During CPU-down hotplug, hrtimers may migrate to isolated CPUs,
compromising CPU isolation. This commit addresses this issue by
masking valid CPUs for hrtimers using housekeeping_cpumask(HK_TYPE_TIMER).

Suggested-by: Waiman Long <longman@redhat.com>
Signed-off-by: Costa Shulyupin <costa.shul@redhat.com>
Reviewed-by: Waiman Long <longman@redhat.com>
Reviewed-by: Frederic Weisbecker <frederic@kernel.org>
---

Changes in v3:
- fixed compilation errors
Changes in v2:
- [v1] https://lore.kernel.org/all/20240211135213.2518068-1-costa.shul@redhat.com/
- reworded and rebased on next

---
 kernel/time/hrtimer.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/kernel/time/hrtimer.c b/kernel/time/hrtimer.c
index edb0f821dcea..6057fe2e179b 100644
--- a/kernel/time/hrtimer.c
+++ b/kernel/time/hrtimer.c
@@ -38,6 +38,7 @@
 #include <linux/sched/deadline.h>
 #include <linux/sched/nohz.h>
 #include <linux/sched/debug.h>
+#include <linux/sched/isolation.h>
 #include <linux/timer.h>
 #include <linux/freezer.h>
 #include <linux/compat.h>
@@ -2223,8 +2224,8 @@ static void migrate_hrtimer_list(struct hrtimer_clock_base *old_base,
 
 int hrtimers_cpu_dying(unsigned int dying_cpu)
 {
+	int i, ncpu = cpumask_any_and(cpu_active_mask, housekeeping_cpumask(HK_TYPE_TIMER));
 	struct hrtimer_cpu_base *old_base, *new_base;
-	int i, ncpu = cpumask_first(cpu_active_mask);
 
 	tick_cancel_sched_timer(dying_cpu);
 
-- 
2.43.0

Costa!

On Mon, Feb 19 2024 at 16:33, Costa Shulyupin wrote:
> During CPU-down hotplug, hrtimers may migrate to isolated CPUs,
> compromising CPU isolation. This commit addresses this issue by
> masking valid CPUs for hrtimers using housekeeping_cpumask(HK_TYPE_TIMER).

Review feedback is meant to be addressed. I fixed it up for you in V2
and told you what needs to be changed and you got the fixed up version
in your inbox. Feel free to ignore me, but don't expect me to mop up
your stuff.

Thanks,

        tglx

During CPU-down hotplug, hrtimers may migrate to isolated CPUs,
compromising CPU isolation.

Address this issue by masking valid CPUs for hrtimers using
housekeeping_cpumask(HK_TYPE_TIMER).

Suggested-by: Waiman Long <longman@redhat.com>
Signed-off-by: Costa Shulyupin <costa.shul@redhat.com>
Reviewed-by: Waiman Long <longman@redhat.com>
---

Changes in v4:
- fixed description
Changes in v3:
- fixed compilation errors
Changes in v2:
- [v1] https://lore.kernel.org/all/20240211135213.2518068-1-costa.shul@redhat.com/
- reworded and rebased on next

---
 kernel/time/hrtimer.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/kernel/time/hrtimer.c b/kernel/time/hrtimer.c
index 5a98b35b0576..1fd106af747d 100644
--- a/kernel/time/hrtimer.c
+++ b/kernel/time/hrtimer.c
@@ -38,6 +38,7 @@
 #include <linux/sched/deadline.h>
 #include <linux/sched/nohz.h>
 #include <linux/sched/debug.h>
+#include <linux/sched/isolation.h>
 #include <linux/timer.h>
 #include <linux/freezer.h>
 #include <linux/compat.h>
@@ -2225,8 +2226,8 @@ static void migrate_hrtimer_list(struct hrtimer_clock_base *old_base,
 
 int hrtimers_cpu_dying(unsigned int dying_cpu)
 {
+	int i, ncpu = cpumask_any_and(cpu_active_mask, housekeeping_cpumask(HK_TYPE_TIMER));
 	struct hrtimer_cpu_base *old_base, *new_base;
-	int i, ncpu = cpumask_first(cpu_active_mask);
 
 	tick_cancel_sched_timer(dying_cpu);
 
-- 
2.43.0

The following commit has been merged into the timers/core branch of tip:

Commit-ID:     56c2cb10120894be40c40a9bf0ce798da14c50f6
Gitweb:        https://git.kernel.org/tip/56c2cb10120894be40c40a9bf0ce798da14c50f6
Author:        Costa Shulyupin <costa.shul@redhat.com>
AuthorDate:    Thu, 22 Feb 2024 22:08:56 +02:00
Committer:     Thomas Gleixner <tglx@linutronix.de>
CommitterDate: Thu, 22 Feb 2024 22:18:21 +01:00

hrtimer: Select housekeeping CPU during migration

During CPU-down hotplug, hrtimers may migrate to isolated CPUs,
compromising CPU isolation.

Address this issue by masking valid CPUs for hrtimers using
housekeeping_cpumask(HK_TYPE_TIMER).

Suggested-by: Waiman Long <longman@redhat.com>
Signed-off-by: Costa Shulyupin <costa.shul@redhat.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Reviewed-by: Waiman Long <longman@redhat.com>
Link: https://lore.kernel.org/r/20240222200856.569036-1-costa.shul@redhat.com
---
 kernel/time/hrtimer.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/kernel/time/hrtimer.c b/kernel/time/hrtimer.c
index 5a98b35..1fd106a 100644
--- a/kernel/time/hrtimer.c
+++ b/kernel/time/hrtimer.c
@@ -38,6 +38,7 @@
 #include <linux/sched/deadline.h>
 #include <linux/sched/nohz.h>
 #include <linux/sched/debug.h>
+#include <linux/sched/isolation.h>
 #include <linux/timer.h>
 #include <linux/freezer.h>
 #include <linux/compat.h>
@@ -2225,8 +2226,8 @@ static void migrate_hrtimer_list(struct hrtimer_clock_base *old_base,
 
 int hrtimers_cpu_dying(unsigned int dying_cpu)
 {
+	int i, ncpu = cpumask_any_and(cpu_active_mask, housekeeping_cpumask(HK_TYPE_TIMER));
 	struct hrtimer_cpu_base *old_base, *new_base;
-	int i, ncpu = cpumask_first(cpu_active_mask);
 
 	tick_cancel_sched_timer(dying_cpu);