Re: [PATCH] perf: Fix data race in perf_event_set_bpf_handler()

[Peter Zijlstra <peterz@infradead.org>]

On Tue, Jan 27, 2026 at 04:37:19PM +0800, Qing Wang wrote:
> On Tue, 27 Jan 2026 at 10:36, Henry Zhang <henryzhangjcle@gmail.com> wrote:
> > diff --git a/kernel/events/core.c b/kernel/events/core.c
> > index a0fa488bce84..1f3ed9e87507 100644
> > --- a/kernel/events/core.c
> > +++ b/kernel/events/core.c
> > @@ -10349,7 +10349,7 @@ static inline int perf_event_set_bpf_handler(struct perf_event *event,
> >  		return -EPROTO;
> >  	}
> >  
> > -	event->prog = prog;
> > +	WRITE_ONCE(event->prog, prog);
> >  	event->bpf_cookie = bpf_cookie;
> >  	return 0;
> >  }
> > @@ -10407,7 +10407,9 @@ static int __perf_event_overflow(struct perf_event *event,
> >  	if (event->attr.aux_pause)
> >  		perf_event_aux_pause(event->aux_event, true);
> >  
> > -	if (event->prog && event->prog->type == BPF_PROG_TYPE_PERF_EVENT &&
> > +	struct bpf_prog *prog = READ_ONCE(event->prog);
> > +
> > +	if (prog && prog->type == BPF_PROG_TYPE_PERF_EVENT &&
> >  	    !bpf_overflow_handler(event, data, regs))
> >  		goto out;
> 
> Looking at this code, I guess there may be an serious issue: a potential
> use-after-free (UAF) risk when accessing event->prog in __perf_event_overflow.
> 
> CPU 0 (interrupt context)               CPU 1 (process context)
> read event->prog
>                                         perf_event_free_bpf_handler()
>                                             put(prog)
>                                                 free(prog)
> access memory pointed to by prog
> 
> This scenario need to be more analysis.

This can only happen if the event can overlap with removal, which it
typically cannot -- but I'll have to audit the software events.

Specifically, events happen in IRQ/NMI context, and event removal
involves an IPI to that very CPU, which by necessity will then have to
wait for event completion.