linux-kernel - [PATCH] sched_ext: idle: Fix cpu_released while RT task is scheduled on an idle core

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives
Hash Suite: Windows password security audit tool. GUI, reports in PDF.
[<prev] [next>] [day] [month] [year] [list]
Message-ID: <ff3aa51349414f169cd550f25b245c38@honor.com>
Date: Tue, 10 Jun 2025 07:51:17 +0000
From: liuwenfang <liuwenfang@...or.com>
To: 'Tejun Heo' <tj@...nel.org>
CC: 'David Vernet' <void@...ifault.com>, 'Andrea Righi' <arighi@...dia.com>,
	'Changwoo Min' <changwoo@...lia.com>, 'Ingo Molnar' <mingo@...hat.com>,
	'Peter Zijlstra' <peterz@...radead.org>, 'Juri Lelli'
	<juri.lelli@...hat.com>, 'Vincent Guittot' <vincent.guittot@...aro.org>,
	'Dietmar Eggemann' <dietmar.eggemann@....com>, 'Steven Rostedt'
	<rostedt@...dmis.org>, 'Ben Segall' <bsegall@...gle.com>, 'Mel Gorman'
	<mgorman@...e.de>, 'Valentin Schneider' <vschneid@...hat.com>,
	"'linux-kernel@...r.kernel.org'" <linux-kernel@...r.kernel.org>
Subject: [PATCH] sched_ext: idle: Fix cpu_released while RT task is scheduled
 on an idle core

Assume task RT1 and RT2 have RT prio, one cpu has scheduled task RT1,
task idle, and task RT2 in order, and rq->scx.cpu_released was true while
task RT1 was running. then rq->scx.cpu_released was changed from true to
false while task RT1 was scheduled out and idle was scheduled in.
But rq->scx.cpu_released could not be changed to true while task RT2 was
scheduled in later.

The sched_class of next task can be observed while update_idle and the
state of rq->scx.cpu_released can be changed properly.

Signed-off-by: liuwenfang liuwenfang@...or.com 
---
 kernel/sched/ext.c      |  2 +-
 kernel/sched/ext.h      | 11 +++++++----
 kernel/sched/ext_idle.c |  6 +++++-
 kernel/sched/ext_idle.h |  1 +
 kernel/sched/idle.c     |  6 +++---
 5 files changed, 17 insertions(+), 9 deletions(-)

diff --git a/kernel/sched/ext.c b/kernel/sched/ext.c
index f5133249f..6bbea0ea1 100644
--- a/kernel/sched/ext.c
+++ b/kernel/sched/ext.c
@@ -3187,7 +3187,7 @@ preempt_reason_from_class(const struct sched_class *class)
 	return SCX_CPU_PREEMPT_UNKNOWN;
 }
 
-static void switch_class(struct rq *rq, struct task_struct *next)
+void switch_class(struct rq *rq, struct task_struct *next)
 {
 	const struct sched_class *next_class = next->sched_class;
 
diff --git a/kernel/sched/ext.h b/kernel/sched/ext.h
index 1bda96b19..eddfcc98e 100644
--- a/kernel/sched/ext.h
+++ b/kernel/sched/ext.h
@@ -67,15 +67,18 @@ static inline void init_sched_ext_class(void) {}
 #endif	/* CONFIG_SCHED_CLASS_EXT */
 
 #if defined(CONFIG_SCHED_CLASS_EXT) && defined(CONFIG_SMP)
-void __scx_update_idle(struct rq *rq, bool idle, bool do_notify);
+void __scx_update_idle(struct rq *rq, struct task_struct *next,
+		       bool idle, bool do_notify);
 
-static inline void scx_update_idle(struct rq *rq, bool idle, bool do_notify)
+static inline void scx_update_idle(struct rq *rq, struct task_struct *next,
+				   bool idle, bool do_notify)
 {
 	if (scx_enabled())
-		__scx_update_idle(rq, idle, do_notify);
+		__scx_update_idle(rq, next, idle, do_notify);
 }
 #else
-static inline void scx_update_idle(struct rq *rq, bool idle, bool do_notify) {}
+static inline void scx_update_idle(struct rq *rq, struct task_struct *next,
+				   bool idle, bool do_notify) {}
 #endif
 
 #ifdef CONFIG_CGROUP_SCHED
diff --git a/kernel/sched/ext_idle.c b/kernel/sched/ext_idle.c
index e67a19a07..9735f1fb1 100644
--- a/kernel/sched/ext_idle.c
+++ b/kernel/sched/ext_idle.c
@@ -660,12 +660,16 @@ static void update_builtin_idle(int cpu, bool idle)
  * while avoiding unnecessary updates and maintaining balanced state
  * transitions.
  */
-void __scx_update_idle(struct rq *rq, bool idle, bool do_notify)
+void __scx_update_idle(struct rq *rq, struct task_struct *next,
+		       bool idle, bool do_notify)
 {
 	int cpu = cpu_of(rq);
 
 	lockdep_assert_rq_held(rq);
 
+	if (!idle && !rq->scx.cpu_released && next)
+		switch_class(rq, next);
+
 	/*
 	 * Trigger ops.update_idle() only when transitioning from a task to
 	 * the idle thread and vice versa.
diff --git a/kernel/sched/ext_idle.h b/kernel/sched/ext_idle.h
index 511cc2221..83d74ea37 100644
--- a/kernel/sched/ext_idle.h
+++ b/kernel/sched/ext_idle.h
@@ -31,5 +31,6 @@ s32 scx_select_cpu_dfl(struct task_struct *p, s32 prev_cpu, u64 wake_flags, u64
 void scx_idle_enable(struct sched_ext_ops *ops);
 void scx_idle_disable(void);
 int scx_idle_init(void);
+void switch_class(struct rq *rq, struct task_struct *next);
 
 #endif /* _KERNEL_SCHED_EXT_IDLE_H */
diff --git a/kernel/sched/idle.c b/kernel/sched/idle.c
index 2c85c86b4..1b77b56bc 100644
--- a/kernel/sched/idle.c
+++ b/kernel/sched/idle.c
@@ -452,20 +452,20 @@ static void wakeup_preempt_idle(struct rq *rq, struct task_struct *p, int flags)
 static void put_prev_task_idle(struct rq *rq, struct task_struct *prev, struct task_struct *next)
 {
 	dl_server_update_idle_time(rq, prev);
-	scx_update_idle(rq, false, true);
+	scx_update_idle(rq, next, false, true);
 }
 
 static void set_next_task_idle(struct rq *rq, struct task_struct *next, bool first)
 {
 	update_idle_core(rq);
-	scx_update_idle(rq, true, true);
+	scx_update_idle(rq, next, true, true);
 	schedstat_inc(rq->sched_goidle);
 	next->se.exec_start = rq_clock_task(rq);
 }
 
 struct task_struct *pick_task_idle(struct rq *rq)
 {
-	scx_update_idle(rq, true, false);
+	scx_update_idle(rq, NULL, true, false);
 	return rq->idle;
 }
 
-- 
2.17.1