diff options
Diffstat (limited to 'kernel/time')
35 files changed, 2847 insertions, 1264 deletions
diff --git a/kernel/time/Kconfig b/kernel/time/Kconfig index fcc42353f125..a41753be1a2b 100644 --- a/kernel/time/Kconfig +++ b/kernel/time/Kconfig @@ -26,13 +26,9 @@ config CLOCKSOURCE_VALIDATE_LAST_CYCLE 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 @@ -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 +190,20 @@ 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 μs)" + depends on CLOCKSOURCE_WATCHDOG + range 50 1000 + default 100 + help + Specify the maximum amount of allowable watchdog skew in + microseconds before reporting the clocksource to be unstable. + endmenu endif diff --git a/kernel/time/Makefile b/kernel/time/Makefile index c8f00168afe8..7e875e63ff3b 100644 --- a/kernel/time/Makefile +++ b/kernel/time/Makefile @@ -16,7 +16,10 @@ 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 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..5897828b9d7e 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> */ @@ -92,7 +92,7 @@ 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; @@ -192,7 +192,7 @@ 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) { @@ -527,8 +527,11 @@ 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) @@ -715,8 +718,11 @@ 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 + * + * Return: ALARMTIMER_NORESTART */ static enum alarmtimer_restart alarmtimer_nsleep_wakeup(struct alarm *alarm, ktime_t now) @@ -733,6 +739,7 @@ static enum alarmtimer_restart alarmtimer_nsleep_wakeup(struct alarm *alarm, * 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. */ @@ -804,9 +811,8 @@ 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 */ @@ -848,9 +854,9 @@ 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 +914,7 @@ 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..5d85014d59b5 100644 --- a/kernel/time/clockevents.c +++ b/kernel/time/clockevents.c @@ -347,8 +347,7 @@ static void clockevents_notify_released(void) 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 +575,7 @@ 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) { @@ -629,6 +627,7 @@ void tick_offline_cpu(unsigned int cpu) /** * tick_cleanup_dead_cpu - Cleanup the tick and clockevents of a dead cpu + * @cpu: The dead CPU */ void tick_cleanup_dead_cpu(int cpu) { @@ -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; @@ -682,16 +681,16 @@ static ssize_t sysfs_show_current_tick_dev(struct device *dev, 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..df922f49d171 --- /dev/null +++ b/kernel/time/clocksource-wdtest.c @@ -0,0 +1,201 @@ +// 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_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; + char *s; + int i; + + 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_ON_ONCE(time_before(j2, j1 + NSEC_PER_USEC)); + + /* Verify tsc-like stability with various numbers of errors injected. */ + for (i = 0; i <= max_cswd_read_retries + 1; i++) { + if (i <= 1 && i < max_cswd_read_retries) + s = ""; + else if (i <= max_cswd_read_retries) + s = ", expect message"; + else + s = ", expect clock skew"; + pr_info("--- Watchdog with %dx error injection, %lu retries%s.\n", i, max_cswd_read_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_cswd_read_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 428beb69426a..8058bec87ace 100644 --- a/kernel/time/clocksource.c +++ b/kernel/time/clocksource.c @@ -14,6 +14,8 @@ #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" @@ -38,7 +40,7 @@ * 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 +95,26 @@ static char override_name[CS_NAME_LEN]; static int finished_booting; static u64 suspend_start; +/* + * Threshold: 0.0312s, when doubled: 0.0625s. + * Also a default for cs->uncertainty_margin when registering clocks. + */ +#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. + */ +#ifdef CONFIG_CLOCKSOURCE_WATCHDOG_MAX_SKEW_US +#define MAX_SKEW_USEC CONFIG_CLOCKSOURCE_WATCHDOG_MAX_SKEW_US +#else +#define MAX_SKEW_USEC 100 +#endif + +#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); @@ -119,10 +141,9 @@ static int clocksource_watchdog_kthread(void *data); static void __clocksource_change_rating(struct clocksource *cs, int rating); /* - * Interval: 0.5sec Threshold: 0.0625s + * Interval: 0.5sec. */ #define WATCHDOG_INTERVAL (HZ >> 1) -#define WATCHDOG_THRESHOLD (NSEC_PER_SEC >> 4) static void clocksource_watchdog_work(struct work_struct *work) { @@ -184,12 +205,193 @@ void clocksource_mark_unstable(struct clocksource *cs) spin_unlock_irqrestore(&watchdog_lock, flags); } +ulong max_cswd_read_retries = 2; +module_param(max_cswd_read_retries, ulong, 0644); +EXPORT_SYMBOL_GPL(max_cswd_read_retries); +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) +{ + unsigned int nretries; + u64 wd_end, wd_end2, wd_delta; + int64_t wd_delay, wd_seq_delay; + + for (nretries = 0; nretries <= max_cswd_read_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_delta = clocksource_delta(wd_end, *wdnow, watchdog->mask); + wd_delay = clocksource_cyc2ns(wd_delta, watchdog->mult, + watchdog->shift); + if (wd_delay <= WATCHDOG_MAX_SKEW) { + if (nretries > 1 || nretries >= max_cswd_read_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 > WATCHDOG_MAX_SKEW/2, + * report system busy, reinit the watchdog and skip the current + * watchdog test. + */ + wd_delta = clocksource_delta(wd_end2, wd_end, watchdog->mask); + wd_seq_delay = clocksource_cyc2ns(wd_delta, watchdog->mult, watchdog->shift); + if (wd_seq_delay > WATCHDOG_MAX_SKEW/2) + goto skip_test; + } + + pr_warn("timekeeping watchdog on CPU%d: %s read-back delay of %lldns, attempt %d, marking unstable\n", + smp_processor_id(), watchdog->name, wd_delay, nretries); + 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 = prandom_u32_max(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(); + preempt_disable(); + clocksource_verify_choose_cpus(); + if (cpumask_empty(&cpus_chosen)) { + preempt_enable(); + cpus_read_unlock(); + pr_warn("Not enough CPUs to check clocksource '%s'.\n", cs->name); + return; + } + testcpu = smp_processor_id(); + pr_warn("Checking clocksource %s synchronization from CPU %d to CPUs %*pbl.\n", cs->name, testcpu, cpumask_pr_args(&cpus_chosen)); + 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); + delta = clocksource_delta(csnow_end, csnow_begin, cs->mask); + cs_nsec = clocksource_cyc2ns(delta, cs->mult, cs->shift); + if (cs_nsec > cs_nsec_max) + cs_nsec_max = cs_nsec; + if (cs_nsec < cs_nsec_min) + cs_nsec_min = cs_nsec; + } + preempt_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 void clocksource_watchdog(struct timer_list *unused) { - struct clocksource *cs; u64 csnow, wdnow, cslast, wdlast, delta; - int64_t wd_nsec, cs_nsec; int next_cpu, reset_pending; + int64_t wd_nsec, cs_nsec; + struct clocksource *cs; + enum wd_read_status read_ret; + u32 md; spin_lock(&watchdog_lock); if (!watchdog_running) @@ -206,10 +408,14 @@ 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_SUCCESS) { + if (read_ret == WD_READ_UNSTABLE) + /* Clock readout unreliable, so give it up. */ + __clocksource_unstable(cs); + continue; + } /* Clocksource initialized ? */ if (!(cs->flags & CLOCK_SOURCE_WATCHDOG) || @@ -235,13 +441,20 @@ static void clocksource_watchdog(struct timer_list *unused) 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) { 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); + 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; } @@ -407,6 +620,12 @@ 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) { @@ -518,7 +737,7 @@ 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. */ @@ -705,8 +924,6 @@ static inline void clocksource_update_max_deferment(struct clocksource *cs) &cs->max_cycles); } -#ifndef CONFIG_ARCH_USES_GETTIMEOFFSET - static struct clocksource *clocksource_find_best(bool oneshot, bool skipcur) { struct clocksource *cs; @@ -798,12 +1015,6 @@ 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 +1095,26 @@ 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. 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. + */ + 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 +1159,15 @@ 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", + cs->name, cs->vdso_clock_mode); + cs->vdso_clock_mode = VDSO_CLOCKMODE_NONE; + } + /* Initialize mult/shift and max_idle_ns */ __clocksource_update_freq_scale(cs, scale, freq); diff --git a/kernel/time/hrtimer.c b/kernel/time/hrtimer.c index 3a609e7344f3..3ae661ab6260 100644 --- a/kernel/time/hrtimer.c +++ b/kernel/time/hrtimer.c @@ -135,7 +135,11 @@ 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] @@ -311,7 +315,7 @@ s64 __ktime_divns(const ktime_t kt, s64 div) div >>= 1; } tmp >>= sft; - do_div(tmp, (unsigned long) div); + do_div(tmp, (u32) div); return dclc < 0 ? -tmp : tmp; } EXPORT_SYMBOL_GPL(__ktime_divns); @@ -338,7 +342,7 @@ 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 +377,7 @@ 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 +401,7 @@ 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, @@ -421,11 +425,6 @@ 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); @@ -547,8 +546,11 @@ 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 +591,37 @@ __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; @@ -619,37 +652,10 @@ 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 +672,31 @@ 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) 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 */ @@ -703,23 +727,7 @@ static inline int hrtimer_is_hres_enabled(void) return hrtimer_hres_enabled; } -/* - * 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); -} +static void retrigger_next_event(void *arg); /* * Switch to high resolution mode @@ -741,29 +749,54 @@ static void hrtimer_switch_to_hres(void) 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 +851,161 @@ 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); + + /* + * If the sequence did not change over the update then the + * remote CPU already handled it. + */ + 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; /* - * Program the timer hardware. We enforce the expiry for - * events which are already in the past. + * 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. */ - tick_program_event(expires, 1); + 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); + /* - * 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 from timekeeping code to reprogram the hrtimer interrupt device + * on all cpus and to notify timerfd. */ -void hrtimers_resume(void) +void clock_was_set_delayed(void) +{ + schedule_work(&hrtimer_work); +} + +/* + * 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_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(); } /* @@ -1002,7 +1121,7 @@ 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 +1132,13 @@ 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 +1151,16 @@ 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; @@ -1086,9 +1214,31 @@ static int __hrtimer_start_range_ns(struct hrtimer *timer, ktime_t tim, struct hrtimer_clock_base *base) { 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_ptr(&hrtimer_bases); + force_local &= base->cpu_base->next_timer == timer; + + /* + * 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 +1248,24 @@ 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; + } + + first = enqueue_hrtimer(timer, new_base, mode); + if (!force_local) + return first; - return enqueue_hrtimer(timer, new_base, mode); + /* + * 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; } /** @@ -1166,7 +1331,7 @@ 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); @@ -1195,7 +1360,7 @@ 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, @@ -1285,7 +1450,7 @@ 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 */ @@ -1381,7 +1546,7 @@ 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(). @@ -1404,7 +1569,7 @@ static void __hrtimer_init(struct hrtimer *timer, clockid_t clock_id, base = softtimer ? HRTIMER_MAX_CLOCK_BASES / 2 : 0; base += hrtimer_clockid_to_base(clock_id); timer->is_soft = softtimer; - timer->is_hard = !softtimer; + timer->is_hard = !!(mode & HRTIMER_MODE_HARD); timer->base = &cpu_base->clock_base[base]; timerqueue_init(&timer->node); } @@ -1413,7 +1578,7 @@ static void __hrtimer_init(struct hrtimer *timer, clockid_t clock_id, * 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 * @@ -1470,7 +1635,7 @@ 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. */ @@ -1480,6 +1645,7 @@ static void __run_hrtimer(struct hrtimer_cpu_base *cpu_base, unsigned long flags) __must_hold(&cpu_base->lock) { enum hrtimer_restart (*fn)(struct hrtimer *); + bool expires_in_hardirq; int restart; lockdep_assert_held(&cpu_base->lock); @@ -1514,7 +1680,11 @@ static void __run_hrtimer(struct hrtimer_cpu_base *cpu_base, */ raw_spin_unlock_irqrestore(&cpu_base->lock, flags); trace_hrtimer_expire_entry(timer, now); + expires_in_hardirq = lockdep_hrtimer_enter(timer); + restart = fn(timer); + + lockdep_hrtimer_exit(expires_in_hardirq); trace_hrtimer_expire_exit(timer); raw_spin_lock_irq(&cpu_base->lock); @@ -1566,7 +1736,7 @@ 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 @@ -1640,8 +1810,8 @@ retry: __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. @@ -1800,7 +1970,7 @@ 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 @@ -1813,7 +1983,7 @@ 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. @@ -1867,11 +2037,11 @@ 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; @@ -1935,9 +2105,9 @@ 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; @@ -1993,8 +2163,11 @@ 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; @@ -2138,6 +2311,7 @@ schedule_hrtimeout_range_clock(ktime_t *expires, u64 delta, return !t.task ? 0 : -EINTR; } +EXPORT_SYMBOL_GPL(schedule_hrtimeout_range_clock); /** * schedule_hrtimeout_range - sleep until timeout diff --git a/kernel/time/itimer.c b/kernel/time/itimer.c index ca4e6d57d68b..00629e658ca1 100644 --- a/kernel/time/itimer.c +++ b/kernel/time/itimer.c @@ -172,10 +172,6 @@ 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 d23b434c2ca7..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; @@ -44,21 +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_SEQLOCK(jiffies_lock); +__cacheline_aligned_in_smp DEFINE_RAW_SPINLOCK(jiffies_lock); +__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) @@ -67,9 +51,9 @@ u64 get_jiffies_64(void) u64 ret; do { - seq = read_seqbegin(&jiffies_lock); + seq = read_seqcount_begin(&jiffies_seq); ret = jiffies_64; - } while (read_seqretry(&jiffies_lock, seq)); + } while (read_seqcount_retry(&jiffies_seq, seq)); return ret; } EXPORT_SYMBOL(get_jiffies_64); diff --git a/kernel/time/namespace.c b/kernel/time/namespace.c index 12858507d75a..aec832801c26 100644 --- a/kernel/time/namespace.c +++ b/kernel/time/namespace.c @@ -8,6 +8,7 @@ #include <linux/user_namespace.h> #include <linux/sched/signal.h> #include <linux/sched/task.h> +#include <linux/clocksource.h> #include <linux/seq_file.h> #include <linux/proc_ns.h> #include <linux/export.h> @@ -87,13 +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; @@ -172,8 +173,8 @@ static struct timens_offset offset_from_ts(struct timespec64 off) * for vdso_data->clock_mode is a non-issue. The task is spin waiting for the * update to finish and for 'seq' to become even anyway. * - * Timens page has vdso_data->clock_mode set to VCLOCK_TIMENS which enforces - * the time namespace handling path. + * Timens page has vdso_data->clock_mode set to VDSO_CLOCKMODE_TIMENS which + * enforces the time namespace handling path. */ static void timens_setup_vdso_data(struct vdso_data *vdata, struct time_namespace *ns) @@ -183,7 +184,7 @@ static void timens_setup_vdso_data(struct vdso_data *vdata, struct timens_offset boottime = offset_from_ts(ns->offsets.boottime); vdata->seq = 1; - vdata->clock_mode = VCLOCK_TIMENS; + vdata->clock_mode = VDSO_CLOCKMODE_TIMENS; offset[CLOCK_MONOTONIC] = monotonic; offset[CLOCK_MONOTONIC_RAW] = monotonic; offset[CLOCK_MONOTONIC_COARSE] = monotonic; @@ -225,11 +226,8 @@ 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); @@ -279,24 +277,24 @@ static void timens_put(struct ns_common *ns) put_time_ns(to_time_ns(ns)); } -static int timens_install(struct nsproxy *nsproxy, struct ns_common *new) +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; if (!ns_capable(ns->user_ns, CAP_SYS_ADMIN) || - !ns_capable(current_user_ns(), CAP_SYS_ADMIN)) + !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; @@ -307,27 +305,20 @@ static int timens_install(struct nsproxy *nsproxy, 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) @@ -337,7 +328,20 @@ static struct user_namespace *timens_owner(struct ns_common *ns) static void show_offset(struct seq_file *m, int clockid, struct timespec64 *ts) { - seq_printf(m, "%d %lld %ld\n", clockid, ts->tv_sec, ts->tv_nsec); + char *clock; + + switch (clockid) { + case CLOCK_BOOTTIME: + clock = "boottime"; + break; + case CLOCK_MONOTONIC: + clock = "monotonic"; + break; + default: + clock = "unknown"; + break; + } + seq_printf(m, "%-10s %10lld %9ld\n", clock, ts->tv_sec, ts->tv_nsec); } void proc_timens_show_offsets(struct task_struct *p, struct seq_file *m) @@ -446,6 +450,7 @@ const struct proc_ns_operations timens_operations = { const struct proc_ns_operations timens_for_children_operations = { .name = "time_for_children", + .real_ns_name = "time", .type = CLONE_NEWTIME, .get = timens_for_children_get, .put = timens_put, @@ -454,15 +459,9 @@ 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..406dccb79c2b 100644 --- a/kernel/time/ntp.c +++ b/kernel/time/ntp.c @@ -494,65 +494,74 @@ 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; - } - - /* 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_work(system_freezable_power_efficient_wq, &sync_work); - 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); + + if (to_set->tv_nsec < TIME_SET_NSEC_FUZZ) { + to_set->tv_nsec = 0; + return true; + } - sched_sync_hw_clock(now, target_nsec, rc); + 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 @@ -560,48 +569,47 @@ 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; } - - sched_sync_hw_clock(now, target_nsec, rc); - return true; +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 /* * If we have an externally synchronized Linux clock, then update RTC clock @@ -613,24 +621,64 @@ 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) { - if (!ntp_synced()) - return; + /* + * 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); +} - if (IS_ENABLED(CONFIG_GENERIC_CMOS_UPDATE) || - IS_ENABLED(CONFIG_RTC_SYSTOHC)) - queue_delayed_work(system_power_efficient_wq, &sync_work, 0); +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: @@ -1044,4 +1092,5 @@ __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..23d1b74c3065 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(void); +#else +static inline void ntp_notify_cmos_timer(void) { } +#endif + #endif /* _LINUX_NTP_INTERNAL_H */ diff --git a/kernel/time/posix-cpu-timers.c b/kernel/time/posix-cpu-timers.c index 8ff6da77a01f..cb925e8ef9a8 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" @@ -34,88 +35,84 @@ 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; } /* * Functions for validating access to tasks. */ -static struct task_struct *lookup_task(const pid_t pid, bool thread, - bool gettime) +static struct pid *pid_for_clock(const clockid_t clock, bool gettime) { - struct task_struct *p; + const bool thread = !!CPUCLOCK_PERTHREAD(clock); + const pid_t upid = CPUCLOCK_PID(clock); + struct pid *pid; + + if (CPUCLOCK_WHICH(clock) >= CPUCLOCK_MAX) + return NULL; /* * If the encoded PID is 0, then the timer is targeted at current * or the process to which current belongs. */ - if (!pid) - return thread ? current : current->group_leader; - - p = find_task_by_vpid(pid); - if (!p) - return p; + if (upid == 0) + return thread ? task_pid(current) : task_tgid(current); - if (thread) - return same_thread_group(p, current) ? p : NULL; + pid = find_vpid(upid); + if (!pid) + return NULL; - if (gettime) { - /* - * For clock_gettime(PROCESS) the task does not need to be - * the actual group leader. tsk->sighand gives - * access to the group's clock. - * - * Timers need the group leader because they take a - * reference on it and store the task pointer until the - * timer is destroyed. - */ - return (p == current || thread_group_leader(p)) ? p : NULL; + if (thread) { + struct task_struct *tsk = pid_task(pid, PIDTYPE_PID); + return (tsk && same_thread_group(tsk, current)) ? pid : NULL; } /* - * For processes require that p is group leader. + * For clock_gettime(PROCESS) allow finding the process by + * with the pid of the current task. The code needs the tgid + * of the process so that pid_task(pid, PIDTYPE_TGID) can be + * used to find the process. */ - return has_group_leader_pid(p) ? p : NULL; + if (gettime && (pid == task_pid(current))) + return task_tgid(current); + + /* + * For processes require that pid identifies a process. + */ + return pid_has_task(pid, PIDTYPE_TGID) ? pid : NULL; } -static struct task_struct *__get_task_for_clock(const clockid_t clock, - bool getref, bool gettime) +static inline int validate_clock_permissions(const clockid_t clock) { - const bool thread = !!CPUCLOCK_PERTHREAD(clock); - const pid_t pid = CPUCLOCK_PID(clock); - struct task_struct *p; - - if (CPUCLOCK_WHICH(clock) >= CPUCLOCK_MAX) - return NULL; + int ret; rcu_read_lock(); - p = lookup_task(pid, thread, gettime); - if (p && getref) - get_task_struct(p); + ret = pid_for_clock(clock, false) ? 0 : -EINVAL; rcu_read_unlock(); - return p; -} -static inline struct task_struct *get_task_for_clock(const clockid_t clock) -{ - return __get_task_for_clock(clock, true, false); + return ret; } -static inline struct task_struct *get_task_for_clock_get(const clockid_t clock) +static inline enum pid_type clock_pid_type(const clockid_t clock) { - return __get_task_for_clock(clock, true, true); + return CPUCLOCK_PERTHREAD(clock) ? PIDTYPE_PID : PIDTYPE_TGID; } -static inline int validate_clock_permissions(const clockid_t clock) +static inline struct task_struct *cpu_timer_task_rcu(struct k_itimer *timer) { - return __get_task_for_clock(clock, false, false) ? 0 : -EINVAL; + return pid_task(timer->it.cpu.pid, clock_pid_type(timer->it_clock)); } /* @@ -289,7 +286,7 @@ 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. @@ -301,6 +298,8 @@ 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; @@ -336,9 +335,7 @@ static void __thread_group_cputime(struct task_struct *tsk, u64 *samples) /* * Sample a process (thread group) clock for the given task clkid. If the * group's cputime accounting is already enabled, read the atomic - * store. Otherwise a full update is required. Task's sighand lock must be - * held to protect the task traversal on a full update. clkid is already - * validated. + * store. Otherwise a full update is required. clkid is already validated. */ static u64 cpu_clock_sample_group(const clockid_t clkid, struct task_struct *p, bool start) @@ -365,15 +362,18 @@ static int posix_cpu_clock_get(const clockid_t clock, struct timespec64 *tp) struct task_struct *tsk; u64 t; - tsk = get_task_for_clock_get(clock); - if (!tsk) + rcu_read_lock(); + tsk = pid_task(pid_for_clock(clock, true), clock_pid_type(clock)); + if (!tsk) { + rcu_read_unlock(); return -EINVAL; + } if (CPUCLOCK_PERTHREAD(clock)) t = cpu_clock_sample(clkid, tsk); else t = cpu_clock_sample_group(clkid, tsk, false); - put_task_struct(tsk); + rcu_read_unlock(); *tp = ns_to_timespec64(t); return 0; @@ -386,17 +386,83 @@ static int posix_cpu_clock_get(const clockid_t clock, struct timespec64 *tp) */ static int posix_cpu_timer_create(struct k_itimer *new_timer) { - struct task_struct *p = get_task_for_clock(new_timer->it_clock); + static struct lock_class_key posix_cpu_timers_key; + struct pid *pid; - if (!p) + rcu_read_lock(); + pid = pid_for_clock(new_timer->it_clock, false); + if (!pid) { + rcu_read_unlock(); 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.task = p; + new_timer->it.cpu.pid = get_pid(pid); + rcu_read_unlock(); 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. @@ -406,13 +472,15 @@ static int posix_cpu_timer_create(struct k_itimer *new_timer) static int posix_cpu_timer_del(struct k_itimer *timer) { struct cpu_timer *ctmr = &timer->it.cpu; - struct task_struct *p = ctmr->task; struct sighand_struct *sighand; + struct task_struct *p; unsigned long flags; int ret = 0; - if (WARN_ON_ONCE(!p)) - return -EINVAL; + rcu_read_lock(); + p = cpu_timer_task_rcu(timer); + if (!p) + goto out; /* * Protect against sighand release/switch in exit/exec and process/ @@ -429,13 +497,15 @@ static int posix_cpu_timer_del(struct k_itimer *timer) if (timer->it.cpu.firing) ret = TIMER_RETRY; else - cpu_timer_dequeue(ctmr); + disarm_timer(timer, p); unlock_task_sighand(p, &flags); } +out: + rcu_read_unlock(); if (!ret) - put_task_struct(p); + put_pid(ctmr->pid); return ret; } @@ -484,18 +554,11 @@ void posix_cpu_timers_exit_group(struct task_struct *tsk) * Insert the timer on the appropriate list before any timers that * expire later. This must be called with the sighand lock held. */ -static void arm_timer(struct k_itimer *timer) +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 task_struct *p = ctmr->task; - 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; if (!cpu_timer_enqueue(&base->tqhead, ctmr)) return; @@ -512,7 +575,7 @@ static void arm_timer(struct k_itimer *timer) 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); } /* @@ -564,13 +627,21 @@ static int posix_cpu_timer_set(struct k_itimer *timer, int timer_flags, clockid_t clkid = CPUCLOCK_WHICH(timer->it_clock); u64 old_expires, new_expires, old_incr, val; struct cpu_timer *ctmr = &timer->it.cpu; - struct task_struct *p = ctmr->task; struct sighand_struct *sighand; + struct task_struct *p; unsigned long flags; int ret = 0; - if (WARN_ON_ONCE(!p)) - return -EINVAL; + rcu_read_lock(); + p = cpu_timer_task_rcu(timer); + if (!p) { + /* + * If p has just been reaped, we can no + * longer get any information about it at all. + */ + rcu_read_unlock(); + return -ESRCH; + } /* * Use the to_ktime conversion because that clamps the maximum @@ -587,8 +658,10 @@ static int posix_cpu_timer_set(struct k_itimer *timer, int timer_flags, * If p has just been reaped, we can no * longer get any information about it at all. */ - if (unlikely(sighand == NULL)) + if (unlikely(sighand == NULL)) { + rcu_read_unlock(); return -ESRCH; + } /* * Disarm any old timer after extracting its expiry time. @@ -662,7 +735,7 @@ static int posix_cpu_timer_set(struct k_itimer *timer, int timer_flags, */ cpu_timer_setexpires(ctmr, new_expires); if (new_expires != 0 && val < new_expires) { - arm_timer(timer); + arm_timer(timer, p); } unlock_task_sighand(p, &flags); @@ -682,17 +755,31 @@ static int posix_cpu_timer_set(struct k_itimer *timer, int timer_flags, timer->it_overrun_last = 0; timer->it_overrun = -1; - if (new_expires != 0 && !(val < new_expires)) { + if (val >= new_expires) { + if (new_expires != 0) { + /* + * The designated time already passed, so we notify + * immediately, even if the thread never runs to + * accumulate more time on this clock. + */ + cpu_timer_fire(timer); + } + /* - * The designated time already passed, so we notify - * immediately, even if the thread never runs to - * accumulate more time on this clock. + * Make sure we don't keep around the process wide cputime + * counter or the tick dependency if they are not necessary. */ - cpu_timer_fire(timer); - } + sighand = lock_task_sighand(p, &flags); + if (!sighand) + goto out; - ret = 0; + if (!cpu_timer_queued(ctmr)) + trigger_base_recalc_expires(timer, p); + + unlock_task_sighand(p, &flags); + } out: + rcu_read_unlock(); if (old) old->it_interval = ns_to_timespec64(old_incr); @@ -704,10 +791,12 @@ static void posix_cpu_timer_get(struct k_itimer *timer, struct itimerspec64 *itp 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 = ctmr->task; + struct task_struct *p; - if (WARN_ON_ONCE(!p)) - return; + rcu_read_lock(); + p = cpu_timer_task_rcu(timer); + if (!p) + goto out; /* * Easy part: convert the reload time. @@ -715,36 +804,15 @@ static void posix_cpu_timer_get(struct k_itimer *timer, struct itimerspec64 *itp itp->it_interval = ktime_to_timespec64(timer->it_interval); if (!expires) - return; + goto out; /* * Sample the clock to take the difference with the expiry time. */ - if (CPUCLOCK_PERTHREAD(timer->it_clock)) { + if (CPUCLOCK_PERTHREAD(timer->it_clock)) now = cpu_clock_sample(clkid, p); - } else { - struct sighand_struct *sighand; - unsigned long flags; - - /* - * Protect against sighand release/switch in exit/exec and - * also make timer sampling safe if it ends up calling - * thread_group_cputime(). - */ - sighand = lock_task_sighand(p, &flags); - if (unlikely(sighand == NULL)) { - /* - * The process has been reaped. - * We can't even collect a sample any more. - * Disarm the timer, nothing else to do. - */ - cpu_timer_setexpires(ctmr, 0); - return; - } else { - now = cpu_clock_sample_group(clkid, p, false); - unlock_task_sighand(p, &flags); - } - } + else + now = cpu_clock_sample_group(clkid, p, false); if (now < expires) { itp->it_value = ns_to_timespec64(expires - now); @@ -756,6 +824,8 @@ static void posix_cpu_timer_get(struct k_itimer *timer, struct itimerspec64 *itp itp->it_value.tv_nsec = 1; itp->it_value.tv_sec = 0; } +out: + rcu_read_unlock(); } #define MAX_COLLECTED 20 @@ -800,7 +870,7 @@ 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); } } @@ -814,7 +884,7 @@ 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; } @@ -888,7 +958,7 @@ 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) @@ -976,56 +1046,38 @@ static void check_process_timers(struct task_struct *tsk, static void posix_cpu_timer_rearm(struct k_itimer *timer) { clockid_t clkid = CPUCLOCK_WHICH(timer->it_clock); - struct cpu_timer *ctmr = &timer->it.cpu; - struct task_struct *p = ctmr->task; + struct task_struct *p; struct sighand_struct *sighand; unsigned long flags; u64 now; - if (WARN_ON_ONCE(!p)) - return; + rcu_read_lock(); + p = cpu_timer_task_rcu(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. */ - if (CPUCLOCK_PERTHREAD(timer->it_clock)) { + if (CPUCLOCK_PERTHREAD(timer->it_clock)) now = cpu_clock_sample(clkid, p); - bump_cpu_timer(timer, now); - if (unlikely(p->exit_state)) - return; - - /* Protect timer list r/w in arm_timer() */ - sighand = lock_task_sighand(p, &flags); - if (!sighand) - return; - } else { - /* - * Protect arm_timer() and timer sampling in case of call to - * thread_group_cputime(). - */ - sighand = lock_task_sighand(p, &flags); - if (unlikely(sighand == NULL)) { - /* - * The process has been reaped. - * We can't even collect a sample any more. - */ - cpu_timer_setexpires(ctmr, 0); - return; - } else if (unlikely(p->exit_state) && thread_group_empty(p)) { - /* If the process is dying, no need to rearm */ - goto unlock; - } + else now = cpu_clock_sample_group(clkid, p, true); - bump_cpu_timer(timer, now); - /* Leave the sighand locked for the call below. */ - } + + bump_cpu_timer(timer, now); /* * Now re-arm for the new expiry time. */ - arm_timer(timer); -unlock: + arm_timer(timer, p); unlock_task_sighand(p, &flags); +out: + rcu_read_unlock(); } /** @@ -1105,40 +1157,178 @@ 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) +{ + handle_posix_cpu_timers(current); +} + /* - * 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. + * Clear existing posix CPU timers task work. */ -void run_posix_cpu_timers(void) +void clear_posix_cputimers_work(struct task_struct *p) { - struct task_struct *tsk = current; - struct k_itimer *timer, *next; - unsigned long flags; - LIST_HEAD(firing); + /* + * A copied work entry from the old task is not meaningful, clear it. + * N.B. init_task_work will not do this. + */ + 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); + p->posix_cputimers_work.scheduled = false; +} - lockdep_assert_irqs_disabled(); +/* + * 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; /* - * The fast path checks that there are no expired thread or thread - * group timers. If that's so, just return. + * 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 (!fastpath_timer_check(tsk)) - return; + if (!IS_ENABLED(CONFIG_PREEMPT_RT)) { + tsk->posix_cputimers_work.scheduled = false; + return true; + } - if (!lock_task_sighand(tsk, &flags)) - return; /* - * 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 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 */ - check_process_timers(tsk, &firing); +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 @@ -1156,6 +1346,13 @@ void run_posix_cpu_timers(void) list_for_each_entry_safe(timer, next, &firing, it.cpu.elist) { int 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; @@ -1172,6 +1369,34 @@ void run_posix_cpu_timers(void) } /* + * 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); +} + +/* * Set one of the process-wide special case CPU timers or RLIMIT_CPU. * The tsk->sighand->siglock must be held by the caller. */ @@ -1201,9 +1426,8 @@ void set_process_cpu_timer(struct task_struct *tsk, unsigned int clkid, } } - if (!*newval) - return; - *newval += now; + if (*newval) + *newval += now; } /* @@ -1213,7 +1437,7 @@ 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, @@ -1335,8 +1559,8 @@ 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; } diff --git a/kernel/time/posix-stubs.c b/kernel/time/posix-stubs.c index fcb3b21d8bdc..90ea5f373e50 100644 --- a/kernel/time/posix-stubs.c +++ b/kernel/time/posix-stubs.c @@ -70,7 +70,7 @@ 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 +90,7 @@ 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) { diff --git a/kernel/time/posix-timers.c b/kernel/time/posix-timers.c index ff0eb30de346..5dead89308b7 100644 --- a/kernel/time/posix-timers.c +++ b/kernel/time/posix-timers.c @@ -121,7 +121,8 @@ static struct k_itimer *__posix_timers_find(struct hlist_head *head, { struct k_itimer *timer; - hlist_for_each_entry_rcu(timer, head, t_hash) { + hlist_for_each_entry_rcu(timer, head, t_hash, + lockdep_is_held(&hash_lock)) { if ((timer->it_signal == sig) && (timer->it_id == id)) return timer; } @@ -272,8 +273,8 @@ static int posix_get_hrtimer_res(clockid_t which_clock, struct timespec64 *tp) 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); @@ -335,7 +336,7 @@ void posixtimer_rearm(struct kernel_siginfo *info) int posix_timer_event(struct k_itimer *timr, int si_private) { enum pid_type type; - int ret = -1; + int ret; /* * FIXME: if ->sigq is queued we can race with * dequeue_signal()->posixtimer_rearm(). @@ -438,12 +439,12 @@ 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: @@ -1050,15 +1051,24 @@ retry_delete: } /* - * This is called by do_exit or de_thread, only when there are no more - * references to the shared signal_struct. + * This is called by do_exit or de_thread, only when nobody else can + * modify the signal->posix_timers list. Yet we need sighand->siglock + * to prevent the race with /proc/pid/timers. */ -void exit_itimers(struct signal_struct *sig) +void exit_itimers(struct task_struct *tsk) { + struct list_head timers; struct k_itimer *tmr; - while (!list_empty(&sig->posix_timers)) { - tmr = list_entry(sig->posix_timers.next, struct k_itimer, list); + if (list_empty(&tsk->signal->posix_timers)) + return; + + spin_lock_irq(&tsk->sighand->siglock); + list_replace_init(&tsk->signal->posix_timers, &timers); + spin_unlock_irq(&tsk->sighand->siglock); + + while (!list_empty(&timers)) { + tmr = list_first_entry(&timers, struct k_itimer, list); itimer_delete(tmr); } } @@ -1190,8 +1200,8 @@ 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; } diff --git a/kernel/time/sched_clock.c b/kernel/time/sched_clock.c index e4332e3e2d56..8464c5acc913 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; @@ -93,6 +69,17 @@ static inline u64 notrace cyc_to_ns(u64 cyc, u32 mult, u32 shift) return (cyc * mult) >> shift; } +notrace struct clock_read_data *sched_clock_read_begin(unsigned int *seq) +{ + *seq = raw_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); +} + unsigned long long notrace sched_clock(void) { u64 cyc, res; @@ -100,13 +87,12 @@ unsigned long long notrace sched_clock(void) struct clock_read_data *rd; do { - seq = raw_read_seqcount(&cd.seq); - rd = cd.read_data + (seq & 1); + rd = sched_clock_read_begin(&seq); 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 (sched_clock_read_retry(seq)); return res; } @@ -208,20 +194,19 @@ sched_clock_register(u64 (*read)(void), int bits, unsigned long rate) if (sched_clock_timer.function != NULL) { /* update timeout for clock wrap */ - hrtimer_start(&sched_clock_timer, cd.wrap_kt, HRTIMER_MODE_REL); + hrtimer_start(&sched_clock_timer, cd.wrap_kt, + HRTIMER_MODE_REL_HARD); } 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 */ @@ -243,7 +228,7 @@ 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); @@ -254,9 +239,9 @@ void __init generic_sched_clock_init(void) * Start the timer to keep sched_clock() properly updated and * sets the initial epoch. */ - hrtimer_init(&sched_clock_timer, CLOCK_MONOTONIC, HRTIMER_MODE_REL); + hrtimer_init(&sched_clock_timer, CLOCK_MONOTONIC, HRTIMER_MODE_REL_HARD); sched_clock_timer.function = sched_clock_poll; - hrtimer_start(&sched_clock_timer, cd.wrap_kt, HRTIMER_MODE_REL); + hrtimer_start(&sched_clock_timer, cd.wrap_kt, HRTIMER_MODE_REL_HARD); } /* @@ -272,7 +257,7 @@ 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 = raw_read_seqcount_latch(&cd.seq); return cd.read_data[seq & 1].epoch_cyc; } @@ -293,7 +278,7 @@ void sched_clock_resume(void) struct clock_read_data *rd = &cd.read_data[0]; rd->epoch_cyc = cd.actual_read_sched_clock(); - hrtimer_start(&sched_clock_timer, cd.wrap_kt, HRTIMER_MODE_REL); + hrtimer_start(&sched_clock_timer, cd.wrap_kt, HRTIMER_MODE_REL_HARD); rd->read_sched_clock = cd.actual_read_sched_clock; } diff --git a/kernel/time/test_udelay.c b/kernel/time/test_udelay.c index 77c63005dc4e..13b11eb62685 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; @@ -150,7 +149,7 @@ module_init(udelay_test_init); static void __exit udelay_test_exit(void) { mutex_lock(&udelay_test_lock); - debugfs_remove(udelay_test_debugfs_file); + debugfs_remove(debugfs_lookup(DEBUGFS_FILENAME, NULL)); mutex_unlock(&udelay_test_lock); } diff --git a/kernel/time/tick-broadcast-hrtimer.c b/kernel/time/tick-broadcast-hrtimer.c index b5a65e212df2..797eb93103ad 100644 --- a/kernel/time/tick-broadcast-hrtimer.c +++ b/kernel/time/tick-broadcast-hrtimer.c @@ -53,7 +53,7 @@ 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. * diff --git a/kernel/time/tick-broadcast.c b/kernel/time/tick-broadcast.c index e51778c312f1..f7fe6fe36173 100644 --- a/kernel/time/tick-broadcast.c +++ b/kernel/time/tick-broadcast.c @@ -33,6 +33,8 @@ static int tick_broadcast_forced; static __cacheline_aligned_in_smp DEFINE_RAW_SPINLOCK(tick_broadcast_lock); #ifdef CONFIG_TICK_ONESHOT +static DEFINE_PER_CPU(struct clock_event_device *, tick_oneshot_wakeup_device); + static void tick_broadcast_setup_oneshot(struct clock_event_device *bc); static void tick_broadcast_clear_oneshot(int cpu); static void tick_resume_broadcast_oneshot(struct clock_event_device *bc); @@ -61,6 +63,13 @@ 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 +97,75 @@ 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 +178,19 @@ 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 +199,7 @@ 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 +240,7 @@ 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) @@ -241,7 +324,6 @@ 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 +338,6 @@ 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 +412,7 @@ 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 +462,7 @@ 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 +472,7 @@ 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) @@ -707,24 +788,16 @@ 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 +926,53 @@ 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 * @@ -877,6 +997,22 @@ 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 */ @@ -905,10 +1041,11 @@ static void tick_broadcast_setup_oneshot(struct clock_event_device *bc) tick_broadcast_oneshot_mask, tmpmask); if (was_periodic && !cpumask_empty(tmpmask)) { + ktime_t nextevt = tick_get_next_period(); + 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); + tick_broadcast_init_next_event(tmpmask, nextevt); + tick_broadcast_set_event(bc, cpu, nextevt); } else bc->next_event = KTIME_MAX; } else { @@ -962,6 +1099,9 @@ 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 7e5d3524e924..46789356f856 100644 --- a/kernel/time/tick-common.c +++ b/kernel/time/tick-common.c @@ -27,10 +27,11 @@ */ 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 @@ -84,13 +85,15 @@ int tick_is_oneshot_available(void) static void tick_periodic(int cpu) { if (tick_do_timer_cpu == cpu) { - write_seqlock(&jiffies_lock); + 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_sequnlock(&jiffies_lock); + write_seqcount_end(&jiffies_seq); + raw_spin_unlock(&jiffies_lock); update_wall_time(); } @@ -125,7 +128,7 @@ 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; @@ -162,16 +165,16 @@ void tick_setup_periodic(struct clock_event_device *dev, int broadcast) ktime_t next; do { - seq = read_seqbegin(&jiffies_lock); + seq = read_seqcount_begin(&jiffies_seq); next = tick_next_period; - } while (read_seqretry(&jiffies_lock, seq)); + } while (read_seqcount_retry(&jiffies_seq, seq)); clockevents_switch_state(dev, CLOCK_EVT_STATE_ONESHOT); for (;;) { if (!clockevents_program_event(dev, next, false)) return; - next = ktime_add(next, tick_period); + next = ktime_add_ns(next, TICK_NSEC); } } } @@ -218,7 +221,6 @@ static void tick_setup_device(struct tick_device *td, 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 @@ -346,12 +348,7 @@ 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)) @@ -376,7 +373,7 @@ out_bc: /* * Can the new device be used as a broadcast device ? */ - tick_install_broadcast_device(newdev); + tick_install_broadcast_device(newdev, cpu); } /** @@ -405,17 +402,13 @@ EXPORT_SYMBOL_GPL(tick_broadcast_oneshot_control); /* * Transfer the do_timer job away from a dying cpu. * - * Called with interrupts disabled. Not locking required. If + * Called with interrupts disabled. No locking required. If * tick_do_timer_cpu is owned by this cpu, nothing can change it. */ void tick_handover_do_timer(void) { - if (tick_do_timer_cpu == smp_processor_id()) { - int cpu = cpumask_first(cpu_online_mask); - - tick_do_timer_cpu = (cpu < nr_cpu_ids) ? cpu : - TICK_DO_TIMER_NONE; - } + if (tick_do_timer_cpu == smp_processor_id()) + tick_do_timer_cpu = cpumask_first(cpu_online_mask); } /* @@ -477,6 +470,13 @@ 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..649f2b48e8f0 100644 --- a/kernel/time/tick-internal.h +++ b/kernel/time/tick-internal.h @@ -15,7 +15,6 @@ 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); @@ -62,7 +61,7 @@ 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 +71,9 @@ 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) { } @@ -165,3 +165,35 @@ DECLARE_PER_CPU(struct hrtimer_cpu_base, hrtimer_bases); extern u64 get_next_timer_interrupt(unsigned long basej, u64 basem); 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 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..475ecceda768 100644 --- a/kernel/time/tick-oneshot.c +++ b/kernel/time/tick-oneshot.c @@ -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) { diff --git a/kernel/time/tick-sched.c b/kernel/time/tick-sched.c index a792d21cac64..b0e3c9205946 100644 --- a/kernel/time/tick-sched.c +++ b/kernel/time/tick-sched.c @@ -20,6 +20,7 @@ #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> @@ -44,7 +45,9 @@ struct tick_sched *tick_get_tick_sched(int 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; @@ -53,48 +56,98 @@ 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. + * 64bit can do a quick check without holding jiffies lock and + * without looking at the sequence count. The smp_load_acquire() + * pairs with the update done later in this function. + * + * 32bit cannot do that because the store of tick_next_period + * consists of two 32bit stores and the first store could move it + * to a random point in the future. */ - delta = ktime_sub(now, READ_ONCE(last_jiffies_update)); - if (delta < tick_period) + 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; + } + + /* Quick check failed, i.e. update is required. */ + raw_spin_lock(&jiffies_lock); + /* + * Reevaluate 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; + } - /* Reevaluate with jiffies_lock held */ - write_seqlock(&jiffies_lock); + 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_sequnlock(&jiffies_lock); - return; + /* + * A plain store is good enough on 32bit as the quick check + * above is protected by the sequence count. + */ + tick_next_period = nextp; } - write_sequnlock(&jiffies_lock); + + /* + * 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(); } @@ -105,15 +158,19 @@ static ktime_t tick_init_jiffy_update(void) { ktime_t period; - write_seqlock(&jiffies_lock); + raw_spin_lock(&jiffies_lock); + write_seqcount_begin(&jiffies_seq); /* Did we start the jiffies update yet ? */ if (last_jiffies_update == 0) last_jiffies_update = tick_next_period; period = last_jiffies_update; - write_sequnlock(&jiffies_lock); + write_seqcount_end(&jiffies_seq); + raw_spin_unlock(&jiffies_lock); return period; } +#define MAX_STALLED_JIFFIES 5 + static void tick_sched_do_timer(struct tick_sched *ts, ktime_t now) { int cpu = smp_processor_id(); @@ -131,7 +188,7 @@ static void tick_sched_do_timer(struct tick_sched *ts, ktime_t now) */ if (unlikely(tick_do_timer_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; } @@ -141,6 +198,21 @@ static void tick_sched_do_timer(struct tick_sched *ts, ktime_t now) if (tick_do_timer_cpu == cpu) tick_do_update_jiffies64(now); + /* + * If 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 (ts->inidle) ts->got_idle_tick = 1; } @@ -175,6 +247,7 @@ static void tick_sched_handle(struct tick_sched *ts, struct pt_regs *regs) #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; @@ -238,9 +311,8 @@ 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, -}; +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 @@ -268,6 +340,46 @@ 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, IPI can then be spared. + * + * 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 + */ + 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. @@ -345,16 +457,13 @@ 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 need 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); @@ -368,9 +477,20 @@ 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) @@ -385,13 +505,10 @@ 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); @@ -400,8 +517,6 @@ void __tick_nohz_task_switch(void) 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. */ @@ -411,7 +526,6 @@ 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) { @@ -456,7 +570,7 @@ 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, @@ -647,7 +761,7 @@ 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) { hrtimer_start_expires(&ts->sched_timer, @@ -670,16 +784,16 @@ static inline bool local_timer_softirq_pending(void) static ktime_t tick_nohz_next_event(struct tick_sched *ts, int cpu) { - u64 basemono, next_tick, next_tmr, next_rcu, delta, expires; + u64 basemono, next_tick, delta, expires; unsigned long basejiff; unsigned int seq; /* Read jiffies and the time when jiffies were updated last */ do { - seq = read_seqbegin(&jiffies_lock); + seq = read_seqcount_begin(&jiffies_seq); basemono = last_jiffies_update; basejiff = jiffies; - } while (read_seqretry(&jiffies_lock, seq)); + } while (read_seqcount_retry(&jiffies_seq, seq)); ts->last_jiffies = basejiff; ts->timer_expires_base = basemono; @@ -689,11 +803,11 @@ static ktime_t tick_nohz_next_event(struct tick_sched *ts, int cpu) * 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 + * 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 { @@ -704,10 +818,8 @@ 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; } /* @@ -815,6 +927,8 @@ static void tick_nohz_stop_tick(struct tick_sched *ts, int cpu) if (unlikely(expires == KTIME_MAX)) { if (ts->nohz_mode == NOHZ_MODE_HIGHRES) hrtimer_cancel(&ts->sched_timer); + else + tick_program_event(KTIME_MAX, 1); return; } @@ -859,27 +973,70 @@ static void tick_nohz_restart_sched_tick(struct tick_sched *ts, ktime_t now) * Cancel the scheduled timer and restore the tick */ ts->tick_stopped = 0; - ts->idle_exittime = now; - 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_stop_sched_tick(ts, cpu); + else if (ts->tick_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) 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()); +} + +/* + * A pending softirq outside an IRQ (or softirq disabled section) context + * should be waiting for ksoftirqd to handle it. Therefore we shouldn't + * reach here 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 below since wakep_softirqd() is ignored. + * + */ +static bool report_idle_softirq(void) +{ + 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 (ratelimit < 10) + return false; + + /* On RT, softirqs handling may be waiting on some lock */ + if (!local_bh_blocked()) + return false; + + pr_warn("NOHZ tick-stop error: local softirq work is pending, handler #%02x!!!\n", + pending); + ratelimit++; + + return true; } static bool can_stop_idle_tick(int cpu, struct tick_sched *ts) @@ -908,17 +1065,8 @@ static bool can_stop_idle_tick(int cpu, struct tick_sched *ts) if (need_resched()) return false; - if (unlikely(local_softirq_pending())) { - static int 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++; - } + if (unlikely(report_idle_softirq())) return false; - } if (tick_nohz_full_enabled()) { /* @@ -927,13 +1075,6 @@ static bool can_stop_idle_tick(int cpu, struct tick_sched *ts) */ 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)) - return false; /* Should not happen for nohz-full */ if (WARN_ON_ONCE(tick_do_timer_cpu == TICK_DO_TIMER_NONE)) @@ -1069,7 +1210,11 @@ 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. */ ktime_t tick_nohz_get_sleep_length(ktime_t *delta_next) { @@ -1129,11 +1274,13 @@ unsigned long tick_nohz_get_idle_calls(void) 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; /* @@ -1147,21 +1294,27 @@ 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 (ts->tick_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); } /** @@ -1193,7 +1346,7 @@ 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(); } @@ -1212,11 +1365,17 @@ static void tick_nohz_handler(struct clock_event_device *dev) 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)) + if (unlikely(ts->tick_stopped)) { + /* + * The clockevent device is not reprogrammed, so change the + * clock event device to ONESHOT_STOPPED to avoid spurious + * interrupts on devices which might not be truly one shot. + */ + tick_program_event(KTIME_MAX, 1); return; + } - hrtimer_forward(&ts->sched_timer, now, tick_period); + hrtimer_forward(&ts->sched_timer, now, TICK_NSEC); tick_program_event(hrtimer_get_expires(&ts->sched_timer), 1); } @@ -1253,7 +1412,7 @@ static void tick_nohz_switch_to_nohz(void) next = tick_init_jiffy_update(); hrtimer_set_expires(&ts->sched_timer, next); - hrtimer_forward_now(&ts->sched_timer, tick_period); + hrtimer_forward_now(&ts->sched_timer, TICK_NSEC); tick_program_event(hrtimer_get_expires(&ts->sched_timer), 1); tick_nohz_activate(ts, NOHZ_MODE_LOWRES); } @@ -1268,6 +1427,13 @@ static inline void tick_nohz_irq_enter(void) now = ktime_get(); if (ts->idle_active) tick_nohz_stop_idle(ts, now); + /* + * 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 (ts->tick_stopped) tick_nohz_update_jiffies(now); } @@ -1319,7 +1485,7 @@ static enum hrtimer_restart tick_sched_timer(struct hrtimer *timer) if (unlikely(ts->tick_stopped)) return HRTIMER_NORESTART; - hrtimer_forward(timer, now, tick_period); + hrtimer_forward(timer, now, TICK_NSEC); return HRTIMER_RESTART; } @@ -1353,13 +1519,13 @@ void tick_setup_sched_timer(void) /* 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_forward(&ts->sched_timer, now, TICK_NSEC); hrtimer_start_expires(&ts->sched_timer, HRTIMER_MODE_ABS_PINNED_HARD); tick_nohz_activate(ts, NOHZ_MODE_HIGHRES); } @@ -1379,7 +1545,7 @@ void tick_cancel_sched_timer(int cpu) } #endif -/** +/* * Async notification about clocksource changes */ void tick_clock_notify(void) @@ -1400,7 +1566,7 @@ void tick_oneshot_notify(void) set_bit(0, &ts->check_clocks); } -/** +/* * Check, if a change happened, which makes oneshot possible. * * Called cyclic from the hrtimer softirq (driven by the timer diff --git a/kernel/time/tick-sched.h b/kernel/time/tick-sched.h index 4fb06527cf64..504649513399 100644 --- a/kernel/time/tick-sched.h +++ b/kernel/time/tick-sched.h @@ -29,7 +29,7 @@ enum 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. + * it is reset 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 @@ -49,6 +49,8 @@ enum tick_nohz_mode { * @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 + * @last_tick_jiffies: Value of jiffies seen on last tick + * @stalled_jiffies: Number of stalled jiffies detected across ticks */ struct tick_sched { struct hrtimer sched_timer; @@ -77,6 +79,8 @@ struct tick_sched { u64 next_timer; ktime_t idle_expires; atomic_t tick_dep_mask; + unsigned long last_tick_jiffies; + unsigned int stalled_jiffies; }; extern struct tick_sched *tick_get_tick_sched(int cpu); diff --git a/kernel/time/time.c b/kernel/time/time.c index 3985b2b32d08..526257b3727c 100644 --- a/kernel/time/time.c +++ b/kernel/time/time.c @@ -449,7 +449,7 @@ 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; @@ -503,7 +503,7 @@ EXPORT_SYMBOL(set_normalized_timespec64); * * Returns 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; @@ -571,7 +571,7 @@ EXPORT_SYMBOL(__usecs_to_jiffies); /* * 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 diff --git a/kernel/time/time_test.c b/kernel/time/time_test.c new file mode 100644 index 000000000000..831e8e779ace --- /dev/null +++ b/kernel/time/time_test.c @@ -0,0 +1,99 @@ +// 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) : %ld", \ + 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(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_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 ca69290bee2a..f72b9f1de178 100644 --- a/kernel/time/timekeeping.c +++ b/kernel/time/timekeeping.c @@ -17,11 +17,13 @@ #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" @@ -39,20 +41,24 @@ enum timekeeping_adv_mode { TK_ADV_FREQ }; +DEFINE_RAW_SPINLOCK(timekeeper_lock); + /* * The most important data for readout fits into a single 64 byte * cache line. */ static struct { - seqcount_t seq; + seqcount_raw_spinlock_t seq; struct timekeeper timekeeper; } tk_core ____cacheline_aligned = { - .seq = SEQCNT_ZERO(tk_core.seq), + .seq = SEQCNT_RAW_SPINLOCK_ZERO(tk_core.seq, &timekeeper_lock), }; -static DEFINE_RAW_SPINLOCK(timekeeper_lock); static struct timekeeper shadow_timekeeper; +/* flag for if timekeeping is suspended */ +int __read_mostly timekeeping_suspended; + /** * struct tk_fast - NMI safe timekeeper * @seq: Sequence counter for protecting updates. The lowest bit @@ -63,7 +69,7 @@ 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]; }; @@ -72,26 +78,42 @@ 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; - static inline void tk_normalize_xtime(struct timekeeper *tk) { while (tk->tkr_mono.xtime_nsec >= ((u64)NSEC_PER_SEC << tk->tkr_mono.shift)) { @@ -157,7 +179,7 @@ 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 @@ -222,10 +244,10 @@ static inline u64 timekeeping_get_delta(const struct tk_read_base *tkr) unsigned int seq; /* - * Since we're called holding a seqlock, the data may shift + * Since we're called holding a seqcount, 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 + * results away. So nest another seqcount here to atomically * grab the points we are checking with. */ do { @@ -349,13 +371,6 @@ 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) { u64 nsec; @@ -363,8 +378,7 @@ static inline u64 timekeeping_delta_to_ns(const struct tk_read_base *tkr, u64 de nsec = delta * tkr->mult + tkr->xtime_nsec; nsec >>= tkr->shift; - /* If arch requires, add in get_arch_timeoffset() */ - return nsec + arch_gettimeoffset(); + return nsec; } static inline u64 timekeeping_get_ns(const struct tk_read_base *tkr) @@ -387,6 +401,7 @@ static inline u64 timekeeping_cycles_to_ns(const struct tk_read_base *tkr, u64 c /** * 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. @@ -416,6 +431,30 @@ static void update_fast_timekeeper(const struct tk_read_base *tkr, memcpy(base + 1, base, sizeof(*base)); } +static __always_inline u64 fast_tk_get_delta_ns(struct tk_read_base *tkr) +{ + u64 delta, cycles = tk_clock_read(tkr); + + delta = clocksource_delta(cycles, tkr->cycle_last, tkr->mask); + return timekeeping_delta_to_ns(tkr, delta); +} + +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 += fast_tk_get_delta_ns(tkr); + } while (read_seqcount_latch_retry(&tkf->seq, seq)); + + return now; +} + /** * ktime_get_mono_fast_ns - Fast NMI safe access to clock monotonic * @@ -442,40 +481,25 @@ 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 +510,7 @@ 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 @@ -501,50 +525,120 @@ EXPORT_SYMBOL_GPL(ktime_get_raw_fast_ns); * (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_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 timekeeper *tk = &tk_core.timekeeper; + + return (ktime_get_mono_fast_ns() + ktime_to_ns(data_race(tk->offs_tai))); +} +EXPORT_SYMBOL_GPL(ktime_get_tai_fast_ns); + +static __always_inline u64 __ktime_get_real_fast(struct tk_fast *tkf, u64 *mono) { struct tk_read_base *tkr; + u64 basem, baser, delta; 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); + basem = ktime_to_ns(tkr->base); + baser = ktime_to_ns(tkr->base_real); + delta = fast_tk_get_delta_ns(tkr); + } while (read_seqcount_latch_retry(&tkf->seq, seq)); - 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; + if (mono) + *mono = basem + delta; + return baser + delta; } /** * ktime_get_real_fast_ns: - NMI safe and fast access to clock realtime. + * + * See ktime_get_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); + return __ktime_get_real_fast(&tk_fast_mono, NULL); } EXPORT_SYMBOL_GPL(ktime_get_real_fast_ns); /** + * ktime_get_fast_timestamps: - NMI safe timestamps + * @snapshot: Pointer to timestamp storage + * + * Stores clock monotonic, boottime and realtime timestamps. + * + * Boot time is a racy access on 32bit systems if the sleep time injection + * happens late during resume and not in timekeeping_resume(). That could + * be avoided by expanding struct tk_read_base with boot offset for 32bit + * and adding more overhead to the update. As this is a hard to observe + * once per resume event which can be filtered with reasonable effort using + * the accurate mono/real timestamps, it's probably not worth the trouble. + * + * Aside of that it might be possible on 32 and 64 bit to observe the + * following when the sleep time injection happens late: + * + * CPU 0 CPU 1 + * timekeeping_resume() + * ktime_get_fast_timestamps() + * mono, real = __ktime_get_real_fast() + * inject_sleep_time() + * update boot offset + * boot = mono + bootoffset; + * + * That means that boot time already has the sleep time adjustment, but + * real time does not. On the next readout both are in sync again. + * + * Preventing this for 64bit is not really feasible without destroying the + * careful cache layout of the timekeeper because the sequence count and + * struct tk_read_base would then need two cache lines instead of one. + * + * Access to the time keeper clock source is disabled across the innermost + * steps of suspend/resume. The accessors still work, but the timestamps + * are frozen until time keeping is resumed which happens very early. + * + * For regular suspend/resume there is no observable difference vs. sched + * clock, but it might affect some of the nasty low level debug printks. + * + * OTOH, access to sched clock is not guaranteed across suspend/resume on + * all systems either so it depends on the hardware in use. + * + * If that turns out to be a real problem then this could be mitigated by + * using sched clock in a similar way as during early boot. But it's not as + * trivial as on early boot because it needs some careful protection + * against the clock monotonic timestamp jumping backwards on resume. + */ +void ktime_get_fast_timestamps(struct ktime_timestamps *snapshot) +{ + struct timekeeper *tk = &tk_core.timekeeper; + + snapshot->real = __ktime_get_real_fast(&tk_fast_mono, &snapshot->mono); + snapshot->boot = snapshot->mono + ktime_to_ns(data_race(tk->offs_boot)); +} + +/** * halt_fast_timekeeper - Prevent fast timekeeper from accessing clocksource. * @tk: Timekeeper to snapshot. * @@ -580,6 +674,7 @@ 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) { @@ -599,6 +694,7 @@ 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) { @@ -689,6 +785,7 @@ static void timekeeping_update(struct timekeeper *tk, unsigned int action) /** * 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, @@ -704,16 +801,8 @@ static void timekeeping_forward_now(struct timekeeper *tk) 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; - /* If arch requires, add in get_arch_timeoffset() */ - tk->tkr_raw.xtime_nsec += (u64)arch_gettimeoffset() << tk->tkr_raw.shift; - tk_normalize_xtime(tk); } @@ -829,7 +918,7 @@ 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 */ @@ -921,8 +1010,7 @@ 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 @@ -953,7 +1041,7 @@ EXPORT_SYMBOL_GPL(ktime_get_real_seconds); * but without the sequence counter protect. This internal function * is called just when timekeeping lock is already held. */ -time64_t __ktime_get_real_seconds(void) +noinstr time64_t __ktime_get_real_seconds(void) { struct timekeeper *tk = &tk_core.timekeeper; @@ -979,6 +1067,7 @@ 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, @@ -1005,9 +1094,8 @@ static int scale64_check_overflow(u64 mult, u64 div, u64 *base) ((int)sizeof(u64)*8 - fls64(mult) < fls64(rem))) return -EOVERFLOW; tmp *= mult; - rem *= mult; - do_div(rem, div); + rem = div64_u64(rem * mult, div); *base = tmp + rem; return 0; } @@ -1237,8 +1325,7 @@ int do_settimeofday64(const struct timespec64 *ts) 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; + ts_delta = timespec64_sub(*ts, xt); if (timespec64_compare(&tk->wall_to_monotonic, &ts_delta) > 0) { ret = -EINVAL; @@ -1254,11 +1341,13 @@ out: write_seqcount_end(&tk_core.seq); raw_spin_unlock_irqrestore(&timekeeper_lock, flags); - /* signal hrtimers about time change */ - clock_was_set(); + /* Signal hrtimers about time change */ + clock_was_set(CLOCK_SET_WALL); - if (!ret) + if (!ret) { audit_tk_injoffset(ts_delta); + add_device_randomness(ts, sizeof(*ts)); + } return ret; } @@ -1266,7 +1355,7 @@ 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. */ @@ -1302,8 +1391,8 @@ error: /* even if we error out, we forwarded the time, so call update */ write_seqcount_end(&tk_core.seq); raw_spin_unlock_irqrestore(&timekeeper_lock, flags); - /* signal hrtimers about time change */ - clock_was_set(); + /* Signal hrtimers about time change */ + clock_was_set(CLOCK_SET_WALL); return ret; } @@ -1342,9 +1431,8 @@ 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) { @@ -1352,7 +1440,7 @@ 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 @@ -1360,35 +1448,45 @@ static void __timekeeping_set_tai_offset(struct timekeeper *tk, s32 tai_offset) static int change_clocksource(void *data) { struct timekeeper *tk = &tk_core.timekeeper; - struct clocksource *new, *old; + struct clocksource *new, *old = NULL; unsigned long flags; + bool change = false; new = (struct clocksource *) data; - 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 (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 { + if (!new->enable || new->enable(new) == 0) + change = true; + else module_put(new->owner); - } } + + raw_spin_lock_irqsave(&timekeeper_lock, flags); + write_seqcount_begin(&tk_core.seq); + + timekeeping_forward_now(tk); + + if (change) { + old = tk->tkr_mono.clock; + tk_setup_internals(tk, new); + } + 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); + if (old) { + if (old->disable) + old->disable(old); + + module_put(old->owner); + } + return 0; } @@ -1475,6 +1573,7 @@ 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. @@ -1493,8 +1592,9 @@ void __weak read_persistent_clock64(struct timespec64 *ts) * from the boot. * * 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 + * @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 @@ -1579,7 +1679,8 @@ 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. @@ -1665,8 +1766,8 @@ void timekeeping_inject_sleeptime64(const struct timespec64 *delta) write_seqcount_end(&tk_core.seq); raw_spin_unlock_irqrestore(&timekeeper_lock, flags); - /* signal hrtimers about time change */ - clock_was_set(); + /* Signal hrtimers about time change */ + clock_was_set(CLOCK_SET_WALL | CLOCK_SET_BOOT); } #endif @@ -1729,8 +1830,10 @@ void timekeeping_resume(void) 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) @@ -1878,7 +1981,7 @@ 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)) { @@ -1950,13 +2053,12 @@ 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) { @@ -1998,11 +2100,11 @@ 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. @@ -2045,7 +2147,7 @@ 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 *real_tk = &tk_core.timekeeper; struct timekeeper *tk = &shadow_timekeeper; @@ -2060,19 +2162,12 @@ static void timekeeping_advance(enum timekeeping_adv_mode mode) if (unlikely(timekeeping_suspended)) goto out; -#ifdef CONFIG_ARCH_USES_GETTIMEOFFSET - offset = real_tk->cycle_interval; - - 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); /* 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); @@ -2123,9 +2218,8 @@ static void timekeeping_advance(enum timekeeping_adv_mode mode) 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(); + + return !!clock_set; } /** @@ -2134,7 +2228,8 @@ out: */ void update_wall_time(void) { - timekeeping_advance(TK_ADV_TICK); + if (timekeeping_advance(TK_ADV_TICK)) + clock_was_set_delayed(); } /** @@ -2194,7 +2289,7 @@ EXPORT_SYMBOL(ktime_get_coarse_ts64); void do_timer(unsigned long ticks) { jiffies_64 += ticks; - calc_global_load(ticks); + calc_global_load(); } /** @@ -2241,7 +2336,7 @@ 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) @@ -2274,7 +2369,7 @@ 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 @@ -2306,6 +2401,20 @@ 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 @@ -2314,8 +2423,9 @@ int do_adjtimex(struct __kernel_timex *txc) { struct timekeeper *tk = &tk_core.timekeeper; struct audit_ntp_data ad; - unsigned long flags; + bool clock_set = false; struct timespec64 ts; + unsigned long flags; s32 orig_tai, tai; int ret; @@ -2323,6 +2433,7 @@ int do_adjtimex(struct __kernel_timex *txc) ret = timekeeping_validate_timex(txc); if (ret) return ret; + add_device_randomness(txc, sizeof(*txc)); if (txc->modes & ADJ_SETOFFSET) { struct timespec64 delta; @@ -2340,6 +2451,7 @@ int do_adjtimex(struct __kernel_timex *txc) 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); @@ -2350,6 +2462,7 @@ int do_adjtimex(struct __kernel_timex *txc) if (tai != orig_tai) { __timekeeping_set_tai_offset(tk, tai); timekeeping_update(tk, TK_MIRROR | TK_CLOCK_WAS_SET); + clock_set = true; } tk_update_leap_state(tk); @@ -2360,10 +2473,10 @@ int do_adjtimex(struct __kernel_timex *txc) /* 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_REALTIME); ntp_notify_cmos_timer(); @@ -2388,17 +2501,3 @@ void hardpps(const struct timespec64 *phase_ts, const struct timespec64 *raw_ts) } 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) -{ - write_seqlock(&jiffies_lock); - do_timer(ticks); - write_sequnlock(&jiffies_lock); - update_wall_time(); -} diff --git a/kernel/time/timekeeping.h b/kernel/time/timekeeping.h index 141ab3ab0354..543beba096c7 100644 --- a/kernel/time/timekeeping.h +++ b/kernel/time/timekeeping.h @@ -22,10 +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 seqlock_t jiffies_lock; +extern raw_spinlock_t jiffies_lock; +extern seqcount_raw_spinlock_t jiffies_seq; #define CS_NAME_LEN 32 diff --git a/kernel/time/timekeeping_internal.h b/kernel/time/timekeeping_internal.h index bcbb52db2256..4ca2787d1642 100644 --- a/kernel/time/timekeeping_internal.h +++ b/kernel/time/timekeeping_internal.h @@ -1,12 +1,14 @@ /* 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 extern void tk_debug_account_sleep_time(const struct timespec64 *t); #else @@ -31,4 +33,7 @@ static inline u64 clocksource_delta(u64 now, u64 last, u64 mask) } #endif +/* Semi public for serialization of non timekeeper VDSO updates. */ +extern raw_spinlock_t timekeeper_lock; + #endif /* _TIMEKEEPING_INTERNAL_H */ diff --git a/kernel/time/timer.c b/kernel/time/timer.c index 4820823515e9..717fcb9fb14a 100644 --- a/kernel/time/timer.c +++ b/kernel/time/timer.c @@ -43,6 +43,8 @@ #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> @@ -156,7 +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)) @@ -203,8 +206,9 @@ 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; @@ -220,7 +224,7 @@ 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); @@ -231,7 +235,42 @@ static void timers_update_migration(void) else static_branch_disable(&timers_migration_enabled); } -#else + +#ifdef CONFIG_SYSCTL +static 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 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 */ @@ -248,20 +287,6 @@ void timers_update_nohz(void) schedule_work(&timer_update_work); } -int timer_migration_handler(struct ctl_table *table, int write, - void __user *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); @@ -488,71 +513,61 @@ 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 * capacity limit of the wheel. */ - if (expires >= WHEEL_TIMEOUT_CUTOFF) - expires = WHEEL_TIMEOUT_MAX; + 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) { @@ -574,35 +589,88 @@ trigger_dyntick_cpu(struct timer_base *base, struct timer_list *timer) * 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: */ - if (!base->is_idle) - return; + if (base->is_idle) + 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, timer->expires, timer->flags); /* - * 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. */ - base->next_expiry = timer->expires; - wake_up_nohz_cpu(base->cpu); + if (time_before(bucket_expiry, base->next_expiry)) { + /* + * Set the next expiry time and kick the CPU so it + * can reevaluate the wheel: + */ + base->next_expiry = bucket_expiry; + base->timers_pending = true; + base->next_expiry_recalc = false; + trigger_dyntick_cpu(base, timer); + } } -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) @@ -653,7 +721,7 @@ 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; } @@ -694,7 +762,7 @@ 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, @@ -719,11 +787,6 @@ 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); @@ -781,6 +844,8 @@ 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); } @@ -826,8 +891,10 @@ 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; @@ -877,31 +944,27 @@ get_target_base(struct timer_base *base, unsigned tflags) static inline void forward_timer_base(struct timer_base *base) { -#ifdef CONFIG_NO_HZ_COMMON - unsigned long jnow; + unsigned long jnow = READ_ONCE(jiffies); /* - * 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. + * No need to forward if we are close enough below jiffies. + * Also while executing timers, base->clk is 1 offset ahead + * of jiffies to avoid endless requeuing to current jiffies. */ - 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 ((long)(jnow - base->clk) < 1) 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)) + if (time_after(base->next_expiry, jnow)) { base->clk = jnow; - else + } else { + if (WARN_ON_ONCE(time_before(base->next_expiry, base->clk))) + return; base->clk = base->next_expiry; -#endif + } } @@ -944,13 +1007,14 @@ static struct timer_base *lock_timer_base(struct timer_list *timer, #define MOD_TIMER_PENDING_ONLY 0x01 #define MOD_TIMER_REDUCE 0x02 +#define MOD_TIMER_NOTPENDING 0x04 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); @@ -960,7 +1024,7 @@ __mod_timer(struct timer_list *timer, unsigned long expires, unsigned int option * the timer is re-modified to have the same timeout or ends up in the * same array bucket then just return: */ - if (timer_pending(timer)) { + if (!(options & MOD_TIMER_NOTPENDING) && timer_pending(timer)) { /* * The downside of this optimization is that it can result in * larger granularity than you would get from adding a new @@ -989,7 +1053,7 @@ __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 @@ -1042,16 +1106,13 @@ __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); @@ -1133,7 +1194,7 @@ EXPORT_SYMBOL(timer_reduce); void add_timer(struct timer_list *timer) { BUG_ON(timer_pending(timer)); - mod_timer(timer, timer->expires); + __mod_timer(timer, timer->expires, MOD_TIMER_NOTPENDING); } EXPORT_SYMBOL(add_timer); @@ -1251,14 +1312,16 @@ 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) { 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); } } @@ -1277,7 +1340,7 @@ 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); /* @@ -1361,6 +1424,13 @@ int del_timer_sync(struct timer_list *timer) */ WARN_ON(in_irq() && !(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); @@ -1442,22 +1512,22 @@ static void expire_timers(struct timer_base *base, struct hlist_head *head) 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; @@ -1479,7 +1549,6 @@ 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 @@ -1512,6 +1581,7 @@ 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; @@ -1519,6 +1589,13 @@ 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 @@ -1556,13 +1633,18 @@ 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; } + + base->next_expiry_recalc = false; + base->timers_pending = !(next == base->clk + NEXT_TIMER_MAX_DELTA); + return next; } +#ifdef CONFIG_NO_HZ_COMMON /* * Check, if the next hrtimer event is before the next timer wheel * event: @@ -1609,7 +1691,6 @@ u64 get_next_timer_interrupt(unsigned long basej, u64 basem) struct timer_base *base = this_cpu_ptr(&timer_bases[BASE_STD]); u64 expires = KTIME_MAX; unsigned long nextevt; - bool is_max_delta; /* * Pretend that there is no timer pending if the cpu is offline. @@ -1619,9 +1700,10 @@ u64 get_next_timer_interrupt(unsigned long basej, u64 basem) return expires; raw_spin_lock(&base->lock); - nextevt = __next_timer_interrupt(base); - is_max_delta = (nextevt == base->clk + NEXT_TIMER_MAX_DELTA); - base->next_expiry = nextevt; + if (base->next_expiry_recalc) + base->next_expiry = __next_timer_interrupt(base); + nextevt = base->next_expiry; + /* * We have a fresh next event. Check whether we can forward the * base. We can only do that when @basej is past base->clk @@ -1638,7 +1720,7 @@ u64 get_next_timer_interrupt(unsigned long basej, u64 basem) expires = basem; base->is_idle = false; } else { - if (!is_max_delta) + if (base->timers_pending) expires = basem + (u64)(nextevt - basej) * TICK_NSEC; /* * If we expect to sleep more than a tick, mark the base idle. @@ -1647,10 +1729,8 @@ u64 get_next_timer_interrupt(unsigned long basej, u64 basem) * logic is only maintained for the BASE_STD base, deferrable * timers may still see large granularity skew (by design). */ - if ((expires - basem) > TICK_NSEC) { - base->must_forward_clk = true; + if ((expires - basem) > TICK_NSEC) base->is_idle = true; - } } raw_spin_unlock(&base->lock); @@ -1674,64 +1754,7 @@ void timer_clear_idle(void) */ base->is_idle = false; } - -static int collect_expired_timers(struct timer_base *base, - struct hlist_head *heads) -{ - unsigned long now = READ_ONCE(jiffies); - - /* - * 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. - */ - if ((long)(now - base->clk) > 2) { - unsigned long next = __next_timer_interrupt(base); - - /* - * If the next timer is ahead of time forward to current - * jiffies, otherwise forward to the next expiry time: - */ - if (time_after(next, now)) { - /* - * The call site will increment base->clk and then - * terminate the expiry loop immediately. - */ - base->clk = now; - return 0; - } - base->clk = next; - } - return __collect_expired_timers(base, heads); -} -#else -static inline int collect_expired_timers(struct timer_base *base, - struct hlist_head *heads) -{ - return __collect_expired_timers(base, heads); -} -#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. - */ -void update_process_times(int user_tick) -{ - struct task_struct *p = current; - - /* 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(); -} /** * __run_timers - run all expired timers (if any) on this CPU. @@ -1742,32 +1765,26 @@ static inline void __run_timers(struct timer_base *base) struct hlist_head heads[LVL_DEPTH]; int levels; - if (!time_after_eq(jiffies, base->clk)) + if (time_before(jiffies, base->next_expiry)) 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; - - while (time_after_eq(jiffies, base->clk)) { - + 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); base->clk++; + base->next_expiry = __next_timer_interrupt(base); while (levels--) expire_timers(base, heads + levels); @@ -1791,24 +1808,45 @@ static __latent_entropy void run_timer_softirq(struct softirq_action *h) /* * Called by the local, per-CPU timer interrupt on SMP. */ -void run_local_timers(void) +static void run_local_timers(void) { struct timer_base *base = this_cpu_ptr(&timer_bases[BASE_STD]); hrtimer_run_queues(); /* Raise the softirq only if required. */ - if (time_before(jiffies, base->clk)) { + if (time_before(jiffies, base->next_expiry)) { if (!IS_ENABLED(CONFIG_NO_HZ_COMMON)) return; /* CPU is awake, so check the deferrable base. */ base++; - if (time_before(jiffies, base->clk)) + if (time_before(jiffies, base->next_expiry)) return; } raise_softirq(TIMER_SOFTIRQ); } /* + * 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. + */ +void update_process_times(int user_tick) +{ + struct task_struct *p = current; + + /* 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(); +} + +/* * Since schedule_timeout()'s timer is defined on the stack, it must store * the target task on the stack as well. */ @@ -1828,21 +1866,23 @@ static void process_timeout(struct timer_list *t) * schedule_timeout - sleep until timeout * @timeout: timeout value in jiffies * - * Make the current task sleep until @timeout jiffies have - * elapsed. The routine will return immediately unless - * the current task state has been set (see set_current_state()). + * 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): * - * You can set the task state as follows - + * %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())". + * 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 + * The current task state is guaranteed to be %TASK_RUNNING when this * routine returns. * * Specifying a @timeout value of %MAX_SCHEDULE_TIMEOUT will schedule @@ -1850,7 +1890,7 @@ static void process_timeout(struct timer_list *t) * 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 + * jiffies will be returned. In all cases the return value is guaranteed * to be non-negative. */ signed long __sched schedule_timeout(signed long timeout) @@ -1882,7 +1922,7 @@ signed long __sched schedule_timeout(signed long timeout) printk(KERN_ERR "schedule_timeout: wrong timeout " "value %lx\n", timeout); dump_stack(); - current->state = TASK_RUNNING; + __set_current_state(TASK_RUNNING); goto out; } } @@ -1891,7 +1931,7 @@ signed long __sched schedule_timeout(signed long timeout) timer.task = current; timer_setup_on_stack(&timer.timer, process_timeout, 0); - __mod_timer(&timer.timer, expire, 0); + __mod_timer(&timer.timer, expire, MOD_TIMER_NOTPENDING); schedule(); del_singleshot_timer_sync(&timer.timer); @@ -1964,8 +2004,9 @@ 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; } @@ -2018,6 +2059,7 @@ 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); } } @@ -2033,6 +2075,7 @@ 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); } @@ -2066,26 +2109,28 @@ unsigned long msleep_interruptible(unsigned int msecs) 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 + * 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 * * In non-atomic context where the exact wakeup time is flexible, use - * usleep_range() instead of udelay(). The sleep improves responsiveness + * usleep_range_state() 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) +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; for (;;) { - __set_current_state(TASK_UNINTERRUPTIBLE); + __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); +EXPORT_SYMBOL(usleep_range_state); diff --git a/kernel/time/timer_list.c b/kernel/time/timer_list.c index acb326f5f50a..ed7d6ad694fb 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)); @@ -218,44 +203,39 @@ 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 +256,7 @@ 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: v0.9\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/vsyscall.c b/kernel/time/vsyscall.c index 9577c89179cd..f0d5062d9cbc 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) { @@ -71,13 +73,15 @@ void update_vsyscall(struct timekeeper *tk) { struct vdso_data *vdata = __arch_get_k_vdso_data(); struct vdso_timestamp *vdso_ts; + s32 clock_mode; u64 nsec; /* copy vsyscall data */ vdso_write_begin(vdata); - vdata[CS_HRES_COARSE].clock_mode = __arch_get_clock_mode(tk); - vdata[CS_RAW].clock_mode = __arch_get_clock_mode(tk); + clock_mode = tk->tkr_mono.clock->vdso_clock_mode; + vdata[CS_HRES_COARSE].clock_mode = clock_mode; + vdata[CS_RAW].clock_mode = clock_mode; /* CLOCK_REALTIME also required for time() */ vdso_ts = &vdata[CS_HRES_COARSE].basetime[CLOCK_REALTIME]; @@ -103,10 +107,10 @@ void update_vsyscall(struct timekeeper *tk) WRITE_ONCE(vdata[CS_HRES_COARSE].hrtimer_res, hrtimer_resolution); /* - * Architectures can opt out of updating the high resolution part - * of the VDSO. + * If the current clocksource is not VDSO capable, then spare the + * update of the high resolution parts. */ - if (__arch_update_vdso_data()) + if (clock_mode != VDSO_CLOCKMODE_NONE) update_vdso_data(vdata, tk); __arch_update_vsyscall(vdata, tk); @@ -125,3 +129,42 @@ 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; + + raw_spin_lock_irqsave(&timekeeper_lock, flags); + 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); + raw_spin_unlock_irqrestore(&timekeeper_lock, flags); +} |