Re: [PATCH v9 8/8] task: rust: rework how current is accessed

[Alice Ryhl]

On Fri, Nov 22, 2024 at 6:55 PM Lorenzo Stoakes
<lorenzo.stoakes@xxxxxxxxxx> wrote:
>
> On Fri, Nov 22, 2024 at 03:40:33PM +0000, Alice Ryhl wrote:
> > Introduce a new type called `CurrentTask` that lets you perform various
> > operations that are only safe on the `current` task. Use the new type to
> > provide a way to access the current mm without incrementing its
> > refcount.
>
> Nice!
>
> >
> > With this change, you can write stuff such as
> >
> >       let vma = current!().mm().lock_vma_under_rcu(addr);
> >
> > without incrementing any refcounts.
> >
> > Signed-off-by: Alice Ryhl <aliceryhl@xxxxxxxxxx>
>
> On assumption that the problem you reference with the rust imports is
> corrected in v10, and that what you are doing with current_raw() is
> sensible, then:
>
> Acked-by: Lorenzo Stoakes <lorenzo.stoakes@xxxxxxxxxx>
>
> Thanks!
>
> > ---
> > Reviewers: Does accessing task->mm on a non-current task require rcu
> > protection?
>
> Hm I am not actually sure, but it seems like you probably do, and I would say
> you need the task lock right?
>
> Looking at find_lock_task_mm() as used by the oomk for instance suggests as much.

Okay, sounds complicated. I'm not going to bother with that right now.

> >  /// The type of process identifiers (PIDs).
> >  type Pid = bindings::pid_t;
> >
> > @@ -121,27 +141,25 @@ pub fn current_raw() -> *mut bindings::task_struct {
> >      /// # Safety
> >      ///
> >      /// Callers must ensure that the returned object doesn't outlive the current task/thread.
> > -    pub unsafe fn current() -> impl Deref<Target = Task> {
> > -        struct TaskRef<'a> {
> > -            task: &'a Task,
> > -            _not_send: NotThreadSafe,
> > +    pub unsafe fn current() -> impl Deref<Target = CurrentTask> {
> > +        struct TaskRef {
> > +            task: *const CurrentTask,
> >          }
>
> Why do we drop the NotThreadSafe bit here? And it seems like the 'a lifetime
> stuff has gone too?
>
> I'm guessing the lifetime stuff is because of the SAFETY comment below about
> assumptions about lifetime?

I dropped the lifetime because it's not doing anything. As for NotThreadSafe:

1. See thread with Boqun.
2. Raw pointers are already considered not thread safe by default, so
the *const CurrentTask field has the same effect.

Alice