Re: [PATCH] fgraph: Convert ret_stack tasklist scanning to rcu

[Oleg Nesterov <oleg@redhat.com>]

On 09/18, Steven Rostedt wrote:
>
> On Mon, 7 Sep 2020 13:43:02 +0200
> Oleg Nesterov <oleg@redhat.com> wrote:
>
> > Afaics, with or without this change alloc_retstack_tasklist() can race
> > with copy_process() and miss the new child; ftrace_graph_init_task()
> > can't help, ftrace_graph_active can be set right after the check and
> > for_each_process_thread() can't see the new process yet.
>
> There's a call in copy_process(): ftrace_graph_init_task() that initializes
> a new tasks ret_stack,

Only if ftrace_graph_active != 0.

register_ftrace_graph() can increment ftrace_graph_active and call
alloc_retstack_tasklist() right after ftrace_graph_init_task() checks
ftrace_graph_active.

> and this loop will ignore it

and this loop won't see it unless the forking process finishes copy_process()
and does list_add_tail_rcu(&p->tasks, &init_task.tasks) which makes it
visible to for_each_process(). Yes, this is very unlikely.

> > Looks like, ftrace_graph_exit_task() was called by the exiting task
> > in the past? Indeed, see 65afa5e603d50 ("tracing/function-return-tracer:
> > free the return stack on free_task()"). I think it makes sense to
> > simplify this function now, it can simply do kfree(t->ret_stack) and
> > nothing more.
>
> Ah, yeah, then you are right. If it can't be called on a running task then
> it can be simplified. Probably need a:
>
>  WARN_ON_ONCE(t->on_rq);
>
> just to make sure this never happens.

Well, ftrace_graph_exit_task(t) is called by free_task(t), right before
kmem_cache_free(t).

> > ACK, but ...
> >
> > > @@ -387,8 +387,8 @@ static int alloc_retstack_tasklist(struct ftrace_ret_stack **ret_stack_list)
> > >  		}
> > >  	}
> > >
> > > -	read_lock(&tasklist_lock);
> >
> > then you should probably rename alloc_retstack_tasklist() ?
> >
>
> tasklist, process thead? Is there a difference?

Aah, please ignore. Somehow I misinterpreted the _tasklist suffix, as if it
refers to tasklist_lock.

Oleg.