On Mon, 12 Jun 2023 19:47:16 +
Wathsala Vithanage wrote:
> For improved performance over the current C11 based ring
> implementation, two ACQUIRE fences between the load of the old_head
> and the load of the const_tail in __rte_ring_move_prod_head and the
> load of old_head and load of the pr
For improved performance over the current C11 based ring
implementation, two ACQUIRE fences between the load of the old_head
and the load of the const_tail in __rte_ring_move_prod_head and the
load of old_head and load of the prod_tail in
__rte_ring_move_cons_head functions have been removed. These
ACQUIRE fence between the two loads established program order to
prevent use of an outdated consumer or producer tail values in the
computation of free_entries and entries in the two functions
respectively. However, (A) the problem described above cannot be solved
by establishing the program order
3 matches
Mail list logo
