Message-ID: <aMlo_K8koZ6HH2kh@localhost.localdomain>
Date: Tue, 16 Sep 2025 15:41:16 +0200
From: Frederic Weisbecker <frederic@...nel.org>
To: Gabriele Monaco <gmonaco@...hat.com>
Cc: linux-kernel@...r.kernel.org,
	Anna-Maria Behnsen <anna-maria@...utronix.de>,
	Thomas Gleixner <tglx@...utronix.de>,
	Waiman Long <longman@...hat.com>, Tejun Heo <tj@...nel.org>,
	Johannes Weiner <hannes@...xchg.org>,
	Michal Koutný <mkoutny@...e.com>,
	cgroups@...r.kernel.org, "John B. Wyatt IV" <jwyatt@...hat.com>,
	"John B. Wyatt IV" <sageofredondo@...il.com>
Subject: Re: [PATCH v12 9/9] timers: Exclude isolated cpus from timer
 migration

Le Mon, Sep 15, 2025 at 04:59:30PM +0200, Gabriele Monaco a écrit :
> Tested-by: John B. Wyatt IV <jwyatt@...hat.com>
> Tested-by: John B. Wyatt IV <sageofredondo@...il.com>

Two people, one tester? :-)

> +/**
> + * tmigr_isolated_exclude_cpumask - Exclude given CPUs from hierarchy
> + * @exclude_cpumask: the cpumask to be excluded from timer migration hierarchy
> + *
> + * This function can be called from cpuset code to provide the new set of
> + * isolated CPUs that should be excluded from the hierarchy.
> + * Online CPUs not present in exclude_cpumask but already excluded are brought
> + * back to the hierarchy.
> + * Functions to isolate/unisolate need to be called locally and can sleep.
> + */
> +int tmigr_isolated_exclude_cpumask(struct cpumask *exclude_cpumask)
> +{
> +	struct work_struct __percpu *works __free(free_percpu) =
> +		alloc_percpu(struct work_struct);
> +	cpumask_var_t cpumask_unisol __free(free_cpumask_var) = CPUMASK_NULL;
> +	cpumask_var_t cpumask_isol __free(free_cpumask_var) = CPUMASK_NULL;
> +	int cpu;
> +
> +	lockdep_assert_cpus_held();
> +
> +	if (!alloc_cpumask_var(&cpumask_isol, GFP_KERNEL))
> +		return -ENOMEM;
> +	if (!alloc_cpumask_var(&cpumask_unisol, GFP_KERNEL))
> +		return -ENOMEM;
> +	if (!works)
> +		return -ENOMEM;
> +
> +	cpumask_andnot(cpumask_unisol, cpu_online_mask, exclude_cpumask);
> +	cpumask_andnot(cpumask_unisol, cpumask_unisol, tmigr_available_cpumask);
> +	/* Set up the mask earlier to avoid races with the migrator CPU */
> +	cpumask_or(tmigr_available_cpumask, tmigr_available_cpumask, cpumask_unisol);
> +	for_each_cpu(cpu, cpumask_unisol) {
> +		struct work_struct *work = per_cpu_ptr(works, cpu);
> +
> +		INIT_WORK(work, tmigr_cpu_unisolate);
> +		schedule_work_on(cpu, work);
> +	}
> +
> +	cpumask_and(cpumask_isol, exclude_cpumask, tmigr_available_cpumask);
> +	cpumask_and(cpumask_isol, cpumask_isol, housekeeping_cpumask(HK_TYPE_KERNEL_NOISE));
> +	/*
> +	 * Handle this here and not in the cpuset code because exclude_cpumask
> +	 * might include also the tick CPU if included in isolcpus.
> +	 */
> +	for_each_cpu(cpu, cpumask_isol) {
> +		if (!tick_nohz_cpu_hotpluggable(cpu)) {
> +			cpumask_clear_cpu(cpu, cpumask_isol);
> +			break;
> +		}
> +	}
> +	/* Set up the mask earlier to avoid races with the migrator CPU */
> +	cpumask_andnot(tmigr_available_cpumask, tmigr_available_cpumask, cpumask_isol);
> +	for_each_cpu(cpu, cpumask_isol) {
> +		struct work_struct *work = per_cpu_ptr(works, cpu);
> +
> +		INIT_WORK(work, tmigr_cpu_isolate);
> +		schedule_work_on(cpu, work);
> +	}

This is racy at various levels:

* The tmigr_available_cpumask clear can race with the cpumask_set_cpu() in
  tmigr_cpu_unisolate(), risking overwrites when CPUs are on the same bitmap
  chunk (bitmap operations aren't atomic).

* tmigr_cpu_isolate() and tmigr_cpu_unisolate() can now run concurrently and
  then cpumask_set_cpu() can race with cpumask_clear_cpu() on
  tmigr_available_cpumask, risking overwrites, though the real problem is
  on the precedent point.
  
* tmigr_cpu_isolate() can race with tmigr_cpu_isolate() on other CPUs so
  the calls to cpumask_clear_cpu() can race. That's fine because
  tmigr_available_cpumask is already set to those CPUs but that still
  leaves un uncomfortable taste. That would leave an excuse for KSCAN to warn
  for example.

* Similar with tmigr_cpu_unisolate() racing together.

So, something like that should be added?

diff --git a/kernel/time/timer_migration.c b/kernel/time/timer_migration.c
index 08e29fc01479..6615e56c8b0d 100644
--- a/kernel/time/timer_migration.c
+++ b/kernel/time/timer_migration.c
@@ -1473,7 +1473,6 @@ static int tmigr_clear_cpu_available(unsigned int cpu)
 	int migrator;
 	u64 firstexp;
 
-	cpumask_clear_cpu(cpu, tmigr_available_cpumask);
 	scoped_guard(raw_spinlock_irq, &tmc->lock) {
 		if (!tmc->available)
 			return 0;
@@ -1489,11 +1488,11 @@ static int tmigr_clear_cpu_available(unsigned int cpu)
 	}
 
 	if (firstexp != KTIME_MAX) {
-		migrator = cpumask_any(tmigr_available_cpumask);
+		migrator = cpumask_any_but(tmigr_available_cpumask, cpu);
 		work_on_cpu(migrator, tmigr_trigger_active, NULL);
 	}
 
-	return 0;
+	return 1;
 }
 
 static int tmigr_set_cpu_available(unsigned int cpu)
@@ -1506,7 +1505,7 @@ static int tmigr_set_cpu_available(unsigned int cpu)
 
 	if (tmigr_is_isolated(cpu))
 		return 0;
-	cpumask_set_cpu(cpu, tmigr_available_cpumask);
+
 	scoped_guard(raw_spinlock_irq, &tmc->lock) {
 		if (tmc->available)
 			return 0;
@@ -1516,7 +1515,19 @@ static int tmigr_set_cpu_available(unsigned int cpu)
 			__tmigr_cpu_activate(tmc);
 		tmc->available = true;
 	}
-	return 0;
+	return 1;
+}
+
+static int tmigr_online_cpu(unsigned int cpu)
+{
+	if (tmigr_set_cpu_available(cpu) > 0)
+		cpumask_set_cpu(cpu, tmigr_available_cpumask);
+}
+
+static int tmigr_offline_cpu(unsigned int cpu)
+{
+	if (tmigr_clear_cpu_available(cpu) > 0)
+		cpumask_clear_cpu(cpu, tmigr_available_cpumask);
 }
 
 static void tmigr_cpu_isolate(struct work_struct *ignored)
@@ -1601,8 +1612,7 @@ int tmigr_isolated_exclude_cpumask(struct cpumask *exclude_cpumask)
 static int __init tmigr_late_init(void)
 {
 	return cpuhp_setup_state(CPUHP_AP_TMIGR_ONLINE, "tmigr:online",
-				 tmigr_set_cpu_available,
-				 tmigr_clear_cpu_available);
+				 tmigr_online_cpu, tmigr_offline_cpu);
 }
 
 static void tmigr_init_group(struct tmigr_group *group, unsigned int lvl,

-- 
Frederic Weisbecker
SUSE Labs

Powered by blists - more mailing lists