On Tue, Aug 13, 2013 at 07:44:55PM -0400, Chris Metcalf wrote:
> int lru_add_drain_all(void)
> {
> static struct cpumask mask;
Instead of cpumask,
> static DEFINE_MUTEX(lock);
you can DEFINE_PER_CPU(struct work_struct, ...).
> for_each_online_cpu(cpu) {
> if (pagevec_count(&per_cpu(lru_add_pvec, cpu)) ||
> pagevec_count(&per_cpu(lru_rotate_pvecs, cpu)) ||
> pagevec_count(&per_cpu(lru_deactivate_pvecs, cpu)) ||
> need_activate_page_drain(cpu))
> cpumask_set_cpu(cpu, &mask);
and schedule the work items directly.
> }
>
> rc = schedule_on_cpu_mask(lru_add_drain_per_cpu, &mask);
Open coding flushing can be a bit bothersome but you can create a
per-cpu workqueue and schedule work items on it and then flush the
workqueue instead too.
No matter how flushing is implemented, the path wouldn't have any
memory allocation, which I thought was the topic of the thread, no?
Thanks.
--
tejun
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majord...@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
