Re: [PATCH v2] kernel/events/core: check return value of task_function_call()

[Peter Zijlstra <peterz@infradead.org>]

On Wed, Sep 21, 2022 at 09:46:40AM +0200, Ingo Molnar wrote:
> 
> * Li Zhong <floridsleeves@gmail.com> wrote:
> 
> > From: lily <floridsleeves@gmail.com>
> > 
> > Check the return value of task_function_call(), which could be error
> > code when the execution fails. We log this on info level.
> > 
> > Signed-off-by: Li Zhong <floridsleeves@gmail.com>
> > ---
> >  kernel/events/core.c | 5 ++++-
> >  1 file changed, 4 insertions(+), 1 deletion(-)
> > 
> > diff --git a/kernel/events/core.c b/kernel/events/core.c
> > index 2621fd24ad26..3848631b009c 100644
> > --- a/kernel/events/core.c
> > +++ b/kernel/events/core.c
> > @@ -13520,7 +13520,10 @@ static void perf_cgroup_attach(struct cgroup_taskset *tset)
> >  	struct cgroup_subsys_state *css;
> >  
> >  	cgroup_taskset_for_each(task, css, tset)
> > -		task_function_call(task, __perf_cgroup_move, task);
> > +		if (!task_function_call(task, __perf_cgroup_move, task)) {
> > +			printk(KERN_INFO "perf: this process isn't running!\n");
> > +			continue;
> 
> If this is a 'should never happen' condition, then this should really be 
> WARN_ON_ONCE().
> 
> (If it can happen, then polluting the syslog is wrong.)

It can happen and is just fine. If this task_function_call() fails, it
means the context switch that made it fail ensured the new values have
been picked up.