WSL2-Linux-Kernel/kernel/time
Frederic Weisbecker c597d8cb0d timers/nohz: Last resort update jiffies on nohz_full IRQ entry
[ Upstream commit 53e87e3cdc ]

When at least one CPU runs in nohz_full mode, a dedicated timekeeper CPU
is guaranteed to stay online and to never stop its tick.

Meanwhile on some rare case, the dedicated timekeeper may be running
with interrupts disabled for a while, such as in stop_machine.

If jiffies stop being updated, a nohz_full CPU may end up endlessly
programming the next tick in the past, taking the last jiffies update
monotonic timestamp as a stale base, resulting in an tick storm.

Here is a scenario where it matters:

0) CPU 0 is the timekeeper and CPU 1 a nohz_full CPU.

1) A stop machine callback is queued to execute somewhere.

2) CPU 0 reaches MULTI_STOP_DISABLE_IRQ while CPU 1 is still in
   MULTI_STOP_PREPARE. Hence CPU 0 can't do its timekeeping duty. CPU 1
   can still take IRQs.

3) CPU 1 receives an IRQ which queues a timer callback one jiffy forward.

4) On IRQ exit, CPU 1 schedules the tick one jiffy forward, taking
   last_jiffies_update as a base. But last_jiffies_update hasn't been
   updated for 2 jiffies since the timekeeper has interrupts disabled.

5) clockevents_program_event(), which relies on ktime_get(), observes
   that the expiration is in the past and therefore programs the min
   delta event on the clock.

6) The tick fires immediately, goto 3)

7) Tick storm, the nohz_full CPU is drown and takes ages to reach
   MULTI_STOP_DISABLE_IRQ, which is the only way out of this situation.

Solve this with unconditionally updating jiffies if the value is stale
on nohz_full IRQ entry. IRQs and other disturbances are expected to be
rare enough on nohz_full for the unconditional call to ktime_get() to
actually matter.

Reported-by: Paul E. McKenney <paulmck@kernel.org>
Signed-off-by: Frederic Weisbecker <frederic@kernel.org>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Tested-by: Paul E. McKenney <paulmck@kernel.org>
Link: https://lore.kernel.org/r/20211026141055.57358-2-frederic@kernel.org
Signed-off-by: Joel Fernandes (Google) <joel@joelfernandes.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2023-08-16 18:22:04 +02:00
..
Kconfig -----BEGIN PGP SIGNATURE----- 2021-06-29 12:31:16 -07:00
Makefile time: Improve performance of time64_to_tm() 2021-06-24 11:51:59 +02:00
alarmtimer.c alarmtimer: Prevent starvation by small intervals and SIG_IGN 2023-02-22 12:57:11 +01:00
clockevents.c clockevents: Use list_move() instead of list_del()/list_add() 2021-06-22 17:16:46 +02:00
clocksource-wdtest.c clocksource: Make clocksource watchdog test safe for slow-HZ systems 2021-08-28 17:01:32 +02:00
clocksource.c clocksource: Suspend the watchdog temporarily when high read latency detected 2023-03-10 09:39:49 +01:00
hrtimer.c timers: Prevent union confusion from unexpected restart_syscall() 2023-03-10 09:39:48 +01:00
itimer.c time: Prevent undefined behaviour in timespec64_to_ns() 2020-10-26 11:48:11 +01:00
jiffies.c clocksource: Make clocksource watchdog test safe for slow-HZ systems 2021-08-28 17:01:32 +02:00
namespace.c memcg: enable accounting for new namesapces and struct nsproxy 2021-09-03 09:58:12 -07:00
ntp.c timekeeping, clocksource: Fix various typos in comments 2021-03-22 23:06:48 +01:00
ntp_internal.h ntp: Make the RTC synchronization more reliable 2020-12-11 10:40:52 +01:00
posix-clock.c posix-clocks: Rename the clock_get() callback to clock_get_timespec() 2020-01-14 12:20:49 +01:00
posix-cpu-timers.c posix-cpu-timers: Implement the missing timer_wait_running callback 2023-05-11 23:00:16 +09:00
posix-stubs.c timers: Prevent union confusion from unexpected restart_syscall() 2023-03-10 09:39:48 +01:00
posix-timers.c posix-timers: Prevent RT livelock in itimer_delete() 2023-07-23 13:46:45 +02:00
posix-timers.h posix-clocks: Introduce clock_get_ktime() callback 2020-01-14 12:20:51 +01:00
sched_clock.c time/sched_clock: Mark sched_clock_read_begin/retry() as notrace 2020-10-26 11:34:31 +01:00
test_udelay.c time/debug: Fix memory leak with using debugfs_lookup() 2023-03-10 09:39:51 +01:00
tick-broadcast-hrtimer.c timekeeping, clocksource: Fix various typos in comments 2021-03-22 23:06:48 +01:00
tick-broadcast.c tick/broadcast: Make broadcast device replacement work correctly 2023-05-24 17:36:41 +01:00
tick-common.c tick/common: Align tick period during sched_timer setup 2023-06-28 10:29:41 +02:00
tick-internal.h clocksource: Make clocksource watchdog test safe for slow-HZ systems 2021-08-28 17:01:32 +02:00
tick-legacy.c timekeeping: remove xtime_update 2020-10-30 21:57:07 +01:00
tick-oneshot.c timekeeping, clocksource: Fix various typos in comments 2021-03-22 23:06:48 +01:00
tick-sched.c timers/nohz: Last resort update jiffies on nohz_full IRQ entry 2023-08-16 18:22:04 +02:00
tick-sched.h tick: Detect and fix jiffies update stall 2023-08-16 18:22:04 +02:00
time.c timekeeping, clocksource: Fix various typos in comments 2021-03-22 23:06:48 +01:00
time_test.c time/kunit: Add missing MODULE_LICENSE() 2021-06-28 07:40:23 +02:00
timeconst.bc time: Add SPDX license identifiers 2018-11-23 11:51:20 +01:00
timeconv.c time: Improve performance of time64_to_tm() 2021-06-24 11:51:59 +02:00
timecounter.c time/timecounter: Mark 1st argument of timecounter_cyc2time() as const 2021-04-16 21:03:50 +02:00
timekeeping.c timekeeping: Fix references to nonexistent ktime_get_fast_ns() 2023-05-11 23:00:37 +09:00
timekeeping.h asm-generic: cross-architecture timer cleanup 2020-12-16 00:07:17 -08:00
timekeeping_debug.c timekeeping/debug: No need to check return value of debugfs_create functions 2019-01-29 20:08:41 +01:00
timekeeping_internal.h timekeeping/vsyscall: Provide vdso_update_begin/end() 2020-08-06 10:57:30 +02:00
timer.c timers: Fix warning condition in __run_timers() 2022-04-20 09:34:21 +02:00
timer_list.c timer_list: Print name of per-cpu wakeup device 2021-05-31 17:04:49 +02:00
vsyscall.c timekeeping, clocksource: Fix various typos in comments 2021-03-22 23:06:48 +01:00