On Fri, 23 Jul 2021 00:43:13 +0200
Stefan Metzmacher [off-list ref] wrote:

Hi Steve,

Hi Stefan,

After some days of training:
https://training.linuxfoundation.org/training/linux-kernel-debugging-and-security/
I was able to get much closer to the problem :-)

In order to reproduce it and get reliable kexec crash dumps,
I needed to give the VM at least 3 cores.

While running './io-uring_cp-forever link-cp.c file' (from:
https://github.com/metze-samba/liburing/commits/io_uring-cp-forever )
in one window, the following simple sequence triggered the problem in most cases:

echo 1 > /sys/kernel/tracing/events/sched/sched_waking/enable
echo 1 > /sys/kernel/tracing/set_event_pid

I was able to reproduce it with running hackbench in a while loop and
in another terminal, executing the above two lines.

I think I found the bug. Can you test this patch?

Thanks,

-- Steve

diff --git a/kernel/tracepoint.c b/kernel/tracepoint.c
index 976bf8ce8039..fc32821f8240 100644
--- a/kernel/tracepoint.c
+++ b/kernel/tracepoint.c

@@ -299,8 +299,8 @@ static int tracepoint_add_func(struct tracepoint *tp,
 	 * a pointer to it.  This array is referenced by __DO_TRACE from
 	 * include/linux/tracepoint.h using rcu_dereference_sched().
 	 */
-	rcu_assign_pointer(tp->funcs, tp_funcs);
 	tracepoint_update_call(tp, tp_funcs, false);
+	rcu_assign_pointer(tp->funcs, tp_funcs);
 	static_key_enable(&tp->key);
 
 	release_probes(old);