On Thu, Oct 9, 2025 at 11:50 PM <[email protected]> wrote:
>
> From: Xuanqiang Luo <[email protected]>
>
> In rculist_nulls.h we can still see ordinary assignments to ->pprev and
> ->next of hlist_nulls.
>
> As noted in the two patches below:
> commit efd04f8a8b45 ("rcu: Use WRITE_ONCE() for assignments to ->next for
> rculist_nulls")
> commit 860c8802ace1 ("rcu: Use WRITE_ONCE() for assignments to ->pprev for
> hlist_nulls")
>
> We should use WRITE_ONCE().
>
> Signed-off-by: Xuanqiang Luo <[email protected]>
> ---
> include/linux/rculist_nulls.h | 6 +++---
> 1 file changed, 3 insertions(+), 3 deletions(-)
>
> diff --git a/include/linux/rculist_nulls.h b/include/linux/rculist_nulls.h
> index 89186c499dd4..d5a656cc4c6a 100644
> --- a/include/linux/rculist_nulls.h
> +++ b/include/linux/rculist_nulls.h
> @@ -138,7 +138,7 @@ static inline void hlist_nulls_add_tail_rcu(struct
> hlist_nulls_node *n,
>
> if (last) {
> WRITE_ONCE(n->next, last->next);
> - n->pprev = &last->next;
> + WRITE_ONCE(n->pprev, &last->next);
n is private to the current thread. No other cpu/thread can see it yet.
Following rcu_assign_pointer() already has the needed barrier making
sure that _when_ the object
can be seen by others, n->pprev value is fully up to date.
> rcu_assign_pointer(hlist_nulls_next_rcu(last), n);
> } else {
> hlist_nulls_add_head_rcu(n, h);
> @@ -148,8 +148,8 @@ static inline void hlist_nulls_add_tail_rcu(struct
> hlist_nulls_node *n,
> /* after that hlist_nulls_del will work */
> static inline void hlist_nulls_add_fake(struct hlist_nulls_node *n)
> {
> - n->pprev = &n->next;
> - n->next = (struct hlist_nulls_node *)NULLS_MARKER(NULL);
> + WRITE_ONCE(n->pprev, &n->next);
> + WRITE_ONCE(n->next, (struct hlist_nulls_node *)NULLS_MARKER(NULL));
Same here, n is private. No need for any WRITE_ONCE().