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

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

[Hui Su]

event_test_stuff() calls kthread_run() and unconditionally passes the
returned task_struct pointer to kthread_stop(). kthread_run() returns an
error pointer such as ERR_PTR(-ENOMEM) when kthread creation fails, for
example under memory pressure during the boot-time event self-test.
kthread_stop() then dereferences the invalid pointer, crashing the kernel.

Check the result of kthread_run() before passing it to kthread_stop(). Use
WARN_ON() so that a failure to create the self-test thread does not go
unnoticed, matching the ring-buffer self-test fix in commit
91542863abad ("ring-buffer: Fix crash passing ERR_PTR to kthread_stop()").

Fixes: e6187007d6c3 ("tracing/events: add startup tests for events")
Cc: stable@vger.kernel.org
Signed-off-by: Hui Su <sh_def@163.com>
---

Changes in v2:
- Use WARN_ON(IS_ERR(test_thread)) so thread creation failures are not
  silently ignored, as suggested by Steven Rostedt.

v1: https://lore.kernel.org/r/20260815190151.568893-4-sh_def@163.com/

 kernel/trace/trace_events.c | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/kernel/trace/trace_events.c b/kernel/trace/trace_events.c
index 3650d84d4f16..28167b229419 100644
--- a/kernel/trace/trace_events.c
+++ b/kernel/trace/trace_events.c
@@ -4901,6 +4901,8 @@ 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;
 	msleep(1);
 	kthread_stop(test_thread);
 }
-- 
2.54.0

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

[sashiko-bot]

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 <sh_def@163.com>

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/20260817120642.668375-3-sh_def@163.com?part=1

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

[Steven Rostedt]

On Mon, 17 Aug 2026 12:15:32 +0000
sashiko-bot@kernel.org wrote:

> > 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?
> 

If that happens than the system is doomed to begin with. We don't care
here. This patch is good enough. It's for a debugging feature, not
anything that would be run in production.

Just look at the name of the function: event_test_stuff()

This is where AI has a hard time with subtle requirements.

Thanks,

-- Steve