On 02/06/2015 09:08 PM, Mathieu Desnoyers wrote:
> A lockless_dereference() appears to be missing in llist_del_first().
> It should only matter for Alpha in practice.
>
> Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@xxxxxxxxxxxx>
> CC: Huang Ying <ying.huang@xxxxxxxxx>
> CC: Paul McKenney <paulmck@xxxxxxxxxxxxxxxxxx>
> CC: David Howells <dhowells@xxxxxxxxxx>
> CC: Pranith Kumar <bobby.prani@xxxxxxxxx>
> CC: stable@xxxxxxxxxxxxxxx # v3.1+
> ---
> lib/llist.c | 8 +++++++-
> 1 file changed, 7 insertions(+), 1 deletion(-)
>
> diff --git a/lib/llist.c b/lib/llist.c
> index f76196d..f34e176 100644
> --- a/lib/llist.c
> +++ b/lib/llist.c
> @@ -26,6 +26,7 @@
> #include <linux/export.h>
> #include <linux/interrupt.h>
> #include <linux/llist.h>
> +#include <linux/rcupdate.h>
Pranith,
I didn't realize you put lockless_dereference() in rcupdate.h
If the point of lockless_reference() is to provide a utility function for
situations _not_ involving RCU, then it doesn't make sense to provide it
in an RCU header file.
Regards,
Peter Hurley
PS - That's not an objection to this patch, though.
> /**
> @@ -67,7 +68,12 @@ struct llist_node *llist_del_first(struct llist_head *head)
> {
> struct llist_node *entry, *old_entry, *next;
>
> - entry = head->first;
> + /*
> + * Load entry before entry->next. Matches the implicit
> + * memory barrier before the cmpxchg in llist_add_batch(),
> + * which ensures entry->next is stored before entry.
> + */
> + entry = lockless_dereference(head->first);
> for (;;) {
> if (entry == NULL)
> return NULL;
>
--
To unsubscribe from this list: send the line "unsubscribe stable" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html