[PATCH v2 4/6] sched: Provide rt_mutex specific scheduler helpers

[Sebastian Andrzej Siewior]

From: Peter Zijlstra <peterz@xxxxxxxxxxxxx>

With PREEMPT_RT there is a rt_mutex recursion problem where
sched_submit_work() can use an rtlock (aka spinlock_t). More
specifically what happens is:

  mutex_lock() /* really rt_mutex */
    ...
      __rt_mutex_slowlock_locked()
	task_blocks_on_rt_mutex()
          // enqueue current task as waiter
          // do PI chain walk
        rt_mutex_slowlock_block()
          schedule()
            sched_submit_work()
              ...
              spin_lock() /* really rtlock */
                ...
                  __rt_mutex_slowlock_locked()
                    task_blocks_on_rt_mutex()
                      // enqueue current task as waiter *AGAIN*
                      // *CONFUSION*

Fix this by making rt_mutex do the sched_submit_work() early, before
it enqueues itself as a waiter -- before it even knows *if* it will
wait.

[[ basically Thomas' patch but with different naming and a few asserts
   added ]]

Originally-by: Thomas Gleixner <tglx@xxxxxxxxxxxxx>
Signed-off-by: "Peter Zijlstra (Intel)" <peterz@xxxxxxxxxxxxx>
Link: https://lore.kernel.org/r/20230815111430.355375399@xxxxxxxxxxxxx
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@xxxxxxxxxxxxx>
---
 include/linux/sched.h    |  3 +++
 include/linux/sched/rt.h |  4 ++++
 kernel/sched/core.c      | 36 ++++++++++++++++++++++++++++++++----
 3 files changed, 39 insertions(+), 4 deletions(-)

diff --git a/include/linux/sched.h b/include/linux/sched.h
index 609bde814cb06..0ea7a023c6c73 100644
--- a/include/linux/sched.h
+++ b/include/linux/sched.h
@@ -906,6 +906,9 @@ struct task_struct {
 	 * ->sched_remote_wakeup gets used, so it can be in this word.
 	 */
 	unsigned			sched_remote_wakeup:1;
+#ifdef CONFIG_RT_MUTEXES
+	unsigned			sched_rt_mutex:1;
+#endif
 
 	/* Bit to tell LSMs we're in execve(): */
 	unsigned			in_execve:1;
diff --git a/include/linux/sched/rt.h b/include/linux/sched/rt.h
index 994c25640e156..b2b9e6eb96830 100644
--- a/include/linux/sched/rt.h
+++ b/include/linux/sched/rt.h
@@ -30,6 +30,10 @@ static inline bool task_is_realtime(struct task_struct *tsk)
 }
 
 #ifdef CONFIG_RT_MUTEXES
+extern void rt_mutex_pre_schedule(void);
+extern void rt_mutex_schedule(void);
+extern void rt_mutex_post_schedule(void);
+
 /*
  * Must hold either p->pi_lock or task_rq(p)->lock.
  */
diff --git a/kernel/sched/core.c b/kernel/sched/core.c
index 901766a88afc3..bba1ed28608ab 100644
--- a/kernel/sched/core.c
+++ b/kernel/sched/core.c
@@ -6738,9 +6738,6 @@ static inline void sched_submit_work(struct task_struct *tsk)
 	static DEFINE_WAIT_OVERRIDE_MAP(sched_map, LD_WAIT_CONFIG);
 	unsigned int task_flags;
 
-	if (task_is_running(tsk))
-		return;
-
 	/*
 	 * Establish LD_WAIT_CONFIG context to ensure none of the code called
 	 * will use a blocking primitive -- which would lead to recursion.
@@ -6798,7 +6795,12 @@ asmlinkage __visible void __sched schedule(void)
 {
 	struct task_struct *tsk = current;
 
-	sched_submit_work(tsk);
+#ifdef CONFIG_RT_MUTEXES
+	lockdep_assert(!tsk->sched_rt_mutex);
+#endif
+
+	if (!task_is_running(tsk))
+		sched_submit_work(tsk);
 	__schedule_loop(SM_NONE);
 	sched_update_worker(tsk);
 }
@@ -7059,6 +7061,32 @@ static void __setscheduler_prio(struct task_struct *p, int prio)
 
 #ifdef CONFIG_RT_MUTEXES
 
+/*
+ * Would be more useful with typeof()/auto_type but they don't mix with
+ * bit-fields. Since it's a local thing, use int. Keep the generic sounding
+ * name such that if someone were to implement this function we get to compare
+ * notes.
+ */
+#define fetch_and_set(x, v) ({ int _x = (x); (x) = (v); _x; })
+
+void rt_mutex_pre_schedule(void)
+{
+	lockdep_assert(!fetch_and_set(current->sched_rt_mutex, 1));
+	sched_submit_work(current);
+}
+
+void rt_mutex_schedule(void)
+{
+	lockdep_assert(current->sched_rt_mutex);
+	__schedule_loop(SM_NONE);
+}
+
+void rt_mutex_post_schedule(void)
+{
+	sched_update_worker(current);
+	lockdep_assert(fetch_and_set(current->sched_rt_mutex, 0));
+}
+
 static inline int __rt_effective_prio(struct task_struct *pi_task, int prio)
 {
 	if (pi_task)
-- 
2.40.1