + + /* + * Do not call when we are not in IRQ context and have + * no valid 'regs' pointer + */ + if (regs) + tick_sched_handle(ts, regs); + else + ts->next_tick = 0; + + /* + * In dynticks mode, tick reprogram is deferred: + * - to the idle task if in dynticks-idle + * - to IRQ exit if in full-dynticks. + */ + if (unlikely(tick_sched_flag_test(ts, TS_FLAG_STOPPED))) + return HRTIMER_NORESTART; + + hrtimer_forward(timer, now, TICK_NSEC); + + return HRTIMER_RESTART; +} #ifdef CONFIG_NO_HZ_FULL cpumask_var_t tick_nohz_full_mask; +EXPORT_SYMBOL_GPL(tick_nohz_full_mask); bool tick_nohz_full_running; EXPORT_SYMBOL_GPL(tick_nohz_full_running); static atomic_t tick_dep_mask; @@ -213,6 +347,11 @@ static bool check_tick_dependency(atomic_t *dep) return true; } + if (val & TICK_DEP_MASK_RCU_EXP) { + trace_tick_stop(0, TICK_DEP_MASK_RCU_EXP); + return true; + } + return false; } @@ -243,10 +382,8 @@ static void nohz_full_kick_func(struct irq_work *work) /* Empty, the tick restart happens on tick_nohz_irq_exit() */ } -static DEFINE_PER_CPU(struct irq_work, nohz_full_kick_work) = { - .func = nohz_full_kick_func, - .flags = ATOMIC_INIT(IRQ_WORK_HARD_IRQ), -}; +static DEFINE_PER_CPU(struct irq_work, nohz_full_kick_work) = + IRQ_WORK_INIT_HARD(nohz_full_kick_func); /* * Kick this CPU if it's full dynticks in order to force it to @@ -274,6 +411,52 @@ void tick_nohz_full_kick_cpu(int cpu) irq_work_queue_on(&per_cpu(nohz_full_kick_work, cpu), cpu); } +static void tick_nohz_kick_task(struct task_struct *tsk) +{ + int cpu; + + /* + * If the task is not running, run_posix_cpu_timers() + * has nothing to elapse, and an IPI can then be optimized out. + * + * activate_task() STORE p->tick_dep_mask + * STORE p->on_rq + * __schedule() (switch to task 'p') smp_mb() (atomic_fetch_or()) + * LOCK rq->lock LOAD p->on_rq + * smp_mb__after_spin_lock() + * tick_nohz_task_switch() + * LOAD p->tick_dep_mask + * + * XXX given a task picks up the dependency on schedule(), should we + * only care about tasks that are currently on the CPU instead of all + * that are on the runqueue? + * + * That is, does this want to be: task_on_cpu() / task_curr()? + */ + if (!sched_task_on_rq(tsk)) + return; + + /* + * If the task concurrently migrates to another CPU, + * we guarantee it sees the new tick dependency upon + * schedule. + * + * set_task_cpu(p, cpu); + * STORE p->cpu = @cpu + * __schedule() (switch to task 'p') + * LOCK rq->lock + * smp_mb__after_spin_lock() STORE p->tick_dep_mask + * tick_nohz_task_switch() smp_mb() (atomic_fetch_or()) + * LOAD p->tick_dep_mask LOAD p->cpu + */ + cpu = task_cpu(tsk); + + preempt_disable(); + if (cpu_online(cpu)) + tick_nohz_full_kick_cpu(cpu); + preempt_enable(); +} + /* * Kick all full dynticks CPUs in order to force these to re-evaluate * their dependency on the tick and restart it if necessary. @@ -303,7 +486,7 @@ static void tick_nohz_dep_set_all(atomic_t *dep, /* * Set a global tick dependency. Used by perf events that rely on freq and - * by unstable clock. + * unstable clocks. */ void tick_nohz_dep_set(enum tick_dep_bits bit) { @@ -317,7 +500,7 @@ void tick_nohz_dep_clear(enum tick_dep_bits bit) /* * Set per-CPU tick dependency. Used by scheduler and perf events in order to - * manage events throttling. + * manage event-throttling. */ void tick_nohz_dep_set_cpu(int cpu, enum tick_dep_bits bit) { @@ -333,7 +516,7 @@ void tick_nohz_dep_set_cpu(int cpu, enum tick_dep_bits bit) if (cpu == smp_processor_id()) { tick_nohz_full_kick(); } else { - /* Remote irq work not NMI-safe */ + /* Remote IRQ work not NMI-safe */ if (!WARN_ON_ONCE(in_nmi())) tick_nohz_full_kick_cpu(cpu); } @@ -351,16 +534,13 @@ void tick_nohz_dep_clear_cpu(int cpu, enum tick_dep_bits bit) EXPORT_SYMBOL_GPL(tick_nohz_dep_clear_cpu); /* - * Set a per-task tick dependency. Posix CPU timers need this in order to elapse - * per task timers. + * Set a per-task tick dependency. RCU needs this. Also posix CPU timers + * in order to elapse per task timers. */ void tick_nohz_dep_set_task(struct task_struct *tsk, enum tick_dep_bits bit) { - /* - * We could optimize this with just kicking the target running the task - * if that noise matters for nohz full users. - */ - tick_nohz_dep_set_all(&tsk->tick_dep_mask, bit); + if (!atomic_fetch_or(BIT(bit), &tsk->tick_dep_mask)) + tick_nohz_kick_task(tsk); } EXPORT_SYMBOL_GPL(tick_nohz_dep_set_task); @@ -374,9 +554,20 @@ EXPORT_SYMBOL_GPL(tick_nohz_dep_clear_task); * Set a per-taskgroup tick dependency. Posix CPU timers need this in order to elapse * per process timers. */ -void tick_nohz_dep_set_signal(struct signal_struct *sig, enum tick_dep_bits bit) +void tick_nohz_dep_set_signal(struct task_struct *tsk, + enum tick_dep_bits bit) { - tick_nohz_dep_set_all(&sig->tick_dep_mask, bit); + int prev; + struct signal_struct *sig = tsk->signal; + + prev = atomic_fetch_or(BIT(bit), &sig->tick_dep_mask); + if (!prev) { + struct task_struct *t; + + lockdep_assert_held(&tsk->sighand->siglock); + __for_each_thread(sig, t) + tick_nohz_kick_task(t); + } } void tick_nohz_dep_clear_signal(struct signal_struct *sig, enum tick_dep_bits bit) @@ -391,23 +582,18 @@ void tick_nohz_dep_clear_signal(struct signal_struct *sig, enum tick_dep_bits bi */ void __tick_nohz_task_switch(void) { - unsigned long flags; struct tick_sched *ts; - local_irq_save(flags); - if (!tick_nohz_full_cpu(smp_processor_id())) - goto out; + return; ts = this_cpu_ptr(&tick_cpu_sched); - if (ts->tick_stopped) { + if (tick_sched_flag_test(ts, TS_FLAG_STOPPED)) { if (atomic_read(¤t->tick_dep_mask) || atomic_read(¤t->signal->tick_dep_mask)) tick_nohz_full_kick(); } -out: - local_irq_restore(flags); } /* Get the boot-time nohz CPU list from the kernel parameters. */ @@ -417,18 +603,22 @@ void __init tick_nohz_full_setup(cpumask_var_t cpumask) cpumask_copy(tick_nohz_full_mask, cpumask); tick_nohz_full_running = true; } -EXPORT_SYMBOL_GPL(tick_nohz_full_setup); -static int tick_nohz_cpu_down(unsigned int cpu) +bool tick_nohz_cpu_hotpluggable(unsigned int cpu) { /* - * The tick_do_timer_cpu CPU handles housekeeping duty (unbound + * The 'tick_do_timer_cpu' CPU handles housekeeping duty (unbound * timers, workqueues, timekeeping, ...) on behalf of full dynticks * CPUs. It must remain online when nohz full is enabled. */ - if (tick_nohz_full_running && tick_do_timer_cpu == cpu) - return -EBUSY; - return 0; + if (tick_nohz_full_running && READ_ONCE(tick_do_timer_cpu) == cpu) + return false; + return true; +} + +static int tick_nohz_cpu_down(unsigned int cpu) +{ + return tick_nohz_cpu_hotpluggable(cpu) ? 0 : -EBUSY; } void __init tick_nohz_init(void) @@ -439,12 +629,12 @@ void __init tick_nohz_init(void) return; /* - * Full dynticks uses irq work to drive the tick rescheduling on safe - * locking contexts. But then we need irq work to raise its own - * interrupts to avoid circular dependency on the tick + * Full dynticks uses IRQ work to drive the tick rescheduling on safe + * locking contexts. But then we need IRQ work to raise its own + * interrupts to avoid circular dependency on the tick. */ if (!arch_irq_work_has_interrupt()) { - pr_warn("NO_HZ: Can't run full dynticks because arch doesn't support irq work self-IPIs\n"); + pr_warn("NO_HZ: Can't run full dynticks because arch doesn't support IRQ work self-IPIs\n"); cpumask_clear(tick_nohz_full_mask); tick_nohz_full_running = false; return; @@ -462,7 +652,7 @@ void __init tick_nohz_init(void) } for_each_cpu(cpu, tick_nohz_full_mask) - context_tracking_cpu_set(cpu); + ct_cpu_track_user(cpu); ret = cpuhp_setup_state_nocalls(CPUHP_AP_ONLINE_DYN, "kernel/nohz:predown", NULL, @@ -471,7 +661,7 @@ void __init tick_nohz_init(void) pr_info("NO_HZ: Full dynticks CPUs: %*pbl.\n", cpumask_pr_args(tick_nohz_full_mask)); } -#endif +#endif /* #ifdef CONFIG_NO_HZ_FULL */ /* * NOHZ - aka dynamic tick functionality @@ -496,25 +686,26 @@ bool tick_nohz_tick_stopped(void) { struct tick_sched *ts = this_cpu_ptr(&tick_cpu_sched); - return ts->tick_stopped; + return tick_sched_flag_test(ts, TS_FLAG_STOPPED); } bool tick_nohz_tick_stopped_cpu(int cpu) { struct tick_sched *ts = per_cpu_ptr(&tick_cpu_sched, cpu); - return ts->tick_stopped; + return tick_sched_flag_test(ts, TS_FLAG_STOPPED); } /** * tick_nohz_update_jiffies - update jiffies when idle was interrupted + * @now: current ktime_t * * Called from interrupt entry when the CPU was idle * * In case the sched_tick was stopped on this CPU, we have to check if jiffies * must be updated. Otherwise an interrupt handler could use a stale jiffy * value. We do this unconditionally on any CPU, as we don't know whether the - * CPU, which has the update task assigned is in a long sleep. + * CPU, which has the update task assigned, is in a long sleep. */ static void tick_nohz_update_jiffies(ktime_t now) { @@ -529,43 +720,67 @@ static void tick_nohz_update_jiffies(ktime_t now) touch_softlockup_watchdog_sched(); } -/* - * Updates the per-CPU time idle statistics counters - */ -static void -update_ts_time_stats(int cpu, struct tick_sched *ts, ktime_t now, u64 *last_update_time) +static void tick_nohz_stop_idle(struct tick_sched *ts, ktime_t now) { ktime_t delta; - if (ts->idle_active) { - delta = ktime_sub(now, ts->idle_entrytime); - if (nr_iowait_cpu(cpu) > 0) - ts->iowait_sleeptime = ktime_add(ts->iowait_sleeptime, delta); - else - ts->idle_sleeptime = ktime_add(ts->idle_sleeptime, delta); - ts->idle_entrytime = now; - } + if (WARN_ON_ONCE(!tick_sched_flag_test(ts, TS_FLAG_IDLE_ACTIVE))) + return; - if (last_update_time) - *last_update_time = ktime_to_us(now); + delta = ktime_sub(now, ts->idle_entrytime); -} + write_seqcount_begin(&ts->idle_sleeptime_seq); + if (nr_iowait_cpu(smp_processor_id()) > 0) + ts->iowait_sleeptime = ktime_add(ts->iowait_sleeptime, delta); + else + ts->idle_sleeptime = ktime_add(ts->idle_sleeptime, delta); -static void tick_nohz_stop_idle(struct tick_sched *ts, ktime_t now) -{ - update_ts_time_stats(smp_processor_id(), ts, now, NULL); - ts->idle_active = 0; + ts->idle_entrytime = now; + tick_sched_flag_clear(ts, TS_FLAG_IDLE_ACTIVE); + write_seqcount_end(&ts->idle_sleeptime_seq); sched_clock_idle_wakeup_event(); } static void tick_nohz_start_idle(struct tick_sched *ts) { + write_seqcount_begin(&ts->idle_sleeptime_seq); ts->idle_entrytime = ktime_get(); - ts->idle_active = 1; + tick_sched_flag_set(ts, TS_FLAG_IDLE_ACTIVE); + write_seqcount_end(&ts->idle_sleeptime_seq); + sched_clock_idle_sleep_event(); } +static u64 get_cpu_sleep_time_us(struct tick_sched *ts, ktime_t *sleeptime, + bool compute_delta, u64 *last_update_time) +{ + ktime_t now, idle; + unsigned int seq; + + if (!tick_nohz_active) + return -1; + + now = ktime_get(); + if (last_update_time) + *last_update_time = ktime_to_us(now); + + do { + seq = read_seqcount_begin(&ts->idle_sleeptime_seq); + + if (tick_sched_flag_test(ts, TS_FLAG_IDLE_ACTIVE) && compute_delta) { + ktime_t delta = ktime_sub(now, ts->idle_entrytime); + + idle = ktime_add(*sleeptime, delta); + } else { + idle = *sleeptime; + } + } while (read_seqcount_retry(&ts->idle_sleeptime_seq, seq)); + + return ktime_to_us(idle); + +} + /** * get_cpu_idle_time_us - get the total idle time of a CPU * @cpu: CPU number to query @@ -573,37 +788,22 @@ static void tick_nohz_start_idle(struct tick_sched *ts) * counters if NULL. * * Return the cumulative idle time (since boot) for a given - * CPU, in microseconds. + * CPU, in microseconds. Note that this is partially broken due to + * the counter of iowait tasks that can be remotely updated without + * any synchronization. Therefore it is possible to observe backward + * values within two consecutive reads. * * This time is measured via accounting rather than sampling, * and is as accurate as ktime_get() is. * - * This function returns -1 if NOHZ is not enabled. + * Return: -1 if NOHZ is not enabled, else total idle time of the @cpu */ u64 get_cpu_idle_time_us(int cpu, u64 *last_update_time) { struct tick_sched *ts = &per_cpu(tick_cpu_sched, cpu); - ktime_t now, idle; - - if (!tick_nohz_active) - return -1; - - now = ktime_get(); - if (last_update_time) { - update_ts_time_stats(cpu, ts, now, last_update_time); - idle = ts->idle_sleeptime; - } else { - if (ts->idle_active && !nr_iowait_cpu(cpu)) { - ktime_t delta = ktime_sub(now, ts->idle_entrytime); - - idle = ktime_add(ts->idle_sleeptime, delta); - } else { - idle = ts->idle_sleeptime; - } - } - - return ktime_to_us(idle); + return get_cpu_sleep_time_us(ts, &ts->idle_sleeptime, + !nr_iowait_cpu(cpu), last_update_time); } EXPORT_SYMBOL_GPL(get_cpu_idle_time_us); @@ -614,36 +814,22 @@ EXPORT_SYMBOL_GPL(get_cpu_idle_time_us); * counters if NULL. * * Return the cumulative iowait time (since boot) for a given - * CPU, in microseconds. + * CPU, in microseconds. Note this is partially broken due to + * the counter of iowait tasks that can be remotely updated without + * any synchronization. Therefore it is possible to observe backward + * values within two consecutive reads. * * This time is measured via accounting rather than sampling, * and is as accurate as ktime_get() is. * - * This function returns -1 if NOHZ is not enabled. + * Return: -1 if NOHZ is not enabled, else total iowait time of @cpu */ u64 get_cpu_iowait_time_us(int cpu, u64 *last_update_time) { struct tick_sched *ts = &per_cpu(tick_cpu_sched, cpu); - ktime_t now, iowait; - - if (!tick_nohz_active) - return -1; - - now = ktime_get(); - if (last_update_time) { - update_ts_time_stats(cpu, ts, now, last_update_time); - iowait = ts->iowait_sleeptime; - } else { - if (ts->idle_active && nr_iowait_cpu(cpu) > 0) { - ktime_t delta = ktime_sub(now, ts->idle_entrytime); - - iowait = ktime_add(ts->iowait_sleeptime, delta); - } else { - iowait = ts->iowait_sleeptime; - } - } - return ktime_to_us(iowait); + return get_cpu_sleep_time_us(ts, &ts->iowait_sleeptime, + nr_iowait_cpu(cpu), last_update_time); } EXPORT_SYMBOL_GPL(get_cpu_iowait_time_us); @@ -653,9 +839,9 @@ static void tick_nohz_restart(struct tick_sched *ts, ktime_t now) hrtimer_set_expires(&ts->sched_timer, ts->last_tick); /* Forward the time to expire in the future */ - hrtimer_forward(&ts->sched_timer, now, tick_period); + hrtimer_forward(&ts->sched_timer, now, TICK_NSEC); - if (ts->nohz_mode == NOHZ_MODE_HIGHRES) { + if (tick_sched_flag_test(ts, TS_FLAG_HIGHRES)) { hrtimer_start_expires(&ts->sched_timer, HRTIMER_MODE_ABS_PINNED_HARD); } else { @@ -663,7 +849,7 @@ static void tick_nohz_restart(struct tick_sched *ts, ktime_t now) } /* - * Reset to make sure next tick stop doesn't get fooled by past + * Reset to make sure the next tick stop doesn't get fooled by past * cached clock deadline. */ ts->next_tick = 0; @@ -671,35 +857,58 @@ static void tick_nohz_restart(struct tick_sched *ts, ktime_t now) static inline bool local_timer_softirq_pending(void) { - return local_softirq_pending() & BIT(TIMER_SOFTIRQ); + return local_timers_pending() & BIT(TIMER_SOFTIRQ); } -static ktime_t tick_nohz_next_event(struct tick_sched *ts, int cpu) +/* + * Read jiffies and the time when jiffies were updated last + */ +u64 get_jiffies_update(unsigned long *basej) { - u64 basemono, next_tick, next_tmr, next_rcu, delta, expires; unsigned long basejiff; unsigned int seq; + u64 basemono; - /* Read jiffies and the time when jiffies were updated last */ do { seq = read_seqcount_begin(&jiffies_seq); basemono = last_jiffies_update; basejiff = jiffies; } while (read_seqcount_retry(&jiffies_seq, seq)); + *basej = basejiff; + return basemono; +} + +/** + * tick_nohz_next_event() - return the clock monotonic based next event + * @ts: pointer to tick_sched struct + * @cpu: CPU number + * + * Return: + * *%0 - When the next event is a maximum of TICK_NSEC in the future + * and the tick is not stopped yet + * *%next_event - Next event based on clock monotonic + */ +static ktime_t tick_nohz_next_event(struct tick_sched *ts, int cpu) +{ + u64 basemono, next_tick, delta, expires; + unsigned long basejiff; + int tick_cpu; + + basemono = get_jiffies_update(&basejiff); ts->last_jiffies = basejiff; ts->timer_expires_base = basemono; /* * Keep the periodic tick, when RCU, architecture or irq_work * requests it. - * Aside of that check whether the local timer softirq is - * pending. If so its a bad idea to call get_next_timer_interrupt() + * Aside of that, check whether the local timer softirq is + * pending. If so, its a bad idea to call get_next_timer_interrupt(), * because there is an already expired timer, so it will request - * immeditate expiry, which rearms the hardware timer with a - * minimal delta which brings us back to this place + * immediate expiry, which rearms the hardware timer with a + * minimal delta, which brings us back to this place * immediately. Lather, rinse and repeat... */ - if (rcu_needs_cpu(basemono, &next_rcu) || arch_needs_cpu() || + if (rcu_needs_cpu() || arch_needs_cpu() || irq_work_needs_cpu() || local_timer_softirq_pending()) { next_tick = basemono + TICK_NSEC; } else { @@ -710,12 +919,14 @@ static ktime_t tick_nohz_next_event(struct tick_sched *ts, int cpu) * disabled this also looks at the next expiring * hrtimer. */ - next_tmr = get_next_timer_interrupt(basejiff, basemono); - ts->next_timer = next_tmr; - /* Take the next rcu event into account */ - next_tick = next_rcu < next_tmr ? next_rcu : next_tmr; + next_tick = get_next_timer_interrupt(basejiff, basemono); + ts->next_timer = next_tick; } + /* Make sure next_tick is never before basemono! */ + if (WARN_ON_ONCE(basemono > next_tick)) + next_tick = basemono; + /* * If the tick is due in the next period, keep it ticking or * force prod the timer. @@ -723,15 +934,10 @@ static ktime_t tick_nohz_next_event(struct tick_sched *ts, int cpu) delta = next_tick - basemono; if (delta <= (u64)TICK_NSEC) { /* - * Tell the timer code that the base is not idle, i.e. undo - * the effect of get_next_timer_interrupt(): - */ - timer_clear_idle(); - /* * We've not stopped the tick yet, and there's a timer in the * next period, so no point in stopping it either, bail. */ - if (!ts->tick_stopped) { + if (!tick_sched_flag_test(ts, TS_FLAG_STOPPED)) { ts->timer_expires = 0; goto out; } @@ -739,12 +945,13 @@ static ktime_t tick_nohz_next_event(struct tick_sched *ts, int cpu) /* * If this CPU is the one which had the do_timer() duty last, we limit - * the sleep time to the timekeeping max_deferment value. + * the sleep time to the timekeeping 'max_deferment' value. * Otherwise we can sleep as long as we want. */ delta = timekeeping_max_deferment(); - if (cpu != tick_do_timer_cpu && - (tick_do_timer_cpu != TICK_DO_TIMER_NONE || !ts->do_timer_last)) + tick_cpu = READ_ONCE(tick_do_timer_cpu); + if (tick_cpu != cpu && + (tick_cpu != TICK_DO_TIMER_NONE || !tick_sched_flag_test(ts, TS_FLAG_DO_TIMER_LAST))) delta = KTIME_MAX; /* Calculate the next expiry time */ @@ -762,74 +969,103 @@ out: static void tick_nohz_stop_tick(struct tick_sched *ts, int cpu) { struct clock_event_device *dev = __this_cpu_read(tick_cpu_device.evtdev); + unsigned long basejiff = ts->last_jiffies; u64 basemono = ts->timer_expires_base; - u64 expires = ts->timer_expires; - ktime_t tick = expires; + bool timer_idle = tick_sched_flag_test(ts, TS_FLAG_STOPPED); + int tick_cpu; + u64 expires; /* Make sure we won't be trying to stop it twice in a row. */ ts->timer_expires_base = 0; /* + * Now the tick should be stopped definitely - so the timer base needs + * to be marked idle as well to not miss a newly queued timer. + */ + expires = timer_base_try_to_set_idle(basejiff, basemono, &timer_idle); + if (expires > ts->timer_expires) { + /* + * This path could only happen when the first timer was removed + * between calculating the possible sleep length and now (when + * high resolution mode is not active, timer could also be a + * hrtimer). + * + * We have to stick to the original calculated expiry value to + * not stop the tick for too long with a shallow C-state (which + * was programmed by cpuidle because of an early next expiration + * value). + */ + expires = ts->timer_expires; + } + + /* If the timer base is not idle, retain the not yet stopped tick. */ + if (!timer_idle) + return; + + /* * If this CPU is the one which updates jiffies, then give up * the assignment and let it be taken by the CPU which runs * the tick timer next, which might be this CPU as well. If we - * don't drop this here the jiffies might be stale and - * do_timer() never invoked. Keep track of the fact that it + * don't drop this here, the jiffies might be stale and + * do_timer() never gets invoked. Keep track of the fact that it * was the one which had the do_timer() duty last. */ - if (cpu == tick_do_timer_cpu) { - tick_do_timer_cpu = TICK_DO_TIMER_NONE; - ts->do_timer_last = 1; - } else if (tick_do_timer_cpu != TICK_DO_TIMER_NONE) { - ts->do_timer_last = 0; + tick_cpu = READ_ONCE(tick_do_timer_cpu); + if (tick_cpu == cpu) { + WRITE_ONCE(tick_do_timer_cpu, TICK_DO_TIMER_NONE); + tick_sched_flag_set(ts, TS_FLAG_DO_TIMER_LAST); + } else if (tick_cpu != TICK_DO_TIMER_NONE) { + tick_sched_flag_clear(ts, TS_FLAG_DO_TIMER_LAST); } - /* Skip reprogram of event if its not changed */ - if (ts->tick_stopped && (expires == ts->next_tick)) { + /* Skip reprogram of event if it's not changed */ + if (tick_sched_flag_test(ts, TS_FLAG_STOPPED) && (expires == ts->next_tick)) { /* Sanity check: make sure clockevent is actually programmed */ - if (tick == KTIME_MAX || ts->next_tick == hrtimer_get_expires(&ts->sched_timer)) + if (expires == KTIME_MAX || ts->next_tick == hrtimer_get_expires(&ts->sched_timer)) return; - WARN_ON_ONCE(1); - printk_once("basemono: %llu ts->next_tick: %llu dev->next_event: %llu timer->active: %d timer->expires: %llu\n", - basemono, ts->next_tick, dev->next_event, - hrtimer_active(&ts->sched_timer), hrtimer_get_expires(&ts->sched_timer)); + WARN_ONCE(1, "basemono: %llu ts->next_tick: %llu dev->next_event: %llu " + "timer->active: %d timer->expires: %llu\n", basemono, ts->next_tick, + dev->next_event, hrtimer_active(&ts->sched_timer), + hrtimer_get_expires(&ts->sched_timer)); } /* - * nohz_stop_sched_tick can be called several times before - * the nohz_restart_sched_tick is called. This happens when - * interrupts arrive which do not cause a reschedule. In the - * first call we save the current tick time, so we can restart - * the scheduler tick in nohz_restart_sched_tick. + * tick_nohz_stop_tick() can be called several times before + * tick_nohz_restart_sched_tick() is called. This happens when + * interrupts arrive which do not cause a reschedule. In the first + * call we save the current tick time, so we can restart the + * scheduler tick in tick_nohz_restart_sched_tick(). */ - if (!ts->tick_stopped) { + if (!tick_sched_flag_test(ts, TS_FLAG_STOPPED)) { calc_load_nohz_start(); quiet_vmstat(); ts->last_tick = hrtimer_get_expires(&ts->sched_timer); - ts->tick_stopped = 1; + tick_sched_flag_set(ts, TS_FLAG_STOPPED); trace_tick_stop(1, TICK_DEP_MASK_NONE); } - ts->next_tick = tick; + ts->next_tick = expires; /* * If the expiration time == KTIME_MAX, then we simply stop * the tick timer. */ if (unlikely(expires == KTIME_MAX)) { - if (ts->nohz_mode == NOHZ_MODE_HIGHRES) + if (tick_sched_flag_test(ts, TS_FLAG_HIGHRES)) hrtimer_cancel(&ts->sched_timer); + else + tick_program_event(KTIME_MAX, 1); return; } - if (ts->nohz_mode == NOHZ_MODE_HIGHRES) { - hrtimer_start(&ts->sched_timer, tick, + if (tick_sched_flag_test(ts, TS_FLAG_HIGHRES)) { + hrtimer_start(&ts->sched_timer, expires, HRTIMER_MODE_ABS_PINNED_HARD); } else { - hrtimer_set_expires(&ts->sched_timer, tick); - tick_program_event(tick, 1); + hrtimer_set_expires(&ts->sched_timer, expires); + tick_program_event(expires, 1); } } @@ -839,7 +1075,7 @@ static void tick_nohz_retain_tick(struct tick_sched *ts) } #ifdef CONFIG_NO_HZ_FULL -static void tick_nohz_stop_sched_tick(struct tick_sched *ts, int cpu) +static void tick_nohz_full_stop_tick(struct tick_sched *ts, int cpu) { if (tick_nohz_next_event(ts, cpu)) tick_nohz_stop_tick(ts, cpu); @@ -861,98 +1097,116 @@ static void tick_nohz_restart_sched_tick(struct tick_sched *ts, ktime_t now) calc_load_nohz_stop(); touch_softlockup_watchdog_sched(); - /* - * Cancel the scheduled timer and restore the tick - */ - ts->tick_stopped = 0; - ts->idle_exittime = now; + /* Cancel the scheduled timer and restore the tick: */ + tick_sched_flag_clear(ts, TS_FLAG_STOPPED); tick_nohz_restart(ts, now); } -static void tick_nohz_full_update_tick(struct tick_sched *ts) +static void __tick_nohz_full_update_tick(struct tick_sched *ts, + ktime_t now) { #ifdef CONFIG_NO_HZ_FULL int cpu = smp_processor_id(); - if (!tick_nohz_full_cpu(cpu)) + if (can_stop_full_tick(cpu, ts)) + tick_nohz_full_stop_tick(ts, cpu); + else if (tick_sched_flag_test(ts, TS_FLAG_STOPPED)) + tick_nohz_restart_sched_tick(ts, now); +#endif +} + +static void tick_nohz_full_update_tick(struct tick_sched *ts) +{ + if (!tick_nohz_full_cpu(smp_processor_id())) return; - if (!ts->tick_stopped && ts->nohz_mode == NOHZ_MODE_INACTIVE) + if (!tick_sched_flag_test(ts, TS_FLAG_NOHZ)) return; - if (can_stop_full_tick(cpu, ts)) - tick_nohz_stop_sched_tick(ts, cpu); - else if (ts->tick_stopped) - tick_nohz_restart_sched_tick(ts, ktime_get()); -#endif + __tick_nohz_full_update_tick(ts, ktime_get()); } -static bool can_stop_idle_tick(int cpu, struct tick_sched *ts) +/* + * A pending softirq outside an IRQ (or softirq disabled section) context + * should be waiting for ksoftirqd to handle it. Therefore we shouldn't + * reach this code due to the need_resched() early check in can_stop_idle_tick(). + * + * However if we are between CPUHP_AP_SMPBOOT_THREADS and CPU_TEARDOWN_CPU on the + * cpu_down() process, softirqs can still be raised while ksoftirqd is parked, + * triggering the code below, since wakep_softirqd() is ignored. + * + */ +static bool report_idle_softirq(void) { - /* - * If this CPU is offline and it is the one which updates - * jiffies, then give up the assignment and let it be taken by - * the CPU which runs the tick timer next. If we don't drop - * this here the jiffies might be stale and do_timer() never - * invoked. - */ - if (unlikely(!cpu_online(cpu))) { - if (cpu == tick_do_timer_cpu) - tick_do_timer_cpu = TICK_DO_TIMER_NONE; - /* - * Make sure the CPU doesn't get fooled by obsolete tick - * deadline if it comes back online later. - */ - ts->next_tick = 0; + static int ratelimit; + unsigned int pending = local_softirq_pending(); + + if (likely(!pending)) return false; + + /* Some softirqs claim to be safe against hotplug and ksoftirqd parking */ + if (!cpu_active(smp_processor_id())) { + pending &= ~SOFTIRQ_HOTPLUG_SAFE_MASK; + if (!pending) + return false; } - if (unlikely(ts->nohz_mode == NOHZ_MODE_INACTIVE)) + if (ratelimit >= 10) return false; - if (need_resched()) + /* On RT, softirq handling may be waiting on some lock */ + if (local_bh_blocked()) return false; - if (unlikely(local_softirq_pending())) { - static int ratelimit; + pr_warn("NOHZ tick-stop error: local softirq work is pending, handler #%02x!!!\n", + pending); + ratelimit++; - if (ratelimit < 10 && - (local_softirq_pending() & SOFTIRQ_STOP_IDLE_MASK)) { - pr_warn("NOHZ: local_softirq_pending %02x\n", - (unsigned int) local_softirq_pending()); - ratelimit++; - } + return true; +} + +static bool can_stop_idle_tick(int cpu, struct tick_sched *ts) +{ + WARN_ON_ONCE(cpu_is_offline(cpu)); + + if (unlikely(!tick_sched_flag_test(ts, TS_FLAG_NOHZ))) + return false; + + if (need_resched()) + return false; + + if (unlikely(report_idle_softirq())) return false; - } if (tick_nohz_full_enabled()) { + int tick_cpu = READ_ONCE(tick_do_timer_cpu); + /* * Keep the tick alive to guarantee timekeeping progression * if there are full dynticks CPUs around */ - if (tick_do_timer_cpu == cpu) - return false; - /* - * Boot safety: make sure the timekeeping duty has been - * assigned before entering dyntick-idle mode, - * tick_do_timer_cpu is TICK_DO_TIMER_BOOT - */ - if (unlikely(tick_do_timer_cpu == TICK_DO_TIMER_BOOT)) + if (tick_cpu == cpu) return false; /* Should not happen for nohz-full */ - if (WARN_ON_ONCE(tick_do_timer_cpu == TICK_DO_TIMER_NONE)) + if (WARN_ON_ONCE(tick_cpu == TICK_DO_TIMER_NONE)) return false; } return true; } -static void __tick_nohz_idle_stop_tick(struct tick_sched *ts) +/** + * tick_nohz_idle_stop_tick - stop the idle tick from the idle task + * + * When the next event is more than a tick into the future, stop the idle tick + */ +void tick_nohz_idle_stop_tick(void) { - ktime_t expires; + struct tick_sched *ts = this_cpu_ptr(&tick_cpu_sched); int cpu = smp_processor_id(); + ktime_t expires; /* * If tick_nohz_get_sleep_length() ran tick_nohz_next_event(), the @@ -968,14 +1222,14 @@ static void __tick_nohz_idle_stop_tick(struct tick_sched *ts) ts->idle_calls++; if (expires > 0LL) { - int was_stopped = ts->tick_stopped; + int was_stopped = tick_sched_flag_test(ts, TS_FLAG_STOPPED); tick_nohz_stop_tick(ts, cpu); ts->idle_sleeps++; ts->idle_expires = expires; - if (!was_stopped && ts->tick_stopped) { + if (!was_stopped && tick_sched_flag_test(ts, TS_FLAG_STOPPED)) { ts->idle_jiffies = ts->last_jiffies; nohz_balance_enter_idle(cpu); } @@ -984,24 +1238,9 @@ static void __tick_nohz_idle_stop_tick(struct tick_sched *ts) } } -/** - * tick_nohz_idle_stop_tick - stop the idle tick from the idle task - * - * When the next event is more than a tick into the future, stop the idle tick - */ -void tick_nohz_idle_stop_tick(void) -{ - __tick_nohz_idle_stop_tick(this_cpu_ptr(&tick_cpu_sched)); -} - void tick_nohz_idle_retain_tick(void) { tick_nohz_retain_tick(this_cpu_ptr(&tick_cpu_sched)); - /* - * Undo the effect of get_next_timer_interrupt() called from - * tick_nohz_next_event(). - */ - timer_clear_idle(); } /** @@ -1021,25 +1260,36 @@ void tick_nohz_idle_enter(void) WARN_ON_ONCE(ts->timer_expires_base); - ts->inidle = 1; + tick_sched_flag_set(ts, TS_FLAG_INIDLE); tick_nohz_start_idle(ts); local_irq_enable(); } /** - * tick_nohz_irq_exit - update next tick event from interrupt exit + * tick_nohz_irq_exit - Notify the tick about IRQ exit + * + * A timer may have been added/modified/deleted either by the current IRQ, + * or by another place using this IRQ as a notification. This IRQ may have + * also updated the RCU callback list. These events may require a + * re-evaluation of the next tick. Depending on the context: + * + * 1) If the CPU is idle and no resched is pending, just proceed with idle + * time accounting. The next tick will be re-evaluated on the next idle + * loop iteration. * - * When an interrupt fires while we are idle and it doesn't cause - * a reschedule, it may still add, modify or delete a timer, enqueue - * an RCU callback, etc... - * So we need to re-calculate and reprogram the next tick event. + * 2) If the CPU is nohz_full: + * + * 2.1) If there is any tick dependency, restart the tick if stopped. + * + * 2.2) If there is no tick dependency, (re-)evaluate the next tick and + * stop/update it accordingly. */ void tick_nohz_irq_exit(void) { struct tick_sched *ts = this_cpu_ptr(&tick_cpu_sched); - if (ts->inidle) + if (tick_sched_flag_test(ts, TS_FLAG_INIDLE)) tick_nohz_start_idle(ts); else tick_nohz_full_update_tick(ts); @@ -1047,6 +1297,8 @@ void tick_nohz_irq_exit(void) /** * tick_nohz_idle_got_tick - Check whether or not the tick handler has run + * + * Return: %true if the tick handler has run, otherwise %false */ bool tick_nohz_idle_got_tick(void) { @@ -1061,10 +1313,12 @@ bool tick_nohz_idle_got_tick(void) /** * tick_nohz_get_next_hrtimer - return the next expiration time for the hrtimer - * or the tick, whatever that expires first. Note that, if the tick has been + * or the tick, whichever expires first. Note that, if the tick has been * stopped, it returns the next hrtimer. * * Called from power state control code with interrupts disabled + * + * Return: the next expiration time */ ktime_t tick_nohz_get_next_hrtimer(void) { @@ -1075,7 +1329,13 @@ ktime_t tick_nohz_get_next_hrtimer(void) * tick_nohz_get_sleep_length - return the expected length of the current sleep * @delta_next: duration until the next event if the tick cannot be stopped * - * Called from power state control code with interrupts disabled + * Called from power state control code with interrupts disabled. + * + * The return value of this function and/or the value returned by it through the + * @delta_next pointer can be negative which must be taken into account by its + * callers. + * + * Return: the expected length of the current sleep */ ktime_t tick_nohz_get_sleep_length(ktime_t *delta_next) { @@ -1089,7 +1349,7 @@ ktime_t tick_nohz_get_sleep_length(ktime_t *delta_next) ktime_t now = ts->idle_entrytime; ktime_t next_event; - WARN_ON_ONCE(!ts->inidle); + WARN_ON_ONCE(!tick_sched_flag_test(ts, TS_FLAG_INIDLE)); *delta_next = ktime_sub(dev->next_event, now); @@ -1101,7 +1361,7 @@ ktime_t tick_nohz_get_sleep_length(ktime_t *delta_next) return *delta_next; /* - * If the next highres timer to expire is earlier than next_event, the + * If the next highres timer to expire is earlier than 'next_event', the * idle governor needs to know that. */ next_event = min_t(u64, next_event, @@ -1113,8 +1373,11 @@ ktime_t tick_nohz_get_sleep_length(ktime_t *delta_next) /** * tick_nohz_get_idle_calls_cpu - return the current idle calls counter value * for a particular CPU. + * @cpu: target CPU number * * Called from the schedutil frequency scaling governor in scheduler context. + * + * Return: the current idle calls counter value for @cpu */ unsigned long tick_nohz_get_idle_calls_cpu(int cpu) { @@ -1123,29 +1386,19 @@ unsigned long tick_nohz_get_idle_calls_cpu(int cpu) return ts->idle_calls; } -/** - * tick_nohz_get_idle_calls - return the current idle calls counter value - * - * Called from the schedutil frequency scaling governor in scheduler context. - */ -unsigned long tick_nohz_get_idle_calls(void) -{ - struct tick_sched *ts = this_cpu_ptr(&tick_cpu_sched); - - return ts->idle_calls; -} - -static void tick_nohz_account_idle_ticks(struct tick_sched *ts) +static void tick_nohz_account_idle_time(struct tick_sched *ts, + ktime_t now) { -#ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE unsigned long ticks; + ts->idle_exittime = now; + if (vtime_accounting_enabled_this_cpu()) return; /* - * We stopped the tick in idle. Update process times would miss the - * time we slept as update_process_times does only a 1 tick - * accounting. Enforce that this is accounted to idle ! + * We stopped the tick in idle. update_process_times() would miss the + * time we slept, as it does only a 1 tick accounting. + * Enforce that this is accounted to idle ! */ ticks = jiffies - ts->idle_jiffies; /* @@ -1153,29 +1406,44 @@ static void tick_nohz_account_idle_ticks(struct tick_sched *ts) */ if (ticks && ticks < LONG_MAX) account_idle_ticks(ticks); -#endif } -static void __tick_nohz_idle_restart_tick(struct tick_sched *ts, ktime_t now) +void tick_nohz_idle_restart_tick(void) { - tick_nohz_restart_sched_tick(ts, now); - tick_nohz_account_idle_ticks(ts); + struct tick_sched *ts = this_cpu_ptr(&tick_cpu_sched); + + if (tick_sched_flag_test(ts, TS_FLAG_STOPPED)) { + ktime_t now = ktime_get(); + tick_nohz_restart_sched_tick(ts, now); + tick_nohz_account_idle_time(ts, now); + } } -void tick_nohz_idle_restart_tick(void) +static void tick_nohz_idle_update_tick(struct tick_sched *ts, ktime_t now) { - struct tick_sched *ts = this_cpu_ptr(&tick_cpu_sched); + if (tick_nohz_full_cpu(smp_processor_id())) + __tick_nohz_full_update_tick(ts, now); + else + tick_nohz_restart_sched_tick(ts, now); - if (ts->tick_stopped) - __tick_nohz_idle_restart_tick(ts, ktime_get()); + tick_nohz_account_idle_time(ts, now); } /** - * tick_nohz_idle_exit - restart the idle tick from the idle task + * tick_nohz_idle_exit - Update the tick upon idle task exit + * + * When the idle task exits, update the tick depending on the + * following situations: + * + * 1) If the CPU is not in nohz_full mode (most cases), then + * restart the tick. + * + * 2) If the CPU is in nohz_full mode (corner case): + * 2.1) If the tick can be kept stopped (no tick dependencies) + * then re-evaluate the next tick and try to keep it stopped + * as long as possible. + * 2.2) If the tick has dependencies, restart the tick. * - * Restart the idle tick when the CPU is woken up from idle - * This also exit the RCU extended quiescent state. The CPU - * can use RCU again after this function is called. */ void tick_nohz_idle_exit(void) { @@ -1185,12 +1453,12 @@ void tick_nohz_idle_exit(void) local_irq_disable(); - WARN_ON_ONCE(!ts->inidle); + WARN_ON_ONCE(!tick_sched_flag_test(ts, TS_FLAG_INIDLE)); WARN_ON_ONCE(ts->timer_expires_base); - ts->inidle = 0; - idle_active = ts->idle_active; - tick_stopped = ts->tick_stopped; + tick_sched_flag_clear(ts, TS_FLAG_INIDLE); + idle_active = tick_sched_flag_test(ts, TS_FLAG_IDLE_ACTIVE); + tick_stopped = tick_sched_flag_test(ts, TS_FLAG_STOPPED); if (idle_active || tick_stopped) now = ktime_get(); @@ -1199,69 +1467,53 @@ void tick_nohz_idle_exit(void) tick_nohz_stop_idle(ts, now); if (tick_stopped) - __tick_nohz_idle_restart_tick(ts, now); + tick_nohz_idle_update_tick(ts, now); local_irq_enable(); } /* - * The nohz low res interrupt handler + * In low-resolution mode, the tick handler must be implemented directly + * at the clockevent level. hrtimer can't be used instead, because its + * infrastructure actually relies on the tick itself as a backend in + * low-resolution mode (see hrtimer_run_queues()). */ -static void tick_nohz_handler(struct clock_event_device *dev) +static void tick_nohz_lowres_handler(struct clock_event_device *dev) { struct tick_sched *ts = this_cpu_ptr(&tick_cpu_sched); - struct pt_regs *regs = get_irq_regs(); - ktime_t now = ktime_get(); dev->next_event = KTIME_MAX; - tick_sched_do_timer(ts, now); - tick_sched_handle(ts, regs); - - /* No need to reprogram if we are running tickless */ - if (unlikely(ts->tick_stopped)) - return; - - hrtimer_forward(&ts->sched_timer, now, tick_period); - tick_program_event(hrtimer_get_expires(&ts->sched_timer), 1); + if (likely(tick_nohz_handler(&ts->sched_timer) == HRTIMER_RESTART)) + tick_program_event(hrtimer_get_expires(&ts->sched_timer), 1); } -static inline void tick_nohz_activate(struct tick_sched *ts, int mode) +static inline void tick_nohz_activate(struct tick_sched *ts) { if (!tick_nohz_enabled) return; - ts->nohz_mode = mode; + tick_sched_flag_set(ts, TS_FLAG_NOHZ); /* One update is enough */ if (!test_and_set_bit(0, &tick_nohz_active)) timers_update_nohz(); } /** - * tick_nohz_switch_to_nohz - switch to nohz mode + * tick_nohz_switch_to_nohz - switch to NOHZ mode */ static void tick_nohz_switch_to_nohz(void) { - struct tick_sched *ts = this_cpu_ptr(&tick_cpu_sched); - ktime_t next; - if (!tick_nohz_enabled) return; - if (tick_switch_to_oneshot(tick_nohz_handler)) + if (tick_switch_to_oneshot(tick_nohz_lowres_handler)) return; /* - * Recycle the hrtimer in ts, so we can share the - * hrtimer_forward with the highres code. + * Recycle the hrtimer in 'ts', so we can share the + * highres code. */ - hrtimer_init(&ts->sched_timer, CLOCK_MONOTONIC, HRTIMER_MODE_ABS_HARD); - /* Get the next period */ - next = tick_init_jiffy_update(); - - hrtimer_set_expires(&ts->sched_timer, next); - hrtimer_forward_now(&ts->sched_timer, tick_period); - tick_program_event(hrtimer_get_expires(&ts->sched_timer), 1); - tick_nohz_activate(ts, NOHZ_MODE_LOWRES); + tick_setup_sched_timer(false); } static inline void tick_nohz_irq_enter(void) @@ -1269,12 +1521,19 @@ static inline void tick_nohz_irq_enter(void) struct tick_sched *ts = this_cpu_ptr(&tick_cpu_sched); ktime_t now; - if (!ts->idle_active && !ts->tick_stopped) + if (!tick_sched_flag_test(ts, TS_FLAG_STOPPED | TS_FLAG_IDLE_ACTIVE)) return; now = ktime_get(); - if (ts->idle_active) + if (tick_sched_flag_test(ts, TS_FLAG_IDLE_ACTIVE)) tick_nohz_stop_idle(ts, now); - if (ts->tick_stopped) + /* + * If all CPUs are idle we may need to update a stale jiffies value. + * Note nohz_full is a special case: a timekeeper is guaranteed to stay + * alive but it might be busy looping with interrupts disabled in some + * rare case (typically stop machine). So we must make sure we have a + * last resort. + */ + if (tick_sched_flag_test(ts, TS_FLAG_STOPPED)) tick_nohz_update_jiffies(now); } @@ -1282,12 +1541,12 @@ static inline void tick_nohz_irq_enter(void) static inline void tick_nohz_switch_to_nohz(void) { } static inline void tick_nohz_irq_enter(void) { } -static inline void tick_nohz_activate(struct tick_sched *ts, int mode) { } +static inline void tick_nohz_activate(struct tick_sched *ts) { } #endif /* CONFIG_NO_HZ_COMMON */ /* - * Called from irq_enter to notify about the possible interruption of idle() + * Called from irq_enter() to notify about the possible interruption of idle() */ void tick_irq_enter(void) { @@ -1295,41 +1554,6 @@ void tick_irq_enter(void) tick_nohz_irq_enter(); } -/* - * High resolution timer specific code - */ -#ifdef CONFIG_HIGH_RES_TIMERS -/* - * We rearm the timer until we get disabled by the idle code. - * Called with interrupts disabled. - */ -static enum hrtimer_restart tick_sched_timer(struct hrtimer *timer) -{ - struct tick_sched *ts = - container_of(timer, struct tick_sched, sched_timer); - struct pt_regs *regs = get_irq_regs(); - ktime_t now = ktime_get(); - - tick_sched_do_timer(ts, now); - - /* - * Do not call, when we are not in irq context and have - * no valid regs pointer - */ - if (regs) - tick_sched_handle(ts, regs); - else - ts->next_tick = 0; - - /* No need to reprogram if we are in idle or full dynticks mode */ - if (unlikely(ts->tick_stopped)) - return HRTIMER_NORESTART; - - hrtimer_forward(timer, now, tick_period); - - return HRTIMER_RESTART; -} - static int sched_skew_tick; static int __init skew_tick(char *str) @@ -1342,50 +1566,65 @@ early_param("skew_tick", skew_tick); /** * tick_setup_sched_timer - setup the tick emulation timer + * @hrtimer: whether to use the hrtimer or not */ -void tick_setup_sched_timer(void) +void tick_setup_sched_timer(bool hrtimer) { struct tick_sched *ts = this_cpu_ptr(&tick_cpu_sched); - ktime_t now = ktime_get(); - /* - * Emulate tick processing via per-CPU hrtimers: - */ + /* Emulate tick processing via per-CPU hrtimers: */ hrtimer_init(&ts->sched_timer, CLOCK_MONOTONIC, HRTIMER_MODE_ABS_HARD); - ts->sched_timer.function = tick_sched_timer; + + if (IS_ENABLED(CONFIG_HIGH_RES_TIMERS) && hrtimer) { + tick_sched_flag_set(ts, TS_FLAG_HIGHRES); + ts->sched_timer.function = tick_nohz_handler; + } /* Get the next period (per-CPU) */ hrtimer_set_expires(&ts->sched_timer, tick_init_jiffy_update()); - /* Offset the tick to avert jiffies_lock contention. */ + /* Offset the tick to avert 'jiffies_lock' contention. */ if (sched_skew_tick) { - u64 offset = ktime_to_ns(tick_period) >> 1; + u64 offset = TICK_NSEC >> 1; do_div(offset, num_possible_cpus()); offset *= smp_processor_id(); hrtimer_add_expires_ns(&ts->sched_timer, offset); } - hrtimer_forward(&ts->sched_timer, now, tick_period); - hrtimer_start_expires(&ts->sched_timer, HRTIMER_MODE_ABS_PINNED_HARD); - tick_nohz_activate(ts, NOHZ_MODE_HIGHRES); + hrtimer_forward_now(&ts->sched_timer, TICK_NSEC); + if (IS_ENABLED(CONFIG_HIGH_RES_TIMERS) && hrtimer) + hrtimer_start_expires(&ts->sched_timer, HRTIMER_MODE_ABS_PINNED_HARD); + else + tick_program_event(hrtimer_get_expires(&ts->sched_timer), 1); + tick_nohz_activate(ts); } -#endif /* HIGH_RES_TIMERS */ -#if defined CONFIG_NO_HZ_COMMON || defined CONFIG_HIGH_RES_TIMERS -void tick_cancel_sched_timer(int cpu) +/* + * Shut down the tick and make sure the CPU won't try to retake the timekeeping + * duty before disabling IRQs in idle for the last time. + */ +void tick_sched_timer_dying(int cpu) { struct tick_sched *ts = &per_cpu(tick_cpu_sched, cpu); + ktime_t idle_sleeptime, iowait_sleeptime; + unsigned long idle_calls, idle_sleeps; -# ifdef CONFIG_HIGH_RES_TIMERS - if (ts->sched_timer.base) + /* This must happen before hrtimers are migrated! */ + if (tick_sched_flag_test(ts, TS_FLAG_HIGHRES)) hrtimer_cancel(&ts->sched_timer); -# endif + idle_sleeptime = ts->idle_sleeptime; + iowait_sleeptime = ts->iowait_sleeptime; + idle_calls = ts->idle_calls; + idle_sleeps = ts->idle_sleeps; memset(ts, 0, sizeof(*ts)); + ts->idle_sleeptime = idle_sleeptime; + ts->iowait_sleeptime = iowait_sleeptime; + ts->idle_calls = idle_calls; + ts->idle_sleeps = idle_sleeps; } -#endif -/** +/* * Async notification about clocksource changes */ void tick_clock_notify(void) @@ -1406,11 +1645,11 @@ void tick_oneshot_notify(void) set_bit(0, &ts->check_clocks); } -/** - * Check, if a change happened, which makes oneshot possible. +/* + * Check if a change happened, which makes oneshot possible. * - * Called cyclic from the hrtimer softirq (driven by the timer - * softirq) allow_nohz signals, that we can switch into low-res nohz + * Called cyclically from the hrtimer softirq (driven by the timer + * softirq). 'allow_nohz' signals that we can switch into low-res NOHZ * mode, because high resolution timers are disabled (either compile * or runtime). Called with interrupts disabled. */ @@ -1421,7 +1660,7 @@ int tick_check_oneshot_change(int allow_nohz) if (!test_and_clear_bit(0, &ts->check_clocks)) return 0; - if (ts->nohz_mode != NOHZ_MODE_INACTIVE) + if (tick_sched_flag_test(ts, TS_FLAG_NOHZ)) return 0; if (!timekeeping_valid_for_hres() || !tick_is_oneshot_available()) diff --git a/kernel/time/tick-sched.h b/kernel/time/tick-sched.h index 4fb06527cf64..b4a7822f495d 100644 --- a/kernel/time/tick-sched.h +++ b/kernel/time/tick-sched.h @@ -14,78 +14,101 @@ struct tick_device { enum tick_device_mode mode; }; -enum tick_nohz_mode { - NOHZ_MODE_INACTIVE, - NOHZ_MODE_LOWRES, - NOHZ_MODE_HIGHRES, -}; +/* The CPU is in the tick idle mode */ +#define TS_FLAG_INIDLE BIT(0) +/* The idle tick has been stopped */ +#define TS_FLAG_STOPPED BIT(1) +/* + * Indicator that the CPU is actively in the tick idle mode; + * it is reset during irq handling phases. + */ +#define TS_FLAG_IDLE_ACTIVE BIT(2) +/* CPU was the last one doing do_timer before going idle */ +#define TS_FLAG_DO_TIMER_LAST BIT(3) +/* NO_HZ is enabled */ +#define TS_FLAG_NOHZ BIT(4) +/* High resolution tick mode */ +#define TS_FLAG_HIGHRES BIT(5) /** * struct tick_sched - sched tick emulation and no idle tick control/stats + * + * @flags: State flags gathering the TS_FLAG_* features + * @got_idle_tick: Tick timer function has run with @inidle set + * @stalled_jiffies: Number of stalled jiffies detected across ticks + * @last_tick_jiffies: Value of jiffies seen on last tick * @sched_timer: hrtimer to schedule the periodic tick in high * resolution mode - * @check_clocks: Notification mechanism about clocksource changes - * @nohz_mode: Mode - one state of tick_nohz_mode - * @inidle: Indicator that the CPU is in the tick idle mode - * @tick_stopped: Indicator that the idle tick has been stopped - * @idle_active: Indicator that the CPU is actively in the tick idle mode; - * it is resetted during irq handling phases. - * @do_timer_lst: CPU was the last one doing do_timer before going idle - * @got_idle_tick: Tick timer function has run with @inidle set * @last_tick: Store the last tick expiry time when the tick * timer is modified for nohz sleeps. This is necessary * to resume the tick timer operation in the timeline * when the CPU returns from nohz sleep. * @next_tick: Next tick to be fired when in dynticks mode. * @idle_jiffies: jiffies at the entry to idle for idle time accounting + * @idle_waketime: Time when the idle was interrupted + * @idle_sleeptime_seq: sequence counter for data consistency + * @idle_entrytime: Time when the idle call was entered + * @last_jiffies: Base jiffies snapshot when next event was last computed + * @timer_expires_base: Base time clock monotonic for @timer_expires + * @timer_expires: Anticipated timer expiration time (in case sched tick is stopped) + * @next_timer: Expiry time of next expiring timer for debugging purpose only + * @idle_expires: Next tick in idle, for debugging purpose only * @idle_calls: Total number of idle calls * @idle_sleeps: Number of idle calls, where the sched tick was stopped - * @idle_entrytime: Time when the idle call was entered - * @idle_waketime: Time when the idle was interrupted * @idle_exittime: Time when the idle state was left * @idle_sleeptime: Sum of the time slept in idle with sched tick stopped * @iowait_sleeptime: Sum of the time slept in idle with sched tick stopped, with IO outstanding - * @timer_expires: Anticipated timer expiration time (in case sched tick is stopped) - * @timer_expires_base: Base time clock monotonic for @timer_expires - * @next_timer: Expiry time of next expiring timer for debugging purpose only * @tick_dep_mask: Tick dependency mask - is set, if someone needs the tick + * @check_clocks: Notification mechanism about clocksource changes */ struct tick_sched { - struct hrtimer sched_timer; - unsigned long check_clocks; - enum tick_nohz_mode nohz_mode; + /* Common flags */ + unsigned long flags; - unsigned int inidle : 1; - unsigned int tick_stopped : 1; - unsigned int idle_active : 1; - unsigned int do_timer_last : 1; - unsigned int got_idle_tick : 1; + /* Tick handling: jiffies stall check */ + unsigned int stalled_jiffies; + unsigned long last_tick_jiffies; + /* Tick handling */ + struct hrtimer sched_timer; ktime_t last_tick; ktime_t next_tick; unsigned long idle_jiffies; - unsigned long idle_calls; - unsigned long idle_sleeps; - ktime_t idle_entrytime; ktime_t idle_waketime; - ktime_t idle_exittime; - ktime_t idle_sleeptime; - ktime_t iowait_sleeptime; + unsigned int got_idle_tick; + + /* Idle entry */ + seqcount_t idle_sleeptime_seq; + ktime_t idle_entrytime; + + /* Tick stop */ unsigned long last_jiffies; - u64 timer_expires; u64 timer_expires_base; + u64 timer_expires; u64 next_timer; ktime_t idle_expires; + unsigned long idle_calls; + unsigned long idle_sleeps; + + /* Idle exit */ + ktime_t idle_exittime; + ktime_t idle_sleeptime; + ktime_t iowait_sleeptime; + + /* Full dynticks handling */ atomic_t tick_dep_mask; + + /* Clocksource changes */ + unsigned long check_clocks; }; extern struct tick_sched *tick_get_tick_sched(int cpu); -extern void tick_setup_sched_timer(void); -#if defined CONFIG_NO_HZ_COMMON || defined CONFIG_HIGH_RES_TIMERS -extern void tick_cancel_sched_timer(int cpu); +extern void tick_setup_sched_timer(bool hrtimer); +#if defined CONFIG_TICK_ONESHOT +extern void tick_sched_timer_dying(int cpu); #else -static inline void tick_cancel_sched_timer(int cpu) { } +static inline void tick_sched_timer_dying(int cpu) { } #endif #ifdef CONFIG_GENERIC_CLOCKEVENTS_BROADCAST diff --git a/kernel/time/time.c b/kernel/time/time.c index 3985b2b32d08..1b69caa87480 100644 --- a/kernel/time/time.c +++ b/kernel/time/time.c @@ -365,11 +365,14 @@ SYSCALL_DEFINE1(adjtimex_time32, struct old_timex32 __user *, utp) } #endif -/* - * Convert jiffies to milliseconds and back. +/** + * jiffies_to_msecs - Convert jiffies to milliseconds + * @j: jiffies value * * Avoid unnecessary multiplications/divisions in the - * two most common HZ cases: + * two most common HZ cases. + * + * Return: milliseconds value */ unsigned int jiffies_to_msecs(const unsigned long j) { @@ -388,6 +391,12 @@ unsigned int jiffies_to_msecs(const unsigned long j) } EXPORT_SYMBOL(jiffies_to_msecs); +/** + * jiffies_to_usecs - Convert jiffies to microseconds + * @j: jiffies value + * + * Return: microseconds value + */ unsigned int jiffies_to_usecs(const unsigned long j) { /* @@ -408,8 +417,15 @@ unsigned int jiffies_to_usecs(const unsigned long j) } EXPORT_SYMBOL(jiffies_to_usecs); -/* +/** * mktime64 - Converts date to seconds. + * @year0: year to convert + * @mon0: month to convert + * @day: day to convert + * @hour: hour to convert + * @min: minute to convert + * @sec: second to convert + * * Converts Gregorian date to seconds since 1970-01-01 00:00:00. * Assumes input in normal date format, i.e. 1980-12-31 23:59:59 * => year=1980, mon=12, day=31, hour=23, min=59, sec=59. @@ -427,6 +443,8 @@ EXPORT_SYMBOL(jiffies_to_usecs); * * An encoding of midnight at the end of the day as 24:00:00 - ie. midnight * tomorrow - (allowable under ISO 8601) is supported. + * + * Return: seconds since the epoch time for the given input date */ time64_t mktime64(const unsigned int year0, const unsigned int mon0, const unsigned int day, const unsigned int hour, @@ -449,7 +467,7 @@ time64_t mktime64(const unsigned int year0, const unsigned int mon0, } EXPORT_SYMBOL(mktime64); -struct __kernel_old_timeval ns_to_kernel_old_timeval(const s64 nsec) +struct __kernel_old_timeval ns_to_kernel_old_timeval(s64 nsec) { struct timespec64 ts = ns_to_timespec64(nsec); struct __kernel_old_timeval tv; @@ -462,7 +480,7 @@ struct __kernel_old_timeval ns_to_kernel_old_timeval(const s64 nsec) EXPORT_SYMBOL(ns_to_kernel_old_timeval); /** - * set_normalized_timespec - set timespec sec and nsec parts and normalize + * set_normalized_timespec64 - set timespec sec and nsec parts and normalize * * @ts: pointer to timespec variable to be set * @sec: seconds to set @@ -471,8 +489,7 @@ EXPORT_SYMBOL(ns_to_kernel_old_timeval); * Set seconds and nanoseconds field of a timespec variable and * normalize to the timespec storage format * - * Note: The tv_nsec part is always in the range of - * 0 <= tv_nsec < NSEC_PER_SEC + * Note: The tv_nsec part is always in the range of 0 <= tv_nsec < NSEC_PER_SEC. * For negative values only the tv_sec field is negative ! */ void set_normalized_timespec64(struct timespec64 *ts, time64_t sec, s64 nsec) @@ -501,9 +518,9 @@ EXPORT_SYMBOL(set_normalized_timespec64); * ns_to_timespec64 - Convert nanoseconds to timespec64 * @nsec: the nanoseconds value to be converted * - * Returns the timespec64 representation of the nsec parameter. + * Return: the timespec64 representation of the nsec parameter. */ -struct timespec64 ns_to_timespec64(const s64 nsec) +struct timespec64 ns_to_timespec64(s64 nsec) { struct timespec64 ts = { 0, 0 }; s32 rem; @@ -526,7 +543,7 @@ struct timespec64 ns_to_timespec64(const s64 nsec) EXPORT_SYMBOL(ns_to_timespec64); /** - * msecs_to_jiffies: - convert milliseconds to jiffies + * __msecs_to_jiffies: - convert milliseconds to jiffies * @m: time in milliseconds * * conversion is done as follows: @@ -539,15 +556,17 @@ EXPORT_SYMBOL(ns_to_timespec64); * - all other values are converted to jiffies by either multiplying * the input value by a factor or dividing it with a factor and * handling any 32-bit overflows. - * for the details see __msecs_to_jiffies() + * for the details see _msecs_to_jiffies() * * msecs_to_jiffies() checks for the passed in value being a constant * via __builtin_constant_p() allowing gcc to eliminate most of the * code, __msecs_to_jiffies() is called if the value passed does not * allow constant folding and the actual conversion must be done at * runtime. - * the _msecs_to_jiffies helpers are the HZ dependent conversion + * The _msecs_to_jiffies helpers are the HZ dependent conversion * routines found in include/linux/jiffies.h + * + * Return: jiffies value */ unsigned long __msecs_to_jiffies(const unsigned int m) { @@ -560,6 +579,12 @@ unsigned long __msecs_to_jiffies(const unsigned int m) } EXPORT_SYMBOL(__msecs_to_jiffies); +/** + * __usecs_to_jiffies: - convert microseconds to jiffies + * @u: time in milliseconds + * + * Return: jiffies value + */ unsigned long __usecs_to_jiffies(const unsigned int u) { if (u > jiffies_to_usecs(MAX_JIFFY_OFFSET)) @@ -568,10 +593,13 @@ unsigned long __usecs_to_jiffies(const unsigned int u) } EXPORT_SYMBOL(__usecs_to_jiffies); -/* +/** + * timespec64_to_jiffies - convert a timespec64 value to jiffies + * @value: pointer to &struct timespec64 + * * The TICK_NSEC - 1 rounds up the value to the next resolution. Note * that a remainder subtract here would not do the right thing as the - * resolution values don't fall on second boundries. I.e. the line: + * resolution values don't fall on second boundaries. I.e. the line: * nsec -= nsec % TICK_NSEC; is NOT a correct resolution rounding. * Note that due to the small error in the multiplier here, this * rounding is incorrect for sufficiently large values of tv_nsec, but @@ -582,8 +610,9 @@ EXPORT_SYMBOL(__usecs_to_jiffies); * * The >> (NSEC_JIFFIE_SC - SEC_JIFFIE_SC) converts the scaled nsec * value to a scaled second value. + * + * Return: jiffies value */ - unsigned long timespec64_to_jiffies(const struct timespec64 *value) { @@ -601,6 +630,11 @@ timespec64_to_jiffies(const struct timespec64 *value) } EXPORT_SYMBOL(timespec64_to_jiffies); +/** + * jiffies_to_timespec64 - convert jiffies value to &struct timespec64 + * @jiffies: jiffies value + * @value: pointer to &struct timespec64 + */ void jiffies_to_timespec64(const unsigned long jiffies, struct timespec64 *value) { @@ -618,6 +652,13 @@ EXPORT_SYMBOL(jiffies_to_timespec64); /* * Convert jiffies/jiffies_64 to clock_t and back. */ + +/** + * jiffies_to_clock_t - Convert jiffies to clock_t + * @x: jiffies value + * + * Return: jiffies converted to clock_t (CLOCKS_PER_SEC) + */ clock_t jiffies_to_clock_t(unsigned long x) { #if (TICK_NSEC % (NSEC_PER_SEC / USER_HZ)) == 0 @@ -632,6 +673,12 @@ clock_t jiffies_to_clock_t(unsigned long x) } EXPORT_SYMBOL(jiffies_to_clock_t); +/** + * clock_t_to_jiffies - Convert clock_t to jiffies + * @x: clock_t value + * + * Return: clock_t value converted to jiffies + */ unsigned long clock_t_to_jiffies(unsigned long x) { #if (HZ % USER_HZ)==0 @@ -649,6 +696,12 @@ unsigned long clock_t_to_jiffies(unsigned long x) } EXPORT_SYMBOL(clock_t_to_jiffies); +/** + * jiffies_64_to_clock_t - Convert jiffies_64 to clock_t + * @x: jiffies_64 value + * + * Return: jiffies_64 value converted to 64-bit "clock_t" (CLOCKS_PER_SEC) + */ u64 jiffies_64_to_clock_t(u64 x) { #if (TICK_NSEC % (NSEC_PER_SEC / USER_HZ)) == 0 @@ -671,6 +724,12 @@ u64 jiffies_64_to_clock_t(u64 x) } EXPORT_SYMBOL(jiffies_64_to_clock_t); +/** + * nsec_to_clock_t - Convert nsec value to clock_t + * @x: nsec value + * + * Return: nsec value converted to 64-bit "clock_t" (CLOCKS_PER_SEC) + */ u64 nsec_to_clock_t(u64 x) { #if (NSEC_PER_SEC % USER_HZ) == 0 @@ -687,6 +746,12 @@ u64 nsec_to_clock_t(u64 x) #endif } +/** + * jiffies64_to_nsecs - Convert jiffies64 to nanoseconds + * @j: jiffies64 value + * + * Return: nanoseconds value + */ u64 jiffies64_to_nsecs(u64 j) { #if !(NSEC_PER_SEC % HZ) @@ -697,6 +762,12 @@ u64 jiffies64_to_nsecs(u64 j) } EXPORT_SYMBOL(jiffies64_to_nsecs); +/** + * jiffies64_to_msecs - Convert jiffies64 to milliseconds + * @j: jiffies64 value + * + * Return: milliseconds value + */ u64 jiffies64_to_msecs(const u64 j) { #if HZ <= MSEC_PER_SEC && !(MSEC_PER_SEC % HZ) @@ -719,6 +790,8 @@ EXPORT_SYMBOL(jiffies64_to_msecs); * note: * NSEC_PER_SEC = 10^9 = (5^9 * 2^9) = (1953125 * 512) * ULLONG_MAX ns = 18446744073.709551615 secs = about 584 years + * + * Return: nsecs converted to jiffies64 value */ u64 nsecs_to_jiffies64(u64 n) { @@ -750,6 +823,8 @@ EXPORT_SYMBOL(nsecs_to_jiffies64); * note: * NSEC_PER_SEC = 10^9 = (5^9 * 2^9) = (1953125 * 512) * ULLONG_MAX ns = 18446744073.709551615 secs = about 584 years + * + * Return: nsecs converted to jiffies value */ unsigned long nsecs_to_jiffies(u64 n) { @@ -757,10 +832,16 @@ unsigned long nsecs_to_jiffies(u64 n) } EXPORT_SYMBOL_GPL(nsecs_to_jiffies); -/* - * Add two timespec64 values and do a safety check for overflow. +/** + * timespec64_add_safe - Add two timespec64 values and do a safety check + * for overflow. + * @lhs: first (left) timespec64 to add + * @rhs: second (right) timespec64 to add + * * It's assumed that both values are valid (>= 0). * And, each timespec64 is in normalized form. + * + * Return: sum of @lhs + @rhs */ struct timespec64 timespec64_add_safe(const struct timespec64 lhs, const struct timespec64 rhs) @@ -778,6 +859,15 @@ struct timespec64 timespec64_add_safe(const struct timespec64 lhs, return res; } +/** + * get_timespec64 - get user's time value into kernel space + * @ts: destination &struct timespec64 + * @uts: user's time value as &struct __kernel_timespec + * + * Handles compat or 32-bit modes. + * + * Return: 0 on success or negative errno on error + */ int get_timespec64(struct timespec64 *ts, const struct __kernel_timespec __user *uts) { @@ -801,6 +891,14 @@ int get_timespec64(struct timespec64 *ts, } EXPORT_SYMBOL_GPL(get_timespec64); +/** + * put_timespec64 - convert timespec64 value to __kernel_timespec format and + * copy the latter to userspace + * @ts: input &struct timespec64 + * @uts: user's &struct __kernel_timespec + * + * Return: 0 on success or negative errno on error + */ int put_timespec64(const struct timespec64 *ts, struct __kernel_timespec __user *uts) { @@ -839,6 +937,15 @@ static int __put_old_timespec32(const struct timespec64 *ts64, return copy_to_user(cts, &ts, sizeof(ts)) ? -EFAULT : 0; } +/** + * get_old_timespec32 - get user's old-format time value into kernel space + * @ts: destination &struct timespec64 + * @uts: user's old-format time value (&struct old_timespec32) + * + * Handles X86_X32_ABI compatibility conversion. + * + * Return: 0 on success or negative errno on error + */ int get_old_timespec32(struct timespec64 *ts, const void __user *uts) { if (COMPAT_USE_64BIT_TIME) @@ -848,6 +955,16 @@ int get_old_timespec32(struct timespec64 *ts, const void __user *uts) } EXPORT_SYMBOL_GPL(get_old_timespec32); +/** + * put_old_timespec32 - convert timespec64 value to &struct old_timespec32 and + * copy the latter to userspace + * @ts: input &struct timespec64 + * @uts: user's &struct old_timespec32 + * + * Handles X86_X32_ABI compatibility conversion. + * + * Return: 0 on success or negative errno on error + */ int put_old_timespec32(const struct timespec64 *ts, void __user *uts) { if (COMPAT_USE_64BIT_TIME) @@ -857,6 +974,13 @@ int put_old_timespec32(const struct timespec64 *ts, void __user *uts) } EXPORT_SYMBOL_GPL(put_old_timespec32); +/** + * get_itimerspec64 - get user's &struct __kernel_itimerspec into kernel space + * @it: destination &struct itimerspec64 + * @uit: user's &struct __kernel_itimerspec + * + * Return: 0 on success or negative errno on error + */ int get_itimerspec64(struct itimerspec64 *it, const struct __kernel_itimerspec __user *uit) { @@ -872,6 +996,14 @@ int get_itimerspec64(struct itimerspec64 *it, } EXPORT_SYMBOL_GPL(get_itimerspec64); +/** + * put_itimerspec64 - convert &struct itimerspec64 to __kernel_itimerspec format + * and copy the latter to userspace + * @it: input &struct itimerspec64 + * @uit: user's &struct __kernel_itimerspec + * + * Return: 0 on success or negative errno on error + */ int put_itimerspec64(const struct itimerspec64 *it, struct __kernel_itimerspec __user *uit) { @@ -887,6 +1019,13 @@ int put_itimerspec64(const struct itimerspec64 *it, } EXPORT_SYMBOL_GPL(put_itimerspec64); +/** + * get_old_itimerspec32 - get user's &struct old_itimerspec32 into kernel space + * @its: destination &struct itimerspec64 + * @uits: user's &struct old_itimerspec32 + * + * Return: 0 on success or negative errno on error + */ int get_old_itimerspec32(struct itimerspec64 *its, const struct old_itimerspec32 __user *uits) { @@ -898,6 +1037,14 @@ int get_old_itimerspec32(struct itimerspec64 *its, } EXPORT_SYMBOL_GPL(get_old_itimerspec32); +/** + * put_old_itimerspec32 - convert &struct itimerspec64 to &struct + * old_itimerspec32 and copy the latter to userspace + * @its: input &struct itimerspec64 + * @uits: user's &struct old_itimerspec32 + * + * Return: 0 on success or negative errno on error + */ int put_old_itimerspec32(const struct itimerspec64 *its, struct old_itimerspec32 __user *uits) { diff --git a/kernel/time/time_test.c b/kernel/time/time_test.c new file mode 100644 index 000000000000..2889763165e5 --- /dev/null +++ b/kernel/time/time_test.c @@ -0,0 +1,100 @@ +// SPDX-License-Identifier: LGPL-2.1+ + +#include <kunit/test.h> +#include <linux/time.h> + +/* + * Traditional implementation of leap year evaluation. + */ +static bool is_leap(long year) +{ + return year % 4 == 0 && (year % 100 != 0 || year % 400 == 0); +} + +/* + * Gets the last day of a month. + */ +static int last_day_of_month(long year, int month) +{ + if (month == 2) + return 28 + is_leap(year); + if (month == 4 || month == 6 || month == 9 || month == 11) + return 30; + return 31; +} + +/* + * Advances a date by one day. + */ +static void advance_date(long *year, int *month, int *mday, int *yday) +{ + if (*mday != last_day_of_month(*year, *month)) { + ++*mday; + ++*yday; + return; + } + + *mday = 1; + if (*month != 12) { + ++*month; + ++*yday; + return; + } + + *month = 1; + *yday = 0; + ++*year; +} + +/* + * Checks every day in a 160000 years interval centered at 1970-01-01 + * against the expected result. + */ +static void time64_to_tm_test_date_range(struct kunit *test) +{ + /* + * 80000 years = (80000 / 400) * 400 years + * = (80000 / 400) * 146097 days + * = (80000 / 400) * 146097 * 86400 seconds + */ + time64_t total_secs = ((time64_t) 80000) / 400 * 146097 * 86400; + long year = 1970 - 80000; + int month = 1; + int mdday = 1; + int yday = 0; + + struct tm result; + time64_t secs; + s64 days; + + for (secs = -total_secs; secs <= total_secs; secs += 86400) { + + time64_to_tm(secs, 0, &result); + + days = div_s64(secs, 86400); + + #define FAIL_MSG "%05ld/%02d/%02d (%2d) : %lld", \ + year, month, mdday, yday, days + + KUNIT_ASSERT_EQ_MSG(test, year - 1900, result.tm_year, FAIL_MSG); + KUNIT_ASSERT_EQ_MSG(test, month - 1, result.tm_mon, FAIL_MSG); + KUNIT_ASSERT_EQ_MSG(test, mdday, result.tm_mday, FAIL_MSG); + KUNIT_ASSERT_EQ_MSG(test, yday, result.tm_yday, FAIL_MSG); + + advance_date(&year, &month, &mdday, &yday); + } +} + +static struct kunit_case time_test_cases[] = { + KUNIT_CASE_SLOW(time64_to_tm_test_date_range), + {} +}; + +static struct kunit_suite time_test_suite = { + .name = "time_test_cases", + .test_cases = time_test_cases, +}; + +kunit_test_suite(time_test_suite); +MODULE_DESCRIPTION("time unit test suite"); +MODULE_LICENSE("GPL"); diff --git a/kernel/time/timeconv.c b/kernel/time/timeconv.c index 589e0a552129..59b922c826e7 100644 --- a/kernel/time/timeconv.c +++ b/kernel/time/timeconv.c @@ -22,47 +22,16 @@ /* * Converts the calendar time to broken-down time representation - * Based on code from glibc-2.6 * * 2009-7-14: * Moved from glibc-2.6 to kernel by Zhaolei<zhaolei@cn.fujitsu.com> + * 2021-06-02: + * Reimplemented by Cassio Neri <cassio.neri@gmail.com> */ #include <linux/time.h> #include <linux/module.h> - -/* - * Nonzero if YEAR is a leap year (every 4 years, - * except every 100th isn't, and every 400th is). - */ -static int __isleap(long year) -{ - return (year) % 4 == 0 && ((year) % 100 != 0 || (year) % 400 == 0); -} - -/* do a mathdiv for long type */ -static long math_div(long a, long b) -{ - return a / b - (a % b < 0); -} - -/* How many leap years between y1 and y2, y1 must less or equal to y2 */ -static long leaps_between(long y1, long y2) -{ - long leaps1 = math_div(y1 - 1, 4) - math_div(y1 - 1, 100) - + math_div(y1 - 1, 400); - long leaps2 = math_div(y2 - 1, 4) - math_div(y2 - 1, 100) - + math_div(y2 - 1, 400); - return leaps2 - leaps1; -} - -/* How many days come before each month (0-12). */ -static const unsigned short __mon_yday[2][13] = { - /* Normal years. */ - {0, 31, 59, 90, 120, 151, 181, 212, 243, 273, 304, 334, 365}, - /* Leap years. */ - {0, 31, 60, 91, 121, 152, 182, 213, 244, 274, 305, 335, 366} -}; +#include <linux/kernel.h> #define SECS_PER_HOUR (60 * 60) #define SECS_PER_DAY (SECS_PER_HOUR * 24) @@ -70,16 +39,18 @@ static const unsigned short __mon_yday[2][13] = { /** * time64_to_tm - converts the calendar time to local broken-down time * - * @totalsecs the number of seconds elapsed since 00:00:00 on January 1, 1970, + * @totalsecs: the number of seconds elapsed since 00:00:00 on January 1, 1970, * Coordinated Universal Time (UTC). - * @offset offset seconds adding to totalsecs. - * @result pointer to struct tm variable to receive broken-down time + * @offset: offset seconds adding to totalsecs. + * @result: pointer to struct tm variable to receive broken-down time */ void time64_to_tm(time64_t totalsecs, int offset, struct tm *result) { - long days, rem, y; + u32 u32tmp, day_of_century, year_of_century, day_of_year, month, day; + u64 u64tmp, udays, century, year; + bool is_Jan_or_Feb, is_leap_year; + long days, rem; int remainder; - const unsigned short *ip; days = div_s64_rem(totalsecs, SECS_PER_DAY, &remainder); rem = remainder; @@ -103,27 +74,68 @@ void time64_to_tm(time64_t totalsecs, int offset, struct tm *result) if (result->tm_wday < 0) result->tm_wday += 7; - y = 1970; - - while (days < 0 || days >= (__isleap(y) ? 366 : 365)) { - /* Guess a corrected year, assuming 365 days per year. */ - long yg = y + math_div(days, 365); - - /* Adjust DAYS and Y to match the guessed year. */ - days -= (yg - y) * 365 + leaps_between(y, yg); - y = yg; - } - - result->tm_year = y - 1900; - - result->tm_yday = days; - - ip = __mon_yday[__isleap(y)]; - for (y = 11; days < ip[y]; y--) - continue; - days -= ip[y]; - - result->tm_mon = y; - result->tm_mday = days + 1; + /* + * The following algorithm is, basically, Proposition 6.3 of Neri + * and Schneider [1]. In a few words: it works on the computational + * (fictitious) calendar where the year starts in March, month = 2 + * (*), and finishes in February, month = 13. This calendar is + * mathematically convenient because the day of the year does not + * depend on whether the year is leap or not. For instance: + * + * March 1st 0-th day of the year; + * ... + * April 1st 31-st day of the year; + * ... + * January 1st 306-th day of the year; (Important!) + * ... + * February 28th 364-th day of the year; + * February 29th 365-th day of the year (if it exists). + * + * After having worked out the date in the computational calendar + * (using just arithmetics) it's easy to convert it to the + * corresponding date in the Gregorian calendar. + * + * [1] "Euclidean Affine Functions and Applications to Calendar + * Algorithms". https://arxiv.org/abs/2102.06959 + * + * (*) The numbering of months follows tm more closely and thus, + * is slightly different from [1]. + */ + + udays = ((u64) days) + 2305843009213814918ULL; + + u64tmp = 4 * udays + 3; + century = div64_u64_rem(u64tmp, 146097, &u64tmp); + day_of_century = (u32) (u64tmp / 4); + + u32tmp = 4 * day_of_century + 3; + u64tmp = 2939745ULL * u32tmp; + year_of_century = upper_32_bits(u64tmp); + day_of_year = lower_32_bits(u64tmp) / 2939745 / 4; + + year = 100 * century + year_of_century; + is_leap_year = year_of_century ? !(year_of_century % 4) : !(century % 4); + + u32tmp = 2141 * day_of_year + 132377; + month = u32tmp >> 16; + day = ((u16) u32tmp) / 2141; + + /* + * Recall that January 1st is the 306-th day of the year in the + * computational (not Gregorian) calendar. + */ + is_Jan_or_Feb = day_of_year >= 306; + + /* Convert to the Gregorian calendar and adjust to Unix time. */ + year = year + is_Jan_or_Feb - 6313183731940000ULL; + month = is_Jan_or_Feb ? month - 12 : month; + day = day + 1; + day_of_year += is_Jan_or_Feb ? -306 : 31 + 28 + is_leap_year; + + /* Convert to tm's format. */ + result->tm_year = (long) (year - 1900); + result->tm_mon = (int) month; + result->tm_mday = (int) day; + result->tm_yday = (int) day_of_year; } EXPORT_SYMBOL(time64_to_tm); diff --git a/kernel/time/timecounter.c b/kernel/time/timecounter.c index 85b98e727306..e6285288d765 100644 --- a/kernel/time/timecounter.c +++ b/kernel/time/timecounter.c @@ -76,7 +76,7 @@ static u64 cc_cyc2ns_backwards(const struct cyclecounter *cc, return ns; } -u64 timecounter_cyc2time(struct timecounter *tc, +u64 timecounter_cyc2time(const struct timecounter *tc, u64 cycle_tstamp) { u64 delta = (cycle_tstamp - tc->cycle_last) & tc->cc->mask; diff --git a/kernel/time/timekeeping.c b/kernel/time/timekeeping.c index d20d489841c8..1e67d076f195 100644 --- a/kernel/time/timekeeping.c +++ b/kernel/time/timekeeping.c @@ -17,19 +17,22 @@ #include <linux/clocksource.h> #include <linux/jiffies.h> #include <linux/time.h> +#include <linux/timex.h> #include <linux/tick.h> #include <linux/stop_machine.h> #include <linux/pvclock_gtod.h> #include <linux/compiler.h> #include <linux/audit.h> +#include <linux/random.h> #include "tick-internal.h" #include "ntp_internal.h" #include "timekeeping_internal.h" #define TK_CLEAR_NTP (1 << 0) -#define TK_MIRROR (1 << 1) -#define TK_CLOCK_WAS_SET (1 << 2) +#define TK_CLOCK_WAS_SET (1 << 1) + +#define TK_UPDATE_ALL (TK_CLEAR_NTP | TK_CLOCK_WAS_SET) enum timekeeping_adv_mode { /* Update timekeeper when a tick has passed */ @@ -43,15 +46,17 @@ enum timekeeping_adv_mode { * The most important data for readout fits into a single 64 byte * cache line. */ -static struct { - seqcount_t seq; +struct tk_data { + seqcount_raw_spinlock_t seq; struct timekeeper timekeeper; -} tk_core ____cacheline_aligned = { - .seq = SEQCNT_ZERO(tk_core.seq), -}; + struct timekeeper shadow_timekeeper; + raw_spinlock_t lock; +} ____cacheline_aligned; -static DEFINE_RAW_SPINLOCK(timekeeper_lock); -static struct timekeeper shadow_timekeeper; +static struct tk_data tk_core; + +/* flag for if timekeeping is suspended */ +int __read_mostly timekeeping_suspended; /** * struct tk_fast - NMI safe timekeeper @@ -63,7 +68,7 @@ static struct timekeeper shadow_timekeeper; * See @update_fast_timekeeper() below. */ struct tk_fast { - seqcount_t seq; + seqcount_latch_t seq; struct tk_read_base base[2]; }; @@ -72,25 +77,71 @@ static u64 cycles_at_suspend; static u64 dummy_clock_read(struct clocksource *cs) { - return cycles_at_suspend; + if (timekeeping_suspended) + return cycles_at_suspend; + return local_clock(); } static struct clocksource dummy_clock = { .read = dummy_clock_read, }; +/* + * Boot time initialization which allows local_clock() to be utilized + * during early boot when clocksources are not available. local_clock() + * returns nanoseconds already so no conversion is required, hence mult=1 + * and shift=0. When the first proper clocksource is installed then + * the fast time keepers are updated with the correct values. + */ +#define FAST_TK_INIT \ + { \ + .clock = &dummy_clock, \ + .mask = CLOCKSOURCE_MASK(64), \ + .mult = 1, \ + .shift = 0, \ + } + static struct tk_fast tk_fast_mono ____cacheline_aligned = { - .base[0] = { .clock = &dummy_clock, }, - .base[1] = { .clock = &dummy_clock, }, + .seq = SEQCNT_LATCH_ZERO(tk_fast_mono.seq), + .base[0] = FAST_TK_INIT, + .base[1] = FAST_TK_INIT, }; static struct tk_fast tk_fast_raw ____cacheline_aligned = { - .base[0] = { .clock = &dummy_clock, }, - .base[1] = { .clock = &dummy_clock, }, + .seq = SEQCNT_LATCH_ZERO(tk_fast_raw.seq), + .base[0] = FAST_TK_INIT, + .base[1] = FAST_TK_INIT, }; -/* flag for if timekeeping is suspended */ -int __read_mostly timekeeping_suspended; +unsigned long timekeeper_lock_irqsave(void) +{ + unsigned long flags; + + raw_spin_lock_irqsave(&tk_core.lock, flags); + return flags; +} + +void timekeeper_unlock_irqrestore(unsigned long flags) +{ + raw_spin_unlock_irqrestore(&tk_core.lock, flags); +} + +/* + * Multigrain timestamps require tracking the latest fine-grained timestamp + * that has been issued, and never returning a coarse-grained timestamp that is + * earlier than that value. + * + * mg_floor represents the latest fine-grained time that has been handed out as + * a file timestamp on the system. This is tracked as a monotonic ktime_t, and + * converted to a realtime clock value on an as-needed basis. + * + * Maintaining mg_floor ensures the multigrain interfaces never issue a + * timestamp earlier than one that has been previously issued. + * + * The exception to this rule is when there is a backward realtime clock jump. If + * such an event occurs, a timestamp can appear to be earlier than a previous one. + */ +static __cacheline_aligned_in_smp atomic64_t mg_floor; static inline void tk_normalize_xtime(struct timekeeper *tk) { @@ -139,13 +190,15 @@ static void tk_set_wall_to_mono(struct timekeeper *tk, struct timespec64 wtm) WARN_ON_ONCE(tk->offs_real != timespec64_to_ktime(tmp)); tk->wall_to_monotonic = wtm; set_normalized_timespec64(&tmp, -wtm.tv_sec, -wtm.tv_nsec); - tk->offs_real = timespec64_to_ktime(tmp); - tk->offs_tai = ktime_add(tk->offs_real, ktime_set(tk->tai_offset, 0)); + /* Paired with READ_ONCE() in ktime_mono_to_any() */ + WRITE_ONCE(tk->offs_real, timespec64_to_ktime(tmp)); + WRITE_ONCE(tk->offs_tai, ktime_add(tk->offs_real, ktime_set(tk->tai_offset, 0))); } static inline void tk_update_sleep_time(struct timekeeper *tk, ktime_t delta) { - tk->offs_boot = ktime_add(tk->offs_boot, delta); + /* Paired with READ_ONCE() in ktime_mono_to_any() */ + WRITE_ONCE(tk->offs_boot, ktime_add(tk->offs_boot, delta)); /* * Timespec representation for VDSO update to avoid 64bit division * on every update. @@ -157,12 +210,12 @@ static inline void tk_update_sleep_time(struct timekeeper *tk, ktime_t delta) * tk_clock_read - atomic clocksource read() helper * * This helper is necessary to use in the read paths because, while the - * seqlock ensures we don't return a bad value while structures are updated, + * seqcount ensures we don't return a bad value while structures are updated, * it doesn't protect from potential crashes. There is the possibility that * the tkr's clocksource may change between the read reference, and the * clock reference passed to the read function. This can cause crashes if * the wrong clocksource is passed to the wrong read function. - * This isn't necessary to use when holding the timekeeper_lock or doing + * This isn't necessary to use when holding the tk_core.lock or doing * a read of the fast-timekeeper tkrs (which is protected by its own locking * and update logic). */ @@ -173,106 +226,6 @@ static inline u64 tk_clock_read(const struct tk_read_base *tkr) return clock->read(clock); } -#ifdef CONFIG_DEBUG_TIMEKEEPING -#define WARNING_FREQ (HZ*300) /* 5 minute rate-limiting */ - -static void timekeeping_check_update(struct timekeeper *tk, u64 offset) -{ - - u64 max_cycles = tk->tkr_mono.clock->max_cycles; - const char *name = tk->tkr_mono.clock->name; - - if (offset > max_cycles) { - printk_deferred("WARNING: timekeeping: Cycle offset (%lld) is larger than allowed by the '%s' clock's max_cycles value (%lld): time overflow danger\n", - offset, name, max_cycles); - printk_deferred(" timekeeping: Your kernel is sick, but tries to cope by capping time updates\n"); - } else { - if (offset > (max_cycles >> 1)) { - printk_deferred("INFO: timekeeping: Cycle offset (%lld) is larger than the '%s' clock's 50%% safety margin (%lld)\n", - offset, name, max_cycles >> 1); - printk_deferred(" timekeeping: Your kernel is still fine, but is feeling a bit nervous\n"); - } - } - - if (tk->underflow_seen) { - if (jiffies - tk->last_warning > WARNING_FREQ) { - printk_deferred("WARNING: Underflow in clocksource '%s' observed, time update ignored.\n", name); - printk_deferred(" Please report this, consider using a different clocksource, if possible.\n"); - printk_deferred(" Your kernel is probably still fine.\n"); - tk->last_warning = jiffies; - } - tk->underflow_seen = 0; - } - - if (tk->overflow_seen) { - if (jiffies - tk->last_warning > WARNING_FREQ) { - printk_deferred("WARNING: Overflow in clocksource '%s' observed, time update capped.\n", name); - printk_deferred(" Please report this, consider using a different clocksource, if possible.\n"); - printk_deferred(" Your kernel is probably still fine.\n"); - tk->last_warning = jiffies; - } - tk->overflow_seen = 0; - } -} - -static inline u64 timekeeping_get_delta(const struct tk_read_base *tkr) -{ - struct timekeeper *tk = &tk_core.timekeeper; - u64 now, last, mask, max, delta; - unsigned int seq; - - /* - * Since we're called holding a seqlock, the data may shift - * under us while we're doing the calculation. This can cause - * false positives, since we'd note a problem but throw the - * results away. So nest another seqlock here to atomically - * grab the points we are checking with. - */ - do { - seq = read_seqcount_begin(&tk_core.seq); - now = tk_clock_read(tkr); - last = tkr->cycle_last; - mask = tkr->mask; - max = tkr->clock->max_cycles; - } while (read_seqcount_retry(&tk_core.seq, seq)); - - delta = clocksource_delta(now, last, mask); - - /* - * Try to catch underflows by checking if we are seeing small - * mask-relative negative values. - */ - if (unlikely((~delta & mask) < (mask >> 3))) { - tk->underflow_seen = 1; - delta = 0; - } - - /* Cap delta value to the max_cycles values to avoid mult overflows */ - if (unlikely(delta > max)) { - tk->overflow_seen = 1; - delta = tkr->clock->max_cycles; - } - - return delta; -} -#else -static inline void timekeeping_check_update(struct timekeeper *tk, u64 offset) -{ -} -static inline u64 timekeeping_get_delta(const struct tk_read_base *tkr) -{ - u64 cycle_now, delta; - - /* read clocksource */ - cycle_now = tk_clock_read(tkr); - - /* calculate the delta since the last update_wall_time */ - delta = clocksource_delta(cycle_now, tkr->cycle_last, tkr->mask); - - return delta; -} -#endif - /** * tk_setup_internals - Set up internals to use clocksource clock. * @@ -348,50 +301,49 @@ static void tk_setup_internals(struct timekeeper *tk, struct clocksource *clock) } /* Timekeeper helper functions. */ - -#ifdef CONFIG_ARCH_USES_GETTIMEOFFSET -static u32 default_arch_gettimeoffset(void) { return 0; } -u32 (*arch_gettimeoffset)(void) = default_arch_gettimeoffset; -#else -static inline u32 arch_gettimeoffset(void) { return 0; } -#endif - -static inline u64 timekeeping_delta_to_ns(const struct tk_read_base *tkr, u64 delta) +static noinline u64 delta_to_ns_safe(const struct tk_read_base *tkr, u64 delta) { - u64 nsec; - - nsec = delta * tkr->mult + tkr->xtime_nsec; - nsec >>= tkr->shift; - - /* If arch requires, add in get_arch_timeoffset() */ - return nsec + arch_gettimeoffset(); + return mul_u64_u32_add_u64_shr(delta, tkr->mult, tkr->xtime_nsec, tkr->shift); } -static inline u64 timekeeping_get_ns(const struct tk_read_base *tkr) +static inline u64 timekeeping_cycles_to_ns(const struct tk_read_base *tkr, u64 cycles) { - u64 delta; + /* Calculate the delta since the last update_wall_time() */ + u64 mask = tkr->mask, delta = (cycles - tkr->cycle_last) & mask; - delta = timekeeping_get_delta(tkr); - return timekeeping_delta_to_ns(tkr, delta); + /* + * This detects both negative motion and the case where the delta + * overflows the multiplication with tkr->mult. + */ + if (unlikely(delta > tkr->clock->max_cycles)) { + /* + * Handle clocksource inconsistency between CPUs to prevent + * time from going backwards by checking for the MSB of the + * mask being set in the delta. + */ + if (delta & ~(mask >> 1)) + return tkr->xtime_nsec >> tkr->shift; + + return delta_to_ns_safe(tkr, delta); + } + + return ((delta * tkr->mult) + tkr->xtime_nsec) >> tkr->shift; } -static inline u64 timekeeping_cycles_to_ns(const struct tk_read_base *tkr, u64 cycles) +static __always_inline u64 timekeeping_get_ns(const struct tk_read_base *tkr) { - u64 delta; - - /* calculate the delta since the last update_wall_time */ - delta = clocksource_delta(cycles, tkr->cycle_last, tkr->mask); - return timekeeping_delta_to_ns(tkr, delta); + return timekeeping_cycles_to_ns(tkr, tk_clock_read(tkr)); } /** * update_fast_timekeeper - Update the fast and NMI safe monotonic timekeeper. * @tkr: Timekeeping readout base from which we take the update + * @tkf: Pointer to NMI safe timekeeper * * We want to use this from any context including NMI and tracing / * instrumenting the timekeeping code itself. * - * Employ the latch technique; see @raw_write_seqcount_latch. + * Employ the latch technique; see @write_seqcount_latch. * * So if a NMI hits the update of base[0] then it will use base[1] * which is still consistent. In the worst case this can result is a @@ -404,16 +356,34 @@ static void update_fast_timekeeper(const struct tk_read_base *tkr, struct tk_read_base *base = tkf->base; /* Force readers off to base[1] */ - raw_write_seqcount_latch(&tkf->seq); + write_seqcount_latch_begin(&tkf->seq); /* Update base[0] */ memcpy(base, tkr, sizeof(*base)); /* Force readers back to base[0] */ - raw_write_seqcount_latch(&tkf->seq); + write_seqcount_latch(&tkf->seq); /* Update base[1] */ memcpy(base + 1, base, sizeof(*base)); + + write_seqcount_latch_end(&tkf->seq); +} + +static __always_inline u64 __ktime_get_fast_ns(struct tk_fast *tkf) +{ + struct tk_read_base *tkr; + unsigned int seq; + u64 now; + + do { + seq = read_seqcount_latch(&tkf->seq); + tkr = tkf->base + (seq & 0x01); + now = ktime_to_ns(tkr->base); + now += timekeeping_get_ns(tkr); + } while (read_seqcount_latch_retry(&tkf->seq, seq)); + + return now; } /** @@ -442,40 +412,25 @@ static void update_fast_timekeeper(const struct tk_read_base *tkr, * * So reader 6 will observe time going backwards versus reader 5. * - * While other CPUs are likely to be able observe that, the only way + * While other CPUs are likely to be able to observe that, the only way * for a CPU local observation is when an NMI hits in the middle of * the update. Timestamps taken from that NMI context might be ahead * of the following timestamps. Callers need to be aware of that and * deal with it. */ -static __always_inline u64 __ktime_get_fast_ns(struct tk_fast *tkf) -{ - struct tk_read_base *tkr; - unsigned int seq; - u64 now; - - do { - seq = raw_read_seqcount_latch(&tkf->seq); - tkr = tkf->base + (seq & 0x01); - now = ktime_to_ns(tkr->base); - - now += timekeeping_delta_to_ns(tkr, - clocksource_delta( - tk_clock_read(tkr), - tkr->cycle_last, - tkr->mask)); - } while (read_seqcount_retry(&tkf->seq, seq)); - - return now; -} - -u64 ktime_get_mono_fast_ns(void) +u64 notrace ktime_get_mono_fast_ns(void) { return __ktime_get_fast_ns(&tk_fast_mono); } EXPORT_SYMBOL_GPL(ktime_get_mono_fast_ns); -u64 ktime_get_raw_fast_ns(void) +/** + * ktime_get_raw_fast_ns - Fast NMI safe access to clock monotonic raw + * + * Contrary to ktime_get_mono_fast_ns() this is always correct because the + * conversion factor is not affected by NTP/PTP correction. + */ +u64 notrace ktime_get_raw_fast_ns(void) { return __ktime_get_fast_ns(&tk_fast_raw); } @@ -486,7 +441,7 @@ EXPORT_SYMBOL_GPL(ktime_get_raw_fast_ns); * * To keep it NMI safe since we're accessing from tracing, we're not using a * separate timekeeper with updates to monotonic clock and boot offset - * protected with seqlocks. This has the following minor side effects: + * protected with seqcounts. This has the following minor side effects: * * (1) Its possible that a timestamp be taken after the boot offset is updated * but before the timekeeper is updated. If this happens, the new boot offset @@ -496,51 +451,60 @@ EXPORT_SYMBOL_GPL(ktime_get_raw_fast_ns); * timekeeping_inject_sleeptime64() * __timekeeping_inject_sleeptime(tk, delta); * timestamp(); - * timekeeping_update(tk, TK_CLEAR_NTP...); + * timekeeping_update_staged(tkd, TK_CLEAR_NTP...); * * (2) On 32-bit systems, the 64-bit boot offset (tk->offs_boot) may be * partially updated. Since the tk->offs_boot update is a rare event, this * should be a rare occurrence which postprocessing should be able to handle. + * + * The caveats vs. timestamp ordering as documented for ktime_get_mono_fast_ns() + * apply as well. */ u64 notrace ktime_get_boot_fast_ns(void) { struct timekeeper *tk = &tk_core.timekeeper; - return (ktime_get_mono_fast_ns() + ktime_to_ns(tk->offs_boot)); + return (ktime_get_mono_fast_ns() + ktime_to_ns(data_race(tk->offs_boot))); } EXPORT_SYMBOL_GPL(ktime_get_boot_fast_ns); - -/* - * See comment for __ktime_get_fast_ns() vs. timestamp ordering +/** + * ktime_get_tai_fast_ns - NMI safe and fast access to tai clock. + * + * The same limitations as described for ktime_get_boot_fast_ns() apply. The + * mono time and the TAI offset are not read atomically which may yield wrong + * readouts. However, an update of the TAI offset is an rare event e.g., caused + * by settime or adjtimex with an offset. The user of this function has to deal + * with the possibility of wrong timestamps in post processing. */ -static __always_inline u64 __ktime_get_real_fast_ns(struct tk_fast *tkf) +u64 notrace ktime_get_tai_fast_ns(void) { - struct tk_read_base *tkr; - unsigned int seq; - u64 now; - - do { - seq = raw_read_seqcount_latch(&tkf->seq); - tkr = tkf->base + (seq & 0x01); - now = ktime_to_ns(tkr->base_real); - - now += timekeeping_delta_to_ns(tkr, - clocksource_delta( - tk_clock_read(tkr), - tkr->cycle_last, - tkr->mask)); - } while (read_seqcount_retry(&tkf->seq, seq)); + struct timekeeper *tk = &tk_core.timekeeper; - return now; + return (ktime_get_mono_fast_ns() + ktime_to_ns(data_race(tk->offs_tai))); } +EXPORT_SYMBOL_GPL(ktime_get_tai_fast_ns); /** * ktime_get_real_fast_ns: - NMI safe and fast access to clock realtime. + * + * See ktime_get_mono_fast_ns() for documentation of the time stamp ordering. */ u64 ktime_get_real_fast_ns(void) { - return __ktime_get_real_fast_ns(&tk_fast_mono); + struct tk_fast *tkf = &tk_fast_mono; + struct tk_read_base *tkr; + u64 baser, delta; + unsigned int seq; + + do { + seq = raw_read_seqcount_latch(&tkf->seq); + tkr = tkf->base + (seq & 0x01); + baser = ktime_to_ns(tkr->base_real); + delta = timekeeping_get_ns(tkr); + } while (raw_read_seqcount_latch_retry(&tkf->seq, seq)); + + return baser + delta; } EXPORT_SYMBOL_GPL(ktime_get_real_fast_ns); @@ -580,17 +544,16 @@ static void update_pvclock_gtod(struct timekeeper *tk, bool was_set) /** * pvclock_gtod_register_notifier - register a pvclock timedata update listener + * @nb: Pointer to the notifier block to register */ int pvclock_gtod_register_notifier(struct notifier_block *nb) { struct timekeeper *tk = &tk_core.timekeeper; - unsigned long flags; int ret; - raw_spin_lock_irqsave(&timekeeper_lock, flags); + guard(raw_spinlock_irqsave)(&tk_core.lock); ret = raw_notifier_chain_register(&pvclock_gtod_chain, nb); update_pvclock_gtod(tk, true); - raw_spin_unlock_irqrestore(&timekeeper_lock, flags); return ret; } @@ -599,17 +562,12 @@ EXPORT_SYMBOL_GPL(pvclock_gtod_register_notifier); /** * pvclock_gtod_unregister_notifier - unregister a pvclock * timedata update listener + * @nb: Pointer to the notifier block to unregister */ int pvclock_gtod_unregister_notifier(struct notifier_block *nb) { - unsigned long flags; - int ret; - - raw_spin_lock_irqsave(&timekeeper_lock, flags); - ret = raw_notifier_chain_unregister(&pvclock_gtod_chain, nb); - raw_spin_unlock_irqrestore(&timekeeper_lock, flags); - - return ret; + guard(raw_spinlock_irqsave)(&tk_core.lock); + return raw_notifier_chain_unregister(&pvclock_gtod_chain, nb); } EXPORT_SYMBOL_GPL(pvclock_gtod_unregister_notifier); @@ -625,6 +583,18 @@ static inline void tk_update_leap_state(struct timekeeper *tk) } /* + * Leap state update for both shadow and the real timekeeper + * Separate to spare a full memcpy() of the timekeeper. + */ +static void tk_update_leap_state_all(struct tk_data *tkd) +{ + write_seqcount_begin(&tkd->seq); + tk_update_leap_state(&tkd->shadow_timekeeper); + tkd->timekeeper.next_leap_ktime = tkd->shadow_timekeeper.next_leap_ktime; + write_seqcount_end(&tkd->seq); +} + +/* * Update the ktime_t based scalar nsec members of the timekeeper */ static inline void tk_update_ktime_data(struct timekeeper *tk) @@ -657,9 +627,30 @@ static inline void tk_update_ktime_data(struct timekeeper *tk) tk->tkr_raw.base = ns_to_ktime(tk->raw_sec * NSEC_PER_SEC); } -/* must hold timekeeper_lock */ -static void timekeeping_update(struct timekeeper *tk, unsigned int action) +/* + * Restore the shadow timekeeper from the real timekeeper. + */ +static void timekeeping_restore_shadow(struct tk_data *tkd) { + lockdep_assert_held(&tkd->lock); + memcpy(&tkd->shadow_timekeeper, &tkd->timekeeper, sizeof(tkd->timekeeper)); +} + +static void timekeeping_update_from_shadow(struct tk_data *tkd, unsigned int action) +{ + struct timekeeper *tk = &tk_core.shadow_timekeeper; + + lockdep_assert_held(&tkd->lock); + + /* + * Block out readers before running the updates below because that + * updates VDSO and other time related infrastructure. Not blocking + * the readers might let a reader see time going backwards when + * reading from the VDSO after the VDSO update and then reading in + * the kernel from the timekeeper before that got updated. + */ + write_seqcount_begin(&tkd->seq); + if (action & TK_CLEAR_NTP) { tk->ntp_error = 0; ntp_clear(); @@ -677,18 +668,22 @@ static void timekeeping_update(struct timekeeper *tk, unsigned int action) if (action & TK_CLOCK_WAS_SET) tk->clock_was_set_seq++; + /* - * The mirroring of the data to the shadow-timekeeper needs - * to happen last here to ensure we don't over-write the - * timekeeper structure on the next update with stale data + * Update the real timekeeper. + * + * We could avoid this memcpy() by switching pointers, but that has + * the downside that the reader side does not longer benefit from + * the cacheline optimized data layout of the timekeeper and requires + * another indirection. */ - if (action & TK_MIRROR) - memcpy(&shadow_timekeeper, &tk_core.timekeeper, - sizeof(tk_core.timekeeper)); + memcpy(&tkd->timekeeper, tk, sizeof(*tk)); + write_seqcount_end(&tkd->seq); } /** * timekeeping_forward_now - update clock to the current time + * @tk: Pointer to the timekeeper to update * * Forward the current clock to update its state since the last call to * update_wall_time(). This is useful before significant clock changes, @@ -699,22 +694,20 @@ static void timekeeping_forward_now(struct timekeeper *tk) u64 cycle_now, delta; cycle_now = tk_clock_read(&tk->tkr_mono); - delta = clocksource_delta(cycle_now, tk->tkr_mono.cycle_last, tk->tkr_mono.mask); + delta = clocksource_delta(cycle_now, tk->tkr_mono.cycle_last, tk->tkr_mono.mask, + tk->tkr_mono.clock->max_raw_delta); tk->tkr_mono.cycle_last = cycle_now; tk->tkr_raw.cycle_last = cycle_now; - tk->tkr_mono.xtime_nsec += delta * tk->tkr_mono.mult; - - /* If arch requires, add in get_arch_timeoffset() */ - tk->tkr_mono.xtime_nsec += (u64)arch_gettimeoffset() << tk->tkr_mono.shift; - - - tk->tkr_raw.xtime_nsec += delta * tk->tkr_raw.mult; + while (delta > 0) { + u64 max = tk->tkr_mono.clock->max_cycles; + u64 incr = delta < max ? delta : max; - /* If arch requires, add in get_arch_timeoffset() */ - tk->tkr_raw.xtime_nsec += (u64)arch_gettimeoffset() << tk->tkr_raw.shift; - - tk_normalize_xtime(tk); + tk->tkr_mono.xtime_nsec += incr * tk->tkr_mono.mult; + tk->tkr_raw.xtime_nsec += incr * tk->tkr_raw.mult; + tk_normalize_xtime(tk); + delta -= incr; + } } /** @@ -829,7 +822,7 @@ ktime_t ktime_get_coarse_with_offset(enum tk_offsets offs) EXPORT_SYMBOL_GPL(ktime_get_coarse_with_offset); /** - * ktime_mono_to_any() - convert mononotic time to any other time + * ktime_mono_to_any() - convert monotonic time to any other time * @tmono: time to convert. * @offs: which offset to use */ @@ -839,6 +832,14 @@ ktime_t ktime_mono_to_any(ktime_t tmono, enum tk_offsets offs) unsigned int seq; ktime_t tconv; + if (IS_ENABLED(CONFIG_64BIT)) { + /* + * Paired with WRITE_ONCE()s in tk_set_wall_to_mono() and + * tk_update_sleep_time(). + */ + return ktime_add(tmono, READ_ONCE(*offset)); + } + do { seq = read_seqcount_begin(&tk_core.seq); tconv = ktime_add(tmono, *offset); @@ -921,8 +922,7 @@ EXPORT_SYMBOL_GPL(ktime_get_seconds); /** * ktime_get_real_seconds - Get the seconds portion of CLOCK_REALTIME * - * Returns the wall clock seconds since 1970. This replaces the - * get_seconds() interface which is not y2038 safe on 32bit systems. + * Returns the wall clock seconds since 1970. * * For 64bit systems the fast access to tk->xtime_sec is preserved. On * 32bit systems the access must be protected with the sequence @@ -970,6 +970,7 @@ void ktime_get_snapshot(struct system_time_snapshot *systime_snapshot) unsigned int seq; ktime_t base_raw; ktime_t base_real; + ktime_t base_boot; u64 nsec_raw; u64 nsec_real; u64 now; @@ -979,10 +980,13 @@ void ktime_get_snapshot(struct system_time_snapshot *systime_snapshot) do { seq = read_seqcount_begin(&tk_core.seq); now = tk_clock_read(&tk->tkr_mono); + systime_snapshot->cs_id = tk->tkr_mono.clock->id; systime_snapshot->cs_was_changed_seq = tk->cs_was_changed_seq; systime_snapshot->clock_was_set_seq = tk->clock_was_set_seq; base_real = ktime_add(tk->tkr_mono.base, tk_core.timekeeper.offs_real); + base_boot = ktime_add(tk->tkr_mono.base, + tk_core.timekeeper.offs_boot); base_raw = tk->tkr_raw.base; nsec_real = timekeeping_cycles_to_ns(&tk->tkr_mono, now); nsec_raw = timekeeping_cycles_to_ns(&tk->tkr_raw, now); @@ -990,6 +994,7 @@ void ktime_get_snapshot(struct system_time_snapshot *systime_snapshot) systime_snapshot->cycles = now; systime_snapshot->real = ktime_add_ns(base_real, nsec_real); + systime_snapshot->boot = ktime_add_ns(base_boot, nsec_real); systime_snapshot->raw = ktime_add_ns(base_raw, nsec_raw); } EXPORT_SYMBOL_GPL(ktime_get_snapshot); @@ -1091,17 +1096,121 @@ static int adjust_historical_crosststamp(struct system_time_snapshot *history, } /* - * cycle_between - true if test occurs chronologically between before and after + * timestamp_in_interval - true if ts is chronologically in [start, end] + * + * True if ts occurs chronologically at or after start, and before or at end. */ -static bool cycle_between(u64 before, u64 test, u64 after) +static bool timestamp_in_interval(u64 start, u64 end, u64 ts) { - if (test > before && test < after) + if (ts >= start && ts <= end) return true; - if (test < before && before > after) + if (start > end && (ts >= start || ts <= end)) return true; return false; } +static bool convert_clock(u64 *val, u32 numerator, u32 denominator) +{ + u64 rem, res; + + if (!numerator || !denominator) + return false; + + res = div64_u64_rem(*val, denominator, &rem) * numerator; + *val = res + div_u64(rem * numerator, denominator); + return true; +} + +static bool convert_base_to_cs(struct system_counterval_t *scv) +{ + struct clocksource *cs = tk_core.timekeeper.tkr_mono.clock; + struct clocksource_base *base; + u32 num, den; + + /* The timestamp was taken from the time keeper clock source */ + if (cs->id == scv->cs_id) + return true; + + /* + * Check whether cs_id matches the base clock. Prevent the compiler from + * re-evaluating @base as the clocksource might change concurrently. + */ + base = READ_ONCE(cs->base); + if (!base || base->id != scv->cs_id) + return false; + + num = scv->use_nsecs ? cs->freq_khz : base->numerator; + den = scv->use_nsecs ? USEC_PER_SEC : base->denominator; + + if (!convert_clock(&scv->cycles, num, den)) + return false; + + scv->cycles += base->offset; + return true; +} + +static bool convert_cs_to_base(u64 *cycles, enum clocksource_ids base_id) +{ + struct clocksource *cs = tk_core.timekeeper.tkr_mono.clock; + struct clocksource_base *base; + + /* + * Check whether base_id matches the base clock. Prevent the compiler from + * re-evaluating @base as the clocksource might change concurrently. + */ + base = READ_ONCE(cs->base); + if (!base || base->id != base_id) + return false; + + *cycles -= base->offset; + if (!convert_clock(cycles, base->denominator, base->numerator)) + return false; + return true; +} + +static bool convert_ns_to_cs(u64 *delta) +{ + struct tk_read_base *tkr = &tk_core.timekeeper.tkr_mono; + + if (BITS_TO_BYTES(fls64(*delta) + tkr->shift) >= sizeof(*delta)) + return false; + + *delta = div_u64((*delta << tkr->shift) - tkr->xtime_nsec, tkr->mult); + return true; +} + +/** + * ktime_real_to_base_clock() - Convert CLOCK_REALTIME timestamp to a base clock timestamp + * @treal: CLOCK_REALTIME timestamp to convert + * @base_id: base clocksource id + * @cycles: pointer to store the converted base clock timestamp + * + * Converts a supplied, future realtime clock value to the corresponding base clock value. + * + * Return: true if the conversion is successful, false otherwise. + */ +bool ktime_real_to_base_clock(ktime_t treal, enum clocksource_ids base_id, u64 *cycles) +{ + struct timekeeper *tk = &tk_core.timekeeper; + unsigned int seq; + u64 delta; + + do { + seq = read_seqcount_begin(&tk_core.seq); + if ((u64)treal < tk->tkr_mono.base_real) + return false; + delta = (u64)treal - tk->tkr_mono.base_real; + if (!convert_ns_to_cs(&delta)) + return false; + *cycles = tk->tkr_mono.cycle_last + delta; + if (!convert_cs_to_base(cycles, base_id)) + return false; + } while (read_seqcount_retry(&tk_core.seq, seq)); + + return true; +} +EXPORT_SYMBOL_GPL(ktime_real_to_base_clock); + /** * get_device_system_crosststamp - Synchronously capture system/device timestamp * @get_time_fn: Callback to get simultaneous device time and @@ -1143,11 +1252,12 @@ int get_device_system_crosststamp(int (*get_time_fn) return ret; /* - * Verify that the clocksource associated with the captured - * system counter value is the same as the currently installed - * timekeeper clocksource + * Verify that the clocksource ID associated with the captured + * system counter value is the same as for the currently + * installed timekeeper clocksource */ - if (tk->tkr_mono.clock != system_counterval.cs) + if (system_counterval.cs_id == CSID_GENERIC || + !convert_base_to_cs(&system_counterval)) return -ENODEV; cycles = system_counterval.cycles; @@ -1157,7 +1267,7 @@ int get_device_system_crosststamp(int (*get_time_fn) */ now = tk_clock_read(&tk->tkr_mono); interval_start = tk->tkr_mono.cycle_last; - if (!cycle_between(interval_start, cycles, now)) { + if (!timestamp_in_interval(interval_start, now, cycles)) { clock_was_set_seq = tk->clock_was_set_seq; cs_was_changed_seq = tk->cs_was_changed_seq; cycles = interval_start; @@ -1170,10 +1280,8 @@ int get_device_system_crosststamp(int (*get_time_fn) tk_core.timekeeper.offs_real); base_raw = tk->tkr_raw.base; - nsec_real = timekeeping_cycles_to_ns(&tk->tkr_mono, - system_counterval.cycles); - nsec_raw = timekeeping_cycles_to_ns(&tk->tkr_raw, - system_counterval.cycles); + nsec_real = timekeeping_cycles_to_ns(&tk->tkr_mono, cycles); + nsec_raw = timekeeping_cycles_to_ns(&tk->tkr_raw, cycles); } while (read_seqcount_retry(&tk_core.seq, seq)); xtstamp->sys_realtime = ktime_add_ns(base_real, nsec_real); @@ -1188,13 +1296,13 @@ int get_device_system_crosststamp(int (*get_time_fn) bool discontinuity; /* - * Check that the counter value occurs after the provided + * Check that the counter value is not before the provided * history reference and that the history doesn't cross a * clocksource change */ if (!history_begin || - !cycle_between(history_begin->cycles, - system_counterval.cycles, cycles) || + !timestamp_in_interval(history_begin->cycles, + cycles, system_counterval.cycles) || history_begin->cs_was_changed_seq != cs_was_changed_seq) return -EINVAL; partial_history_cycles = cycles - system_counterval.cycles; @@ -1215,6 +1323,30 @@ int get_device_system_crosststamp(int (*get_time_fn) EXPORT_SYMBOL_GPL(get_device_system_crosststamp); /** + * timekeeping_clocksource_has_base - Check whether the current clocksource + * is based on given a base clock + * @id: base clocksource ID + * + * Note: The return value is a snapshot which can become invalid right + * after the function returns. + * + * Return: true if the timekeeper clocksource has a base clock with @id, + * false otherwise + */ +bool timekeeping_clocksource_has_base(enum clocksource_ids id) +{ + /* + * This is a snapshot, so no point in using the sequence + * count. Just prevent the compiler from re-evaluating @base as the + * clocksource might change concurrently. + */ + struct clocksource_base *base = READ_ONCE(tk_core.timekeeper.tkr_mono.clock->base); + + return base ? base->id == id : false; +} +EXPORT_SYMBOL_GPL(timekeeping_clocksource_has_base); + +/** * do_settimeofday64 - Sets the time of day. * @ts: pointer to the timespec64 variable containing the new time * @@ -1222,89 +1354,71 @@ EXPORT_SYMBOL_GPL(get_device_system_crosststamp); */ int do_settimeofday64(const struct timespec64 *ts) { - struct timekeeper *tk = &tk_core.timekeeper; struct timespec64 ts_delta, xt; - unsigned long flags; - int ret = 0; if (!timespec64_valid_settod(ts)) return -EINVAL; - raw_spin_lock_irqsave(&timekeeper_lock, flags); - write_seqcount_begin(&tk_core.seq); - - timekeeping_forward_now(tk); - - xt = tk_xtime(tk); - ts_delta.tv_sec = ts->tv_sec - xt.tv_sec; - ts_delta.tv_nsec = ts->tv_nsec - xt.tv_nsec; + scoped_guard (raw_spinlock_irqsave, &tk_core.lock) { + struct timekeeper *tks = &tk_core.shadow_timekeeper; - if (timespec64_compare(&tk->wall_to_monotonic, &ts_delta) > 0) { - ret = -EINVAL; - goto out; - } - - tk_set_wall_to_mono(tk, timespec64_sub(tk->wall_to_monotonic, ts_delta)); + timekeeping_forward_now(tks); - tk_set_xtime(tk, ts); -out: - timekeeping_update(tk, TK_CLEAR_NTP | TK_MIRROR | TK_CLOCK_WAS_SET); + xt = tk_xtime(tks); + ts_delta = timespec64_sub(*ts, xt); - write_seqcount_end(&tk_core.seq); - raw_spin_unlock_irqrestore(&timekeeper_lock, flags); + if (timespec64_compare(&tks->wall_to_monotonic, &ts_delta) > 0) { + timekeeping_restore_shadow(&tk_core); + return -EINVAL; + } - /* signal hrtimers about time change */ - clock_was_set(); + tk_set_wall_to_mono(tks, timespec64_sub(tks->wall_to_monotonic, ts_delta)); + tk_set_xtime(tks, ts); + timekeeping_update_from_shadow(&tk_core, TK_UPDATE_ALL); + } - if (!ret) - audit_tk_injoffset(ts_delta); + /* Signal hrtimers about time change */ + clock_was_set(CLOCK_SET_WALL); - return ret; + audit_tk_injoffset(ts_delta); + add_device_randomness(ts, sizeof(*ts)); + return 0; } EXPORT_SYMBOL(do_settimeofday64); /** * timekeeping_inject_offset - Adds or subtracts from the current time. - * @tv: pointer to the timespec variable containing the offset + * @ts: Pointer to the timespec variable containing the offset * * Adds or subtracts an offset value from the current time. */ static int timekeeping_inject_offset(const struct timespec64 *ts) { - struct timekeeper *tk = &tk_core.timekeeper; - unsigned long flags; - struct timespec64 tmp; - int ret = 0; - if (ts->tv_nsec < 0 || ts->tv_nsec >= NSEC_PER_SEC) return -EINVAL; - raw_spin_lock_irqsave(&timekeeper_lock, flags); - write_seqcount_begin(&tk_core.seq); - - timekeeping_forward_now(tk); - - /* Make sure the proposed value is valid */ - tmp = timespec64_add(tk_xtime(tk), *ts); - if (timespec64_compare(&tk->wall_to_monotonic, ts) > 0 || - !timespec64_valid_settod(&tmp)) { - ret = -EINVAL; - goto error; - } - - tk_xtime_add(tk, ts); - tk_set_wall_to_mono(tk, timespec64_sub(tk->wall_to_monotonic, *ts)); + scoped_guard (raw_spinlock_irqsave, &tk_core.lock) { + struct timekeeper *tks = &tk_core.shadow_timekeeper; + struct timespec64 tmp; -error: /* even if we error out, we forwarded the time, so call update */ - timekeeping_update(tk, TK_CLEAR_NTP | TK_MIRROR | TK_CLOCK_WAS_SET); + timekeeping_forward_now(tks); - write_seqcount_end(&tk_core.seq); - raw_spin_unlock_irqrestore(&timekeeper_lock, flags); + /* Make sure the proposed value is valid */ + tmp = timespec64_add(tk_xtime(tks), *ts); + if (timespec64_compare(&tks->wall_to_monotonic, ts) > 0 || + !timespec64_valid_settod(&tmp)) { + timekeeping_restore_shadow(&tk_core); + return -EINVAL; + } - /* signal hrtimers about time change */ - clock_was_set(); + tk_xtime_add(tks, ts); + tk_set_wall_to_mono(tks, timespec64_sub(tks->wall_to_monotonic, *ts)); + timekeeping_update_from_shadow(&tk_core, TK_UPDATE_ALL); + } - return ret; + /* Signal hrtimers about time change */ + clock_was_set(CLOCK_SET_WALL); + return 0; } /* @@ -1341,9 +1455,8 @@ void timekeeping_warp_clock(void) } } -/** +/* * __timekeeping_set_tai_offset - Sets the TAI offset from UTC and monotonic - * */ static void __timekeeping_set_tai_offset(struct timekeeper *tk, s32 tai_offset) { @@ -1351,42 +1464,43 @@ static void __timekeeping_set_tai_offset(struct timekeeper *tk, s32 tai_offset) tk->offs_tai = ktime_add(tk->offs_real, ktime_set(tai_offset, 0)); } -/** +/* * change_clocksource - Swaps clocksources if a new one is available * * Accumulates current time interval and initializes new clocksource */ static int change_clocksource(void *data) { - struct timekeeper *tk = &tk_core.timekeeper; - struct clocksource *new, *old; - unsigned long flags; - - new = (struct clocksource *) data; + struct clocksource *new = data, *old = NULL; - raw_spin_lock_irqsave(&timekeeper_lock, flags); - write_seqcount_begin(&tk_core.seq); - - timekeeping_forward_now(tk); /* - * If the cs is in module, get a module reference. Succeeds - * for built-in code (owner == NULL) as well. + * If the clocksource is in a module, get a module reference. + * Succeeds for built-in code (owner == NULL) as well. Abort if the + * reference can't be acquired. */ - if (try_module_get(new->owner)) { - if (!new->enable || new->enable(new) == 0) { - old = tk->tkr_mono.clock; - tk_setup_internals(tk, new); - if (old->disable) - old->disable(old); - module_put(old->owner); - } else { - module_put(new->owner); - } + if (!try_module_get(new->owner)) + return 0; + + /* Abort if the device can't be enabled */ + if (new->enable && new->enable(new) != 0) { + module_put(new->owner); + return 0; } - timekeeping_update(tk, TK_CLEAR_NTP | TK_MIRROR | TK_CLOCK_WAS_SET); - write_seqcount_end(&tk_core.seq); - raw_spin_unlock_irqrestore(&timekeeper_lock, flags); + scoped_guard (raw_spinlock_irqsave, &tk_core.lock) { + struct timekeeper *tks = &tk_core.shadow_timekeeper; + + timekeeping_forward_now(tks); + old = tks->tkr_mono.clock; + tk_setup_internals(tks, new); + timekeeping_update_from_shadow(&tk_core, TK_UPDATE_ALL); + } + + if (old) { + if (old->disable) + old->disable(old); + module_put(old->owner); + } return 0; } @@ -1474,6 +1588,7 @@ u64 timekeeping_max_deferment(void) /** * read_persistent_clock64 - Return time from the persistent clock. + * @ts: Pointer to the storage for the readout value * * Weak dummy function for arches that do not yet support it. * Reads the time from the battery backed persistent clock. @@ -1490,10 +1605,11 @@ void __weak read_persistent_clock64(struct timespec64 *ts) /** * read_persistent_wall_and_boot_offset - Read persistent clock, and also offset * from the boot. + * @wall_time: current time as returned by persistent clock + * @boot_offset: offset that is defined as wall_time - boot_time * * Weak dummy function for arches that do not yet support it. - * wall_time - current time as returned by persistent clock - * boot_offset - offset that is defined as wall_time - boot_time + * * The default function calculates offset based on the current value of * local_clock(). This way architectures that support sched_clock() but don't * support dedicated boot time clock will provide the best estimate of the @@ -1507,6 +1623,12 @@ read_persistent_wall_and_boot_offset(struct timespec64 *wall_time, *boot_offset = ns_to_timespec64(local_clock()); } +static __init void tkd_basic_setup(struct tk_data *tkd) +{ + raw_spin_lock_init(&tkd->lock); + seqcount_raw_spinlock_init(&tkd->seq, &tkd->lock); +} + /* * Flag reflecting whether timekeeping_resume() has injected sleeptime. * @@ -1531,9 +1653,10 @@ static bool persistent_clock_exists; void __init timekeeping_init(void) { struct timespec64 wall_time, boot_offset, wall_to_mono; - struct timekeeper *tk = &tk_core.timekeeper; + struct timekeeper *tks = &tk_core.shadow_timekeeper; struct clocksource *clock; - unsigned long flags; + + tkd_basic_setup(&tk_core); read_persistent_wall_and_boot_offset(&wall_time, &boot_offset); if (timespec64_valid_settod(&wall_time) && @@ -1553,24 +1676,21 @@ void __init timekeeping_init(void) */ wall_to_mono = timespec64_sub(boot_offset, wall_time); - raw_spin_lock_irqsave(&timekeeper_lock, flags); - write_seqcount_begin(&tk_core.seq); + guard(raw_spinlock_irqsave)(&tk_core.lock); + ntp_init(); clock = clocksource_default_clock(); if (clock->enable) clock->enable(clock); - tk_setup_internals(tk, clock); + tk_setup_internals(tks, clock); - tk_set_xtime(tk, &wall_time); - tk->raw_sec = 0; + tk_set_xtime(tks, &wall_time); + tks->raw_sec = 0; - tk_set_wall_to_mono(tk, wall_to_mono); + tk_set_wall_to_mono(tks, wall_to_mono); - timekeeping_update(tk, TK_MIRROR | TK_CLOCK_WAS_SET); - - write_seqcount_end(&tk_core.seq); - raw_spin_unlock_irqrestore(&timekeeper_lock, flags); + timekeeping_update_from_shadow(&tk_core, TK_CLOCK_WAS_SET); } /* time in seconds when suspend began for persistent clock */ @@ -1578,7 +1698,8 @@ static struct timespec64 timekeeping_suspend_time; /** * __timekeeping_inject_sleeptime - Internal function to add sleep interval - * @delta: pointer to a timespec delta value + * @tk: Pointer to the timekeeper to be updated + * @delta: Pointer to the delta value in timespec64 format * * Takes a timespec offset measuring a suspend interval and properly * adds the sleep offset to the timekeeping variables. @@ -1599,7 +1720,7 @@ static void __timekeeping_inject_sleeptime(struct timekeeper *tk, } #if defined(CONFIG_PM_SLEEP) && defined(CONFIG_RTC_HCTOSYS_DEVICE) -/** +/* * We have three kinds of time sources to use for sleep time * injection, the preference order is: * 1) non-stop clocksource @@ -1620,7 +1741,7 @@ bool timekeeping_rtc_skipresume(void) return !suspend_timing_needed; } -/** +/* * 1) can be determined whether to use or not only when doing * timekeeping_resume() which is invoked after rtc_suspend(), * so we can't skip rtc_suspend() surely if system has 1). @@ -1647,25 +1768,17 @@ bool timekeeping_rtc_skipsuspend(void) */ void timekeeping_inject_sleeptime64(const struct timespec64 *delta) { - struct timekeeper *tk = &tk_core.timekeeper; - unsigned long flags; - - raw_spin_lock_irqsave(&timekeeper_lock, flags); - write_seqcount_begin(&tk_core.seq); - - suspend_timing_needed = false; - - timekeeping_forward_now(tk); - - __timekeeping_inject_sleeptime(tk, delta); - - timekeeping_update(tk, TK_CLEAR_NTP | TK_MIRROR | TK_CLOCK_WAS_SET); + scoped_guard(raw_spinlock_irqsave, &tk_core.lock) { + struct timekeeper *tks = &tk_core.shadow_timekeeper; - write_seqcount_end(&tk_core.seq); - raw_spin_unlock_irqrestore(&timekeeper_lock, flags); + suspend_timing_needed = false; + timekeeping_forward_now(tks); + __timekeeping_inject_sleeptime(tks, delta); + timekeeping_update_from_shadow(&tk_core, TK_UPDATE_ALL); + } - /* signal hrtimers about time change */ - clock_was_set(); + /* Signal hrtimers about time change */ + clock_was_set(CLOCK_SET_WALL | CLOCK_SET_BOOT); } #endif @@ -1674,20 +1787,19 @@ void timekeeping_inject_sleeptime64(const struct timespec64 *delta) */ void timekeeping_resume(void) { - struct timekeeper *tk = &tk_core.timekeeper; - struct clocksource *clock = tk->tkr_mono.clock; - unsigned long flags; + struct timekeeper *tks = &tk_core.shadow_timekeeper; + struct clocksource *clock = tks->tkr_mono.clock; struct timespec64 ts_new, ts_delta; - u64 cycle_now, nsec; bool inject_sleeptime = false; + u64 cycle_now, nsec; + unsigned long flags; read_persistent_clock64(&ts_new); clockevents_resume(); clocksource_resume(); - raw_spin_lock_irqsave(&timekeeper_lock, flags); - write_seqcount_begin(&tk_core.seq); + raw_spin_lock_irqsave(&tk_core.lock, flags); /* * After system resumes, we need to calculate the suspended time and @@ -1701,7 +1813,7 @@ void timekeeping_resume(void) * The less preferred source will only be tried if there is no better * usable source. The rtc part is handled separately in rtc core code. */ - cycle_now = tk_clock_read(&tk->tkr_mono); + cycle_now = tk_clock_read(&tks->tkr_mono); nsec = clocksource_stop_suspend_timing(clock, cycle_now); if (nsec > 0) { ts_delta = ns_to_timespec64(nsec); @@ -1713,32 +1825,33 @@ void timekeeping_resume(void) if (inject_sleeptime) { suspend_timing_needed = false; - __timekeeping_inject_sleeptime(tk, &ts_delta); + __timekeeping_inject_sleeptime(tks, &ts_delta); } /* Re-base the last cycle value */ - tk->tkr_mono.cycle_last = cycle_now; - tk->tkr_raw.cycle_last = cycle_now; + tks->tkr_mono.cycle_last = cycle_now; + tks->tkr_raw.cycle_last = cycle_now; - tk->ntp_error = 0; + tks->ntp_error = 0; timekeeping_suspended = 0; - timekeeping_update(tk, TK_MIRROR | TK_CLOCK_WAS_SET); - write_seqcount_end(&tk_core.seq); - raw_spin_unlock_irqrestore(&timekeeper_lock, flags); + timekeeping_update_from_shadow(&tk_core, TK_CLOCK_WAS_SET); + raw_spin_unlock_irqrestore(&tk_core.lock, flags); touch_softlockup_watchdog(); + /* Resume the clockevent device(s) and hrtimers */ tick_resume(); - hrtimers_resume(); + /* Notify timerfd as resume is equivalent to clock_was_set() */ + timerfd_resume(); } int timekeeping_suspend(void) { - struct timekeeper *tk = &tk_core.timekeeper; - unsigned long flags; - struct timespec64 delta, delta_delta; - static struct timespec64 old_delta; + struct timekeeper *tks = &tk_core.shadow_timekeeper; + struct timespec64 delta, delta_delta; + static struct timespec64 old_delta; struct clocksource *curr_clock; + unsigned long flags; u64 cycle_now; read_persistent_clock64(&timekeeping_suspend_time); @@ -1753,9 +1866,8 @@ int timekeeping_suspend(void) suspend_timing_needed = true; - raw_spin_lock_irqsave(&timekeeper_lock, flags); - write_seqcount_begin(&tk_core.seq); - timekeeping_forward_now(tk); + raw_spin_lock_irqsave(&tk_core.lock, flags); + timekeeping_forward_now(tks); timekeeping_suspended = 1; /* @@ -1763,8 +1875,8 @@ int timekeeping_suspend(void) * just read from the current clocksource. Save this to potentially * use in suspend timing. */ - curr_clock = tk->tkr_mono.clock; - cycle_now = tk->tkr_mono.cycle_last; + curr_clock = tks->tkr_mono.clock; + cycle_now = tks->tkr_mono.cycle_last; clocksource_start_suspend_timing(curr_clock, cycle_now); if (persistent_clock_exists) { @@ -1774,7 +1886,7 @@ int timekeeping_suspend(void) * try to compensate so the difference in system time * and persistent_clock time stays close to constant. */ - delta = timespec64_sub(tk_xtime(tk), timekeeping_suspend_time); + delta = timespec64_sub(tk_xtime(tks), timekeeping_suspend_time); delta_delta = timespec64_sub(delta, old_delta); if (abs(delta_delta.tv_sec) >= 2) { /* @@ -1789,10 +1901,9 @@ int timekeeping_suspend(void) } } - timekeeping_update(tk, TK_MIRROR); - halt_fast_timekeeper(tk); - write_seqcount_end(&tk_core.seq); - raw_spin_unlock_irqrestore(&timekeeper_lock, flags); + timekeeping_update_from_shadow(&tk_core, 0); + halt_fast_timekeeper(tks); + raw_spin_unlock_irqrestore(&tk_core.lock, flags); tick_suspend(); clocksource_suspend(); @@ -1877,7 +1988,7 @@ static __always_inline void timekeeping_apply_adjustment(struct timekeeper *tk, * xtime_nsec_1 = offset + xtime_nsec_2 * Which gives us: * xtime_nsec_2 = xtime_nsec_1 - offset - * Which simplfies to: + * Which simplifies to: * xtime_nsec -= offset */ if ((mult_adj > 0) && (tk->tkr_mono.mult + mult_adj < mult_adj)) { @@ -1897,16 +2008,17 @@ static __always_inline void timekeeping_apply_adjustment(struct timekeeper *tk, */ static void timekeeping_adjust(struct timekeeper *tk, s64 offset) { + u64 ntp_tl = ntp_tick_length(); u32 mult; /* * Determine the multiplier from the current NTP tick length. * Avoid expensive division when the tick length doesn't change. */ - if (likely(tk->ntp_tick == ntp_tick_length())) { + if (likely(tk->ntp_tick == ntp_tl)) { mult = tk->tkr_mono.mult - tk->ntp_err_mult; } else { - tk->ntp_tick = ntp_tick_length(); + tk->ntp_tick = ntp_tl; mult = div64_u64((tk->ntp_tick >> tk->ntp_error_shift) - tk->xtime_remainder, tk->cycle_interval); } @@ -1949,13 +2061,12 @@ static void timekeeping_adjust(struct timekeeper *tk, s64 offset) } } -/** +/* * accumulate_nsecs_to_secs - Accumulates nsecs into secs * * Helper function that accumulates the nsecs greater than a second * from the xtime_nsec field to the xtime_secs field. * It also calls into the NTP code to handle leapsecond processing. - * */ static inline unsigned int accumulate_nsecs_to_secs(struct timekeeper *tk) { @@ -1997,11 +2108,11 @@ static inline unsigned int accumulate_nsecs_to_secs(struct timekeeper *tk) return clock_set; } -/** +/* * logarithmic_accumulation - shifted accumulation of cycles * * This functions accumulates a shifted interval of cycles into - * into a shifted interval nanoseconds. Allows for O(log) accumulation + * a shifted interval nanoseconds. Allows for O(log) accumulation * loop. * * Returns the unconsumed cycles. @@ -2044,37 +2155,27 @@ static u64 logarithmic_accumulation(struct timekeeper *tk, u64 offset, * timekeeping_advance - Updates the timekeeper to the current time and * current NTP tick length */ -static void timekeeping_advance(enum timekeeping_adv_mode mode) +static bool timekeeping_advance(enum timekeeping_adv_mode mode) { + struct timekeeper *tk = &tk_core.shadow_timekeeper; struct timekeeper *real_tk = &tk_core.timekeeper; - struct timekeeper *tk = &shadow_timekeeper; - u64 offset; - int shift = 0, maxshift; unsigned int clock_set = 0; - unsigned long flags; + int shift = 0, maxshift; + u64 offset; - raw_spin_lock_irqsave(&timekeeper_lock, flags); + guard(raw_spinlock_irqsave)(&tk_core.lock); /* Make sure we're fully resumed: */ if (unlikely(timekeeping_suspended)) - goto out; - -#ifdef CONFIG_ARCH_USES_GETTIMEOFFSET - offset = real_tk->cycle_interval; + return false; - if (mode != TK_ADV_TICK) - goto out; -#else offset = clocksource_delta(tk_clock_read(&tk->tkr_mono), - tk->tkr_mono.cycle_last, tk->tkr_mono.mask); + tk->tkr_mono.cycle_last, tk->tkr_mono.mask, + tk->tkr_mono.clock->max_raw_delta); /* Check if there's really nothing to do */ if (offset < real_tk->cycle_interval && mode == TK_ADV_TICK) - goto out; -#endif - - /* Do some additional sanity checking */ - timekeeping_check_update(tk, offset); + return false; /* * With NO_HZ we may have to accumulate many cycle_intervals @@ -2090,8 +2191,7 @@ static void timekeeping_advance(enum timekeeping_adv_mode mode) maxshift = (64 - (ilog2(ntp_tick_length())+1)) - 1; shift = min(shift, maxshift); while (offset >= tk->cycle_interval) { - offset = logarithmic_accumulation(tk, offset, shift, - &clock_set); + offset = logarithmic_accumulation(tk, offset, shift, &clock_set); if (offset < tk->cycle_interval<<shift) shift--; } @@ -2105,26 +2205,9 @@ static void timekeeping_advance(enum timekeeping_adv_mode mode) */ clock_set |= accumulate_nsecs_to_secs(tk); - write_seqcount_begin(&tk_core.seq); - /* - * Update the real timekeeper. - * - * We could avoid this memcpy by switching pointers, but that - * requires changes to all other timekeeper usage sites as - * well, i.e. move the timekeeper pointer getter into the - * spinlocked/seqcount protected sections. And we trade this - * memcpy under the tk_core.seq against one before we start - * updating. - */ - timekeeping_update(tk, clock_set); - memcpy(real_tk, tk, sizeof(*tk)); - /* The memcpy must come last. Do not put anything here! */ - write_seqcount_end(&tk_core.seq); -out: - raw_spin_unlock_irqrestore(&timekeeper_lock, flags); - if (clock_set) - /* Have to call _delayed version, since in irq context*/ - clock_was_set_delayed(); + timekeeping_update_from_shadow(&tk_core, clock_set); + + return !!clock_set; } /** @@ -2133,7 +2216,8 @@ out: */ void update_wall_time(void) { - timekeeping_advance(TK_ADV_TICK); + if (timekeeping_advance(TK_ADV_TICK)) + clock_was_set_delayed(); } /** @@ -2169,6 +2253,94 @@ void ktime_get_coarse_real_ts64(struct timespec64 *ts) } EXPORT_SYMBOL(ktime_get_coarse_real_ts64); +/** + * ktime_get_coarse_real_ts64_mg - return latter of coarse grained time or floor + * @ts: timespec64 to be filled + * + * Fetch the global mg_floor value, convert it to realtime and compare it + * to the current coarse-grained time. Fill @ts with whichever is + * latest. Note that this is a filesystem-specific interface and should be + * avoided outside of that context. + */ +void ktime_get_coarse_real_ts64_mg(struct timespec64 *ts) +{ + struct timekeeper *tk = &tk_core.timekeeper; + u64 floor = atomic64_read(&mg_floor); + ktime_t f_real, offset, coarse; + unsigned int seq; + + do { + seq = read_seqcount_begin(&tk_core.seq); + *ts = tk_xtime(tk); + offset = tk_core.timekeeper.offs_real; + } while (read_seqcount_retry(&tk_core.seq, seq)); + + coarse = timespec64_to_ktime(*ts); + f_real = ktime_add(floor, offset); + if (ktime_after(f_real, coarse)) + *ts = ktime_to_timespec64(f_real); +} + +/** + * ktime_get_real_ts64_mg - attempt to update floor value and return result + * @ts: pointer to the timespec to be set + * + * Get a monotonic fine-grained time value and attempt to swap it into + * mg_floor. If that succeeds then accept the new floor value. If it fails + * then another task raced in during the interim time and updated the + * floor. Since any update to the floor must be later than the previous + * floor, either outcome is acceptable. + * + * Typically this will be called after calling ktime_get_coarse_real_ts64_mg(), + * and determining that the resulting coarse-grained timestamp did not effect + * a change in ctime. Any more recent floor value would effect a change to + * ctime, so there is no need to retry the atomic64_try_cmpxchg() on failure. + * + * @ts will be filled with the latest floor value, regardless of the outcome of + * the cmpxchg. Note that this is a filesystem specific interface and should be + * avoided outside of that context. + */ +void ktime_get_real_ts64_mg(struct timespec64 *ts) +{ + struct timekeeper *tk = &tk_core.timekeeper; + ktime_t old = atomic64_read(&mg_floor); + ktime_t offset, mono; + unsigned int seq; + u64 nsecs; + + do { + seq = read_seqcount_begin(&tk_core.seq); + + ts->tv_sec = tk->xtime_sec; + mono = tk->tkr_mono.base; + nsecs = timekeeping_get_ns(&tk->tkr_mono); + offset = tk_core.timekeeper.offs_real; + } while (read_seqcount_retry(&tk_core.seq, seq)); + + mono = ktime_add_ns(mono, nsecs); + + /* + * Attempt to update the floor with the new time value. As any + * update must be later then the existing floor, and would effect + * a change to ctime from the perspective of the current task, + * accept the resulting floor value regardless of the outcome of + * the swap. + */ + if (atomic64_try_cmpxchg(&mg_floor, &old, mono)) { + ts->tv_nsec = 0; + timespec64_add_ns(ts, nsecs); + timekeeping_inc_mg_floor_swaps(); + } else { + /* + * Another task changed mg_floor since "old" was fetched. + * "old" has been updated with the latest value of "mg_floor". + * That value is newer than the previous floor value, which + * is enough to effect a change to ctime. Accept it. + */ + *ts = ktime_to_timespec64(ktime_add(old, offset)); + } +} + void ktime_get_coarse_ts64(struct timespec64 *ts) { struct timekeeper *tk = &tk_core.timekeeper; @@ -2193,7 +2365,7 @@ EXPORT_SYMBOL(ktime_get_coarse_ts64); void do_timer(unsigned long ticks) { jiffies_64 += ticks; - calc_global_load(ticks); + calc_global_load(); } /** @@ -2240,7 +2412,7 @@ ktime_t ktime_get_update_offsets_now(unsigned int *cwsseq, ktime_t *offs_real, return base; } -/** +/* * timekeeping_validate_timex - Ensures the timex is ok for use in do_adjtimex */ static int timekeeping_validate_timex(const struct __kernel_timex *txc) @@ -2273,7 +2445,7 @@ static int timekeeping_validate_timex(const struct __kernel_timex *txc) /* * Validate if a timespec/timeval used to inject a time - * offset is valid. Offsets can be postive or negative, so + * offset is valid. Offsets can be positive or negative, so * we don't check tv_sec. The value of the timeval/timespec * is the sum of its fields,but *NOTE*: * The field tv_usec/tv_nsec must always be non-negative and @@ -2305,26 +2477,42 @@ static int timekeeping_validate_timex(const struct __kernel_timex *txc) return 0; } +/** + * random_get_entropy_fallback - Returns the raw clock source value, + * used by random.c for platforms with no valid random_get_entropy(). + */ +unsigned long random_get_entropy_fallback(void) +{ + struct tk_read_base *tkr = &tk_core.timekeeper.tkr_mono; + struct clocksource *clock = READ_ONCE(tkr->clock); + + if (unlikely(timekeeping_suspended || !clock)) + return 0; + return clock->read(clock); +} +EXPORT_SYMBOL_GPL(random_get_entropy_fallback); /** * do_adjtimex() - Accessor function to NTP __do_adjtimex function + * @txc: Pointer to kernel_timex structure containing NTP parameters */ int do_adjtimex(struct __kernel_timex *txc) { - struct timekeeper *tk = &tk_core.timekeeper; struct audit_ntp_data ad; - unsigned long flags; + bool offset_set = false; + bool clock_set = false; struct timespec64 ts; - s32 orig_tai, tai; int ret; /* Validate the data before disabling interrupts */ ret = timekeeping_validate_timex(txc); if (ret) return ret; + add_device_randomness(txc, sizeof(*txc)); if (txc->modes & ADJ_SETOFFSET) { struct timespec64 delta; + delta.tv_sec = txc->time.tv_sec; delta.tv_nsec = txc->time.tv_usec; if (!(txc->modes & ADJ_NANO)) @@ -2333,38 +2521,41 @@ int do_adjtimex(struct __kernel_timex *txc) if (ret) return ret; + offset_set = delta.tv_sec != 0; audit_tk_injoffset(delta); } audit_ntp_init(&ad); ktime_get_real_ts64(&ts); + add_device_randomness(&ts, sizeof(ts)); - raw_spin_lock_irqsave(&timekeeper_lock, flags); - write_seqcount_begin(&tk_core.seq); + scoped_guard (raw_spinlock_irqsave, &tk_core.lock) { + struct timekeeper *tks = &tk_core.shadow_timekeeper; + s32 orig_tai, tai; - orig_tai = tai = tk->tai_offset; - ret = __do_adjtimex(txc, &ts, &tai, &ad); + orig_tai = tai = tks->tai_offset; + ret = __do_adjtimex(txc, &ts, &tai, &ad); - if (tai != orig_tai) { - __timekeeping_set_tai_offset(tk, tai); - timekeeping_update(tk, TK_MIRROR | TK_CLOCK_WAS_SET); + if (tai != orig_tai) { + __timekeeping_set_tai_offset(tks, tai); + timekeeping_update_from_shadow(&tk_core, TK_CLOCK_WAS_SET); + clock_set = true; + } else { + tk_update_leap_state_all(&tk_core); + } } - tk_update_leap_state(tk); - - write_seqcount_end(&tk_core.seq); - raw_spin_unlock_irqrestore(&timekeeper_lock, flags); audit_ntp_log(&ad); /* Update the multiplier immediately if frequency was set directly */ if (txc->modes & (ADJ_FREQUENCY | ADJ_TICK)) - timekeeping_advance(TK_ADV_FREQ); + clock_set |= timekeeping_advance(TK_ADV_FREQ); - if (tai != orig_tai) - clock_was_set(); + if (clock_set) + clock_was_set(CLOCK_SET_WALL); - ntp_notify_cmos_timer(); + ntp_notify_cmos_timer(offset_set); return ret; } @@ -2372,34 +2563,13 @@ int do_adjtimex(struct __kernel_timex *txc) #ifdef CONFIG_NTP_PPS /** * hardpps() - Accessor function to NTP __hardpps function + * @phase_ts: Pointer to timespec64 structure representing phase timestamp + * @raw_ts: Pointer to timespec64 structure representing raw timestamp */ void hardpps(const struct timespec64 *phase_ts, const struct timespec64 *raw_ts) { - unsigned long flags; - - raw_spin_lock_irqsave(&timekeeper_lock, flags); - write_seqcount_begin(&tk_core.seq); - + guard(raw_spinlock_irqsave)(&tk_core.lock); __hardpps(phase_ts, raw_ts); - - write_seqcount_end(&tk_core.seq); - raw_spin_unlock_irqrestore(&timekeeper_lock, flags); } EXPORT_SYMBOL(hardpps); #endif /* CONFIG_NTP_PPS */ - -/** - * xtime_update() - advances the timekeeping infrastructure - * @ticks: number of ticks, that have elapsed since the last call. - * - * Must be called with interrupts disabled. - */ -void xtime_update(unsigned long ticks) -{ - raw_spin_lock(&jiffies_lock); - write_seqcount_begin(&jiffies_seq); - do_timer(ticks); - write_seqcount_end(&jiffies_seq); - raw_spin_unlock(&jiffies_lock); - update_wall_time(); -} diff --git a/kernel/time/timekeeping.h b/kernel/time/timekeeping.h index 099737f6f10c..543beba096c7 100644 --- a/kernel/time/timekeeping.h +++ b/kernel/time/timekeeping.h @@ -22,11 +22,12 @@ static inline int sched_clock_suspend(void) { return 0; } static inline void sched_clock_resume(void) { } #endif +extern void update_process_times(int user); extern void do_timer(unsigned long ticks); extern void update_wall_time(void); extern raw_spinlock_t jiffies_lock; -extern seqcount_t jiffies_seq; +extern seqcount_raw_spinlock_t jiffies_seq; #define CS_NAME_LEN 32 diff --git a/kernel/time/timekeeping_debug.c b/kernel/time/timekeeping_debug.c index b73e8850e58d..badeb222eab9 100644 --- a/kernel/time/timekeeping_debug.c +++ b/kernel/time/timekeeping_debug.c @@ -17,6 +17,9 @@ #define NUM_BINS 32 +/* Incremented every time mg_floor is updated */ +DEFINE_PER_CPU(unsigned long, timekeeping_mg_floor_swaps); + static unsigned int sleep_time_bin[NUM_BINS] = {0}; static int tk_debug_sleep_time_show(struct seq_file *s, void *data) @@ -53,3 +56,13 @@ void tk_debug_account_sleep_time(const struct timespec64 *t) (s64)t->tv_sec, t->tv_nsec / NSEC_PER_MSEC); } +unsigned long timekeeping_get_mg_floor_swaps(void) +{ + unsigned long sum = 0; + int cpu; + + for_each_possible_cpu(cpu) + sum += data_race(per_cpu(timekeeping_mg_floor_swaps, cpu)); + + return sum; +} diff --git a/kernel/time/timekeeping_internal.h b/kernel/time/timekeeping_internal.h index bcbb52db2256..8c9079108ffb 100644 --- a/kernel/time/timekeeping_internal.h +++ b/kernel/time/timekeeping_internal.h @@ -1,34 +1,48 @@ /* SPDX-License-Identifier: GPL-2.0 */ #ifndef _TIMEKEEPING_INTERNAL_H #define _TIMEKEEPING_INTERNAL_H -/* - * timekeeping debug functions - */ + #include <linux/clocksource.h> +#include <linux/spinlock.h> #include <linux/time.h> +/* + * timekeeping debug functions + */ #ifdef CONFIG_DEBUG_FS + +DECLARE_PER_CPU(unsigned long, timekeeping_mg_floor_swaps); + +static inline void timekeeping_inc_mg_floor_swaps(void) +{ + this_cpu_inc(timekeeping_mg_floor_swaps); +} + extern void tk_debug_account_sleep_time(const struct timespec64 *t); + #else + #define tk_debug_account_sleep_time(x) + +static inline void timekeeping_inc_mg_floor_swaps(void) +{ +} + #endif -#ifdef CONFIG_CLOCKSOURCE_VALIDATE_LAST_CYCLE -static inline u64 clocksource_delta(u64 now, u64 last, u64 mask) +static inline u64 clocksource_delta(u64 now, u64 last, u64 mask, u64 max_delta) { u64 ret = (now - last) & mask; /* - * Prevent time going backwards by checking the MSB of mask in - * the result. If set, return 0. + * Prevent time going backwards by checking the result against + * @max_delta. If greater, return 0. */ - return ret & ~(mask >> 1) ? 0 : ret; + return ret > max_delta ? 0 : ret; } -#else -static inline u64 clocksource_delta(u64 now, u64 last, u64 mask) -{ - return (now - last) & mask; -} -#endif + +/* Semi public for serialization of non timekeeper VDSO updates. */ +unsigned long timekeeper_lock_irqsave(void); +void timekeeper_unlock_irqrestore(unsigned long flags); #endif /* _TIMEKEEPING_INTERNAL_H */ diff --git a/kernel/time/timer.c b/kernel/time/timer.c index 026ac01af9da..c8f776dc6ee0 100644 --- a/kernel/time/timer.c +++ b/kernel/time/timer.c @@ -37,13 +37,13 @@ #include <linux/tick.h> #include <linux/kallsyms.h> #include <linux/irq_work.h> -#include <linux/sched/signal.h> #include <linux/sched/sysctl.h> #include <linux/sched/nohz.h> #include <linux/sched/debug.h> #include <linux/slab.h> #include <linux/compat.h> #include <linux/random.h> +#include <linux/sysctl.h> #include <linux/uaccess.h> #include <asm/unistd.h> @@ -52,6 +52,7 @@ #include <asm/io.h> #include "tick-internal.h" +#include "timer_migration.h" #define CREATE_TRACE_POINTS #include <trace/events/timer.h> @@ -62,15 +63,15 @@ EXPORT_SYMBOL(jiffies_64); /* * The timer wheel has LVL_DEPTH array levels. Each level provides an array of - * LVL_SIZE buckets. Each level is driven by its own clock and therefor each + * LVL_SIZE buckets. Each level is driven by its own clock and therefore each * level has a different granularity. * - * The level granularity is: LVL_CLK_DIV ^ lvl + * The level granularity is: LVL_CLK_DIV ^ level * The level clock frequency is: HZ / (LVL_CLK_DIV ^ level) * * The array level of a newly armed timer depends on the relative expiry * time. The farther the expiry time is away the higher the array level and - * therefor the granularity becomes. + * therefore the granularity becomes. * * Contrary to the original timer wheel implementation, which aims for 'exact' * expiry of the timers, this implementation removes the need for recascading @@ -157,7 +158,8 @@ EXPORT_SYMBOL(jiffies_64); /* * The time start value for each level to select the bucket at enqueue - * time. + * time. We start from the last possible delta of the previous level + * so that we can later add an extra LVL_GRAN(n) to n (see calc_index()). */ #define LVL_START(n) ((LVL_SIZE - 1) << (((n) - 1) * LVL_CLK_SHIFT)) @@ -185,15 +187,66 @@ EXPORT_SYMBOL(jiffies_64); #define WHEEL_SIZE (LVL_SIZE * LVL_DEPTH) #ifdef CONFIG_NO_HZ_COMMON -# define NR_BASES 2 -# define BASE_STD 0 -# define BASE_DEF 1 +/* + * If multiple bases need to be locked, use the base ordering for lock + * nesting, i.e. lowest number first. + */ +# define NR_BASES 3 +# define BASE_LOCAL 0 +# define BASE_GLOBAL 1 +# define BASE_DEF 2 #else # define NR_BASES 1 -# define BASE_STD 0 +# define BASE_LOCAL 0 +# define BASE_GLOBAL 0 # define BASE_DEF 0 #endif +/** + * struct timer_base - Per CPU timer base (number of base depends on config) + * @lock: Lock protecting the timer_base + * @running_timer: When expiring timers, the lock is dropped. To make + * sure not to race against deleting/modifying a + * currently running timer, the pointer is set to the + * timer, which expires at the moment. If no timer is + * running, the pointer is NULL. + * @expiry_lock: PREEMPT_RT only: Lock is taken in softirq around + * timer expiry callback execution and when trying to + * delete a running timer and it wasn't successful in + * the first glance. It prevents priority inversion + * when callback was preempted on a remote CPU and a + * caller tries to delete the running timer. It also + * prevents a life lock, when the task which tries to + * delete a timer preempted the softirq thread which + * is running the timer callback function. + * @timer_waiters: PREEMPT_RT only: Tells, if there is a waiter + * waiting for the end of the timer callback function + * execution. + * @clk: clock of the timer base; is updated before enqueue + * of a timer; during expiry, it is 1 offset ahead of + * jiffies to avoid endless requeuing to current + * jiffies + * @next_expiry: expiry value of the first timer; it is updated when + * finding the next timer and during enqueue; the + * value is not valid, when next_expiry_recalc is set + * @cpu: Number of CPU the timer base belongs to + * @next_expiry_recalc: States, whether a recalculation of next_expiry is + * required. Value is set true, when a timer was + * deleted. + * @is_idle: Is set, when timer_base is idle. It is triggered by NOHZ + * code. This state is only used in standard + * base. Deferrable timers, which are enqueued remotely + * never wake up an idle CPU. So no matter of supporting it + * for this base. + * @timers_pending: Is set, when a timer is pending in the base. It is only + * reliable when next_expiry_recalc is not set. + * @pending_map: bitmap of the timer wheel; each bit reflects a + * bucket of the wheel. When a bit is set, at least a + * single timer is enqueued in the related bucket. + * @vectors: Array of lists; Each array member reflects a bucket + * of the timer wheel. The list contains all timers + * which are enqueued into a specific bucket. + */ struct timer_base { raw_spinlock_t lock; struct timer_list *running_timer; @@ -204,8 +257,9 @@ struct timer_base { unsigned long clk; unsigned long next_expiry; unsigned int cpu; + bool next_expiry_recalc; bool is_idle; - bool must_forward_clk; + bool timers_pending; DECLARE_BITMAP(pending_map, WHEEL_SIZE); struct hlist_head vectors[WHEEL_SIZE]; } ____cacheline_aligned; @@ -221,7 +275,7 @@ static void timer_update_keys(struct work_struct *work); static DECLARE_WORK(timer_update_work, timer_update_keys); #ifdef CONFIG_SMP -unsigned int sysctl_timer_migration = 1; +static unsigned int sysctl_timer_migration = 1; DEFINE_STATIC_KEY_FALSE(timers_migration_enabled); @@ -232,7 +286,41 @@ static void timers_update_migration(void) else static_branch_disable(&timers_migration_enabled); } -#else + +#ifdef CONFIG_SYSCTL +static int timer_migration_handler(const struct ctl_table *table, int write, + void *buffer, size_t *lenp, loff_t *ppos) +{ + int ret; + + mutex_lock(&timer_keys_mutex); + ret = proc_dointvec_minmax(table, write, buffer, lenp, ppos); + if (!ret && write) + timers_update_migration(); + mutex_unlock(&timer_keys_mutex); + return ret; +} + +static const struct ctl_table timer_sysctl[] = { + { + .procname = "timer_migration", + .data = &sysctl_timer_migration, + .maxlen = sizeof(unsigned int), + .mode = 0644, + .proc_handler = timer_migration_handler, + .extra1 = SYSCTL_ZERO, + .extra2 = SYSCTL_ONE, + }, +}; + +static int __init timer_sysctl_init(void) +{ + register_sysctl("kernel", timer_sysctl); + return 0; +} +device_initcall(timer_sysctl_init); +#endif /* CONFIG_SYSCTL */ +#else /* CONFIG_SMP */ static inline void timers_update_migration(void) { } #endif /* !CONFIG_SMP */ @@ -249,19 +337,6 @@ void timers_update_nohz(void) schedule_work(&timer_update_work); } -int timer_migration_handler(struct ctl_table *table, int write, - void *buffer, size_t *lenp, loff_t *ppos) -{ - int ret; - - mutex_lock(&timer_keys_mutex); - ret = proc_dointvec_minmax(table, write, buffer, lenp, ppos); - if (!ret && write) - timers_update_migration(); - mutex_unlock(&timer_keys_mutex); - return ret; -} - static inline bool is_timers_nohz_active(void) { return static_branch_unlikely(&timers_nohz_active); @@ -289,7 +364,7 @@ static unsigned long round_jiffies_common(unsigned long j, int cpu, rem = j % HZ; /* - * If the target jiffie is just after a whole second (which can happen + * If the target jiffy is just after a whole second (which can happen * due to delays of the timer irq, long irq off times etc etc) then * we should round down to the whole second, not up. Use 1/4th second * as cutoff for this rounding as an extreme upper bound for this. @@ -488,35 +563,48 @@ static inline void timer_set_idx(struct timer_list *timer, unsigned int idx) * Helper function to calculate the array index for a given expiry * time. */ -static inline unsigned calc_index(unsigned expires, unsigned lvl) +static inline unsigned calc_index(unsigned long expires, unsigned lvl, + unsigned long *bucket_expiry) { - expires = (expires + LVL_GRAN(lvl)) >> LVL_SHIFT(lvl); + + /* + * The timer wheel has to guarantee that a timer does not fire + * early. Early expiry can happen due to: + * - Timer is armed at the edge of a tick + * - Truncation of the expiry time in the outer wheel levels + * + * Round up with level granularity to prevent this. + */ + expires = (expires >> LVL_SHIFT(lvl)) + 1; + *bucket_expiry = expires << LVL_SHIFT(lvl); return LVL_OFFS(lvl) + (expires & LVL_MASK); } -static int calc_wheel_index(unsigned long expires, unsigned long clk) +static int calc_wheel_index(unsigned long expires, unsigned long clk, + unsigned long *bucket_expiry) { unsigned long delta = expires - clk; unsigned int idx; if (delta < LVL_START(1)) { - idx = calc_index(expires, 0); + idx = calc_index(expires, 0, bucket_expiry); } else if (delta < LVL_START(2)) { - idx = calc_index(expires, 1); + idx = calc_index(expires, 1, bucket_expiry); } else if (delta < LVL_START(3)) { - idx = calc_index(expires, 2); + idx = calc_index(expires, 2, bucket_expiry); } else if (delta < LVL_START(4)) { - idx = calc_index(expires, 3); + idx = calc_index(expires, 3, bucket_expiry); } else if (delta < LVL_START(5)) { - idx = calc_index(expires, 4); + idx = calc_index(expires, 4, bucket_expiry); } else if (delta < LVL_START(6)) { - idx = calc_index(expires, 5); + idx = calc_index(expires, 5, bucket_expiry); } else if (delta < LVL_START(7)) { - idx = calc_index(expires, 6); + idx = calc_index(expires, 6, bucket_expiry); } else if (LVL_DEPTH > 8 && delta < LVL_START(8)) { - idx = calc_index(expires, 7); + idx = calc_index(expires, 7, bucket_expiry); } else if ((long) delta < 0) { idx = clk & LVL_MASK; + *bucket_expiry = clk; } else { /* * Force expire obscene large timeouts to expire at the @@ -525,92 +613,117 @@ static int calc_wheel_index(unsigned long expires, unsigned long clk) if (delta >= WHEEL_TIMEOUT_CUTOFF) expires = clk + WHEEL_TIMEOUT_MAX; - idx = calc_index(expires, LVL_DEPTH - 1); + idx = calc_index(expires, LVL_DEPTH - 1, bucket_expiry); } return idx; } -/* - * Enqueue the timer into the hash bucket, mark it pending in - * the bitmap and store the index in the timer flags. - */ -static void enqueue_timer(struct timer_base *base, struct timer_list *timer, - unsigned int idx) -{ - hlist_add_head(&timer->entry, base->vectors + idx); - __set_bit(idx, base->pending_map); - timer_set_idx(timer, idx); - - trace_timer_start(timer, timer->expires, timer->flags); -} - -static void -__internal_add_timer(struct timer_base *base, struct timer_list *timer) -{ - unsigned int idx; - - idx = calc_wheel_index(timer->expires, base->clk); - enqueue_timer(base, timer, idx); -} - static void trigger_dyntick_cpu(struct timer_base *base, struct timer_list *timer) { - if (!is_timers_nohz_active()) - return; - /* - * TODO: This wants some optimizing similar to the code below, but we - * will do that when we switch from push to pull for deferrable timers. + * Deferrable timers do not prevent the CPU from entering dynticks and + * are not taken into account on the idle/nohz_full path. An IPI when a + * new deferrable timer is enqueued will wake up the remote CPU but + * nothing will be done with the deferrable timer base. Therefore skip + * the remote IPI for deferrable timers completely. */ - if (timer->flags & TIMER_DEFERRABLE) { - if (tick_nohz_full_cpu(base->cpu)) - wake_up_nohz_cpu(base->cpu); + if (!is_timers_nohz_active() || timer->flags & TIMER_DEFERRABLE) return; - } /* * We might have to IPI the remote CPU if the base is idle and the - * timer is not deferrable. If the other CPU is on the way to idle - * then it can't set base->is_idle as we hold the base lock: + * timer is pinned. If it is a non pinned timer, it is only queued + * on the remote CPU, when timer was running during queueing. Then + * everything is handled by remote CPU anyway. If the other CPU is + * on the way to idle then it can't set base->is_idle as we hold + * the base lock: */ - if (!base->is_idle) - return; + if (base->is_idle) { + WARN_ON_ONCE(!(timer->flags & TIMER_PINNED || + tick_nohz_full_cpu(base->cpu))); + wake_up_nohz_cpu(base->cpu); + } +} - /* Check whether this is the new first expiring timer: */ - if (time_after_eq(timer->expires, base->next_expiry)) - return; +/* + * Enqueue the timer into the hash bucket, mark it pending in + * the bitmap, store the index in the timer flags then wake up + * the target CPU if needed. + */ +static void enqueue_timer(struct timer_base *base, struct timer_list *timer, + unsigned int idx, unsigned long bucket_expiry) +{ + + hlist_add_head(&timer->entry, base->vectors + idx); + __set_bit(idx, base->pending_map); + timer_set_idx(timer, idx); + + trace_timer_start(timer, bucket_expiry); /* - * Set the next expiry time and kick the CPU so it can reevaluate the - * wheel: + * Check whether this is the new first expiring timer. The + * effective expiry time of the timer is required here + * (bucket_expiry) instead of timer->expires. */ - if (time_before(timer->expires, base->clk)) { + if (time_before(bucket_expiry, base->next_expiry)) { /* - * Prevent from forward_timer_base() moving the base->clk - * backward + * Set the next expiry time and kick the CPU so it + * can reevaluate the wheel: */ - base->next_expiry = base->clk; - } else { - base->next_expiry = timer->expires; + WRITE_ONCE(base->next_expiry, bucket_expiry); + base->timers_pending = true; + base->next_expiry_recalc = false; + trigger_dyntick_cpu(base, timer); } - wake_up_nohz_cpu(base->cpu); } -static void -internal_add_timer(struct timer_base *base, struct timer_list *timer) +static void internal_add_timer(struct timer_base *base, struct timer_list *timer) { - __internal_add_timer(base, timer); - trigger_dyntick_cpu(base, timer); + unsigned long bucket_expiry; + unsigned int idx; + + idx = calc_wheel_index(timer->expires, base->clk, &bucket_expiry); + enqueue_timer(base, timer, idx, bucket_expiry); } #ifdef CONFIG_DEBUG_OBJECTS_TIMERS -static struct debug_obj_descr timer_debug_descr; +static const struct debug_obj_descr timer_debug_descr; + +struct timer_hint { + void (*function)(struct timer_list *t); + long offset; +}; + +#define TIMER_HINT(fn, container, timr, hintfn) \ + { \ + .function = fn, \ + .offset = offsetof(container, hintfn) - \ + offsetof(container, timr) \ + } + +static const struct timer_hint timer_hints[] = { + TIMER_HINT(delayed_work_timer_fn, + struct delayed_work, timer, work.func), + TIMER_HINT(kthread_delayed_work_timer_fn, + struct kthread_delayed_work, timer, work.func), +}; static void *timer_debug_hint(void *addr) { - return ((struct timer_list *) addr)->function; + struct timer_list *timer = addr; + int i; + + for (i = 0; i < ARRAY_SIZE(timer_hints); i++) { + if (timer_hints[i].function == timer->function) { + void (**fn)(void) = addr + timer_hints[i].offset; + + return *fn; + } + } + + return timer->function; } static bool timer_is_static_object(void *addr) @@ -622,7 +735,7 @@ static bool timer_is_static_object(void *addr) } /* - * fixup_init is called when: + * timer_fixup_init is called when: * - an active object is initialized */ static bool timer_fixup_init(void *addr, enum debug_obj_state state) @@ -646,7 +759,7 @@ static void stub_timer(struct timer_list *unused) } /* - * fixup_activate is called when: + * timer_fixup_activate is called when: * - an active object is activated * - an unknown non-static object is activated */ @@ -661,14 +774,14 @@ static bool timer_fixup_activate(void *addr, enum debug_obj_state state) case ODEBUG_STATE_ACTIVE: WARN_ON(1); - /* fall through */ + fallthrough; default: return false; } } /* - * fixup_free is called when: + * timer_fixup_free is called when: * - an active object is freed */ static bool timer_fixup_free(void *addr, enum debug_obj_state state) @@ -686,7 +799,7 @@ static bool timer_fixup_free(void *addr, enum debug_obj_state state) } /* - * fixup_assert_init is called when: + * timer_fixup_assert_init is called when: * - an untracked/uninit-ed object is found */ static bool timer_fixup_assert_init(void *addr, enum debug_obj_state state) @@ -702,7 +815,7 @@ static bool timer_fixup_assert_init(void *addr, enum debug_obj_state state) } } -static struct debug_obj_descr timer_debug_descr = { +static const struct debug_obj_descr timer_debug_descr = { .name = "timer_list", .debug_hint = timer_debug_hint, .is_static_object = timer_is_static_object, @@ -727,11 +840,6 @@ static inline void debug_timer_deactivate(struct timer_list *timer) debug_object_deactivate(timer, &timer_debug_descr); } -static inline void debug_timer_free(struct timer_list *timer) -{ - debug_object_free(timer, &timer_debug_descr); -} - static inline void debug_timer_assert_init(struct timer_list *timer) { debug_object_assert_init(timer, &timer_debug_descr); @@ -789,6 +897,8 @@ static void do_init_timer(struct timer_list *timer, { timer->entry.pprev = NULL; timer->function = func; + if (WARN_ON_ONCE(flags & ~TIMER_INIT_FLAGS)) + flags &= TIMER_INIT_FLAGS; timer->flags = flags | raw_smp_processor_id(); lockdep_init_map(&timer->lockdep_map, name, key, 0); } @@ -802,7 +912,7 @@ static void do_init_timer(struct timer_list *timer, * @key: lockdep class key of the fake lock used for tracking timer * sync lock dependencies * - * init_timer_key() must be done to a timer prior calling *any* of the + * init_timer_key() must be done to a timer prior to calling *any* of the * other timer functions. */ void init_timer_key(struct timer_list *timer, @@ -834,8 +944,10 @@ static int detach_if_pending(struct timer_list *timer, struct timer_base *base, if (!timer_pending(timer)) return 0; - if (hlist_is_singular_node(&timer->entry, base->vectors + idx)) + if (hlist_is_singular_node(&timer->entry, base->vectors + idx)) { __clear_bit(idx, base->pending_map); + base->next_expiry_recalc = true; + } detach_timer(timer, clear_pending); return 1; @@ -843,28 +955,30 @@ static int detach_if_pending(struct timer_list *timer, struct timer_base *base, static inline struct timer_base *get_timer_cpu_base(u32 tflags, u32 cpu) { - struct timer_base *base = per_cpu_ptr(&timer_bases[BASE_STD], cpu); + int index = tflags & TIMER_PINNED ? BASE_LOCAL : BASE_GLOBAL; /* * If the timer is deferrable and NO_HZ_COMMON is set then we need * to use the deferrable base. */ if (IS_ENABLED(CONFIG_NO_HZ_COMMON) && (tflags & TIMER_DEFERRABLE)) - base = per_cpu_ptr(&timer_bases[BASE_DEF], cpu); - return base; + index = BASE_DEF; + + return per_cpu_ptr(&timer_bases[index], cpu); } static inline struct timer_base *get_timer_this_cpu_base(u32 tflags) { - struct timer_base *base = this_cpu_ptr(&timer_bases[BASE_STD]); + int index = tflags & TIMER_PINNED ? BASE_LOCAL : BASE_GLOBAL; /* * If the timer is deferrable and NO_HZ_COMMON is set then we need * to use the deferrable base. */ if (IS_ENABLED(CONFIG_NO_HZ_COMMON) && (tflags & TIMER_DEFERRABLE)) - base = this_cpu_ptr(&timer_bases[BASE_DEF]); - return base; + index = BASE_DEF; + + return this_cpu_ptr(&timer_bases[index]); } static inline struct timer_base *get_timer_base(u32 tflags) @@ -872,49 +986,34 @@ static inline struct timer_base *get_timer_base(u32 tflags) return get_timer_cpu_base(tflags, tflags & TIMER_CPUMASK); } -static inline struct timer_base * -get_target_base(struct timer_base *base, unsigned tflags) +static inline void __forward_timer_base(struct timer_base *base, + unsigned long basej) { -#if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON) - if (static_branch_likely(&timers_migration_enabled) && - !(tflags & TIMER_PINNED)) - return get_timer_cpu_base(tflags, get_nohz_timer_target()); -#endif - return get_timer_this_cpu_base(tflags); -} - -static inline void forward_timer_base(struct timer_base *base) -{ -#ifdef CONFIG_NO_HZ_COMMON - unsigned long jnow; - /* - * We only forward the base when we are idle or have just come out of - * idle (must_forward_clk logic), and have a delta between base clock - * and jiffies. In the common case, run_timers will take care of it. + * Check whether we can forward the base. We can only do that when + * @basej is past base->clk otherwise we might rewind base->clk. */ - if (likely(!base->must_forward_clk)) - return; - - jnow = READ_ONCE(jiffies); - base->must_forward_clk = base->is_idle; - if ((long)(jnow - base->clk) < 2) + if (time_before_eq(basej, base->clk)) return; /* * If the next expiry value is > jiffies, then we fast forward to * jiffies otherwise we forward to the next expiry value. */ - if (time_after(base->next_expiry, jnow)) { - base->clk = jnow; + if (time_after(base->next_expiry, basej)) { + base->clk = basej; } else { if (WARN_ON_ONCE(time_before(base->next_expiry, base->clk))) return; base->clk = base->next_expiry; } -#endif + } +static inline void forward_timer_base(struct timer_base *base) +{ + __forward_timer_base(base, READ_ONCE(jiffies)); +} /* * We are using hashed locking: Holding per_cpu(timer_bases[x]).lock means @@ -960,12 +1059,12 @@ static struct timer_base *lock_timer_base(struct timer_list *timer, static inline int __mod_timer(struct timer_list *timer, unsigned long expires, unsigned int options) { + unsigned long clk = 0, flags, bucket_expiry; struct timer_base *base, *new_base; unsigned int idx = UINT_MAX; - unsigned long clk = 0, flags; int ret = 0; - BUG_ON(!timer->function); + debug_assert_init(timer); /* * This is a common optimization triggered by the networking code - if @@ -992,6 +1091,14 @@ __mod_timer(struct timer_list *timer, unsigned long expires, unsigned int option * dequeue/enqueue dance. */ base = lock_timer_base(timer, &flags); + /* + * Has @timer been shutdown? This needs to be evaluated + * while holding base lock to prevent a race against the + * shutdown code. + */ + if (!timer->function) + goto out_unlock; + forward_timer_base(base); if (timer_pending(timer) && (options & MOD_TIMER_REDUCE) && @@ -1001,7 +1108,7 @@ __mod_timer(struct timer_list *timer, unsigned long expires, unsigned int option } clk = base->clk; - idx = calc_wheel_index(expires, clk); + idx = calc_wheel_index(expires, clk, &bucket_expiry); /* * Retrieve and compare the array index of the pending @@ -1018,6 +1125,14 @@ __mod_timer(struct timer_list *timer, unsigned long expires, unsigned int option } } else { base = lock_timer_base(timer, &flags); + /* + * Has @timer been shutdown? This needs to be evaluated + * while holding base lock to prevent a race against the + * shutdown code. + */ + if (!timer->function) + goto out_unlock; + forward_timer_base(base); } @@ -1025,13 +1140,13 @@ __mod_timer(struct timer_list *timer, unsigned long expires, unsigned int option if (!ret && (options & MOD_TIMER_PENDING_ONLY)) goto out_unlock; - new_base = get_target_base(base, timer->flags); + new_base = get_timer_this_cpu_base(timer->flags); if (base != new_base) { /* * We are trying to schedule the timer on the new base. * However we can't change timer's base while it is running, - * otherwise del_timer_sync() can't detect that the timer's + * otherwise timer_delete_sync() can't detect that the timer's * handler yet has not finished. This also guarantees that the * timer is serialized wrt itself. */ @@ -1054,16 +1169,13 @@ __mod_timer(struct timer_list *timer, unsigned long expires, unsigned int option /* * If 'idx' was calculated above and the base time did not advance * between calculating 'idx' and possibly switching the base, only - * enqueue_timer() and trigger_dyntick_cpu() is required. Otherwise - * we need to (re)calculate the wheel index via - * internal_add_timer(). + * enqueue_timer() is required. Otherwise we need to (re)calculate + * the wheel index via internal_add_timer(). */ - if (idx != UINT_MAX && clk == base->clk) { - enqueue_timer(base, timer, idx); - trigger_dyntick_cpu(base, timer); - } else { + if (idx != UINT_MAX && clk == base->clk) + enqueue_timer(base, timer, idx, bucket_expiry); + else internal_add_timer(base, timer); - } out_unlock: raw_spin_unlock_irqrestore(&base->lock, flags); @@ -1072,14 +1184,20 @@ out_unlock: } /** - * mod_timer_pending - modify a pending timer's timeout - * @timer: the pending timer to be modified - * @expires: new timeout in jiffies + * mod_timer_pending - Modify a pending timer's timeout + * @timer: The pending timer to be modified + * @expires: New absolute timeout in jiffies + * + * mod_timer_pending() is the same for pending timers as mod_timer(), but + * will not activate inactive timers. * - * mod_timer_pending() is the same for pending timers as mod_timer(), - * but will not re-activate and modify already deleted timers. + * If @timer->function == NULL then the start operation is silently + * discarded. * - * It is useful for unserialized use of timers. + * Return: + * * %0 - The timer was inactive and not modified or was in + * shutdown state and the operation was discarded + * * %1 - The timer was active and requeued to expire at @expires */ int mod_timer_pending(struct timer_list *timer, unsigned long expires) { @@ -1088,24 +1206,31 @@ int mod_timer_pending(struct timer_list *timer, unsigned long expires) EXPORT_SYMBOL(mod_timer_pending); /** - * mod_timer - modify a timer's timeout - * @timer: the timer to be modified - * @expires: new timeout in jiffies - * - * mod_timer() is a more efficient way to update the expire field of an - * active timer (if the timer is inactive it will be activated) + * mod_timer - Modify a timer's timeout + * @timer: The timer to be modified + * @expires: New absolute timeout in jiffies * * mod_timer(timer, expires) is equivalent to: * * del_timer(timer); timer->expires = expires; add_timer(timer); * + * mod_timer() is more efficient than the above open coded sequence. In + * case that the timer is inactive, the del_timer() part is a NOP. The + * timer is in any case activated with the new expiry time @expires. + * * Note that if there are multiple unserialized concurrent users of the * same timer, then mod_timer() is the only safe way to modify the timeout, * since add_timer() cannot modify an already running timer. * - * The function returns whether it has modified a pending timer or not. - * (ie. mod_timer() of an inactive timer returns 0, mod_timer() of an - * active timer returns 1.) + * If @timer->function == NULL then the start operation is silently + * discarded. In this case the return value is 0 and meaningless. + * + * Return: + * * %0 - The timer was inactive and started or was in shutdown + * state and the operation was discarded + * * %1 - The timer was active and requeued to expire at @expires or + * the timer was active and not modified because @expires did + * not change the effective expiry time */ int mod_timer(struct timer_list *timer, unsigned long expires) { @@ -1116,11 +1241,22 @@ EXPORT_SYMBOL(mod_timer); /** * timer_reduce - Modify a timer's timeout if it would reduce the timeout * @timer: The timer to be modified - * @expires: New timeout in jiffies + * @expires: New absolute timeout in jiffies * * timer_reduce() is very similar to mod_timer(), except that it will only - * modify a running timer if that would reduce the expiration time (it will - * start a timer that isn't running). + * modify an enqueued timer if that would reduce the expiration time. If + * @timer is not enqueued it starts the timer. + * + * If @timer->function == NULL then the start operation is silently + * discarded. + * + * Return: + * * %0 - The timer was inactive and started or was in shutdown + * state and the operation was discarded + * * %1 - The timer was active and requeued to expire at @expires or + * the timer was active and not modified because @expires + * did not change the effective expiry time such that the + * timer would expire earlier than already scheduled */ int timer_reduce(struct timer_list *timer, unsigned long expires) { @@ -1129,39 +1265,91 @@ int timer_reduce(struct timer_list *timer, unsigned long expires) EXPORT_SYMBOL(timer_reduce); /** - * add_timer - start a timer - * @timer: the timer to be added + * add_timer - Start a timer + * @timer: The timer to be started * - * The kernel will do a ->function(@timer) callback from the - * timer interrupt at the ->expires point in the future. The - * current time is 'jiffies'. + * Start @timer to expire at @timer->expires in the future. @timer->expires + * is the absolute expiry time measured in 'jiffies'. When the timer expires + * timer->function(timer) will be invoked from soft interrupt context. * - * The timer's ->expires, ->function fields must be set prior calling this - * function. + * The @timer->expires and @timer->function fields must be set prior + * to calling this function. + * + * If @timer->function == NULL then the start operation is silently + * discarded. + * + * If @timer->expires is already in the past @timer will be queued to + * expire at the next timer tick. * - * Timers with an ->expires field in the past will be executed in the next - * timer tick. + * This can only operate on an inactive timer. Attempts to invoke this on + * an active timer are rejected with a warning. */ void add_timer(struct timer_list *timer) { - BUG_ON(timer_pending(timer)); + if (WARN_ON_ONCE(timer_pending(timer))) + return; __mod_timer(timer, timer->expires, MOD_TIMER_NOTPENDING); } EXPORT_SYMBOL(add_timer); /** - * add_timer_on - start a timer on a particular CPU - * @timer: the timer to be added - * @cpu: the CPU to start it on + * add_timer_local() - Start a timer on the local CPU + * @timer: The timer to be started + * + * Same as add_timer() except that the timer flag TIMER_PINNED is set. * - * This is not very scalable on SMP. Double adds are not possible. + * See add_timer() for further details. + */ +void add_timer_local(struct timer_list *timer) +{ + if (WARN_ON_ONCE(timer_pending(timer))) + return; + timer->flags |= TIMER_PINNED; + __mod_timer(timer, timer->expires, MOD_TIMER_NOTPENDING); +} +EXPORT_SYMBOL(add_timer_local); + +/** + * add_timer_global() - Start a timer without TIMER_PINNED flag set + * @timer: The timer to be started + * + * Same as add_timer() except that the timer flag TIMER_PINNED is unset. + * + * See add_timer() for further details. + */ +void add_timer_global(struct timer_list *timer) +{ + if (WARN_ON_ONCE(timer_pending(timer))) + return; + timer->flags &= ~TIMER_PINNED; + __mod_timer(timer, timer->expires, MOD_TIMER_NOTPENDING); +} +EXPORT_SYMBOL(add_timer_global); + +/** + * add_timer_on - Start a timer on a particular CPU + * @timer: The timer to be started + * @cpu: The CPU to start it on + * + * Same as add_timer() except that it starts the timer on the given CPU and + * the TIMER_PINNED flag is set. When timer shouldn't be a pinned timer in + * the next round, add_timer_global() should be used instead as it unsets + * the TIMER_PINNED flag. + * + * See add_timer() for further details. */ void add_timer_on(struct timer_list *timer, int cpu) { struct timer_base *new_base, *base; unsigned long flags; - BUG_ON(timer_pending(timer) || !timer->function); + debug_assert_init(timer); + + if (WARN_ON_ONCE(timer_pending(timer))) + return; + + /* Make sure timer flags have TIMER_PINNED flag set */ + timer->flags |= TIMER_PINNED; new_base = get_timer_cpu_base(timer->flags, cpu); @@ -1171,6 +1359,13 @@ void add_timer_on(struct timer_list *timer, int cpu) * wrong base locked. See lock_timer_base(). */ base = lock_timer_base(timer, &flags); + /* + * Has @timer been shutdown? This needs to be evaluated while + * holding base lock to prevent a race against the shutdown code. + */ + if (!timer->function) + goto out_unlock; + if (base != new_base) { timer->flags |= TIMER_MIGRATING; @@ -1184,22 +1379,27 @@ void add_timer_on(struct timer_list *timer, int cpu) debug_timer_activate(timer); internal_add_timer(base, timer); +out_unlock: raw_spin_unlock_irqrestore(&base->lock, flags); } EXPORT_SYMBOL_GPL(add_timer_on); /** - * del_timer - deactivate a timer. - * @timer: the timer to be deactivated - * - * del_timer() deactivates a timer - this works on both active and inactive - * timers. - * - * The function returns whether it has deactivated a pending timer or not. - * (ie. del_timer() of an inactive timer returns 0, del_timer() of an - * active timer returns 1.) + * __timer_delete - Internal function: Deactivate a timer + * @timer: The timer to be deactivated + * @shutdown: If true, this indicates that the timer is about to be + * shutdown permanently. + * + * If @shutdown is true then @timer->function is set to NULL under the + * timer base lock which prevents further rearming of the time. In that + * case any attempt to rearm @timer after this function returns will be + * silently ignored. + * + * Return: + * * %0 - The timer was not pending + * * %1 - The timer was pending and deactivated */ -int del_timer(struct timer_list *timer) +static int __timer_delete(struct timer_list *timer, bool shutdown) { struct timer_base *base; unsigned long flags; @@ -1207,24 +1407,90 @@ int del_timer(struct timer_list *timer) debug_assert_init(timer); - if (timer_pending(timer)) { + /* + * If @shutdown is set then the lock has to be taken whether the + * timer is pending or not to protect against a concurrent rearm + * which might hit between the lockless pending check and the lock + * acquisition. By taking the lock it is ensured that such a newly + * enqueued timer is dequeued and cannot end up with + * timer->function == NULL in the expiry code. + * + * If timer->function is currently executed, then this makes sure + * that the callback cannot requeue the timer. + */ + if (timer_pending(timer) || shutdown) { base = lock_timer_base(timer, &flags); ret = detach_if_pending(timer, base, true); + if (shutdown) + timer->function = NULL; raw_spin_unlock_irqrestore(&base->lock, flags); } return ret; } -EXPORT_SYMBOL(del_timer); /** - * try_to_del_timer_sync - Try to deactivate a timer - * @timer: timer to delete + * timer_delete - Deactivate a timer + * @timer: The timer to be deactivated + * + * The function only deactivates a pending timer, but contrary to + * timer_delete_sync() it does not take into account whether the timer's + * callback function is concurrently executed on a different CPU or not. + * It neither prevents rearming of the timer. If @timer can be rearmed + * concurrently then the return value of this function is meaningless. + * + * Return: + * * %0 - The timer was not pending + * * %1 - The timer was pending and deactivated + */ +int timer_delete(struct timer_list *timer) +{ + return __timer_delete(timer, false); +} +EXPORT_SYMBOL(timer_delete); + +/** + * timer_shutdown - Deactivate a timer and prevent rearming + * @timer: The timer to be deactivated + * + * The function does not wait for an eventually running timer callback on a + * different CPU but it prevents rearming of the timer. Any attempt to arm + * @timer after this function returns will be silently ignored. * - * This function tries to deactivate a timer. Upon successful (ret >= 0) - * exit the timer is not queued and the handler is not running on any CPU. + * This function is useful for teardown code and should only be used when + * timer_shutdown_sync() cannot be invoked due to locking or context constraints. + * + * Return: + * * %0 - The timer was not pending + * * %1 - The timer was pending */ -int try_to_del_timer_sync(struct timer_list *timer) +int timer_shutdown(struct timer_list *timer) +{ + return __timer_delete(timer, true); +} +EXPORT_SYMBOL_GPL(timer_shutdown); + +/** + * __try_to_del_timer_sync - Internal function: Try to deactivate a timer + * @timer: Timer to deactivate + * @shutdown: If true, this indicates that the timer is about to be + * shutdown permanently. + * + * If @shutdown is true then @timer->function is set to NULL under the + * timer base lock which prevents further rearming of the timer. Any + * attempt to rearm @timer after this function returns will be silently + * ignored. + * + * This function cannot guarantee that the timer cannot be rearmed + * right after dropping the base lock if @shutdown is false. That + * needs to be prevented by the calling code if necessary. + * + * Return: + * * %0 - The timer was not pending + * * %1 - The timer was pending and deactivated + * * %-1 - The timer callback function is running on a different CPU + */ +static int __try_to_del_timer_sync(struct timer_list *timer, bool shutdown) { struct timer_base *base; unsigned long flags; @@ -1236,11 +1502,34 @@ int try_to_del_timer_sync(struct timer_list *timer) if (base->running_timer != timer) ret = detach_if_pending(timer, base, true); + if (shutdown) + timer->function = NULL; raw_spin_unlock_irqrestore(&base->lock, flags); return ret; } + +/** + * try_to_del_timer_sync - Try to deactivate a timer + * @timer: Timer to deactivate + * + * This function tries to deactivate a timer. On success the timer is not + * queued and the timer callback function is not running on any CPU. + * + * This function does not guarantee that the timer cannot be rearmed right + * after dropping the base lock. That needs to be prevented by the calling + * code if necessary. + * + * Return: + * * %0 - The timer was not pending + * * %1 - The timer was pending and deactivated + * * %-1 - The timer callback function is running on a different CPU + */ +int try_to_del_timer_sync(struct timer_list *timer) +{ + return __try_to_del_timer_sync(timer, false); +} EXPORT_SYMBOL(try_to_del_timer_sync); #ifdef CONFIG_PREEMPT_RT @@ -1263,14 +1552,18 @@ static inline void timer_base_unlock_expiry(struct timer_base *base) * The counterpart to del_timer_wait_running(). * * If there is a waiter for base->expiry_lock, then it was waiting for the - * timer callback to finish. Drop expiry_lock and reaquire it. That allows + * timer callback to finish. Drop expiry_lock and reacquire it. That allows * the waiter to acquire the lock and make progress. */ static void timer_sync_wait_running(struct timer_base *base) + __releases(&base->lock) __releases(&base->expiry_lock) + __acquires(&base->expiry_lock) __acquires(&base->lock) { if (atomic_read(&base->timer_waiters)) { + raw_spin_unlock_irq(&base->lock); spin_unlock(&base->expiry_lock); spin_lock(&base->expiry_lock); + raw_spin_lock_irq(&base->lock); } } @@ -1289,7 +1582,7 @@ static void del_timer_wait_running(struct timer_list *timer) u32 tf; tf = READ_ONCE(timer->flags); - if (!(tf & TIMER_MIGRATING)) { + if (!(tf & (TIMER_MIGRATING | TIMER_IRQSAFE))) { struct timer_base *base = get_timer_base(tf); /* @@ -1314,44 +1607,29 @@ static inline void timer_sync_wait_running(struct timer_base *base) { } static inline void del_timer_wait_running(struct timer_list *timer) { } #endif -#if defined(CONFIG_SMP) || defined(CONFIG_PREEMPT_RT) /** - * del_timer_sync - deactivate a timer and wait for the handler to finish. - * @timer: the timer to be deactivated - * - * This function only differs from del_timer() on SMP: besides deactivating - * the timer it also makes sure the handler has finished executing on other - * CPUs. - * - * Synchronization rules: Callers must prevent restarting of the timer, - * otherwise this function is meaningless. It must not be called from - * interrupt contexts unless the timer is an irqsafe one. The caller must - * not hold locks which would prevent completion of the timer's - * handler. The timer's handler must not call add_timer_on(). Upon exit the - * timer is not queued and the handler is not running on any CPU. - * - * Note: For !irqsafe timers, you must not hold locks that are held in - * interrupt context while calling this function. Even if the lock has - * nothing to do with the timer in question. Here's why:: - * - * CPU0 CPU1 - * ---- ---- - * <SOFTIRQ> - * call_timer_fn(); - * base->running_timer = mytimer; - * spin_lock_irq(somelock); - * <IRQ> - * spin_lock(somelock); - * del_timer_sync(mytimer); - * while (base->running_timer == mytimer); - * - * Now del_timer_sync() will never return and never release somelock. - * The interrupt on the other CPU is waiting to grab somelock but - * it has interrupted the softirq that CPU0 is waiting to finish. - * - * The function returns whether it has deactivated a pending timer or not. + * __timer_delete_sync - Internal function: Deactivate a timer and wait + * for the handler to finish. + * @timer: The timer to be deactivated + * @shutdown: If true, @timer->function will be set to NULL under the + * timer base lock which prevents rearming of @timer + * + * If @shutdown is not set the timer can be rearmed later. If the timer can + * be rearmed concurrently, i.e. after dropping the base lock then the + * return value is meaningless. + * + * If @shutdown is set then @timer->function is set to NULL under timer + * base lock which prevents rearming of the timer. Any attempt to rearm + * a shutdown timer is silently ignored. + * + * If the timer should be reused after shutdown it has to be initialized + * again. + * + * Return: + * * %0 - The timer was not pending + * * %1 - The timer was pending and deactivated */ -int del_timer_sync(struct timer_list *timer) +static int __timer_delete_sync(struct timer_list *timer, bool shutdown) { int ret; @@ -1371,10 +1649,17 @@ int del_timer_sync(struct timer_list *timer) * don't use it in hardirq context, because it * could lead to deadlock. */ - WARN_ON(in_irq() && !(timer->flags & TIMER_IRQSAFE)); + WARN_ON(in_hardirq() && !(timer->flags & TIMER_IRQSAFE)); + + /* + * Must be able to sleep on PREEMPT_RT because of the slowpath in + * del_timer_wait_running(). + */ + if (IS_ENABLED(CONFIG_PREEMPT_RT) && !(timer->flags & TIMER_IRQSAFE)) + lockdep_assert_preemption_enabled(); do { - ret = try_to_del_timer_sync(timer); + ret = __try_to_del_timer_sync(timer, shutdown); if (unlikely(ret < 0)) { del_timer_wait_running(timer); @@ -1384,8 +1669,96 @@ int del_timer_sync(struct timer_list *timer) return ret; } -EXPORT_SYMBOL(del_timer_sync); -#endif + +/** + * timer_delete_sync - Deactivate a timer and wait for the handler to finish. + * @timer: The timer to be deactivated + * + * Synchronization rules: Callers must prevent restarting of the timer, + * otherwise this function is meaningless. It must not be called from + * interrupt contexts unless the timer is an irqsafe one. The caller must + * not hold locks which would prevent completion of the timer's callback + * function. The timer's handler must not call add_timer_on(). Upon exit + * the timer is not queued and the handler is not running on any CPU. + * + * For !irqsafe timers, the caller must not hold locks that are held in + * interrupt context. Even if the lock has nothing to do with the timer in + * question. Here's why:: + * + * CPU0 CPU1 + * ---- ---- + * <SOFTIRQ> + * call_timer_fn(); + * base->running_timer = mytimer; + * spin_lock_irq(somelock); + * <IRQ> + * spin_lock(somelock); + * timer_delete_sync(mytimer); + * while (base->running_timer == mytimer); + * + * Now timer_delete_sync() will never return and never release somelock. + * The interrupt on the other CPU is waiting to grab somelock but it has + * interrupted the softirq that CPU0 is waiting to finish. + * + * This function cannot guarantee that the timer is not rearmed again by + * some concurrent or preempting code, right after it dropped the base + * lock. If there is the possibility of a concurrent rearm then the return + * value of the function is meaningless. + * + * If such a guarantee is needed, e.g. for teardown situations then use + * timer_shutdown_sync() instead. + * + * Return: + * * %0 - The timer was not pending + * * %1 - The timer was pending and deactivated + */ +int timer_delete_sync(struct timer_list *timer) +{ + return __timer_delete_sync(timer, false); +} +EXPORT_SYMBOL(timer_delete_sync); + +/** + * timer_shutdown_sync - Shutdown a timer and prevent rearming + * @timer: The timer to be shutdown + * + * When the function returns it is guaranteed that: + * - @timer is not queued + * - The callback function of @timer is not running + * - @timer cannot be enqueued again. Any attempt to rearm + * @timer is silently ignored. + * + * See timer_delete_sync() for synchronization rules. + * + * This function is useful for final teardown of an infrastructure where + * the timer is subject to a circular dependency problem. + * + * A common pattern for this is a timer and a workqueue where the timer can + * schedule work and work can arm the timer. On shutdown the workqueue must + * be destroyed and the timer must be prevented from rearming. Unless the + * code has conditionals like 'if (mything->in_shutdown)' to prevent that + * there is no way to get this correct with timer_delete_sync(). + * + * timer_shutdown_sync() is solving the problem. The correct ordering of + * calls in this case is: + * + * timer_shutdown_sync(&mything->timer); + * workqueue_destroy(&mything->workqueue); + * + * After this 'mything' can be safely freed. + * + * This obviously implies that the timer is not required to be functional + * for the rest of the shutdown operation. + * + * Return: + * * %0 - The timer was not pending + * * %1 - The timer was pending + */ +int timer_shutdown_sync(struct timer_list *timer) +{ + return __timer_delete_sync(timer, true); +} +EXPORT_SYMBOL_GPL(timer_shutdown_sync); static void call_timer_fn(struct timer_list *timer, void (*fn)(struct timer_list *), @@ -1407,8 +1780,8 @@ static void call_timer_fn(struct timer_list *timer, #endif /* * Couple the lock chain with the lock chain at - * del_timer_sync() by acquiring the lock_map around the fn() - * call here and in del_timer_sync(). + * timer_delete_sync() by acquiring the lock_map around the fn() + * call here and in timer_delete_sync(). */ lock_map_acquire(&lockdep_map); @@ -1451,25 +1824,31 @@ static void expire_timers(struct timer_base *base, struct hlist_head *head) fn = timer->function; + if (WARN_ON_ONCE(!fn)) { + /* Should never happen. Emphasis on should! */ + base->running_timer = NULL; + continue; + } + if (timer->flags & TIMER_IRQSAFE) { raw_spin_unlock(&base->lock); call_timer_fn(timer, fn, baseclk); - base->running_timer = NULL; raw_spin_lock(&base->lock); + base->running_timer = NULL; } else { raw_spin_unlock_irq(&base->lock); call_timer_fn(timer, fn, baseclk); + raw_spin_lock_irq(&base->lock); base->running_timer = NULL; timer_sync_wait_running(base); - raw_spin_lock_irq(&base->lock); } } } -static int __collect_expired_timers(struct timer_base *base, - struct hlist_head *heads) +static int collect_expired_timers(struct timer_base *base, + struct hlist_head *heads) { - unsigned long clk = base->clk; + unsigned long clk = base->clk = base->next_expiry; struct hlist_head *vec; int i, levels = 0; unsigned int idx; @@ -1491,7 +1870,6 @@ static int __collect_expired_timers(struct timer_base *base, return levels; } -#ifdef CONFIG_NO_HZ_COMMON /* * Find the next pending bucket of a level. Search from level start (@offset) * + @clk upwards and if nothing there, search from start of the level @@ -1514,8 +1892,10 @@ static int next_pending_bucket(struct timer_base *base, unsigned offset, /* * Search the first expiring timer in the various clock levels. Caller must * hold base->lock. + * + * Store next expiry time in base->next_expiry. */ -static unsigned long __next_timer_interrupt(struct timer_base *base) +static void timer_recalc_next_expiry(struct timer_base *base) { unsigned long clk, next, adj; unsigned lvl, offset = 0; @@ -1524,6 +1904,7 @@ static unsigned long __next_timer_interrupt(struct timer_base *base) clk = base->clk; for (lvl = 0; lvl < LVL_DEPTH; lvl++, offset += LVL_SIZE) { int pos = next_pending_bucket(base, offset, clk & LVL_MASK); + unsigned long lvl_clk = clk & LVL_CLK_MASK; if (pos >= 0) { unsigned long tmp = clk + (unsigned long) pos; @@ -1531,13 +1912,20 @@ static unsigned long __next_timer_interrupt(struct timer_base *base) tmp <<= LVL_SHIFT(lvl); if (time_before(tmp, next)) next = tmp; + + /* + * If the next expiration happens before we reach + * the next level, no need to check further. + */ + if (pos <= ((LVL_CLK_DIV - lvl_clk) & LVL_CLK_MASK)) + break; } /* * Clock for the next level. If the current level clock lower * bits are zero, we look at the next level as is. If not we * need to advance it by one because that's going to be the * next expiring bucket in that level. base->clk is the next - * expiring jiffie. So in case of: + * expiring jiffy. So in case of: * * LVL5 LVL4 LVL3 LVL2 LVL1 LVL0 * 0 0 0 0 0 0 @@ -1568,13 +1956,17 @@ static unsigned long __next_timer_interrupt(struct timer_base *base) * So the simple check whether the lower bits of the current * level are 0 or not is sufficient for all cases. */ - adj = clk & LVL_CLK_MASK ? 1 : 0; + adj = lvl_clk ? 1 : 0; clk >>= LVL_CLK_SHIFT; clk += adj; } - return next; + + WRITE_ONCE(base->next_expiry, next); + base->next_expiry_recalc = false; + base->timers_pending = !(next == base->clk + NEXT_TIMER_MAX_DELTA); } +#ifdef CONFIG_NO_HZ_COMMON /* * Check, if the next hrtimer event is before the next timer wheel * event: @@ -1598,7 +1990,7 @@ static u64 cmp_next_hrtimer_event(u64 basem, u64 expires) return basem; /* - * Round up to the next jiffie. High resolution timers are + * Round up to the next jiffy. High resolution timers are * off, so the hrtimers are expired in the tick and we need to * make sure that this tick really expires the timer to avoid * a ping pong of the nohz stop code. @@ -1608,149 +2000,382 @@ static u64 cmp_next_hrtimer_event(u64 basem, u64 expires) return DIV_ROUND_UP_ULL(nextevt, TICK_NSEC) * TICK_NSEC; } -/** - * get_next_timer_interrupt - return the time (clock mono) of the next timer - * @basej: base time jiffies - * @basem: base time clock monotonic - * - * Returns the tick aligned clock monotonic time of the next pending - * timer or KTIME_MAX if no timer is pending. - */ -u64 get_next_timer_interrupt(unsigned long basej, u64 basem) +static unsigned long next_timer_interrupt(struct timer_base *base, + unsigned long basej) { - struct timer_base *base = this_cpu_ptr(&timer_bases[BASE_STD]); - u64 expires = KTIME_MAX; - unsigned long nextevt; - bool is_max_delta; + if (base->next_expiry_recalc) + timer_recalc_next_expiry(base); /* - * Pretend that there is no timer pending if the cpu is offline. - * Possible pending timers will be migrated later to an active cpu. + * Move next_expiry for the empty base into the future to prevent an + * unnecessary raise of the timer softirq when the next_expiry value + * will be reached even if there is no timer pending. + * + * This update is also required to make timer_base::next_expiry values + * easy comparable to find out which base holds the first pending timer. */ - if (cpu_is_offline(smp_processor_id())) - return expires; + if (!base->timers_pending) + WRITE_ONCE(base->next_expiry, basej + NEXT_TIMER_MAX_DELTA); + + return base->next_expiry; +} + +static unsigned long fetch_next_timer_interrupt(unsigned long basej, u64 basem, + struct timer_base *base_local, + struct timer_base *base_global, + struct timer_events *tevt) +{ + unsigned long nextevt, nextevt_local, nextevt_global; + bool local_first; + + nextevt_local = next_timer_interrupt(base_local, basej); + nextevt_global = next_timer_interrupt(base_global, basej); + + local_first = time_before_eq(nextevt_local, nextevt_global); + + nextevt = local_first ? nextevt_local : nextevt_global; - raw_spin_lock(&base->lock); - nextevt = __next_timer_interrupt(base); - is_max_delta = (nextevt == base->clk + NEXT_TIMER_MAX_DELTA); - base->next_expiry = nextevt; /* - * We have a fresh next event. Check whether we can forward the - * base. We can only do that when @basej is past base->clk - * otherwise we might rewind base->clk. + * If the @nextevt is at max. one tick away, use @nextevt and store + * it in the local expiry value. The next global event is irrelevant in + * this case and can be left as KTIME_MAX. */ - if (time_after(basej, base->clk)) { - if (time_after(nextevt, basej)) - base->clk = basej; - else if (time_after(nextevt, base->clk)) - base->clk = nextevt; - } + if (time_before_eq(nextevt, basej + 1)) { + /* If we missed a tick already, force 0 delta */ + if (time_before(nextevt, basej)) + nextevt = basej; + tevt->local = basem + (u64)(nextevt - basej) * TICK_NSEC; - if (time_before_eq(nextevt, basej)) { - expires = basem; - base->is_idle = false; - } else { - if (!is_max_delta) - expires = basem + (u64)(nextevt - basej) * TICK_NSEC; /* - * If we expect to sleep more than a tick, mark the base idle. - * Also the tick is stopped so any added timer must forward - * the base clk itself to keep granularity small. This idle - * logic is only maintained for the BASE_STD base, deferrable - * timers may still see large granularity skew (by design). + * This is required for the remote check only but it doesn't + * hurt, when it is done for both call sites: + * + * * The remote callers will only take care of the global timers + * as local timers will be handled by CPU itself. When not + * updating tevt->global with the already missed first global + * timer, it is possible that it will be missed completely. + * + * * The local callers will ignore the tevt->global anyway, when + * nextevt is max. one tick away. */ - if ((expires - basem) > TICK_NSEC) { - base->must_forward_clk = true; - base->is_idle = true; - } + if (!local_first) + tevt->global = tevt->local; + return nextevt; } - raw_spin_unlock(&base->lock); - return cmp_next_hrtimer_event(basem, expires); + /* + * Update tevt.* values: + * + * If the local queue expires first, then the global event can be + * ignored. If the global queue is empty, nothing to do either. + */ + if (!local_first && base_global->timers_pending) + tevt->global = basem + (u64)(nextevt_global - basej) * TICK_NSEC; + + if (base_local->timers_pending) + tevt->local = basem + (u64)(nextevt_local - basej) * TICK_NSEC; + + return nextevt; } +# ifdef CONFIG_SMP /** - * timer_clear_idle - Clear the idle state of the timer base + * fetch_next_timer_interrupt_remote() - Store next timers into @tevt + * @basej: base time jiffies + * @basem: base time clock monotonic + * @tevt: Pointer to the storage for the expiry values + * @cpu: Remote CPU * - * Called with interrupts disabled + * Stores the next pending local and global timer expiry values in the + * struct pointed to by @tevt. If a queue is empty the corresponding + * field is set to KTIME_MAX. If local event expires before global + * event, global event is set to KTIME_MAX as well. + * + * Caller needs to make sure timer base locks are held (use + * timer_lock_remote_bases() for this purpose). */ -void timer_clear_idle(void) +void fetch_next_timer_interrupt_remote(unsigned long basej, u64 basem, + struct timer_events *tevt, + unsigned int cpu) +{ + struct timer_base *base_local, *base_global; + + /* Preset local / global events */ + tevt->local = tevt->global = KTIME_MAX; + + base_local = per_cpu_ptr(&timer_bases[BASE_LOCAL], cpu); + base_global = per_cpu_ptr(&timer_bases[BASE_GLOBAL], cpu); + + lockdep_assert_held(&base_local->lock); + lockdep_assert_held(&base_global->lock); + + fetch_next_timer_interrupt(basej, basem, base_local, base_global, tevt); +} + +/** + * timer_unlock_remote_bases - unlock timer bases of cpu + * @cpu: Remote CPU + * + * Unlocks the remote timer bases. + */ +void timer_unlock_remote_bases(unsigned int cpu) + __releases(timer_bases[BASE_LOCAL]->lock) + __releases(timer_bases[BASE_GLOBAL]->lock) +{ + struct timer_base *base_local, *base_global; + + base_local = per_cpu_ptr(&timer_bases[BASE_LOCAL], cpu); + base_global = per_cpu_ptr(&timer_bases[BASE_GLOBAL], cpu); + + raw_spin_unlock(&base_global->lock); + raw_spin_unlock(&base_local->lock); +} + +/** + * timer_lock_remote_bases - lock timer bases of cpu + * @cpu: Remote CPU + * + * Locks the remote timer bases. + */ +void timer_lock_remote_bases(unsigned int cpu) + __acquires(timer_bases[BASE_LOCAL]->lock) + __acquires(timer_bases[BASE_GLOBAL]->lock) +{ + struct timer_base *base_local, *base_global; + + base_local = per_cpu_ptr(&timer_bases[BASE_LOCAL], cpu); + base_global = per_cpu_ptr(&timer_bases[BASE_GLOBAL], cpu); + + lockdep_assert_irqs_disabled(); + + raw_spin_lock(&base_local->lock); + raw_spin_lock_nested(&base_global->lock, SINGLE_DEPTH_NESTING); +} + +/** + * timer_base_is_idle() - Return whether timer base is set idle + * + * Returns value of local timer base is_idle value. + */ +bool timer_base_is_idle(void) { - struct timer_base *base = this_cpu_ptr(&timer_bases[BASE_STD]); + return __this_cpu_read(timer_bases[BASE_LOCAL].is_idle); +} + +static void __run_timer_base(struct timer_base *base); + +/** + * timer_expire_remote() - expire global timers of cpu + * @cpu: Remote CPU + * + * Expire timers of global base of remote CPU. + */ +void timer_expire_remote(unsigned int cpu) +{ + struct timer_base *base = per_cpu_ptr(&timer_bases[BASE_GLOBAL], cpu); + + __run_timer_base(base); +} + +static void timer_use_tmigr(unsigned long basej, u64 basem, + unsigned long *nextevt, bool *tick_stop_path, + bool timer_base_idle, struct timer_events *tevt) +{ + u64 next_tmigr; + + if (timer_base_idle) + next_tmigr = tmigr_cpu_new_timer(tevt->global); + else if (tick_stop_path) + next_tmigr = tmigr_cpu_deactivate(tevt->global); + else + next_tmigr = tmigr_quick_check(tevt->global); /* - * We do this unlocked. The worst outcome is a remote enqueue sending - * a pointless IPI, but taking the lock would just make the window for - * sending the IPI a few instructions smaller for the cost of taking - * the lock in the exit from idle path. + * If the CPU is the last going idle in timer migration hierarchy, make + * sure the CPU will wake up in time to handle remote timers. + * next_tmigr == KTIME_MAX if other CPUs are still active. */ - base->is_idle = false; + if (next_tmigr < tevt->local) { + u64 tmp; + + /* If we missed a tick already, force 0 delta */ + if (next_tmigr < basem) + next_tmigr = basem; + + tmp = div_u64(next_tmigr - basem, TICK_NSEC); + + *nextevt = basej + (unsigned long)tmp; + tevt->local = next_tmigr; + } } +# else +static void timer_use_tmigr(unsigned long basej, u64 basem, + unsigned long *nextevt, bool *tick_stop_path, + bool timer_base_idle, struct timer_events *tevt) +{ + /* + * Make sure first event is written into tevt->local to not miss a + * timer on !SMP systems. + */ + tevt->local = min_t(u64, tevt->local, tevt->global); +} +# endif /* CONFIG_SMP */ -static int collect_expired_timers(struct timer_base *base, - struct hlist_head *heads) +static inline u64 __get_next_timer_interrupt(unsigned long basej, u64 basem, + bool *idle) { - unsigned long now = READ_ONCE(jiffies); + struct timer_events tevt = { .local = KTIME_MAX, .global = KTIME_MAX }; + struct timer_base *base_local, *base_global; + unsigned long nextevt; + bool idle_is_possible; + + /* + * When the CPU is offline, the tick is cancelled and nothing is supposed + * to try to stop it. + */ + if (WARN_ON_ONCE(cpu_is_offline(smp_processor_id()))) { + if (idle) + *idle = true; + return tevt.local; + } + + base_local = this_cpu_ptr(&timer_bases[BASE_LOCAL]); + base_global = this_cpu_ptr(&timer_bases[BASE_GLOBAL]); + + raw_spin_lock(&base_local->lock); + raw_spin_lock_nested(&base_global->lock, SINGLE_DEPTH_NESTING); + + nextevt = fetch_next_timer_interrupt(basej, basem, base_local, + base_global, &tevt); + + /* + * If the next event is only one jiffy ahead there is no need to call + * timer migration hierarchy related functions. The value for the next + * global timer in @tevt struct equals then KTIME_MAX. This is also + * true, when the timer base is idle. + * + * The proper timer migration hierarchy function depends on the callsite + * and whether timer base is idle or not. @nextevt will be updated when + * this CPU needs to handle the first timer migration hierarchy + * event. See timer_use_tmigr() for detailed information. + */ + idle_is_possible = time_after(nextevt, basej + 1); + if (idle_is_possible) + timer_use_tmigr(basej, basem, &nextevt, idle, + base_local->is_idle, &tevt); /* - * NOHZ optimization. After a long idle sleep we need to forward the - * base to current jiffies. Avoid a loop by searching the bitfield for - * the next expiring timer. + * We have a fresh next event. Check whether we can forward the + * base. */ - if ((long)(now - base->clk) > 2) { - unsigned long next = __next_timer_interrupt(base); + __forward_timer_base(base_local, basej); + __forward_timer_base(base_global, basej); + /* + * Set base->is_idle only when caller is timer_base_try_to_set_idle() + */ + if (idle) { /* - * If the next timer is ahead of time forward to current - * jiffies, otherwise forward to the next expiry time: + * Bases are idle if the next event is more than a tick + * away. Caution: @nextevt could have changed by enqueueing a + * global timer into timer migration hierarchy. Therefore a new + * check is required here. + * + * If the base is marked idle then any timer add operation must + * forward the base clk itself to keep granularity small. This + * idle logic is only maintained for the BASE_LOCAL and + * BASE_GLOBAL base, deferrable timers may still see large + * granularity skew (by design). */ - if (time_after(next, now)) { + if (!base_local->is_idle && time_after(nextevt, basej + 1)) { + base_local->is_idle = true; /* - * The call site will increment base->clk and then - * terminate the expiry loop immediately. + * Global timers queued locally while running in a task + * in nohz_full mode need a self-IPI to kick reprogramming + * in IRQ tail. */ - base->clk = now; - return 0; + if (tick_nohz_full_cpu(base_local->cpu)) + base_global->is_idle = true; + trace_timer_base_idle(true, base_local->cpu); } - base->clk = next; + *idle = base_local->is_idle; + + /* + * When timer base is not set idle, undo the effect of + * tmigr_cpu_deactivate() to prevent inconsistent states - active + * timer base but inactive timer migration hierarchy. + * + * When timer base was already marked idle, nothing will be + * changed here. + */ + if (!base_local->is_idle && idle_is_possible) + tmigr_cpu_activate(); } - return __collect_expired_timers(base, heads); + + raw_spin_unlock(&base_global->lock); + raw_spin_unlock(&base_local->lock); + + return cmp_next_hrtimer_event(basem, tevt.local); } -#else -static inline int collect_expired_timers(struct timer_base *base, - struct hlist_head *heads) + +/** + * get_next_timer_interrupt() - return the time (clock mono) of the next timer + * @basej: base time jiffies + * @basem: base time clock monotonic + * + * Returns the tick aligned clock monotonic time of the next pending timer or + * KTIME_MAX if no timer is pending. If timer of global base was queued into + * timer migration hierarchy, first global timer is not taken into account. If + * it was the last CPU of timer migration hierarchy going idle, first global + * event is taken into account. + */ +u64 get_next_timer_interrupt(unsigned long basej, u64 basem) { - return __collect_expired_timers(base, heads); + return __get_next_timer_interrupt(basej, basem, NULL); } -#endif -/* - * 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. +/** + * timer_base_try_to_set_idle() - Try to set the idle state of the timer bases + * @basej: base time jiffies + * @basem: base time clock monotonic + * @idle: pointer to store the value of timer_base->is_idle on return; + * *idle contains the information whether tick was already stopped + * + * Returns the tick aligned clock monotonic time of the next pending timer or + * KTIME_MAX if no timer is pending. When tick was already stopped KTIME_MAX is + * returned as well. */ -void update_process_times(int user_tick) +u64 timer_base_try_to_set_idle(unsigned long basej, u64 basem, bool *idle) { - struct task_struct *p = current; + if (*idle) + return KTIME_MAX; - /* Note: this timer irq context must be accounted for as well. */ - account_process_tick(p, user_tick); - run_local_timers(); - rcu_sched_clock_irq(user_tick); -#ifdef CONFIG_IRQ_WORK - if (in_irq()) - irq_work_tick(); -#endif - scheduler_tick(); - if (IS_ENABLED(CONFIG_POSIX_TIMERS)) - run_posix_cpu_timers(); + return __get_next_timer_interrupt(basej, basem, idle); +} - /* The current CPU might make use of net randoms without receiving IRQs - * to renew them often enough. Let's update the net_rand_state from a - * non-constant value that's not affine to the number of calls to make - * sure it's updated when there's some activity (we don't care in idle). +/** + * timer_clear_idle - Clear the idle state of the timer base + * + * Called with interrupts disabled + */ +void timer_clear_idle(void) +{ + /* + * We do this unlocked. The worst outcome is a remote pinned timer + * enqueue sending a pointless IPI, but taking the lock would just + * make the window for sending the IPI a few instructions smaller + * for the cost of taking the lock in the exit from idle + * path. Required for BASE_LOCAL only. */ - this_cpu_add(net_rand_state.s1, rol32(jiffies, 24) + user_tick); + __this_cpu_write(timer_bases[BASE_LOCAL].is_idle, false); + if (tick_nohz_full_cpu(smp_processor_id())) + __this_cpu_write(timer_bases[BASE_GLOBAL].is_idle, false); + trace_timer_base_idle(false, smp_processor_id()); + + /* Activate without holding the timer_base->lock */ + tmigr_cpu_activate(); } +#endif /** * __run_timers - run all expired timers (if any) on this CPU. @@ -1761,206 +2386,141 @@ static inline void __run_timers(struct timer_base *base) struct hlist_head heads[LVL_DEPTH]; int levels; - if (!time_after_eq(jiffies, base->clk)) - return; - - timer_base_lock_expiry(base); - raw_spin_lock_irq(&base->lock); - - /* - * timer_base::must_forward_clk must be cleared before running - * timers so that any timer functions that call mod_timer() will - * not try to forward the base. Idle tracking / clock forwarding - * logic is only used with BASE_STD timers. - * - * The must_forward_clk flag is cleared unconditionally also for - * the deferrable base. The deferrable base is not affected by idle - * tracking and never forwarded, so clearing the flag is a NOOP. - * - * The fact that the deferrable base is never forwarded can cause - * large variations in granularity for deferrable timers, but they - * can be deferred for long periods due to idle anyway. - */ - base->must_forward_clk = false; + lockdep_assert_held(&base->lock); - while (time_after_eq(jiffies, base->clk)) { + if (base->running_timer) + return; + while (time_after_eq(jiffies, base->clk) && + time_after_eq(jiffies, base->next_expiry)) { levels = collect_expired_timers(base, heads); + /* + * The two possible reasons for not finding any expired + * timer at this clk are that all matching timers have been + * dequeued or no timer has been queued since + * base::next_expiry was set to base::clk + + * NEXT_TIMER_MAX_DELTA. + */ + WARN_ON_ONCE(!levels && !base->next_expiry_recalc + && base->timers_pending); + /* + * While executing timers, base->clk is set 1 offset ahead of + * jiffies to avoid endless requeuing to current jiffies. + */ base->clk++; + timer_recalc_next_expiry(base); while (levels--) expire_timers(base, heads + levels); } - raw_spin_unlock_irq(&base->lock); - timer_base_unlock_expiry(base); } -/* - * This function runs timers and the timer-tq in bottom half context. - */ -static __latent_entropy void run_timer_softirq(struct softirq_action *h) +static void __run_timer_base(struct timer_base *base) { - struct timer_base *base = this_cpu_ptr(&timer_bases[BASE_STD]); + /* Can race against a remote CPU updating next_expiry under the lock */ + if (time_before(jiffies, READ_ONCE(base->next_expiry))) + return; + timer_base_lock_expiry(base); + raw_spin_lock_irq(&base->lock); __run_timers(base); - if (IS_ENABLED(CONFIG_NO_HZ_COMMON)) - __run_timers(this_cpu_ptr(&timer_bases[BASE_DEF])); + raw_spin_unlock_irq(&base->lock); + timer_base_unlock_expiry(base); } -/* - * Called by the local, per-CPU timer interrupt on SMP. - */ -void run_local_timers(void) +static void run_timer_base(int index) { - struct timer_base *base = this_cpu_ptr(&timer_bases[BASE_STD]); + struct timer_base *base = this_cpu_ptr(&timer_bases[index]); - hrtimer_run_queues(); - /* Raise the softirq only if required. */ - if (time_before(jiffies, base->clk)) { - if (!IS_ENABLED(CONFIG_NO_HZ_COMMON)) - return; - /* CPU is awake, so check the deferrable base. */ - base++; - if (time_before(jiffies, base->clk)) - return; - } - raise_softirq(TIMER_SOFTIRQ); + __run_timer_base(base); } /* - * Since schedule_timeout()'s timer is defined on the stack, it must store - * the target task on the stack as well. + * This function runs timers and the timer-tq in bottom half context. */ -struct process_timer { - struct timer_list timer; - struct task_struct *task; -}; - -static void process_timeout(struct timer_list *t) +static __latent_entropy void run_timer_softirq(void) { - struct process_timer *timeout = from_timer(timeout, t, timer); + run_timer_base(BASE_LOCAL); + if (IS_ENABLED(CONFIG_NO_HZ_COMMON)) { + run_timer_base(BASE_GLOBAL); + run_timer_base(BASE_DEF); - wake_up_process(timeout->task); + if (is_timers_nohz_active()) + tmigr_handle_remote(); + } } -/** - * schedule_timeout - sleep until timeout - * @timeout: timeout value in jiffies - * - * Make the current task sleep until @timeout jiffies have elapsed. - * The function behavior depends on the current task state - * (see also set_current_state() description): - * - * %TASK_RUNNING - the scheduler is called, but the task does not sleep - * at all. That happens because sched_submit_work() does nothing for - * tasks in %TASK_RUNNING state. - * - * %TASK_UNINTERRUPTIBLE - at least @timeout jiffies are guaranteed to - * pass before the routine returns unless the current task is explicitly - * woken up, (e.g. by wake_up_process()). - * - * %TASK_INTERRUPTIBLE - the routine may return early if a signal is - * delivered to the current task or the current task is explicitly woken - * up. - * - * The current task state is guaranteed to be %TASK_RUNNING when this - * routine returns. - * - * Specifying a @timeout value of %MAX_SCHEDULE_TIMEOUT will schedule - * the CPU away without a bound on the timeout. In this case the return - * value will be %MAX_SCHEDULE_TIMEOUT. - * - * Returns 0 when the timer has expired otherwise the remaining time in - * jiffies will be returned. In all cases the return value is guaranteed - * to be non-negative. +/* + * Called by the local, per-CPU timer interrupt on SMP. */ -signed long __sched schedule_timeout(signed long timeout) +static void run_local_timers(void) { - struct process_timer timer; - unsigned long expire; + struct timer_base *base = this_cpu_ptr(&timer_bases[BASE_LOCAL]); - switch (timeout) - { - case MAX_SCHEDULE_TIMEOUT: - /* - * These two special cases are useful to be comfortable - * in the caller. Nothing more. We could take - * MAX_SCHEDULE_TIMEOUT from one of the negative value - * but I' d like to return a valid offset (>=0) to allow - * the caller to do everything it want with the retval. - */ - schedule(); - goto out; - default: + hrtimer_run_queues(); + + for (int i = 0; i < NR_BASES; i++, base++) { /* - * Another bit of PARANOID. Note that the retval will be - * 0 since no piece of kernel is supposed to do a check - * for a negative retval of schedule_timeout() (since it - * should never happens anyway). You just have the printk() - * that will tell you if something is gone wrong and where. + * Raise the softirq only if required. + * + * timer_base::next_expiry can be written by a remote CPU while + * holding the lock. If this write happens at the same time than + * the lockless local read, sanity checker could complain about + * data corruption. + * + * There are two possible situations where + * timer_base::next_expiry is written by a remote CPU: + * + * 1. Remote CPU expires global timers of this CPU and updates + * timer_base::next_expiry of BASE_GLOBAL afterwards in + * next_timer_interrupt() or timer_recalc_next_expiry(). The + * worst outcome is a superfluous raise of the timer softirq + * when the not yet updated value is read. + * + * 2. A new first pinned timer is enqueued by a remote CPU + * and therefore timer_base::next_expiry of BASE_LOCAL is + * updated. When this update is missed, this isn't a + * problem, as an IPI is executed nevertheless when the CPU + * was idle before. When the CPU wasn't idle but the update + * is missed, then the timer would expire one jiffy late - + * bad luck. + * + * Those unlikely corner cases where the worst outcome is only a + * one jiffy delay or a superfluous raise of the softirq are + * not that expensive as doing the check always while holding + * the lock. + * + * Possible remote writers are using WRITE_ONCE(). Local reader + * uses therefore READ_ONCE(). */ - if (timeout < 0) { - printk(KERN_ERR "schedule_timeout: wrong timeout " - "value %lx\n", timeout); - dump_stack(); - current->state = TASK_RUNNING; - goto out; + if (time_after_eq(jiffies, READ_ONCE(base->next_expiry)) || + (i == BASE_DEF && tmigr_requires_handle_remote())) { + raise_timer_softirq(TIMER_SOFTIRQ); + return; } } - - expire = timeout + jiffies; - - timer.task = current; - timer_setup_on_stack(&timer.timer, process_timeout, 0); - __mod_timer(&timer.timer, expire, MOD_TIMER_NOTPENDING); - schedule(); - del_singleshot_timer_sync(&timer.timer); - - /* Remove the timer from the object tracker */ - destroy_timer_on_stack(&timer.timer); - - timeout = expire - jiffies; - - out: - return timeout < 0 ? 0 : timeout; } -EXPORT_SYMBOL(schedule_timeout); /* - * We can use __set_current_state() here because schedule_timeout() calls - * schedule() unconditionally. + * 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. */ -signed long __sched schedule_timeout_interruptible(signed long timeout) -{ - __set_current_state(TASK_INTERRUPTIBLE); - return schedule_timeout(timeout); -} -EXPORT_SYMBOL(schedule_timeout_interruptible); - -signed long __sched schedule_timeout_killable(signed long timeout) -{ - __set_current_state(TASK_KILLABLE); - return schedule_timeout(timeout); -} -EXPORT_SYMBOL(schedule_timeout_killable); - -signed long __sched schedule_timeout_uninterruptible(signed long timeout) +void update_process_times(int user_tick) { - __set_current_state(TASK_UNINTERRUPTIBLE); - return schedule_timeout(timeout); -} -EXPORT_SYMBOL(schedule_timeout_uninterruptible); + struct task_struct *p = current; -/* - * Like schedule_timeout_uninterruptible(), except this task will not contribute - * to load average. - */ -signed long __sched schedule_timeout_idle(signed long timeout) -{ - __set_current_state(TASK_IDLE); - return schedule_timeout(timeout); + /* Note: this timer irq context must be accounted for as well. */ + account_process_tick(p, user_tick); + run_local_timers(); + rcu_sched_clock_irq(user_tick); +#ifdef CONFIG_IRQ_WORK + if (in_irq()) + irq_work_tick(); +#endif + sched_tick(); + if (IS_ENABLED(CONFIG_POSIX_TIMERS)) + run_posix_cpu_timers(); } -EXPORT_SYMBOL(schedule_timeout_idle); #ifdef CONFIG_HOTPLUG_CPU static void migrate_timer_list(struct timer_base *new_base, struct hlist_head *head) @@ -1985,8 +2545,9 @@ int timers_prepare_cpu(unsigned int cpu) base = per_cpu_ptr(&timer_bases[b], cpu); base->clk = jiffies; base->next_expiry = base->clk + NEXT_TIMER_MAX_DELTA; + base->next_expiry_recalc = false; + base->timers_pending = false; base->is_idle = false; - base->must_forward_clk = true; } return 0; } @@ -1997,8 +2558,6 @@ int timers_dead_cpu(unsigned int cpu) struct timer_base *new_base; int b, i; - BUG_ON(cpu_online(cpu)); - for (b = 0; b < NR_BASES; b++) { old_base = per_cpu_ptr(&timer_bases[b], cpu); new_base = get_cpu_ptr(&timer_bases[b]); @@ -2015,7 +2574,8 @@ int timers_dead_cpu(unsigned int cpu) */ forward_timer_base(new_base); - BUG_ON(old_base->running_timer); + WARN_ON_ONCE(old_base->running_timer); + old_base->running_timer = NULL; for (i = 0; i < WHEEL_SIZE; i++) migrate_timer_list(new_base, old_base->vectors + i); @@ -2039,6 +2599,7 @@ static void __init init_timer_cpu(int cpu) base->cpu = cpu; raw_spin_lock_init(&base->lock); base->clk = jiffies; + base->next_expiry = base->clk + NEXT_TIMER_MAX_DELTA; timer_base_init_expiry_lock(base); } } @@ -2054,59 +2615,6 @@ static void __init init_timer_cpus(void) void __init init_timers(void) { init_timer_cpus(); + posix_cputimers_init_work(); open_softirq(TIMER_SOFTIRQ, run_timer_softirq); } - -/** - * msleep - sleep safely even with waitqueue interruptions - * @msecs: Time in milliseconds to sleep for - */ -void msleep(unsigned int msecs) -{ - unsigned long timeout = msecs_to_jiffies(msecs) + 1; - - while (timeout) - timeout = schedule_timeout_uninterruptible(timeout); -} - -EXPORT_SYMBOL(msleep); - -/** - * msleep_interruptible - sleep waiting for signals - * @msecs: Time in milliseconds to sleep for - */ -unsigned long msleep_interruptible(unsigned int msecs) -{ - unsigned long timeout = msecs_to_jiffies(msecs) + 1; - - while (timeout && !signal_pending(current)) - timeout = schedule_timeout_interruptible(timeout); - return jiffies_to_msecs(timeout); -} - -EXPORT_SYMBOL(msleep_interruptible); - -/** - * usleep_range - Sleep for an approximate time - * @min: Minimum time in usecs to sleep - * @max: Maximum time in usecs to sleep - * - * In non-atomic context where the exact wakeup time is flexible, use - * usleep_range() instead of udelay(). The sleep improves responsiveness - * by avoiding the CPU-hogging busy-wait of udelay(), and the range reduces - * power usage by allowing hrtimers to take advantage of an already- - * scheduled interrupt instead of scheduling a new one just for this sleep. - */ -void __sched usleep_range(unsigned long min, unsigned long max) -{ - ktime_t exp = ktime_add_us(ktime_get(), min); - u64 delta = (u64)(max - min) * NSEC_PER_USEC; - - for (;;) { - __set_current_state(TASK_UNINTERRUPTIBLE); - /* Do not return before the requested sleep time has elapsed */ - if (!schedule_hrtimeout_range(&exp, delta, HRTIMER_MODE_ABS)) - break; - } -} -EXPORT_SYMBOL(usleep_range); diff --git a/kernel/time/timer_list.c b/kernel/time/timer_list.c index acb326f5f50a..1c311c46da50 100644 --- a/kernel/time/timer_list.c +++ b/kernel/time/timer_list.c @@ -42,24 +42,11 @@ static void SEQ_printf(struct seq_file *m, const char *fmt, ...) va_end(args); } -static void print_name_offset(struct seq_file *m, void *sym) -{ - char symname[KSYM_NAME_LEN]; - - if (lookup_symbol_name((unsigned long)sym, symname) < 0) - SEQ_printf(m, "<%pK>", sym); - else - SEQ_printf(m, "%s", symname); -} - static void print_timer(struct seq_file *m, struct hrtimer *taddr, struct hrtimer *timer, int idx, u64 now) { - SEQ_printf(m, " #%d: ", idx); - print_name_offset(m, taddr); - SEQ_printf(m, ", "); - print_name_offset(m, timer->function); + SEQ_printf(m, " #%d: <%pK>, %ps", idx, taddr, timer->function); SEQ_printf(m, ", S:%02x", timer->state); SEQ_printf(m, "\n"); SEQ_printf(m, " # expires at %Lu-%Lu nsecs [in %Ld to %Ld nsecs]\n", @@ -116,9 +103,7 @@ print_base(struct seq_file *m, struct hrtimer_clock_base *base, u64 now) SEQ_printf(m, " .resolution: %u nsecs\n", hrtimer_resolution); - SEQ_printf(m, " .get_time: "); - print_name_offset(m, base->get_time); - SEQ_printf(m, "\n"); + SEQ_printf(m, " .get_time: %ps\n", base->get_time); #ifdef CONFIG_HIGH_RES_TIMERS SEQ_printf(m, " .offset: %Lu nsecs\n", (unsigned long long) ktime_to_ns(base->offset)); @@ -162,11 +147,15 @@ static void print_cpu(struct seq_file *m, int cpu, u64 now) # define P_ns(x) \ SEQ_printf(m, " .%-15s: %Lu nsecs\n", #x, \ (unsigned long long)(ktime_to_ns(ts->x))) +# define P_flag(x, f) \ + SEQ_printf(m, " .%-15s: %d\n", #x, !!(ts->flags & (f))) + { struct tick_sched *ts = tick_get_tick_sched(cpu); - P(nohz_mode); + P_flag(nohz, TS_FLAG_NOHZ); + P_flag(highres, TS_FLAG_HIGHRES); P_ns(last_tick); - P(tick_stopped); + P_flag(tick_stopped, TS_FLAG_STOPPED); P(idle_jiffies); P(idle_calls); P(idle_sleeps); @@ -218,44 +207,39 @@ print_tickdevice(struct seq_file *m, struct tick_device *td, int cpu) SEQ_printf(m, " next_event: %Ld nsecs\n", (unsigned long long) ktime_to_ns(dev->next_event)); - SEQ_printf(m, " set_next_event: "); - print_name_offset(m, dev->set_next_event); - SEQ_printf(m, "\n"); + SEQ_printf(m, " set_next_event: %ps\n", dev->set_next_event); - if (dev->set_state_shutdown) { - SEQ_printf(m, " shutdown: "); - print_name_offset(m, dev->set_state_shutdown); - SEQ_printf(m, "\n"); - } + if (dev->set_state_shutdown) + SEQ_printf(m, " shutdown: %ps\n", + dev->set_state_shutdown); - if (dev->set_state_periodic) { - SEQ_printf(m, " periodic: "); - print_name_offset(m, dev->set_state_periodic); - SEQ_printf(m, "\n"); - } + if (dev->set_state_periodic) + SEQ_printf(m, " periodic: %ps\n", + dev->set_state_periodic); - if (dev->set_state_oneshot) { - SEQ_printf(m, " oneshot: "); - print_name_offset(m, dev->set_state_oneshot); - SEQ_printf(m, "\n"); - } + if (dev->set_state_oneshot) + SEQ_printf(m, " oneshot: %ps\n", + dev->set_state_oneshot); - if (dev->set_state_oneshot_stopped) { - SEQ_printf(m, " oneshot stopped: "); - print_name_offset(m, dev->set_state_oneshot_stopped); - SEQ_printf(m, "\n"); - } + if (dev->set_state_oneshot_stopped) + SEQ_printf(m, " oneshot stopped: %ps\n", + dev->set_state_oneshot_stopped); - if (dev->tick_resume) { - SEQ_printf(m, " resume: "); - print_name_offset(m, dev->tick_resume); - SEQ_printf(m, "\n"); - } + if (dev->tick_resume) + SEQ_printf(m, " resume: %ps\n", + dev->tick_resume); - SEQ_printf(m, " event_handler: "); - print_name_offset(m, dev->event_handler); + SEQ_printf(m, " event_handler: %ps\n", dev->event_handler); SEQ_printf(m, "\n"); SEQ_printf(m, " retries: %lu\n", dev->retries); + +#ifdef CONFIG_GENERIC_CLOCKEVENTS_BROADCAST + if (cpu >= 0) { + const struct clock_event_device *wd = tick_get_wakeup_device(cpu); + + SEQ_printf(m, "Wakeup Device: %s\n", wd ? wd->name : "<NULL>"); + } +#endif SEQ_printf(m, "\n"); } @@ -276,7 +260,7 @@ static void timer_list_show_tickdevices_header(struct seq_file *m) static inline void timer_list_header(struct seq_file *m, u64 now) { - SEQ_printf(m, "Timer List Version: v0.8\n"); + SEQ_printf(m, "Timer List Version: v0.10\n"); SEQ_printf(m, "HRTIMER_MAX_CLOCK_BASES: %d\n", HRTIMER_MAX_CLOCK_BASES); SEQ_printf(m, "now at %Ld nsecs\n", (unsigned long long)now); SEQ_printf(m, "\n"); diff --git a/kernel/time/timer_migration.c b/kernel/time/timer_migration.c new file mode 100644 index 000000000000..2f6330831f08 --- /dev/null +++ b/kernel/time/timer_migration.c @@ -0,0 +1,1863 @@ +// SPDX-License-Identifier: GPL-2.0-only +/* + * Infrastructure for migratable timers + * + * Copyright(C) 2022 linutronix GmbH + */ +#include <linux/cpuhotplug.h> +#include <linux/slab.h> +#include <linux/smp.h> +#include <linux/spinlock.h> +#include <linux/timerqueue.h> +#include <trace/events/ipi.h> + +#include "timer_migration.h" +#include "tick-internal.h" + +#define CREATE_TRACE_POINTS +#include <trace/events/timer_migration.h> + +/* + * The timer migration mechanism is built on a hierarchy of groups. The + * lowest level group contains CPUs, the next level groups of CPU groups + * and so forth. The CPU groups are kept per node so for the normal case + * lock contention won't happen across nodes. Depending on the number of + * CPUs per node even the next level might be kept as groups of CPU groups + * per node and only the levels above cross the node topology. + * + * Example topology for a two node system with 24 CPUs each. + * + * LVL 2 [GRP2:0] + * GRP1:0 = GRP1:M + * + * LVL 1 [GRP1:0] [GRP1:1] + * GRP0:0 - GRP0:2 GRP0:3 - GRP0:5 + * + * LVL 0 [GRP0:0] [GRP0:1] [GRP0:2] [GRP0:3] [GRP0:4] [GRP0:5] + * CPUS 0-7 8-15 16-23 24-31 32-39 40-47 + * + * The groups hold a timer queue of events sorted by expiry time. These + * queues are updated when CPUs go in idle. When they come out of idle + * ignore flag of events is set. + * + * Each group has a designated migrator CPU/group as long as a CPU/group is + * active in the group. This designated role is necessary to avoid that all + * active CPUs in a group try to migrate expired timers from other CPUs, + * which would result in massive lock bouncing. + * + * When a CPU is awake, it checks in it's own timer tick the group + * hierarchy up to the point where it is assigned the migrator role or if + * no CPU is active, it also checks the groups where no migrator is set + * (TMIGR_NONE). + * + * If it finds expired timers in one of the group queues it pulls them over + * from the idle CPU and runs the timer function. After that it updates the + * group and the parent groups if required. + * + * CPUs which go idle arm their CPU local timer hardware for the next local + * (pinned) timer event. If the next migratable timer expires after the + * next local timer or the CPU has no migratable timer pending then the + * CPU does not queue an event in the LVL0 group. If the next migratable + * timer expires before the next local timer then the CPU queues that timer + * in the LVL0 group. In both cases the CPU marks itself idle in the LVL0 + * group. + * + * When CPU comes out of idle and when a group has at least a single active + * child, the ignore flag of the tmigr_event is set. This indicates, that + * the event is ignored even if it is still enqueued in the parent groups + * timer queue. It will be removed when touching the timer queue the next + * time. This spares locking in active path as the lock protects (after + * setup) only event information. For more information about locking, + * please read the section "Locking rules". + * + * If the CPU is the migrator of the group then it delegates that role to + * the next active CPU in the group or sets migrator to TMIGR_NONE when + * there is no active CPU in the group. This delegation needs to be + * propagated up the hierarchy so hand over from other leaves can happen at + * all hierarchy levels w/o doing a search. + * + * When the last CPU in the system goes idle, then it drops all migrator + * duties up to the top level of the hierarchy (LVL2 in the example). It + * then has to make sure, that it arms it's own local hardware timer for + * the earliest event in the system. + * + * + * Lifetime rules: + * --------------- + * + * The groups are built up at init time or when CPUs come online. They are + * not destroyed when a group becomes empty due to offlining. The group + * just won't participate in the hierarchy management anymore. Destroying + * groups would result in interesting race conditions which would just make + * the whole mechanism slow and complex. + * + * + * Locking rules: + * -------------- + * + * For setting up new groups and handling events it's required to lock both + * child and parent group. The lock ordering is always bottom up. This also + * includes the per CPU locks in struct tmigr_cpu. For updating the migrator and + * active CPU/group information atomic_try_cmpxchg() is used instead and only + * the per CPU tmigr_cpu->lock is held. + * + * During the setup of groups tmigr_level_list is required. It is protected by + * @tmigr_mutex. + * + * When @timer_base->lock as well as tmigr related locks are required, the lock + * ordering is: first @timer_base->lock, afterwards tmigr related locks. + * + * + * Protection of the tmigr group state information: + * ------------------------------------------------ + * + * The state information with the list of active children and migrator needs to + * be protected by a sequence counter. It prevents a race when updates in child + * groups are propagated in changed order. The state update is performed + * lockless and group wise. The following scenario describes what happens + * without updating the sequence counter: + * + * Therefore, let's take three groups and four CPUs (CPU2 and CPU3 as well + * as GRP0:1 will not change during the scenario): + * + * LVL 1 [GRP1:0] + * migrator = GRP0:1 + * active = GRP0:0, GRP0:1 + * / \ + * LVL 0 [GRP0:0] [GRP0:1] + * migrator = CPU0 migrator = CPU2 + * active = CPU0 active = CPU2 + * / \ / \ + * CPUs 0 1 2 3 + * active idle active idle + * + * + * 1. CPU0 goes idle. As the update is performed group wise, in the first step + * only GRP0:0 is updated. The update of GRP1:0 is pending as CPU0 has to + * walk the hierarchy. + * + * LVL 1 [GRP1:0] + * migrator = GRP0:1 + * active = GRP0:0, GRP0:1 + * / \ + * LVL 0 [GRP0:0] [GRP0:1] + * --> migrator = TMIGR_NONE migrator = CPU2 + * --> active = active = CPU2 + * / \ / \ + * CPUs 0 1 2 3 + * --> idle idle active idle + * + * 2. While CPU0 goes idle and continues to update the state, CPU1 comes out of + * idle. CPU1 updates GRP0:0. The update for GRP1:0 is pending as CPU1 also + * has to walk the hierarchy. Both CPUs (CPU0 and CPU1) now walk the + * hierarchy to perform the needed update from their point of view. The + * currently visible state looks the following: + * + * LVL 1 [GRP1:0] + * migrator = GRP0:1 + * active = GRP0:0, GRP0:1 + * / \ + * LVL 0 [GRP0:0] [GRP0:1] + * --> migrator = CPU1 migrator = CPU2 + * --> active = CPU1 active = CPU2 + * / \ / \ + * CPUs 0 1 2 3 + * idle --> active active idle + * + * 3. Here is the race condition: CPU1 managed to propagate its changes (from + * step 2) through the hierarchy to GRP1:0 before CPU0 (step 1) did. The + * active members of GRP1:0 remain unchanged after the update since it is + * still valid from CPU1 current point of view: + * + * LVL 1 [GRP1:0] + * --> migrator = GRP0:1 + * --> active = GRP0:0, GRP0:1 + * / \ + * LVL 0 [GRP0:0] [GRP0:1] + * migrator = CPU1 migrator = CPU2 + * active = CPU1 active = CPU2 + * / \ / \ + * CPUs 0 1 2 3 + * idle active active idle + * + * 4. Now CPU0 finally propagates its changes (from step 1) to GRP1:0. + * + * LVL 1 [GRP1:0] + * --> migrator = GRP0:1 + * --> active = GRP0:1 + * / \ + * LVL 0 [GRP0:0] [GRP0:1] + * migrator = CPU1 migrator = CPU2 + * active = CPU1 active = CPU2 + * / \ / \ + * CPUs 0 1 2 3 + * idle active active idle + * + * + * The race of CPU0 vs. CPU1 led to an inconsistent state in GRP1:0. CPU1 is + * active and is correctly listed as active in GRP0:0. However GRP1:0 does not + * have GRP0:0 listed as active, which is wrong. The sequence counter has been + * added to avoid inconsistent states during updates. The state is updated + * atomically only if all members, including the sequence counter, match the + * expected value (compare-and-exchange). + * + * Looking back at the previous example with the addition of the sequence + * counter: The update as performed by CPU0 in step 4 will fail. CPU1 changed + * the sequence number during the update in step 3 so the expected old value (as + * seen by CPU0 before starting the walk) does not match. + * + * Prevent race between new event and last CPU going inactive + * ---------------------------------------------------------- + * + * When the last CPU is going idle and there is a concurrent update of a new + * first global timer of an idle CPU, the group and child states have to be read + * while holding the lock in tmigr_update_events(). The following scenario shows + * what happens, when this is not done. + * + * 1. Only CPU2 is active: + * + * LVL 1 [GRP1:0] + * migrator = GRP0:1 + * active = GRP0:1 + * next_expiry = KTIME_MAX + * / \ + * LVL 0 [GRP0:0] [GRP0:1] + * migrator = TMIGR_NONE migrator = CPU2 + * active = active = CPU2 + * next_expiry = KTIME_MAX next_expiry = KTIME_MAX + * / \ / \ + * CPUs 0 1 2 3 + * idle idle active idle + * + * 2. Now CPU 2 goes idle (and has no global timer, that has to be handled) and + * propagates that to GRP0:1: + * + * LVL 1 [GRP1:0] + * migrator = GRP0:1 + * active = GRP0:1 + * next_expiry = KTIME_MAX + * / \ + * LVL 0 [GRP0:0] [GRP0:1] + * migrator = TMIGR_NONE --> migrator = TMIGR_NONE + * active = --> active = + * next_expiry = KTIME_MAX next_expiry = KTIME_MAX + * / \ / \ + * CPUs 0 1 2 3 + * idle idle --> idle idle + * + * 3. Now the idle state is propagated up to GRP1:0. As this is now the last + * child going idle in top level group, the expiry of the next group event + * has to be handed back to make sure no event is lost. As there is no event + * enqueued, KTIME_MAX is handed back to CPU2. + * + * LVL 1 [GRP1:0] + * --> migrator = TMIGR_NONE + * --> active = + * next_expiry = KTIME_MAX + * / \ + * LVL 0 [GRP0:0] [GRP0:1] + * migrator = TMIGR_NONE migrator = TMIGR_NONE + * active = active = + * next_expiry = KTIME_MAX next_expiry = KTIME_MAX + * / \ / \ + * CPUs 0 1 2 3 + * idle idle --> idle idle + * + * 4. CPU 0 has a new timer queued from idle and it expires at TIMER0. CPU0 + * propagates that to GRP0:0: + * + * LVL 1 [GRP1:0] + * migrator = TMIGR_NONE + * active = + * next_expiry = KTIME_MAX + * / \ + * LVL 0 [GRP0:0] [GRP0:1] + * migrator = TMIGR_NONE migrator = TMIGR_NONE + * active = active = + * --> next_expiry = TIMER0 next_expiry = KTIME_MAX + * / \ / \ + * CPUs 0 1 2 3 + * idle idle idle idle + * + * 5. GRP0:0 is not active, so the new timer has to be propagated to + * GRP1:0. Therefore the GRP1:0 state has to be read. When the stalled value + * (from step 2) is read, the timer is enqueued into GRP1:0, but nothing is + * handed back to CPU0, as it seems that there is still an active child in + * top level group. + * + * LVL 1 [GRP1:0] + * migrator = TMIGR_NONE + * active = + * --> next_expiry = TIMER0 + * / \ + * LVL 0 [GRP0:0] [GRP0:1] + * migrator = TMIGR_NONE migrator = TMIGR_NONE + * active = active = + * next_expiry = TIMER0 next_expiry = KTIME_MAX + * / \ / \ + * CPUs 0 1 2 3 + * idle idle idle idle + * + * This is prevented by reading the state when holding the lock (when a new + * timer has to be propagated from idle path):: + * + * CPU2 (tmigr_inactive_up()) CPU0 (tmigr_new_timer_up()) + * -------------------------- --------------------------- + * // step 3: + * cmpxchg(&GRP1:0->state); + * tmigr_update_events() { + * spin_lock(&GRP1:0->lock); + * // ... update events ... + * // hand back first expiry when GRP1:0 is idle + * spin_unlock(&GRP1:0->lock); + * // ^^^ release state modification + * } + * tmigr_update_events() { + * spin_lock(&GRP1:0->lock) + * // ^^^ acquire state modification + * group_state = atomic_read(&GRP1:0->state) + * // .... update events ... + * // hand back first expiry when GRP1:0 is idle + * spin_unlock(&GRP1:0->lock) <3> + * // ^^^ makes state visible for other + * // callers of tmigr_new_timer_up() + * } + * + * When CPU0 grabs the lock directly after cmpxchg, the first timer is reported + * back to CPU0 and also later on to CPU2. So no timer is missed. A concurrent + * update of the group state from active path is no problem, as the upcoming CPU + * will take care of the group events. + * + * Required event and timerqueue update after a remote expiry: + * ----------------------------------------------------------- + * + * After expiring timers of a remote CPU, a walk through the hierarchy and + * update of events and timerqueues is required. It is obviously needed if there + * is a 'new' global timer but also if there is no new global timer but the + * remote CPU is still idle. + * + * 1. CPU0 and CPU1 are idle and have both a global timer expiring at the same + * time. So both have an event enqueued in the timerqueue of GRP0:0. CPU3 is + * also idle and has no global timer pending. CPU2 is the only active CPU and + * thus also the migrator: + * + * LVL 1 [GRP1:0] + * migrator = GRP0:1 + * active = GRP0:1 + * --> timerqueue = evt-GRP0:0 + * / \ + * LVL 0 [GRP0:0] [GRP0:1] + * migrator = TMIGR_NONE migrator = CPU2 + * active = active = CPU2 + * groupevt.ignore = false groupevt.ignore = true + * groupevt.cpu = CPU0 groupevt.cpu = + * timerqueue = evt-CPU0, timerqueue = + * evt-CPU1 + * / \ / \ + * CPUs 0 1 2 3 + * idle idle active idle + * + * 2. CPU2 starts to expire remote timers. It starts with LVL0 group + * GRP0:1. There is no event queued in the timerqueue, so CPU2 continues with + * the parent of GRP0:1: GRP1:0. In GRP1:0 it dequeues the first event. It + * looks at tmigr_event::cpu struct member and expires the pending timer(s) + * of CPU0. + * + * LVL 1 [GRP1:0] + * migrator = GRP0:1 + * active = GRP0:1 + * --> timerqueue = + * / \ + * LVL 0 [GRP0:0] [GRP0:1] + * migrator = TMIGR_NONE migrator = CPU2 + * active = active = CPU2 + * groupevt.ignore = false groupevt.ignore = true + * --> groupevt.cpu = CPU0 groupevt.cpu = + * timerqueue = evt-CPU0, timerqueue = + * evt-CPU1 + * / \ / \ + * CPUs 0 1 2 3 + * idle idle active idle + * + * 3. Some work has to be done after expiring the timers of CPU0. If we stop + * here, then CPU1's pending global timer(s) will not expire in time and the + * timerqueue of GRP0:0 has still an event for CPU0 enqueued which has just + * been processed. So it is required to walk the hierarchy from CPU0's point + * of view and update it accordingly. CPU0's event will be removed from the + * timerqueue because it has no pending timer. If CPU0 would have a timer + * pending then it has to expire after CPU1's first timer because all timers + * from this period were just expired. Either way CPU1's event will be first + * in GRP0:0's timerqueue and therefore set in the CPU field of the group + * event which is then enqueued in GRP1:0's timerqueue as GRP0:0 is still not + * active: + * + * LVL 1 [GRP1:0] + * migrator = GRP0:1 + * active = GRP0:1 + * --> timerqueue = evt-GRP0:0 + * / \ + * LVL 0 [GRP0:0] [GRP0:1] + * migrator = TMIGR_NONE migrator = CPU2 + * active = active = CPU2 + * groupevt.ignore = false groupevt.ignore = true + * --> groupevt.cpu = CPU1 groupevt.cpu = + * --> timerqueue = evt-CPU1 timerqueue = + * / \ / \ + * CPUs 0 1 2 3 + * idle idle active idle + * + * Now CPU2 (migrator) will continue step 2 at GRP1:0 and will expire the + * timer(s) of CPU1. + * + * The hierarchy walk in step 3 can be skipped if the migrator notices that a + * CPU of GRP0:0 is active again. The CPU will mark GRP0:0 active and take care + * of the group as migrator and any needed updates within the hierarchy. + */ + +static DEFINE_MUTEX(tmigr_mutex); +static struct list_head *tmigr_level_list __read_mostly; + +static unsigned int tmigr_hierarchy_levels __read_mostly; +static unsigned int tmigr_crossnode_level __read_mostly; + +static DEFINE_PER_CPU(struct tmigr_cpu, tmigr_cpu); + +#define TMIGR_NONE 0xFF +#define BIT_CNT 8 + +static inline bool tmigr_is_not_available(struct tmigr_cpu *tmc) +{ + return !(tmc->tmgroup && tmc->online); +} + +/* + * Returns true, when @childmask corresponds to the group migrator or when the + * group is not active - so no migrator is set. + */ +static bool tmigr_check_migrator(struct tmigr_group *group, u8 childmask) +{ + union tmigr_state s; + + s.state = atomic_read(&group->migr_state); + + if ((s.migrator == childmask) || (s.migrator == TMIGR_NONE)) + return true; + + return false; +} + +static bool tmigr_check_migrator_and_lonely(struct tmigr_group *group, u8 childmask) +{ + bool lonely, migrator = false; + unsigned long active; + union tmigr_state s; + + s.state = atomic_read(&group->migr_state); + + if ((s.migrator == childmask) || (s.migrator == TMIGR_NONE)) + migrator = true; + + active = s.active; + lonely = bitmap_weight(&active, BIT_CNT) <= 1; + + return (migrator && lonely); +} + +static bool tmigr_check_lonely(struct tmigr_group *group) +{ + unsigned long active; + union tmigr_state s; + + s.state = atomic_read(&group->migr_state); + + active = s.active; + + return bitmap_weight(&active, BIT_CNT) <= 1; +} + +/** + * struct tmigr_walk - data required for walking the hierarchy + * @nextexp: Next CPU event expiry information which is handed into + * the timer migration code by the timer code + * (get_next_timer_interrupt()) + * @firstexp: Contains the first event expiry information when + * hierarchy is completely idle. When CPU itself was the + * last going idle, information makes sure, that CPU will + * be back in time. When using this value in the remote + * expiry case, firstexp is stored in the per CPU tmigr_cpu + * struct of CPU which expires remote timers. It is updated + * in top level group only. Be aware, there could occur a + * new top level of the hierarchy between the 'top level + * call' in tmigr_update_events() and the check for the + * parent group in walk_groups(). Then @firstexp might + * contain a value != KTIME_MAX even if it was not the + * final top level. This is not a problem, as the worst + * outcome is a CPU which might wake up a little early. + * @evt: Pointer to tmigr_event which needs to be queued (of idle + * child group) + * @childmask: groupmask of child group + * @remote: Is set, when the new timer path is executed in + * tmigr_handle_remote_cpu() + * @basej: timer base in jiffies + * @now: timer base monotonic + * @check: is set if there is the need to handle remote timers; + * required in tmigr_requires_handle_remote() only + * @tmc_active: this flag indicates, whether the CPU which triggers + * the hierarchy walk is !idle in the timer migration + * hierarchy. When the CPU is idle and the whole hierarchy is + * idle, only the first event of the top level has to be + * considered. + */ +struct tmigr_walk { + u64 nextexp; + u64 firstexp; + struct tmigr_event *evt; + u8 childmask; + bool remote; + unsigned long basej; + u64 now; + bool check; + bool tmc_active; +}; + +typedef bool (*up_f)(struct tmigr_group *, struct tmigr_group *, struct tmigr_walk *); + +static void __walk_groups(up_f up, struct tmigr_walk *data, + struct tmigr_cpu *tmc) +{ + struct tmigr_group *child = NULL, *group = tmc->tmgroup; + + do { + WARN_ON_ONCE(group->level >= tmigr_hierarchy_levels); + + if (up(group, child, data)) + break; + + child = group; + /* + * Pairs with the store release on group connection + * to make sure group initialization is visible. + */ + group = READ_ONCE(group->parent); + data->childmask = child->groupmask; + WARN_ON_ONCE(!data->childmask); + } while (group); +} + +static void walk_groups(up_f up, struct tmigr_walk *data, struct tmigr_cpu *tmc) +{ + lockdep_assert_held(&tmc->lock); + + __walk_groups(up, data, tmc); +} + +/* + * Returns the next event of the timerqueue @group->events + * + * Removes timers with ignore flag and update next_expiry of the group. Values + * of the group event are updated in tmigr_update_events() only. + */ +static struct tmigr_event *tmigr_next_groupevt(struct tmigr_group *group) +{ + struct timerqueue_node *node = NULL; + struct tmigr_event *evt = NULL; + + lockdep_assert_held(&group->lock); + + WRITE_ONCE(group->next_expiry, KTIME_MAX); + + while ((node = timerqueue_getnext(&group->events))) { + evt = container_of(node, struct tmigr_event, nextevt); + + if (!READ_ONCE(evt->ignore)) { + WRITE_ONCE(group->next_expiry, evt->nextevt.expires); + return evt; + } + + /* + * Remove next timers with ignore flag, because the group lock + * is held anyway + */ + if (!timerqueue_del(&group->events, node)) + break; + } + + return NULL; +} + +/* + * Return the next event (with the expiry equal or before @now) + * + * Event, which is returned, is also removed from the queue. + */ +static struct tmigr_event *tmigr_next_expired_groupevt(struct tmigr_group *group, + u64 now) +{ + struct tmigr_event *evt = tmigr_next_groupevt(group); + + if (!evt || now < evt->nextevt.expires) + return NULL; + + /* + * The event is ready to expire. Remove it and update next group event. + */ + timerqueue_del(&group->events, &evt->nextevt); + tmigr_next_groupevt(group); + + return evt; +} + +static u64 tmigr_next_groupevt_expires(struct tmigr_group *group) +{ + struct tmigr_event *evt; + + evt = tmigr_next_groupevt(group); + + if (!evt) + return KTIME_MAX; + else + return evt->nextevt.expires; +} + +static bool tmigr_active_up(struct tmigr_group *group, + struct tmigr_group *child, + struct tmigr_walk *data) +{ + union tmigr_state curstate, newstate; + bool walk_done; + u8 childmask; + + childmask = data->childmask; + /* + * No memory barrier is required here in contrast to + * tmigr_inactive_up(), as the group state change does not depend on the + * child state. + */ + curstate.state = atomic_read(&group->migr_state); + + do { + newstate = curstate; + walk_done = true; + + if (newstate.migrator == TMIGR_NONE) { + newstate.migrator = childmask; + + /* Changes need to be propagated */ + walk_done = false; + } + + newstate.active |= childmask; + newstate.seq++; + + } while (!atomic_try_cmpxchg(&group->migr_state, &curstate.state, newstate.state)); + + trace_tmigr_group_set_cpu_active(group, newstate, childmask); + + /* + * The group is active (again). The group event might be still queued + * into the parent group's timerqueue but can now be handled by the + * migrator of this group. Therefore the ignore flag for the group event + * is updated to reflect this. + * + * The update of the ignore flag in the active path is done lockless. In + * worst case the migrator of the parent group observes the change too + * late and expires remotely all events belonging to this group. The + * lock is held while updating the ignore flag in idle path. So this + * state change will not be lost. + */ + WRITE_ONCE(group->groupevt.ignore, true); + + return walk_done; +} + +static void __tmigr_cpu_activate(struct tmigr_cpu *tmc) +{ + struct tmigr_walk data; + + data.childmask = tmc->groupmask; + + trace_tmigr_cpu_active(tmc); + + tmc->cpuevt.ignore = true; + WRITE_ONCE(tmc->wakeup, KTIME_MAX); + + walk_groups(&tmigr_active_up, &data, tmc); +} + +/** + * tmigr_cpu_activate() - set this CPU active in timer migration hierarchy + * + * Call site timer_clear_idle() is called with interrupts disabled. + */ +void tmigr_cpu_activate(void) +{ + struct tmigr_cpu *tmc = this_cpu_ptr(&tmigr_cpu); + + if (tmigr_is_not_available(tmc)) + return; + + if (WARN_ON_ONCE(!tmc->idle)) + return; + + raw_spin_lock(&tmc->lock); + tmc->idle = false; + __tmigr_cpu_activate(tmc); + raw_spin_unlock(&tmc->lock); +} + +/* + * Returns true, if there is nothing to be propagated to the next level + * + * @data->firstexp is set to expiry of first gobal event of the (top level of + * the) hierarchy, but only when hierarchy is completely idle. + * + * The child and group states need to be read under the lock, to prevent a race + * against a concurrent tmigr_inactive_up() run when the last CPU goes idle. See + * also section "Prevent race between new event and last CPU going inactive" in + * the documentation at the top. + * + * This is the only place where the group event expiry value is set. + */ +static +bool tmigr_update_events(struct tmigr_group *group, struct tmigr_group *child, + struct tmigr_walk *data) +{ + struct tmigr_event *evt, *first_childevt; + union tmigr_state childstate, groupstate; + bool remote = data->remote; + bool walk_done = false; + bool ignore; + u64 nextexp; + + if (child) { + raw_spin_lock(&child->lock); + raw_spin_lock_nested(&group->lock, SINGLE_DEPTH_NESTING); + + childstate.state = atomic_read(&child->migr_state); + groupstate.state = atomic_read(&group->migr_state); + + if (childstate.active) { + walk_done = true; + goto unlock; + } + + first_childevt = tmigr_next_groupevt(child); + nextexp = child->next_expiry; + evt = &child->groupevt; + + /* + * This can race with concurrent idle exit (activate). + * If the current writer wins, a useless remote expiration may + * be scheduled. If the activate wins, the event is properly + * ignored. + */ + ignore = (nextexp == KTIME_MAX) ? true : false; + WRITE_ONCE(evt->ignore, ignore); + } else { + nextexp = data->nextexp; + + first_childevt = evt = data->evt; + ignore = evt->ignore; + + /* + * Walking the hierarchy is required in any case when a + * remote expiry was done before. This ensures to not lose + * already queued events in non active groups (see section + * "Required event and timerqueue update after a remote + * expiry" in the documentation at the top). + * + * The two call sites which are executed without a remote expiry + * before, are not prevented from propagating changes through + * the hierarchy by the return: + * - When entering this path by tmigr_new_timer(), @evt->ignore + * is never set. + * - tmigr_inactive_up() takes care of the propagation by + * itself and ignores the return value. But an immediate + * return is possible if there is a parent, sparing group + * locking at this level, because the upper walking call to + * the parent will take care about removing this event from + * within the group and update next_expiry accordingly. + * + * However if there is no parent, ie: the hierarchy has only a + * single level so @group is the top level group, make sure the + * first event information of the group is updated properly and + * also handled properly, so skip this fast return path. + */ + if (ignore && !remote && group->parent) + return true; + + raw_spin_lock(&group->lock); + + childstate.state = 0; + groupstate.state = atomic_read(&group->migr_state); + } + + /* + * If the child event is already queued in the group, remove it from the + * queue when the expiry time changed only or when it could be ignored. + */ + if (timerqueue_node_queued(&evt->nextevt)) { + if ((evt->nextevt.expires == nextexp) && !ignore) { + /* Make sure not to miss a new CPU event with the same expiry */ + evt->cpu = first_childevt->cpu; + goto check_toplvl; + } + + if (!timerqueue_del(&group->events, &evt->nextevt)) + WRITE_ONCE(group->next_expiry, KTIME_MAX); + } + + if (ignore) { + /* + * When the next child event could be ignored (nextexp is + * KTIME_MAX) and there was no remote timer handling before or + * the group is already active, there is no need to walk the + * hierarchy even if there is a parent group. + * + * The other way round: even if the event could be ignored, but + * if a remote timer handling was executed before and the group + * is not active, walking the hierarchy is required to not miss + * an enqueued timer in the non active group. The enqueued timer + * of the group needs to be propagated to a higher level to + * ensure it is handled. + */ + if (!remote || groupstate.active) + walk_done = true; + } else { + evt->nextevt.expires = nextexp; + evt->cpu = first_childevt->cpu; + + if (timerqueue_add(&group->events, &evt->nextevt)) + WRITE_ONCE(group->next_expiry, nextexp); + } + +check_toplvl: + if (!group->parent && (groupstate.migrator == TMIGR_NONE)) { + walk_done = true; + + /* + * Nothing to do when update was done during remote timer + * handling. First timer in top level group which needs to be + * handled when top level group is not active, is calculated + * directly in tmigr_handle_remote_up(). + */ + if (remote) + goto unlock; + + /* + * The top level group is idle and it has to be ensured the + * global timers are handled in time. (This could be optimized + * by keeping track of the last global scheduled event and only + * arming it on the CPU if the new event is earlier. Not sure if + * its worth the complexity.) + */ + data->firstexp = tmigr_next_groupevt_expires(group); + } + + trace_tmigr_update_events(child, group, childstate, groupstate, + nextexp); + +unlock: + raw_spin_unlock(&group->lock); + + if (child) + raw_spin_unlock(&child->lock); + + return walk_done; +} + +static bool tmigr_new_timer_up(struct tmigr_group *group, + struct tmigr_group *child, + struct tmigr_walk *data) +{ + return tmigr_update_events(group, child, data); +} + +/* + * Returns the expiry of the next timer that needs to be handled. KTIME_MAX is + * returned, if an active CPU will handle all the timer migration hierarchy + * timers. + */ +static u64 tmigr_new_timer(struct tmigr_cpu *tmc, u64 nextexp) +{ + struct tmigr_walk data = { .nextexp = nextexp, + .firstexp = KTIME_MAX, + .evt = &tmc->cpuevt }; + + lockdep_assert_held(&tmc->lock); + + if (tmc->remote) + return KTIME_MAX; + + trace_tmigr_cpu_new_timer(tmc); + + tmc->cpuevt.ignore = false; + data.remote = false; + + walk_groups(&tmigr_new_timer_up, &data, tmc); + + /* If there is a new first global event, make sure it is handled */ + return data.firstexp; +} + +static void tmigr_handle_remote_cpu(unsigned int cpu, u64 now, + unsigned long jif) +{ + struct timer_events tevt; + struct tmigr_walk data; + struct tmigr_cpu *tmc; + + tmc = per_cpu_ptr(&tmigr_cpu, cpu); + + raw_spin_lock_irq(&tmc->lock); + + /* + * If the remote CPU is offline then the timers have been migrated to + * another CPU. + * + * If tmigr_cpu::remote is set, at the moment another CPU already + * expires the timers of the remote CPU. + * + * If tmigr_event::ignore is set, then the CPU returns from idle and + * takes care of its timers. + * + * If the next event expires in the future, then the event has been + * updated and there are no timers to expire right now. The CPU which + * updated the event takes care when hierarchy is completely + * idle. Otherwise the migrator does it as the event is enqueued. + */ + if (!tmc->online || tmc->remote || tmc->cpuevt.ignore || + now < tmc->cpuevt.nextevt.expires) { + raw_spin_unlock_irq(&tmc->lock); + return; + } + + trace_tmigr_handle_remote_cpu(tmc); + + tmc->remote = true; + WRITE_ONCE(tmc->wakeup, KTIME_MAX); + + /* Drop the lock to allow the remote CPU to exit idle */ + raw_spin_unlock_irq(&tmc->lock); + + if (cpu != smp_processor_id()) + timer_expire_remote(cpu); + + /* + * Lock ordering needs to be preserved - timer_base locks before tmigr + * related locks (see section "Locking rules" in the documentation at + * the top). During fetching the next timer interrupt, also tmc->lock + * needs to be held. Otherwise there is a possible race window against + * the CPU itself when it comes out of idle, updates the first timer in + * the hierarchy and goes back to idle. + * + * timer base locks are dropped as fast as possible: After checking + * whether the remote CPU went offline in the meantime and after + * fetching the next remote timer interrupt. Dropping the locks as fast + * as possible keeps the locking region small and prevents holding + * several (unnecessary) locks during walking the hierarchy for updating + * the timerqueue and group events. + */ + local_irq_disable(); + timer_lock_remote_bases(cpu); + raw_spin_lock(&tmc->lock); + + /* + * When the CPU went offline in the meantime, no hierarchy walk has to + * be done for updating the queued events, because the walk was + * already done during marking the CPU offline in the hierarchy. + * + * When the CPU is no longer idle, the CPU takes care of the timers and + * also of the timers in the hierarchy. + * + * (See also section "Required event and timerqueue update after a + * remote expiry" in the documentation at the top) + */ + if (!tmc->online || !tmc->idle) { + timer_unlock_remote_bases(cpu); + goto unlock; + } + + /* next event of CPU */ + fetch_next_timer_interrupt_remote(jif, now, &tevt, cpu); + timer_unlock_remote_bases(cpu); + + data.nextexp = tevt.global; + data.firstexp = KTIME_MAX; + data.evt = &tmc->cpuevt; + data.remote = true; + + /* + * The update is done even when there is no 'new' global timer pending + * on the remote CPU (see section "Required event and timerqueue update + * after a remote expiry" in the documentation at the top) + */ + walk_groups(&tmigr_new_timer_up, &data, tmc); + +unlock: + tmc->remote = false; + raw_spin_unlock_irq(&tmc->lock); +} + +static bool tmigr_handle_remote_up(struct tmigr_group *group, + struct tmigr_group *child, + struct tmigr_walk *data) +{ + struct tmigr_event *evt; + unsigned long jif; + u8 childmask; + u64 now; + + jif = data->basej; + now = data->now; + + childmask = data->childmask; + + trace_tmigr_handle_remote(group); +again: + /* + * Handle the group only if @childmask is the migrator or if the + * group has no migrator. Otherwise the group is active and is + * handled by its own migrator. + */ + if (!tmigr_check_migrator(group, childmask)) + return true; + + raw_spin_lock_irq(&group->lock); + + evt = tmigr_next_expired_groupevt(group, now); + + if (evt) { + unsigned int remote_cpu = evt->cpu; + + raw_spin_unlock_irq(&group->lock); + + tmigr_handle_remote_cpu(remote_cpu, now, jif); + + /* check if there is another event, that needs to be handled */ + goto again; + } + + /* + * Keep track of the expiry of the first event that needs to be handled + * (group->next_expiry was updated by tmigr_next_expired_groupevt(), + * next was set by tmigr_handle_remote_cpu()). + */ + data->firstexp = group->next_expiry; + + raw_spin_unlock_irq(&group->lock); + + return false; +} + +/** + * tmigr_handle_remote() - Handle global timers of remote idle CPUs + * + * Called from the timer soft interrupt with interrupts enabled. + */ +void tmigr_handle_remote(void) +{ + struct tmigr_cpu *tmc = this_cpu_ptr(&tmigr_cpu); + struct tmigr_walk data; + + if (tmigr_is_not_available(tmc)) + return; + + data.childmask = tmc->groupmask; + data.firstexp = KTIME_MAX; + + /* + * NOTE: This is a doubled check because the migrator test will be done + * in tmigr_handle_remote_up() anyway. Keep this check to speed up the + * return when nothing has to be done. + */ + if (!tmigr_check_migrator(tmc->tmgroup, tmc->groupmask)) { + /* + * If this CPU was an idle migrator, make sure to clear its wakeup + * value so it won't chase timers that have already expired elsewhere. + * This avoids endless requeue from tmigr_new_timer(). + */ + if (READ_ONCE(tmc->wakeup) == KTIME_MAX) + return; + } + + data.now = get_jiffies_update(&data.basej); + + /* + * Update @tmc->wakeup only at the end and do not reset @tmc->wakeup to + * KTIME_MAX. Even if tmc->lock is not held during the whole remote + * handling, tmc->wakeup is fine to be stale as it is called in + * interrupt context and tick_nohz_next_event() is executed in interrupt + * exit path only after processing the last pending interrupt. + */ + + __walk_groups(&tmigr_handle_remote_up, &data, tmc); + + raw_spin_lock_irq(&tmc->lock); + WRITE_ONCE(tmc->wakeup, data.firstexp); + raw_spin_unlock_irq(&tmc->lock); +} + +static bool tmigr_requires_handle_remote_up(struct tmigr_group *group, + struct tmigr_group *child, + struct tmigr_walk *data) +{ + u8 childmask; + + childmask = data->childmask; + + /* + * Handle the group only if the child is the migrator or if the group + * has no migrator. Otherwise the group is active and is handled by its + * own migrator. + */ + if (!tmigr_check_migrator(group, childmask)) + return true; + + /* + * When there is a parent group and the CPU which triggered the + * hierarchy walk is not active, proceed the walk to reach the top level + * group before reading the next_expiry value. + */ + if (group->parent && !data->tmc_active) + return false; + + /* + * The lock is required on 32bit architectures to read the variable + * consistently with a concurrent writer. On 64bit the lock is not + * required because the read operation is not split and so it is always + * consistent. + */ + if (IS_ENABLED(CONFIG_64BIT)) { + data->firstexp = READ_ONCE(group->next_expiry); + if (data->now >= data->firstexp) { + data->check = true; + return true; + } + } else { + raw_spin_lock(&group->lock); + data->firstexp = group->next_expiry; + if (data->now >= group->next_expiry) { + data->check = true; + raw_spin_unlock(&group->lock); + return true; + } + raw_spin_unlock(&group->lock); + } + + return false; +} + +/** + * tmigr_requires_handle_remote() - Check the need of remote timer handling + * + * Must be called with interrupts disabled. + */ +bool tmigr_requires_handle_remote(void) +{ + struct tmigr_cpu *tmc = this_cpu_ptr(&tmigr_cpu); + struct tmigr_walk data; + unsigned long jif; + bool ret = false; + + if (tmigr_is_not_available(tmc)) + return ret; + + data.now = get_jiffies_update(&jif); + data.childmask = tmc->groupmask; + data.firstexp = KTIME_MAX; + data.tmc_active = !tmc->idle; + data.check = false; + + /* + * If the CPU is active, walk the hierarchy to check whether a remote + * expiry is required. + * + * Check is done lockless as interrupts are disabled and @tmc->idle is + * set only by the local CPU. + */ + if (!tmc->idle) { + __walk_groups(&tmigr_requires_handle_remote_up, &data, tmc); + + return data.check; + } + + /* + * When the CPU is idle, compare @tmc->wakeup with @data.now. The lock + * is required on 32bit architectures to read the variable consistently + * with a concurrent writer. On 64bit the lock is not required because + * the read operation is not split and so it is always consistent. + */ + if (IS_ENABLED(CONFIG_64BIT)) { + if (data.now >= READ_ONCE(tmc->wakeup)) + return true; + } else { + raw_spin_lock(&tmc->lock); + if (data.now >= tmc->wakeup) + ret = true; + raw_spin_unlock(&tmc->lock); + } + + return ret; +} + +/** + * tmigr_cpu_new_timer() - enqueue next global timer into hierarchy (idle tmc) + * @nextexp: Next expiry of global timer (or KTIME_MAX if not) + * + * The CPU is already deactivated in the timer migration + * hierarchy. tick_nohz_get_sleep_length() calls tick_nohz_next_event() + * and thereby the timer idle path is executed once more. @tmc->wakeup + * holds the first timer, when the timer migration hierarchy is + * completely idle. + * + * Returns the first timer that needs to be handled by this CPU or KTIME_MAX if + * nothing needs to be done. + */ +u64 tmigr_cpu_new_timer(u64 nextexp) +{ + struct tmigr_cpu *tmc = this_cpu_ptr(&tmigr_cpu); + u64 ret; + + if (tmigr_is_not_available(tmc)) + return nextexp; + + raw_spin_lock(&tmc->lock); + + ret = READ_ONCE(tmc->wakeup); + if (nextexp != KTIME_MAX) { + if (nextexp != tmc->cpuevt.nextevt.expires || + tmc->cpuevt.ignore) { + ret = tmigr_new_timer(tmc, nextexp); + /* + * Make sure the reevaluation of timers in idle path + * will not miss an event. + */ + WRITE_ONCE(tmc->wakeup, ret); + } + } + trace_tmigr_cpu_new_timer_idle(tmc, nextexp); + raw_spin_unlock(&tmc->lock); + return ret; +} + +static bool tmigr_inactive_up(struct tmigr_group *group, + struct tmigr_group *child, + struct tmigr_walk *data) +{ + union tmigr_state curstate, newstate, childstate; + bool walk_done; + u8 childmask; + + childmask = data->childmask; + childstate.state = 0; + + /* + * The memory barrier is paired with the cmpxchg() in tmigr_active_up() + * to make sure the updates of child and group states are ordered. The + * ordering is mandatory, as the group state change depends on the child + * state. + */ + curstate.state = atomic_read_acquire(&group->migr_state); + + for (;;) { + if (child) + childstate.state = atomic_read(&child->migr_state); + + newstate = curstate; + walk_done = true; + + /* Reset active bit when the child is no longer active */ + if (!childstate.active) + newstate.active &= ~childmask; + + if (newstate.migrator == childmask) { + /* + * Find a new migrator for the group, because the child + * group is idle! + */ + if (!childstate.active) { + unsigned long new_migr_bit, active = newstate.active; + + new_migr_bit = find_first_bit(&active, BIT_CNT); + + if (new_migr_bit != BIT_CNT) { + newstate.migrator = BIT(new_migr_bit); + } else { + newstate.migrator = TMIGR_NONE; + + /* Changes need to be propagated */ + walk_done = false; + } + } + } + + newstate.seq++; + + WARN_ON_ONCE((newstate.migrator != TMIGR_NONE) && !(newstate.active)); + + if (atomic_try_cmpxchg(&group->migr_state, &curstate.state, newstate.state)) { + trace_tmigr_group_set_cpu_inactive(group, newstate, childmask); + break; + } + + /* + * The memory barrier is paired with the cmpxchg() in + * tmigr_active_up() to make sure the updates of child and group + * states are ordered. It is required only when the above + * try_cmpxchg() fails. + */ + smp_mb__after_atomic(); + } + + data->remote = false; + + /* Event Handling */ + tmigr_update_events(group, child, data); + + return walk_done; +} + +static u64 __tmigr_cpu_deactivate(struct tmigr_cpu *tmc, u64 nextexp) +{ + struct tmigr_walk data = { .nextexp = nextexp, + .firstexp = KTIME_MAX, + .evt = &tmc->cpuevt, + .childmask = tmc->groupmask }; + + /* + * If nextexp is KTIME_MAX, the CPU event will be ignored because the + * local timer expires before the global timer, no global timer is set + * or CPU goes offline. + */ + if (nextexp != KTIME_MAX) + tmc->cpuevt.ignore = false; + + walk_groups(&tmigr_inactive_up, &data, tmc); + return data.firstexp; +} + +/** + * tmigr_cpu_deactivate() - Put current CPU into inactive state + * @nextexp: The next global timer expiry of the current CPU + * + * Must be called with interrupts disabled. + * + * Return: the next event expiry of the current CPU or the next event expiry + * from the hierarchy if this CPU is the top level migrator or the hierarchy is + * completely idle. + */ +u64 tmigr_cpu_deactivate(u64 nextexp) +{ + struct tmigr_cpu *tmc = this_cpu_ptr(&tmigr_cpu); + u64 ret; + + if (tmigr_is_not_available(tmc)) + return nextexp; + + raw_spin_lock(&tmc->lock); + + ret = __tmigr_cpu_deactivate(tmc, nextexp); + + tmc->idle = true; + + /* + * Make sure the reevaluation of timers in idle path will not miss an + * event. + */ + WRITE_ONCE(tmc->wakeup, ret); + + trace_tmigr_cpu_idle(tmc, nextexp); + raw_spin_unlock(&tmc->lock); + return ret; +} + +/** + * tmigr_quick_check() - Quick forecast of next tmigr event when CPU wants to + * go idle + * @nextevt: The next global timer expiry of the current CPU + * + * Return: + * * KTIME_MAX - when it is probable that nothing has to be done (not + * the only one in the level 0 group; and if it is the + * only one in level 0 group, but there are more than a + * single group active on the way to top level) + * * nextevt - when CPU is offline and has to handle timer on its own + * or when on the way to top in every group only a single + * child is active but @nextevt is before the lowest + * next_expiry encountered while walking up to top level. + * * next_expiry - value of lowest expiry encountered while walking groups + * if only a single child is active on each and @nextevt + * is after this lowest expiry. + */ +u64 tmigr_quick_check(u64 nextevt) +{ + struct tmigr_cpu *tmc = this_cpu_ptr(&tmigr_cpu); + struct tmigr_group *group = tmc->tmgroup; + + if (tmigr_is_not_available(tmc)) + return nextevt; + + if (WARN_ON_ONCE(tmc->idle)) + return nextevt; + + if (!tmigr_check_migrator_and_lonely(tmc->tmgroup, tmc->groupmask)) + return KTIME_MAX; + + do { + if (!tmigr_check_lonely(group)) { + return KTIME_MAX; + } else { + /* + * Since current CPU is active, events may not be sorted + * from bottom to the top because the CPU's event is ignored + * up to the top and its sibling's events not propagated upwards. + * Thus keep track of the lowest observed expiry. + */ + nextevt = min_t(u64, nextevt, READ_ONCE(group->next_expiry)); + if (!group->parent) + return nextevt; + } + group = group->parent; + } while (group); + + return KTIME_MAX; +} + +/* + * tmigr_trigger_active() - trigger a CPU to become active again + * + * This function is executed on a CPU which is part of cpu_online_mask, when the + * last active CPU in the hierarchy is offlining. With this, it is ensured that + * the other CPU is active and takes over the migrator duty. + */ +static long tmigr_trigger_active(void *unused) +{ + struct tmigr_cpu *tmc = this_cpu_ptr(&tmigr_cpu); + + WARN_ON_ONCE(!tmc->online || tmc->idle); + + return 0; +} + +static int tmigr_cpu_offline(unsigned int cpu) +{ + struct tmigr_cpu *tmc = this_cpu_ptr(&tmigr_cpu); + int migrator; + u64 firstexp; + + raw_spin_lock_irq(&tmc->lock); + tmc->online = false; + WRITE_ONCE(tmc->wakeup, KTIME_MAX); + + /* + * CPU has to handle the local events on his own, when on the way to + * offline; Therefore nextevt value is set to KTIME_MAX + */ + firstexp = __tmigr_cpu_deactivate(tmc, KTIME_MAX); + trace_tmigr_cpu_offline(tmc); + raw_spin_unlock_irq(&tmc->lock); + + if (firstexp != KTIME_MAX) { + migrator = cpumask_any_but(cpu_online_mask, cpu); + work_on_cpu(migrator, tmigr_trigger_active, NULL); + } + + return 0; +} + +static int tmigr_cpu_online(unsigned int cpu) +{ + struct tmigr_cpu *tmc = this_cpu_ptr(&tmigr_cpu); + + /* Check whether CPU data was successfully initialized */ + if (WARN_ON_ONCE(!tmc->tmgroup)) + return -EINVAL; + + raw_spin_lock_irq(&tmc->lock); + trace_tmigr_cpu_online(tmc); + tmc->idle = timer_base_is_idle(); + if (!tmc->idle) + __tmigr_cpu_activate(tmc); + tmc->online = true; + raw_spin_unlock_irq(&tmc->lock); + return 0; +} + +static void tmigr_init_group(struct tmigr_group *group, unsigned int lvl, + int node) +{ + union tmigr_state s; + + raw_spin_lock_init(&group->lock); + + group->level = lvl; + group->numa_node = lvl < tmigr_crossnode_level ? node : NUMA_NO_NODE; + + group->num_children = 0; + + s.migrator = TMIGR_NONE; + s.active = 0; + s.seq = 0; + atomic_set(&group->migr_state, s.state); + + /* + * If this is a new top-level, prepare its groupmask in advance. + * This avoids accidents where yet another new top-level is + * created in the future and made visible before the current groupmask. + */ + if (list_empty(&tmigr_level_list[lvl])) { + group->groupmask = BIT(0); + /* + * The previous top level has prepared its groupmask already, + * simply account it as the first child. + */ + if (lvl > 0) + group->num_children = 1; + } + + timerqueue_init_head(&group->events); + timerqueue_init(&group->groupevt.nextevt); + group->groupevt.nextevt.expires = KTIME_MAX; + WRITE_ONCE(group->next_expiry, KTIME_MAX); + group->groupevt.ignore = true; +} + +static struct tmigr_group *tmigr_get_group(unsigned int cpu, int node, + unsigned int lvl) +{ + struct tmigr_group *tmp, *group = NULL; + + lockdep_assert_held(&tmigr_mutex); + + /* Try to attach to an existing group first */ + list_for_each_entry(tmp, &tmigr_level_list[lvl], list) { + /* + * If @lvl is below the cross NUMA node level, check whether + * this group belongs to the same NUMA node. + */ + if (lvl < tmigr_crossnode_level && tmp->numa_node != node) + continue; + + /* Capacity left? */ + if (tmp->num_children >= TMIGR_CHILDREN_PER_GROUP) + continue; + + /* + * TODO: A possible further improvement: Make sure that all CPU + * siblings end up in the same group of the lowest level of the + * hierarchy. Rely on the topology sibling mask would be a + * reasonable solution. + */ + + group = tmp; + break; + } + + if (group) + return group; + + /* Allocate and set up a new group */ + group = kzalloc_node(sizeof(*group), GFP_KERNEL, node); + if (!group) + return ERR_PTR(-ENOMEM); + + tmigr_init_group(group, lvl, node); + + /* Setup successful. Add it to the hierarchy */ + list_add(&group->list, &tmigr_level_list[lvl]); + trace_tmigr_group_set(group); + return group; +} + +static void tmigr_connect_child_parent(struct tmigr_group *child, + struct tmigr_group *parent, + bool activate) +{ + struct tmigr_walk data; + + raw_spin_lock_irq(&child->lock); + raw_spin_lock_nested(&parent->lock, SINGLE_DEPTH_NESTING); + + if (activate) { + /* + * @child is the old top and @parent the new one. In this + * case groupmask is pre-initialized and @child already + * accounted, along with its new sibling corresponding to the + * CPU going up. + */ + WARN_ON_ONCE(child->groupmask != BIT(0) || parent->num_children != 2); + } else { + /* Adding @child for the CPU going up to @parent. */ + child->groupmask = BIT(parent->num_children++); + } + + /* + * Make sure parent initialization is visible before publishing it to a + * racing CPU entering/exiting idle. This RELEASE barrier enforces an + * address dependency that pairs with the READ_ONCE() in __walk_groups(). + */ + smp_store_release(&child->parent, parent); + + raw_spin_unlock(&parent->lock); + raw_spin_unlock_irq(&child->lock); + + trace_tmigr_connect_child_parent(child); + + if (!activate) + return; + + /* + * To prevent inconsistent states, active children need to be active in + * the new parent as well. Inactive children are already marked inactive + * in the parent group: + * + * * When new groups were created by tmigr_setup_groups() starting from + * the lowest level (and not higher then one level below the current + * top level), then they are not active. They will be set active when + * the new online CPU comes active. + * + * * But if a new group above the current top level is required, it is + * mandatory to propagate the active state of the already existing + * child to the new parent. So tmigr_connect_child_parent() is + * executed with the formerly top level group (child) and the newly + * created group (parent). + * + * * It is ensured that the child is active, as this setup path is + * executed in hotplug prepare callback. This is exectued by an + * already connected and !idle CPU. Even if all other CPUs go idle, + * the CPU executing the setup will be responsible up to current top + * level group. And the next time it goes inactive, it will release + * the new childmask and parent to subsequent walkers through this + * @child. Therefore propagate active state unconditionally. + */ + data.childmask = child->groupmask; + + /* + * There is only one new level per time (which is protected by + * tmigr_mutex). When connecting the child and the parent and set the + * child active when the parent is inactive, the parent needs to be the + * uppermost level. Otherwise there went something wrong! + */ + WARN_ON(!tmigr_active_up(parent, child, &data) && parent->parent); +} + +static int tmigr_setup_groups(unsigned int cpu, unsigned int node) +{ + struct tmigr_group *group, *child, **stack; + int top = 0, err = 0, i = 0; + struct list_head *lvllist; + + stack = kcalloc(tmigr_hierarchy_levels, sizeof(*stack), GFP_KERNEL); + if (!stack) + return -ENOMEM; + + do { + group = tmigr_get_group(cpu, node, i); + if (IS_ERR(group)) { + err = PTR_ERR(group); + break; + } + + top = i; + stack[i++] = group; + + /* + * When booting only less CPUs of a system than CPUs are + * available, not all calculated hierarchy levels are required. + * + * The loop is aborted as soon as the highest level, which might + * be different from tmigr_hierarchy_levels, contains only a + * single group. + */ + if (group->parent || list_is_singular(&tmigr_level_list[i - 1])) + break; + + } while (i < tmigr_hierarchy_levels); + + /* Assert single root */ + WARN_ON_ONCE(!err && !group->parent && !list_is_singular(&tmigr_level_list[top])); + + while (i > 0) { + group = stack[--i]; + + if (err < 0) { + list_del(&group->list); + kfree(group); + continue; + } + + WARN_ON_ONCE(i != group->level); + + /* + * Update tmc -> group / child -> group connection + */ + if (i == 0) { + struct tmigr_cpu *tmc = per_cpu_ptr(&tmigr_cpu, cpu); + + raw_spin_lock_irq(&group->lock); + + tmc->tmgroup = group; + tmc->groupmask = BIT(group->num_children++); + + raw_spin_unlock_irq(&group->lock); + + trace_tmigr_connect_cpu_parent(tmc); + + /* There are no children that need to be connected */ + continue; + } else { + child = stack[i - 1]; + /* Will be activated at online time */ + tmigr_connect_child_parent(child, group, false); + } + + /* check if uppermost level was newly created */ + if (top != i) + continue; + + WARN_ON_ONCE(top == 0); + + lvllist = &tmigr_level_list[top]; + + /* + * Newly created root level should have accounted the upcoming + * CPU's child group and pre-accounted the old root. + */ + if (group->num_children == 2 && list_is_singular(lvllist)) { + /* + * The target CPU must never do the prepare work, except + * on early boot when the boot CPU is the target. Otherwise + * it may spuriously activate the old top level group inside + * the new one (nevertheless whether old top level group is + * active or not) and/or release an uninitialized childmask. + */ + WARN_ON_ONCE(cpu == raw_smp_processor_id()); + + lvllist = &tmigr_level_list[top - 1]; + list_for_each_entry(child, lvllist, list) { + if (child->parent) + continue; + + tmigr_connect_child_parent(child, group, true); + } + } + } + + kfree(stack); + + return err; +} + +static int tmigr_add_cpu(unsigned int cpu) +{ + int node = cpu_to_node(cpu); + int ret; + + mutex_lock(&tmigr_mutex); + ret = tmigr_setup_groups(cpu, node); + mutex_unlock(&tmigr_mutex); + + return ret; +} + +static int tmigr_cpu_prepare(unsigned int cpu) +{ + struct tmigr_cpu *tmc = per_cpu_ptr(&tmigr_cpu, cpu); + int ret = 0; + + /* Not first online attempt? */ + if (tmc->tmgroup) + return ret; + + raw_spin_lock_init(&tmc->lock); + timerqueue_init(&tmc->cpuevt.nextevt); + tmc->cpuevt.nextevt.expires = KTIME_MAX; + tmc->cpuevt.ignore = true; + tmc->cpuevt.cpu = cpu; + tmc->remote = false; + WRITE_ONCE(tmc->wakeup, KTIME_MAX); + + ret = tmigr_add_cpu(cpu); + if (ret < 0) + return ret; + + if (tmc->groupmask == 0) + return -EINVAL; + + return ret; +} + +static int __init tmigr_init(void) +{ + unsigned int cpulvl, nodelvl, cpus_per_node, i; + unsigned int nnodes = num_possible_nodes(); + unsigned int ncpus = num_possible_cpus(); + int ret = -ENOMEM; + + BUILD_BUG_ON_NOT_POWER_OF_2(TMIGR_CHILDREN_PER_GROUP); + + /* Nothing to do if running on UP */ + if (ncpus == 1) + return 0; + + /* + * Calculate the required hierarchy levels. Unfortunately there is no + * reliable information available, unless all possible CPUs have been + * brought up and all NUMA nodes are populated. + * + * Estimate the number of levels with the number of possible nodes and + * the number of possible CPUs. Assume CPUs are spread evenly across + * nodes. We cannot rely on cpumask_of_node() because it only works for + * online CPUs. + */ + cpus_per_node = DIV_ROUND_UP(ncpus, nnodes); + + /* Calc the hierarchy levels required to hold the CPUs of a node */ + cpulvl = DIV_ROUND_UP(order_base_2(cpus_per_node), + ilog2(TMIGR_CHILDREN_PER_GROUP)); + + /* Calculate the extra levels to connect all nodes */ + nodelvl = DIV_ROUND_UP(order_base_2(nnodes), + ilog2(TMIGR_CHILDREN_PER_GROUP)); + + tmigr_hierarchy_levels = cpulvl + nodelvl; + + /* + * If a NUMA node spawns more than one CPU level group then the next + * level(s) of the hierarchy contains groups which handle all CPU groups + * of the same NUMA node. The level above goes across NUMA nodes. Store + * this information for the setup code to decide in which level node + * matching is no longer required. + */ + tmigr_crossnode_level = cpulvl; + + tmigr_level_list = kcalloc(tmigr_hierarchy_levels, sizeof(struct list_head), GFP_KERNEL); + if (!tmigr_level_list) + goto err; + + for (i = 0; i < tmigr_hierarchy_levels; i++) + INIT_LIST_HEAD(&tmigr_level_list[i]); + + pr_info("Timer migration: %d hierarchy levels; %d children per group;" + " %d crossnode level\n", + tmigr_hierarchy_levels, TMIGR_CHILDREN_PER_GROUP, + tmigr_crossnode_level); + + ret = cpuhp_setup_state(CPUHP_TMIGR_PREPARE, "tmigr:prepare", + tmigr_cpu_prepare, NULL); + if (ret) + goto err; + + ret = cpuhp_setup_state(CPUHP_AP_TMIGR_ONLINE, "tmigr:online", + tmigr_cpu_online, tmigr_cpu_offline); + if (ret) + goto err; + + return 0; + +err: + pr_err("Timer migration setup failed\n"); + return ret; +} +early_initcall(tmigr_init); diff --git a/kernel/time/timer_migration.h b/kernel/time/timer_migration.h new file mode 100644 index 000000000000..ae19f70f8170 --- /dev/null +++ b/kernel/time/timer_migration.h @@ -0,0 +1,146 @@ +/* SPDX-License-Identifier: GPL-2.0-only */ +#ifndef _KERNEL_TIME_MIGRATION_H +#define _KERNEL_TIME_MIGRATION_H + +/* Per group capacity. Must be a power of 2! */ +#define TMIGR_CHILDREN_PER_GROUP 8 + +/** + * struct tmigr_event - a timer event associated to a CPU + * @nextevt: The node to enqueue an event in the parent group queue + * @cpu: The CPU to which this event belongs + * @ignore: Hint whether the event could be ignored; it is set when + * CPU or group is active; + */ +struct tmigr_event { + struct timerqueue_node nextevt; + unsigned int cpu; + bool ignore; +}; + +/** + * struct tmigr_group - timer migration hierarchy group + * @lock: Lock protecting the event information and group hierarchy + * information during setup + * @parent: Pointer to the parent group. Pointer is updated when a + * new hierarchy level is added because of a CPU coming + * online the first time. Once it is set, the pointer will + * not be removed or updated. When accessing parent pointer + * lock less to decide whether to abort a propagation or + * not, it is not a problem. The worst outcome is an + * unnecessary/early CPU wake up. But do not access parent + * pointer several times in the same 'action' (like + * activation, deactivation, check for remote expiry,...) + * without holding the lock as it is not ensured that value + * will not change. + * @groupevt: Next event of the group which is only used when the + * group is !active. The group event is then queued into + * the parent timer queue. + * Ignore bit of @groupevt is set when the group is active. + * @next_expiry: Base monotonic expiry time of the next event of the + * group; It is used for the racy lockless check whether a + * remote expiry is required; it is always reliable + * @events: Timer queue for child events queued in the group + * @migr_state: State of the group (see union tmigr_state) + * @level: Hierarchy level of the group; Required during setup + * @numa_node: Required for setup only to make sure CPU and low level + * group information is NUMA local. It is set to NUMA node + * as long as the group level is per NUMA node (level < + * tmigr_crossnode_level); otherwise it is set to + * NUMA_NO_NODE + * @num_children: Counter of group children to make sure the group is only + * filled with TMIGR_CHILDREN_PER_GROUP; Required for setup + * only + * @groupmask: mask of the group in the parent group; is set during + * setup and will never change; can be read lockless + * @list: List head that is added to the per level + * tmigr_level_list; is required during setup when a + * new group needs to be connected to the existing + * hierarchy groups + */ +struct tmigr_group { + raw_spinlock_t lock; + struct tmigr_group *parent; + struct tmigr_event groupevt; + u64 next_expiry; + struct timerqueue_head events; + atomic_t migr_state; + unsigned int level; + int numa_node; + unsigned int num_children; + u8 groupmask; + struct list_head list; +}; + +/** + * struct tmigr_cpu - timer migration per CPU group + * @lock: Lock protecting the tmigr_cpu group information + * @online: Indicates whether the CPU is online; In deactivate path + * it is required to know whether the migrator in the top + * level group is to be set offline, while a timer is + * pending. Then another online CPU needs to be notified to + * take over the migrator role. Furthermore the information + * is required in CPU hotplug path as the CPU is able to go + * idle before the timer migration hierarchy hotplug AP is + * reached. During this phase, the CPU has to handle the + * global timers on its own and must not act as a migrator. + * @idle: Indicates whether the CPU is idle in the timer migration + * hierarchy + * @remote: Is set when timers of the CPU are expired remotely + * @tmgroup: Pointer to the parent group + * @groupmask: mask of tmigr_cpu in the parent group + * @wakeup: Stores the first timer when the timer migration + * hierarchy is completely idle and remote expiry was done; + * is returned to timer code in the idle path and is only + * used in idle path. + * @cpuevt: CPU event which could be enqueued into the parent group + */ +struct tmigr_cpu { + raw_spinlock_t lock; + bool online; + bool idle; + bool remote; + struct tmigr_group *tmgroup; + u8 groupmask; + u64 wakeup; + struct tmigr_event cpuevt; +}; + +/** + * union tmigr_state - state of tmigr_group + * @state: Combined version of the state - only used for atomic + * read/cmpxchg function + * &anon struct: Split version of the state - only use the struct members to + * update information to stay independent of endianness + * @active: Contains each mask bit of the active children + * @migrator: Contains mask of the child which is migrator + * @seq: Sequence counter needs to be increased when an update + * to the tmigr_state is done. It prevents a race when + * updates in the child groups are propagated in changed + * order. Detailed information about the scenario is + * given in the documentation at the begin of + * timer_migration.c. + */ +union tmigr_state { + u32 state; + struct { + u8 active; + u8 migrator; + u16 seq; + } __packed; +}; + +#if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON) +extern void tmigr_handle_remote(void); +extern bool tmigr_requires_handle_remote(void); +extern void tmigr_cpu_activate(void); +extern u64 tmigr_cpu_deactivate(u64 nextevt); +extern u64 tmigr_cpu_new_timer(u64 nextevt); +extern u64 tmigr_quick_check(u64 nextevt); +#else +static inline void tmigr_handle_remote(void) { } +static inline bool tmigr_requires_handle_remote(void) { return false; } +static inline void tmigr_cpu_activate(void) { } +#endif + +#endif diff --git a/kernel/time/vsyscall.c b/kernel/time/vsyscall.c index 54ce6eb2ca36..05d383143165 100644 --- a/kernel/time/vsyscall.c +++ b/kernel/time/vsyscall.c @@ -13,6 +13,8 @@ #include <vdso/helpers.h> #include <vdso/vsyscall.h> +#include "timekeeping_internal.h" + static inline void update_vdso_data(struct vdso_data *vdata, struct timekeeper *tk) { @@ -20,10 +22,16 @@ static inline void update_vdso_data(struct vdso_data *vdata, u64 nsec, sec; vdata[CS_HRES_COARSE].cycle_last = tk->tkr_mono.cycle_last; +#ifdef CONFIG_GENERIC_VDSO_OVERFLOW_PROTECT + vdata[CS_HRES_COARSE].max_cycles = tk->tkr_mono.clock->max_cycles; +#endif vdata[CS_HRES_COARSE].mask = tk->tkr_mono.mask; vdata[CS_HRES_COARSE].mult = tk->tkr_mono.mult; vdata[CS_HRES_COARSE].shift = tk->tkr_mono.shift; vdata[CS_RAW].cycle_last = tk->tkr_raw.cycle_last; +#ifdef CONFIG_GENERIC_VDSO_OVERFLOW_PROTECT + vdata[CS_RAW].max_cycles = tk->tkr_raw.clock->max_cycles; +#endif vdata[CS_RAW].mask = tk->tkr_raw.mask; vdata[CS_RAW].mult = tk->tkr_raw.mult; vdata[CS_RAW].shift = tk->tkr_raw.shift; @@ -106,12 +114,12 @@ void update_vsyscall(struct timekeeper *tk) /* * If the current clocksource is not VDSO capable, then spare the - * update of the high reolution parts. + * update of the high resolution parts. */ if (clock_mode != VDSO_CLOCKMODE_NONE) update_vdso_data(vdata, tk); - __arch_update_vsyscall(vdata, tk); + __arch_update_vsyscall(vdata); vdso_write_end(vdata); @@ -127,3 +135,41 @@ void update_vsyscall_tz(void) __arch_sync_vdso_data(vdata); } + +/** + * vdso_update_begin - Start of a VDSO update section + * + * Allows architecture code to safely update the architecture specific VDSO + * data. Disables interrupts, acquires timekeeper lock to serialize against + * concurrent updates from timekeeping and invalidates the VDSO data + * sequence counter to prevent concurrent readers from accessing + * inconsistent data. + * + * Returns: Saved interrupt flags which need to be handed in to + * vdso_update_end(). + */ +unsigned long vdso_update_begin(void) +{ + struct vdso_data *vdata = __arch_get_k_vdso_data(); + unsigned long flags = timekeeper_lock_irqsave(); + + vdso_write_begin(vdata); + return flags; +} + +/** + * vdso_update_end - End of a VDSO update section + * @flags: Interrupt flags as returned from vdso_update_begin() + * + * Pairs with vdso_update_begin(). Marks vdso data consistent, invokes data + * synchronization if the architecture requires it, drops timekeeper lock + * and restores interrupt flags. + */ +void vdso_update_end(unsigned long flags) +{ + struct vdso_data *vdata = __arch_get_k_vdso_data(); + + vdso_write_end(vdata); + __arch_sync_vdso_data(vdata); + timekeeper_unlock_irqrestore(flags); +} |