Re: [PATCH v2] perf/bpf: Fix data races in BPF perf event handling

On Thu, Aug 13, 2026 at 4:35 PM Peter Zijlstra <[email protected]> wrote:
>

> As I said yesterday, how can perf_event_detach_bpf_prog() run
> concurrently with __perf_event_overflow()? Unless you answer that, this
> patch ain't moving nowhere.
>


You are right to question this. The KCSAN report from syzbot shows the
race was detected at runtime:

I don't have evidence to prove this race condition.

> > 3. perf_event_alloc() reads parent_event->prog locklessly during
> >    fork() which can race with a concurrent detach clearing and freeing
> >    the prog, potentially causing a NULL pointer dereference or
> >    use-after-free in bpf_prog_inc(). Fix by holding bpf_event_mutex
> >    when inheriting the BPF program. Make bpf_event_mutex non-static
> >    and declare it extern in perf_event.h so it is accessible from
> >    kernel/events/core.c.
>
> This seems like a separate issue and should thus be a separate patch.

Agreed. I will split the perf_event_alloc() fix into a separate patch.

Thanks

Deepanshu