[PATCH 2/3] scx_central: Cancel BPF timer during ops.exit

[Zhao Mengmeng]

From: Zhao Mengmeng <zhaomengmeng@xxxxxxxxxx>

Cancel the BPF timer in ops.exit, which is called before
free_kick_syncs(), so that new cpu kick will not come and hit the NULL
check in kick_cpus_irq_workfn().

Signed-off-by: Zhao Mengmeng <zhaomengmeng@xxxxxxxxxx>
---
 tools/sched_ext/scx_central.bpf.c | 7 +++++++
 1 file changed, 7 insertions(+)

diff --git a/tools/sched_ext/scx_central.bpf.c b/tools/sched_ext/scx_central.bpf.c
index 4efcce099bd5..6c2cd5511a25 100644
--- a/tools/sched_ext/scx_central.bpf.c
+++ b/tools/sched_ext/scx_central.bpf.c
@@ -359,7 +359,14 @@ int BPF_STRUCT_OPS_SLEEPABLE(central_init)
 
 void BPF_STRUCT_OPS(central_exit, struct scx_exit_info *ei)
 {
+	u32 key = 0;
+	struct bpf_timer *timer;
+
 	UEI_RECORD(uei, ei);
+
+	timer = bpf_map_lookup_elem(&central_timer, &key);
+	if (timer)
+		bpf_timer_cancel(timer);
 }
 
 SCX_OPS_DEFINE(central_ops,
-- 
2.43.0