Re: [PATCH v2 04/10] ftrace: Store direct called addresses in their ops

[Florent Revest]

On Thu, Mar 16, 2023 at 12:43 AM Steven Rostedt <rostedt@xxxxxxxxxxx> wrote:
>
> On Tue,  7 Feb 2023 19:21:29 +0100
> Florent Revest <revest@xxxxxxxxxxxx> wrote:
>
> > @@ -5445,6 +5445,7 @@ __modify_ftrace_direct(struct ftrace_ops *ops, unsigned long addr)
> >       /* Enable the tmp_ops to have the same functions as the direct ops */
> >       ftrace_ops_init(&tmp_ops);
> >       tmp_ops.func_hash = ops->func_hash;
> > +     tmp_ops.direct_call = addr;
> >
> >       err = register_ftrace_function_nolock(&tmp_ops);
> >       if (err)
> > @@ -5466,6 +5467,7 @@ __modify_ftrace_direct(struct ftrace_ops *ops, unsigned long addr)
> >                       entry->direct = addr;
> >               }
> >       }
> > +     WRITE_ONCE(ops->direct_call, addr);
>
> I'm curious about the use of WRITE_ONCE(). It should not go outside the
> mutex barrier.

This WRITE_ONCE was originally suggested by Mark here:
https://lore.kernel.org/all/Y9vW99htjOphDXqY@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxx/#t

My understanding is that it's not so much about avoiding re-ordering
but rather about avoiding store tearing since a ftrace_caller
trampoline could concurrently read ops->direct_call. Does that make
sense ?

> -- Steve
>
> >
> >       mutex_unlock(&ftrace_lock);
> >