Re: [PATCH v2] tracing: Fix crash passing ERR_PTR to kthread_stop()

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [High] Using `WARN_ON` inside `event_test_stuff()` causes a massive warning storm and potential soft lockup if kthread creation repeatedly fails.
--

--- Patch [1]: [PATCH v2] tracing: Fix crash passing ERR_PTR to kthread_stop() ---
commit 692df106ab8d8e56701ad9c14c135871ee02df4a
Author: Hui Su <[email protected]>

tracing: Fix crash passing ERR_PTR to kthread_stop()

This commit adds a check for the return value of kthread_run() in
event_test_stuff() to prevent passing an ERR_PTR to kthread_stop() if thread
creation fails under memory pressure during boot-time trace self tests.

> diff --git a/kernel/trace/trace_events.c b/kernel/trace/trace_events.c
> index 0356cae0cf74e..9e941e5805f5e 100644
> --- a/kernel/trace/trace_events.c
> +++ b/kernel/trace/trace_events.c
> @@ -3866,6 +3866,8 @@ static __init void event_test_stuff(void)
>  	struct task_struct *test_thread;
>  
>  	test_thread = kthread_run(event_test_thread, NULL, "test-events");
> +	if (WARN_ON(IS_ERR(test_thread)))
> +		return;

[Severity: High]
Does this risk causing a console flood and soft lockup if thread creation
persistently fails?

During boot-time trace self tests, event_test_stuff() is called within
large loops spanning all trace events:

kernel/trace/trace_events.c:event_trace_self_tests() {
    ...
	list_for_each_entry(file, &tr->events, list) {
        ...
		ftrace_event_enable_disable(file, 1);
		event_test_stuff();
		ftrace_event_enable_disable(file, 0);
    ...
}

If kthread creation fails, WARN_ON prints a full stack trace and returns.
Since event_test_stuff() returns void, the caller blindly continues to the
next trace event, potentially leading to thousands of consecutive stack
dumps.

Could this be changed to WARN_ON_ONCE() or modified to abort the test loop
gracefully?

>  	msleep(1);
>  	kthread_stop(test_thread);
>  }

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=1