diff mbox series

[2/2] timers: Fix get_next_timer_interrupt() with no timers pending

Message ID 20210715104218.81276-3-frederic@kernel.org
State Accepted
Commit aebacb7f6ca1926918734faae14d1f0b6fae5cb7
Headers show
Series [1/2] posix-cpu-timers: Fix rearm racing against process tick | expand

Commit Message

Frederic Weisbecker July 15, 2021, 10:42 a.m. UTC 
From: Nicolas Saenz Julienne <nsaenzju@redhat.com>

31cd0e119d50 ("timers: Recalculate next timer interrupt only when
necessary") subtly altered get_next_timer_interrupt()'s behaviour. The
function no longer consistently returns KTIME_MAX with no timers
pending.

In order to decide if there are any timers pending we check whether the
next expiry will happen NEXT_TIMER_MAX_DELTA jiffies from now.
Unfortunately, the next expiry time and the timer base clock are no
longer updated in unison. The former changes upon certain timer
operations (enqueue, expire, detach), whereas the latter keeps track of
jiffies as they move forward. Ultimately breaking the logic above.

A simplified example:

- Upon entering get_next_timer_interrupt() with:

	jiffies = 1
	base->clk = 0;
	base->next_expiry = NEXT_TIMER_MAX_DELTA;

  'base->next_expiry == base->clk + NEXT_TIMER_MAX_DELTA', the function
  returns KTIME_MAX.

- 'base->clk' is updated to the jiffies value.

- The next time we enter get_next_timer_interrupt(), taking into account
  no timer operations happened:

	base->clk = 1;
	base->next_expiry = NEXT_TIMER_MAX_DELTA;

  'base->next_expiry != base->clk + NEXT_TIMER_MAX_DELTA', the function
  returns a valid expire time, which is incorrect.

This ultimately might unnecessarily rearm sched's timer on nohz_full
setups, and add latency to the system[1].

So, introduce 'base->timers_pending'[2], update it every time
'base->next_expiry' changes, and use it in get_next_timer_interrupt().

[1] See tick_nohz_stop_tick().
[2] A quick pahole check on x86_64 and arm64 shows it doesn't make
    'struct timer_base' any bigger.

Fixes: 31cd0e119d50 ("timers: Recalculate next timer interrupt only when necessary")
Signed-off-by: Nicolas Saenz Julienne <nsaenzju@redhat.com>
Signed-off-by: Frederic Weisbecker <frederic@kernel.org>
---
 kernel/time/timer.c | 8 +++++---
 1 file changed, 5 insertions(+), 3 deletions(-)
diff mbox series

Patch

diff --git a/kernel/time/timer.c b/kernel/time/timer.c
index 3fadb58fc9d7..9eb11c2209e5 100644
--- a/kernel/time/timer.c
+++ b/kernel/time/timer.c
@@ -207,6 +207,7 @@  struct timer_base {
 	unsigned int		cpu;
 	bool			next_expiry_recalc;
 	bool			is_idle;
+	bool			timers_pending;
 	DECLARE_BITMAP(pending_map, WHEEL_SIZE);
 	struct hlist_head	vectors[WHEEL_SIZE];
 } ____cacheline_aligned;
@@ -595,6 +596,7 @@  static void enqueue_timer(struct timer_base *base, struct timer_list *timer,
 		 * can reevaluate the wheel:
 		 */
 		base->next_expiry = bucket_expiry;
+		base->timers_pending = true;
 		base->next_expiry_recalc = false;
 		trigger_dyntick_cpu(base, timer);
 	}
@@ -1582,6 +1584,7 @@  static unsigned long __next_timer_interrupt(struct timer_base *base)
 	}
 
 	base->next_expiry_recalc = false;
+	base->timers_pending = !(next == base->clk + NEXT_TIMER_MAX_DELTA);
 
 	return next;
 }
@@ -1633,7 +1636,6 @@  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.
@@ -1646,7 +1648,6 @@  u64 get_next_timer_interrupt(unsigned long basej, u64 basem)
 	if (base->next_expiry_recalc)
 		base->next_expiry = __next_timer_interrupt(base);
 	nextevt = base->next_expiry;
-	is_max_delta = (nextevt == base->clk + NEXT_TIMER_MAX_DELTA);
 
 	/*
 	 * We have a fresh next event. Check whether we can forward the
@@ -1664,7 +1665,7 @@  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.
@@ -1947,6 +1948,7 @@  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->timers_pending = false;
 		base->is_idle = false;
 	}
 	return 0;