On Tue, 15 Jul 2025 12:29:12 +0200
Peter Zijlstra <peterz@xxxxxxxxxxxxx> wrote:
> The below is the last four patches rolled into one. Not been near a
> compiler.
And it shows ;-)
> @@ -117,13 +138,13 @@ static void unwind_deferred_task_work(st
> struct unwind_task_info *info = container_of(head, struct unwind_task_info, work);
> struct unwind_stacktrace trace;
> struct unwind_work *work;
> + unsigned long bits;
> u64 cookie;
>
> - if (WARN_ON_ONCE(!info->pending))
> + if (WARN_ON_ONCE(!unwind_pending(info)))
> return;
>
> - /* Allow work to come in again */
> - WRITE_ONCE(info->pending, 0);
> + bits = atomic_long_fetch_andnot(UNWIND_PENDING, &info->unwind_mask);
I may need to do what other parts of the kernel has done and turn the
above into:
bits = atomic_long_fetch_andnot(UNWIND_PENDING, (atomic_long_t *)&info->unwind_mask);
As there's other bit manipulations that atomic_long does not take care
of and it's making the code more confusing. When I looked to see how
other users of atomic_long_andnot() did things, most just typecasted
the value to use that function :-/
-- Steve
>
> /*
> * From here on out, the callback must always be called, even if it's
> @@ -136,9 +157,11 @@ static void unwind_deferred_task_work(st
>
> cookie = info->id.id;
>
> - guard(mutex)(&callback_mutex);
> - list_for_each_entry(work, &callbacks, list) {
> - work->func(work, &trace, cookie);
> + guard(srcu_lite)(&unwind_srcu);
> + list_for_each_entry_srcu(work, &callbacks, list,
> + srcu_read_lock_held(&unwind_srcu)) {
> + if (test_bit(work->bit, &bits))
> + work->func(work, &trace, cookie);
> }
> }
>
> @@ -162,7 +185,7 @@ static void unwind_deferred_task_work(st
> * because it has already been previously called for the same entry context,
> * it will be called again with the same stack trace and cookie.
> *
> - * Return: 1 if the the callback was already queued.
> + * Return: 1 if the callback was already queued.
> * 0 if the callback successfully was queued.
> * Negative if there's an error.
> * @cookie holds the cookie of the first request by any user
