hrtimer: Fix timers queued locally from offline CPUs

hrtimers are migrated away from the dying CPU to any online target at the CPUHP_AP_HRTIMERS_DYING stage in order not to delay bandwidth timers handling tasks involved in the CPU hotplug forward progress.

However wake ups can still be performed by the outgoing CPU after CPUHP_AP_HRTIMERS_DYING. Those can result again in bandwidth timers being armed. Depending on several considerations (crystal ball power management based election, earliest timer already enqueued, timer migration enabled or not), the target may eventually be the current CPU even if offline. If that happens, the timer is eventually ignored.

The most notable example is RCU which had to deal with each an every of those wake-ups by deferring them to an online CPU, along with related workarounds:

_ e787644caf76 (rcu: Defer RCU kthreads wakeup when CPU is dying) _ 9139f93209d1 (rcu/nocb: Fix RT throttling hrtimer armed from offline CPU) _ f7345ccc62a4 (rcu/nocb: Fix rcuog wake-up from offline softirq)

The problem isn't confined to RCU though as the stop machine kthread (which runs CPUHP_AP_HRTIMERS_DYING) reports its completion at the end and performs a wake up that eventually arms the deadline server timer:

           WARNING: CPU: 94 PID: 588 at kernel/time/hrtimer.c:1086 hrtimer_start_range_ns+0x289/0x2d0
           Modules linked in:
           CPU: 94 UID: 0 PID: 588 Comm: migration/94 Not tainted
           Stopper: multi_cpu_stop+0x0/0x120 <- stop_machine_cpuslocked+0x66/0xc0
           RIP: 0010:hrtimer_start_range_ns+0x289/0x2d0
           Code: 41 5c 41 5d 41 5e 41 5f 5d e9 63 94 ea 00 0f 0b 48 83 c4 10 5b 41 5c 41 5d 41 5e 41 5f 5d e9 39 fc 15 01 0f 0b e9 c1 fd ff ff <0f> 0b 48 8b
+45 00 e9 59 ff ff ff f3 0f 1e fa 65 8b 05 1d ec e8 7e
           RSP: 0018:ffffc900019cbcc8 EFLAGS: 00010046
           RAX: ffff88bf449a4c40 RBX: 0000000000000082 RCX: 0000000000000001
           RDX: 0000000000000001 RSI: ffff88bf43224c80 RDI: ffff88bf449a4c40
           RBP: ffff88bf449a4c80 R08: ffff888280970090 R09: 0000000000000000
           R10: ffff88bf432252e0 R11: ffffffff811abf70 R12: ffff88bf449a4c40
           R13: ffff88bf43234b28 R14: ffff88bf43224c80 R15: 0000000000000000
           FS:  0000000000000000(0000) GS:ffff88bf44980000(0000) knlGS:0000000000000000
           CR2: 0000000000000000 CR3: 000000404b230001 CR4: 0000000000770ef0
           PKRU: 55555554
           Call Trace:
            <TASK>
            ? __warn+0xcf/0x1b0
            ? hrtimer_start_range_ns+0x289/0x2d0
            ? report_bug+0x120/0x1a0
            ? handle_bug+0x60/0x90
            ? exc_invalid_op+0x1a/0x50
            ? asm_exc_invalid_op+0x1a/0x20
            ? register_refined_jiffies+0xb0/0xb0
            ? hrtimer_start_range_ns+0x289/0x2d0
            ? hrtimer_start_range_ns+0x186/0x2d0
            start_dl_timer+0xfc/0x150
            enqueue_dl_entity+0x367/0x640
            dl_server_start+0x53/0xa0
            enqueue_task_fair+0x363/0x460
            enqueue_task+0x3c/0x200
            ttwu_do_activate+0x94/0x240
            try_to_wake_up+0x315/0x600
            complete+0x4b/0x80

Instead of providing yet another bandaid to work around the situation, fix it from hrtimers infrastructure instead: always migrate away a timer to an online target whenever it is enqueued from an offline CPU.

This will also allow us to revert all the above RCU disgraceful hacks.

Reported-by Reported-by: Vlad Poenaru <[email protected]>

Reported-by Usama Arif <[email protected]>

Fixes 5c0930ccaad5 ("hrtimers: Push pending hrtimers away from outgoing CPU earlier")

Closes [email protected]

Signed-off-by Frederic Weisbecker <[email protected]>

---
 include/linux/hrtimer_defs.h |  1 +
 kernel/time/hrtimer.c        | 60 +++++++++++++++++++++++++++++++-----
 2 files changed, 54 insertions(+), 7 deletions(-)

include/linux/hrtimer_defs.h +1 -0

index c3b4b7ed7c16..84a5045f80f3 100644
@@ -125,6 +125,7 @@ struct hrtimer_cpu_base {
 	ktime_t				softirq_expires_next;
 	struct hrtimer			*softirq_next_timer;
 	struct hrtimer_clock_base	clock_base[HRTIMER_MAX_CLOCK_BASES];
+	call_single_data_t		csd;
 } ____cacheline_aligned;
 
 

kernel/time/hrtimer.c +47 -7

index 80fe3749d2db..48c0078d2c4f 100644
@@ -58,6 +58,8 @@
 #define HRTIMER_ACTIVE_SOFT	(HRTIMER_ACTIVE_HARD << MASK_SHIFT)
 #define HRTIMER_ACTIVE_ALL	(HRTIMER_ACTIVE_SOFT | HRTIMER_ACTIVE_HARD)
 
+static void retrigger_next_event(void *arg);
+
 /*
  * The timer bases:
  *

@@ -111,7 +113,8 @@ DEFINE_PER_CPU(struct hrtimer_cpu_base, hrtimer_bases) =
 			.clockid = CLOCK_TAI,
 			.get_time = &ktime_get_clocktai,
 		},
-	}
+	},
+	.csd = CSD_INIT(retrigger_next_event, NULL)
 };
 
 static const int hrtimer_clock_to_base_table[MAX_CLOCKS] = {

@@ -716,8 +719,6 @@ static inline int hrtimer_is_hres_enabled(void)
 	return hrtimer_hres_enabled;
 }
 
-static void retrigger_next_event(void *arg);
-
 /*
  * Switch to high resolution mode
  */

@@ -761,6 +762,8 @@ static void retrigger_next_event(void *arg)
 {
 	struct hrtimer_cpu_base *base = this_cpu_ptr(&hrtimer_bases);
 
+	guard(raw_spinlock)(&base->lock);
+
 	/*
 	 * When high resolution mode or nohz is active, then the offsets of
 	 * CLOCK_REALTIME/TAI/BOOTTIME have to be updated. Otherwise the

@@ -773,18 +776,17 @@ static void retrigger_next_event(void *arg)
 	 * 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.
+	 * CPU was in a NOHZ idle sleep. base->lock must still be held to
+	 * to release and synchronize against the csd unlock.
 	 */
 	if (!hrtimer_hres_active(base) && !tick_nohz_active)
 		return;
 
-	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);
 }
 
 /*

@@ -1202,10 +1204,48 @@ hrtimer_update_softirq_timer(struct hrtimer_cpu_base *cpu_base, bool reprogram)
 	hrtimer_reprogram(cpu_base->softirq_next_timer, reprogram);
 }
 
+/*
+ * If the current CPU is offline and timers have been already
+ * migrated away, make sure not to enqueue locally and perform
+ * a remote retrigger as a last resort.
+ */
+static void enqueue_hrtimer_offline(struct hrtimer *timer,
+				    struct hrtimer_clock_base *base,
+				    const enum hrtimer_mode mode)
+{
+	struct hrtimer_cpu_base *new_cpu_base, *old_cpu_base, *this_cpu_base;
+	struct hrtimer_clock_base *new_base;
+	int cpu;
+
+	old_cpu_base = base->cpu_base;
+	this_cpu_base = this_cpu_ptr(&hrtimer_bases);
+
+	if (old_cpu_base == this_cpu_base || !old_cpu_base->online) {
+		WARN_ON_ONCE(hrtimer_callback_running(timer));
+		cpu = cpumask_any_and(cpu_online_mask,
+				      housekeeping_cpumask(HK_TYPE_TIMER));
+		new_cpu_base = &per_cpu(hrtimer_bases, cpu);
+		new_base = &new_cpu_base->clock_base[base->index];
+		WRITE_ONCE(timer->base, &migration_base);
+		raw_spin_unlock(&old_cpu_base->lock);
+		raw_spin_lock(&new_cpu_base->lock);
+		WRITE_ONCE(timer->base, new_base);
+	} else {
+		new_base = base;
+		new_cpu_base = new_base->cpu_base;
+		cpu = new_cpu_base->cpu;
+	}
+
+	if (enqueue_hrtimer(timer, new_base, mode))
+		smp_call_function_single_async(cpu, &new_cpu_base->csd);
+}
+
+
 static int __hrtimer_start_range_ns(struct hrtimer *timer, ktime_t tim,
 				    u64 delta_ns, const enum hrtimer_mode mode,
 				    struct hrtimer_clock_base *base)
 {
+	struct hrtimer_cpu_base *this_cpu_base = this_cpu_ptr(&hrtimer_bases);
 	struct hrtimer_clock_base *new_base;
 	bool force_local, first;
 

@@ -1217,7 +1257,7 @@ static int __hrtimer_start_range_ns(struct hrtimer *timer, ktime_t tim,
 	 * 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 == this_cpu_base;
 	force_local &= base->cpu_base->next_timer == timer;
 
 	/*

kernel/time/hrtimer.c +0 -0

 	/* Switch the timer base, if necessary: */
 	if (!force_local) {
 		new_base = switch_hrtimer_base(timer, base,

kernel/time/hrtimer.c static int __hrtimer_start_range_ns(struct hrtimer *timer, ktime_t tim, · line 1284

Quoted 6 lines

kernel/time/hrtimer.c +2 -0

@@ -1240,6 +1280,12 @@ static int __hrtimer_start_range_ns(struct hrtimer *timer, ktime_t tim,
 
 	hrtimer_set_expires_range_ns(timer, tim, delta_ns);
 
+	if (unlikely(!this_cpu_base->online)) {
+		enqueue_hrtimer_offline(timer, base, mode);

Thanks for the fix!

It looks good to me, maybe as a follow up, we could rename switch_hrtimer_base to enqueue_hrtimer_local? (or maybe something more appropriate)
There are now 2 different paths that switch hrtimer base (enqueue_hrtimer_offline and switch_hrtimer_base).

kernel/time/hrtimer.c static int __hrtimer_start_range_ns(struct hrtimer *timer, ktime_t tim, · line 1288

Quoted 5 lines

kernel/time/hrtimer.c +4 -0

+		return 0;
+	}
+
+

nit: extra new line above.

Quoted 3 lines

kernel/time/hrtimer.c +0 -0

 	/* Switch the timer base, if necessary: */
 	if (!force_local) {
 		new_base = switch_hrtimer_base(timer, base,

5c0930ccaad5 ("hrtimers: Push pending hrtimers away from outgoing CPU earlier") was introduced to fix stalls with scheduler bandwidth timers getting migrated while some kthreads handling CPU hotplug rely on bandwidth.

However this has introduced several other issues which used to be confined to RCU. But not anymore as it is spreading to hotplug code itself (https://lore.kernel.org/all/[email protected]/)

Instead of introducing yet another new hackery, fix the problem in hrtimers for everyone.

Frederic Weisbecker (3):
hrtimers: Force migrate away hrtimers queued after
CPUHP_AP_HRTIMERS_DYING
rcu: Remove swake_up_one_online() bandaid
Revert "rcu/nocb: Fix rcuog wake-up from offline softirq"

 include/linux/hrtimer_defs.h |  1 +
 kernel/rcu/tree.c            | 34 +-------------------
 kernel/rcu/tree_exp.h        |  2 +-
 kernel/rcu/tree_nocb.h       | 10 ++----
 kernel/time/hrtimer.c        | 60 +++++++++++++++++++++++++++++++-----
 5 files changed, 58 insertions(+), 49 deletions(-)

Signature · 1 line

2.46.0

Quoted 10 lines · Frederic Weisbecker

5c0930ccaad5 ("hrtimers: Push pending hrtimers away from outgoing CPU earlier") was introduced to fix stalls with scheduler bandwidth timers getting migrated while some kthreads handling CPU hotplug rely on bandwidth.

However this has introduced several other issues which used to be confined to RCU. But not anymore as it is spreading to hotplug code itself (https://lore.kernel.org/all/[email protected]/)

Instead of introducing yet another new hackery, fix the problem in hrtimers for everyone.

The good news is that this passes 12 hours of 400*TREE03. (Yay!!!)

The so-so news is that this gives only about 70% confidence that these patches help, but on the other hand, it also gives much higher confidence that these patches are not hurting anything.

At least for TREE03.

The not-so-good news is that this series causes build failures for rcutorture scenarios (such as SRCU-T) that build with CONFIG_SMP=n:

------------------------------------------------------------------------
kernel/time/hrtimer.c: In function ‘enqueue_hrtimer_offline’:
kernel/time/hrtimer.c:1229:42: error: ‘migration_base’ undeclared (first use in this function); did you mean ‘is_migration_base’?
------------------------------------------------------------------------

When built with KCSAN enabled (--kcsan to kvm.sh), there is this additional build failure on that same line of code:

------------------------------------------------------------------------
kernel/time/hrtimer.c:1229:3: error: incompatible pointer types assigning to 'volatile typeof (timer->base)' (aka 'struct hrtimer_clock_base *volatile') from 'bool (*)(struct hrtimer_clock_base *)' (aka '_Bool (*)(struct hrtimer_clock_base *)') [-Werror,-Wincompatible-pointer-types]

 1229 |                 WRITE_ONCE(timer->base, &migration_base);
------------------------------------------------------------------------

Me, I am a bit surprised that enqueue_hrtimer_offline() is even built in a CONFIG_SMP=n kernel.

But there might be some reason why #ifdef-ing out that function's body would be a bad idea, so over to you! ;-)

Quoted 18 lines · Paul E. McKenney

Quoted 10 lines · Frederic Weisbecker

5c0930ccaad5 ("hrtimers: Push pending hrtimers away from outgoing CPU earlier") was introduced to fix stalls with scheduler bandwidth timers getting migrated while some kthreads handling CPU hotplug rely on bandwidth.

However this has introduced several other issues which used to be confined to RCU. But not anymore as it is spreading to hotplug code itself (https://lore.kernel.org/all/[email protected]/)

Instead of introducing yet another new hackery, fix the problem in hrtimers for everyone.

The good news is that this passes 12 hours of 400*TREE03. (Yay!!!)

The so-so news is that this gives only about 70% confidence that these patches help, but on the other hand, it also gives much higher confidence that these patches are not hurting anything.

Quoted 1 line

At least for TREE03.

Another day, another 12 hours of 400*TREE03 passed. This gets us up beyond 90% confidence that these patches help, and even more confidence that they are not too severely hurting anything.

Quoted 15 lines

The not-so-good news is that this series causes build failures for rcutorture scenarios (such as SRCU-T) that build with CONFIG_SMP=n:

------------------------------------------------------------------------
kernel/time/hrtimer.c: In function ‘enqueue_hrtimer_offline’:
kernel/time/hrtimer.c:1229:42: error: ‘migration_base’ undeclared (first use in this function); did you mean ‘is_migration_base’?
------------------------------------------------------------------------

When built with KCSAN enabled (--kcsan to kvm.sh), there is this additional build failure on that same line of code:

------------------------------------------------------------------------
kernel/time/hrtimer.c:1229:3: error: incompatible pointer types assigning to 'volatile typeof (timer->base)' (aka 'struct hrtimer_clock_base *volatile') from 'bool (*)(struct hrtimer_clock_base *)' (aka '_Bool (*)(struct hrtimer_clock_base *)') [-Werror,-Wincompatible-pointer-types]

Quoted 8 lines

 1229 |                 WRITE_ONCE(timer->base, &migration_base);
------------------------------------------------------------------------

Me, I am a bit surprised that enqueue_hrtimer_offline() is even built in a CONFIG_SMP=n kernel.

But there might be some reason why #ifdef-ing out that function's body would be a bad idea, so over to you! ;-)

Curiosity overcame me, and the #ifdef makes things at least appear to work for CONFIG_SMP=n kernels. Experimental patch below, which I intend to use for further testing, but I have no plans to push it upstream.

Thanx, Paul

------------------------------------------------------------------------

commit 61a0b92b3a3cfef69e3848806e51d1b99a9e9406

Author: Paul E. McKenney <[email protected]>
Date:   Fri Dec 20 15:10:26 2024 -0800

EXP hrtimers: No-op enqueue_hrtimer_offline() if !HOTPLUG_CPU

In kernels built with CONFIG_HOTPLUG_CPU=n, this experimental commit uses #ifdef to remove the body of the enqueue_hrtimer_offline() function. This might or might not be the correct fix to the build complaint about migration_base being undeclared.

Signed-off-by Paul E. McKenney <[email protected]>

Cc Vlad Poenaru <[email protected]>

Cc Usama Arif <[email protected]>

Cc Frederic Weisbecker <[email protected]>

kernel/time/hrtimer.c +1 -0

index 48c0078d2c4f2..4235b7825b152 100644
@@ -1213,6 +1213,7 @@ static void enqueue_hrtimer_offline(struct hrtimer *timer,
 				    struct hrtimer_clock_base *base,
 				    const enum hrtimer_mode mode)
 {
+#ifdef CONFIG_HOTPLUG_CPU
 	struct hrtimer_cpu_base *new_cpu_base, *old_cpu_base, *this_cpu_base;
 	struct hrtimer_clock_base *new_base;
 	int cpu;

Quoted 19 lines · Paul E. McKenney

Quoted 18 lines · Paul E. McKenney

Quoted 10 lines · Frederic Weisbecker

5c0930ccaad5 ("hrtimers: Push pending hrtimers away from outgoing CPU earlier") was introduced to fix stalls with scheduler bandwidth timers getting migrated while some kthreads handling CPU hotplug rely on bandwidth.

However this has introduced several other issues which used to be confined to RCU. But not anymore as it is spreading to hotplug code itself (https://lore.kernel.org/all/[email protected]/)

Instead of introducing yet another new hackery, fix the problem in hrtimers for everyone.

The good news is that this passes 12 hours of 400*TREE03. (Yay!!!)

The so-so news is that this gives only about 70% confidence that these patches help, but on the other hand, it also gives much higher confidence that these patches are not hurting anything.

Quoted 5 lines

Quoted 1 line

At least for TREE03.

Another day, another 12 hours of 400*TREE03 passed. This gets us up beyond 90% confidence that these patches help, and even more confidence that they are not too severely hurting anything.

Cool :-)

Quoted 62 lines

Quoted 15 lines

The not-so-good news is that this series causes build failures for rcutorture scenarios (such as SRCU-T) that build with CONFIG_SMP=n:

------------------------------------------------------------------------
kernel/time/hrtimer.c: In function ‘enqueue_hrtimer_offline’:
kernel/time/hrtimer.c:1229:42: error: ‘migration_base’ undeclared (first use in this function); did you mean ‘is_migration_base’?
------------------------------------------------------------------------

When built with KCSAN enabled (--kcsan to kvm.sh), there is this additional build failure on that same line of code:

------------------------------------------------------------------------
kernel/time/hrtimer.c:1229:3: error: incompatible pointer types assigning to 'volatile typeof (timer->base)' (aka 'struct hrtimer_clock_base *volatile') from 'bool (*)(struct hrtimer_clock_base *)' (aka '_Bool (*)(struct hrtimer_clock_base *)') [-Werror,-Wincompatible-pointer-types]

Quoted 8 lines

 1229 |                 WRITE_ONCE(timer->base, &migration_base);
------------------------------------------------------------------------

Me, I am a bit surprised that enqueue_hrtimer_offline() is even built in a CONFIG_SMP=n kernel.

But there might be some reason why #ifdef-ing out that function's body would be a bad idea, so over to you! ;-)

Curiosity overcame me, and the #ifdef makes things at least appear to work for CONFIG_SMP=n kernels. Experimental patch below, which I intend to use for further testing, but I have no plans to push it upstream.

Thanx, Paul

------------------------------------------------------------------------

commit 61a0b92b3a3cfef69e3848806e51d1b99a9e9406

Author: Paul E. McKenney <[email protected]>
Date:   Fri Dec 20 15:10:26 2024 -0800

EXP hrtimers: No-op enqueue_hrtimer_offline() if !HOTPLUG_CPU

In kernels built with CONFIG_HOTPLUG_CPU=n, this experimental commit uses #ifdef to remove the body of the enqueue_hrtimer_offline() function. This might or might not be the correct fix to the build complaint about migration_base being undeclared.

Signed-off-by Paul E. McKenney <[email protected]>

Cc Vlad Poenaru <[email protected]>

Cc Usama Arif <[email protected]>

Cc Frederic Weisbecker <[email protected]>

kernel/time/hrtimer.c +1 -0

index 48c0078d2c4f2..4235b7825b152 100644
@@ -1213,6 +1213,7 @@ static void enqueue_hrtimer_offline(struct hrtimer *timer,
 				    struct hrtimer_clock_base *base,
 				    const enum hrtimer_mode mode)
 {
+#ifdef CONFIG_HOTPLUG_CPU
 	struct hrtimer_cpu_base *new_cpu_base, *old_cpu_base, *this_cpu_base;
 	struct hrtimer_clock_base *new_base;
 	int cpu;

@@ -1238,6 +1239,7 @@ static void enqueue_hrtimer_offline(struct hrtimer *timer,
 

kernel/time/hrtimer.c static void enqueue_hrtimer_offline(struct hrtimer *timer, · line 1243

Quoted 4 lines

kernel/time/hrtimer.c +1 -0

 	if (enqueue_hrtimer(timer, new_base, mode))
 		smp_call_function_single_async(cpu, &new_cpu_base->csd);
+#endif
 }

I was thinking about something like that. Can I fold that and add your SOB?