[PATCH 2/2] sched_ext: Fix exit_cpu accuracy for lockup paths

[Cheng-Yang Chou <yphbchou0911@gmail.com>]

handle_lockup() uses raw_smp_processor_id() for exit_cpu, which is wrong
for two paths:

- scx_hardlockup_irq_workfn() has the hung CPU in a local variable but
  irq_work may run elsewhere; pass the local cpu explicitly.
- scx_rcu_cpu_stall() takes no CPU, recording the detector rather than
  the stalled one; add stalled_cpu to its signature and
  panic_on_rcu_stall(), threading it from print_cpu_stall() (current
  CPU) and print_other_cpu_stall() (first detected stalled CPU).

Add an exit_cpu parameter to handle_lockup() so callers pass the correct
CPU directly. Remove the now-unused scx_verror() macro.

Signed-off-by: Cheng-Yang Chou <yphbchou0911@gmail.com>
---
 include/linux/sched/ext.h   |  4 ++--
 kernel/rcu/tree_exp.h       |  2 +-
 kernel/rcu/tree_stall.h     | 11 +++++++----
 kernel/sched/ext.c          | 14 ++++++++------
 kernel/sched/ext_internal.h |  2 --
 5 files changed, 18 insertions(+), 15 deletions(-)

diff --git a/include/linux/sched/ext.h b/include/linux/sched/ext.h
index d05efcac794d..16bbf24089ca 100644
--- a/include/linux/sched/ext.h
+++ b/include/linux/sched/ext.h
@@ -259,7 +259,7 @@ void sched_ext_dead(struct task_struct *p);
 void print_scx_info(const char *log_lvl, struct task_struct *p);
 void scx_softlockup(u32 dur_s);
 bool scx_hardlockup(int cpu);
-bool scx_rcu_cpu_stall(void);
+bool scx_rcu_cpu_stall(int stalled_cpu);
 
 #else	/* !CONFIG_SCHED_CLASS_EXT */
 
@@ -267,7 +267,7 @@ static inline void sched_ext_dead(struct task_struct *p) {}
 static inline void print_scx_info(const char *log_lvl, struct task_struct *p) {}
 static inline void scx_softlockup(u32 dur_s) {}
 static inline bool scx_hardlockup(int cpu) { return false; }
-static inline bool scx_rcu_cpu_stall(void) { return false; }
+static inline bool scx_rcu_cpu_stall(int stalled_cpu) { return false; }
 
 #endif	/* CONFIG_SCHED_CLASS_EXT */
 
diff --git a/kernel/rcu/tree_exp.h b/kernel/rcu/tree_exp.h
index 82cada459e5d..2c9a99a9b68e 100644
--- a/kernel/rcu/tree_exp.h
+++ b/kernel/rcu/tree_exp.h
@@ -675,7 +675,7 @@ static void synchronize_rcu_expedited_wait(void)
 
 		nbcon_cpu_emergency_exit();
 
-		panic_on_rcu_stall();
+		panic_on_rcu_stall(smp_processor_id());
 	}
 }
 
diff --git a/kernel/rcu/tree_stall.h b/kernel/rcu/tree_stall.h
index b67532cb8770..172ac08e673d 100644
--- a/kernel/rcu/tree_stall.h
+++ b/kernel/rcu/tree_stall.h
@@ -159,7 +159,7 @@ static int __init check_cpu_stall_init(void)
 early_initcall(check_cpu_stall_init);
 
 /* If so specified via sysctl, panic, yielding cleaner stall-warning output. */
-static void panic_on_rcu_stall(void)
+static void panic_on_rcu_stall(int stalled_cpu)
 {
 	static int cpu_stall;
 
@@ -167,7 +167,7 @@ static void panic_on_rcu_stall(void)
 	 * Attempt to kick out the BPF scheduler if it's installed and defer
 	 * the panic to give the system a chance to recover.
 	 */
-	if (scx_rcu_cpu_stall())
+	if (scx_rcu_cpu_stall(stalled_cpu))
 		return;
 
 	if (++cpu_stall < sysctl_max_rcu_stall_to_panic)
@@ -631,6 +631,7 @@ static void rcu_check_gp_kthread_expired_fqs_timer(void)
 static void print_other_cpu_stall(unsigned long gp_seq, unsigned long gps)
 {
 	int cpu;
+	int first_stalled_cpu = -1;
 	unsigned long flags;
 	unsigned long gpa;
 	unsigned long j;
@@ -660,6 +661,8 @@ static void print_other_cpu_stall(unsigned long gp_seq, unsigned long gps)
 			for_each_leaf_node_possible_cpu(rnp, cpu)
 				if (rnp->qsmask & leaf_node_cpu_bit(rnp, cpu)) {
 					print_cpu_stall_info(cpu);
+					if (first_stalled_cpu < 0)
+						first_stalled_cpu = cpu;
 					ndetected++;
 				}
 		}
@@ -701,7 +704,7 @@ static void print_other_cpu_stall(unsigned long gp_seq, unsigned long gps)
 
 	nbcon_cpu_emergency_exit();
 
-	panic_on_rcu_stall();
+	panic_on_rcu_stall(first_stalled_cpu);
 
 	rcu_force_quiescent_state();  /* Kick them all. */
 }
@@ -754,7 +757,7 @@ static void print_cpu_stall(unsigned long gp_seq, unsigned long gps)
 
 	nbcon_cpu_emergency_exit();
 
-	panic_on_rcu_stall();
+	panic_on_rcu_stall(smp_processor_id());
 
 	/*
 	 * Attempt to revive the RCU machinery by forcing a context switch.
diff --git a/kernel/sched/ext.c b/kernel/sched/ext.c
index 48c65ac8e230..8a0b1662a75a 100644
--- a/kernel/sched/ext.c
+++ b/kernel/sched/ext.c
@@ -5069,6 +5069,7 @@ bool scx_allow_ttwu_queue(const struct task_struct *p)
 
 /**
  * handle_lockup - sched_ext common lockup handler
+ * @exit_cpu: CPU to record in exit_info; pass the stalled/hung CPU, not current
  * @fmt: format string
  *
  * Called on system stall or lockup condition and initiates abort of sched_ext
@@ -5078,7 +5079,7 @@ bool scx_allow_ttwu_queue(const struct task_struct *p)
  * resolve the lockup. %false if sched_ext is not enabled or abort was already
  * initiated by someone else.
  */
-static __printf(1, 2) bool handle_lockup(const char *fmt, ...)
+static __printf(2, 3) bool handle_lockup(int exit_cpu, const char *fmt, ...)
 {
 	struct scx_sched *sch;
 	va_list args;
@@ -5094,7 +5095,7 @@ static __printf(1, 2) bool handle_lockup(const char *fmt, ...)
 	case SCX_ENABLING:
 	case SCX_ENABLED:
 		va_start(args, fmt);
-		ret = scx_verror(sch, fmt, args);
+		ret = scx_vexit(sch, SCX_EXIT_ERROR, 0, exit_cpu, fmt, args);
 		va_end(args);
 		return ret;
 	default:
@@ -5114,9 +5115,9 @@ static __printf(1, 2) bool handle_lockup(const char *fmt, ...)
  * resolve the reported RCU stall. %false if sched_ext is not enabled or someone
  * else already initiated abort.
  */
-bool scx_rcu_cpu_stall(void)
+bool scx_rcu_cpu_stall(int stalled_cpu)
 {
-	return handle_lockup("RCU CPU stall detected!");
+	return handle_lockup(stalled_cpu, "RCU CPU stall detected!");
 }
 
 /**
@@ -5131,7 +5132,8 @@ bool scx_rcu_cpu_stall(void)
  */
 void scx_softlockup(u32 dur_s)
 {
-	if (!handle_lockup("soft lockup - CPU %d stuck for %us", smp_processor_id(), dur_s))
+	if (!handle_lockup(smp_processor_id(), "soft lockup - CPU %d stuck for %us",
+			   smp_processor_id(), dur_s))
 		return;
 
 	printk_deferred(KERN_ERR "sched_ext: Soft lockup - CPU %d stuck for %us, disabling BPF scheduler\n",
@@ -5150,7 +5152,7 @@ static void scx_hardlockup_irq_workfn(struct irq_work *work)
 {
 	int cpu = atomic_xchg(&scx_hardlockup_cpu, -1);
 
-	if (cpu >= 0 && handle_lockup("hard lockup - CPU %d", cpu))
+	if (cpu >= 0 && handle_lockup(cpu, "hard lockup - CPU %d", cpu))
 		printk_deferred(KERN_ERR "sched_ext: Hard lockup - CPU %d, disabling BPF scheduler\n",
 				cpu);
 }
diff --git a/kernel/sched/ext_internal.h b/kernel/sched/ext_internal.h
index b4f5dd28855e..2d04f27cf7c5 100644
--- a/kernel/sched/ext_internal.h
+++ b/kernel/sched/ext_internal.h
@@ -1486,8 +1486,6 @@ __printf(5, 6) bool __scx_exit(struct scx_sched *sch, enum scx_exit_kind kind,
 	__scx_exit(sch, kind, exit_code, raw_smp_processor_id(), fmt, ##args)
 #define scx_error(sch, fmt, args...)						\
 	scx_exit((sch), SCX_EXIT_ERROR, 0, fmt, ##args)
-#define scx_verror(sch, fmt, args)						\
-	scx_vexit((sch), SCX_EXIT_ERROR, 0, raw_smp_processor_id(), fmt, args)
 
 /*
  * Return the rq currently locked from an scx callback, or NULL if no rq is
-- 
2.48.1