回复: [PATCH v2] kvfree_rcu: Release page cache under memory pressure

[Zhang, Qiang]

________________________________________
发件人: Uladzislau Rezki <urezki@xxxxxxxxx>
发送时间: 2021年1月29日 22:19
收件人: Zhang, Qiang
抄送: urezki@xxxxxxxxx; paulmck@xxxxxxxxxx; joel@xxxxxxxxxxxxxxxxx; rcu@xxxxxxxxxxxxxxx; linux-kernel@xxxxxxxxxxxxxxx
主题: Re: [PATCH v2] kvfree_rcu: Release page cache under memory pressure

[Please note: This e-mail is from an EXTERNAL e-mail address]

On Fri, Jan 29, 2021 at 04:04:42PM +0800, qiang.zhang@xxxxxxxxxxxxx wrote:
> From: Zqiang <qiang.zhang@xxxxxxxxxxxxx>
>
> Add free per-cpu existing krcp's page cache operation, when
> the system is under memory pressure.
>
> Signed-off-by: Zqiang <qiang.zhang@xxxxxxxxxxxxx>
> ---
>  kernel/rcu/tree.c | 25 +++++++++++++++++++++++++
>  1 file changed, 25 insertions(+)
>
> diff --git a/kernel/rcu/tree.c b/kernel/rcu/tree.c
> index c1ae1e52f638..ec098910d80b 100644
> --- a/kernel/rcu/tree.c
> +++ b/kernel/rcu/tree.c
> @@ -3571,17 +3571,40 @@ void kvfree_call_rcu(struct rcu_head *head, rcu_callback_t func)
>  }
>  EXPORT_SYMBOL_GPL(kvfree_call_rcu);
>
> +static int free_krc_page_cache(struct kfree_rcu_cpu *krcp)
> +{
> +     unsigned long flags;
> +     struct kvfree_rcu_bulk_data *bnode;
> +     int i;
> +
> +     for (i = 0; i < rcu_min_cached_objs; i++) {
> +             raw_spin_lock_irqsave(&krcp->lock, flags);
>I am not sure why we should disable IRQs. I think it can be >avoided.

Suppose in multi CPU system, the kfree_rcu_shrink_scan function is runing on CPU2,
and we just traverse to CPU2, and then call free_krc_page_cache function,
if not disable irq, a interrupt may be occurs on CPU2 after the CPU2 corresponds to krcp variable 's lock be acquired,  if the interrupt or softirq handler function to call kvfree_rcu function, in this function , acquire CPU2 corresponds to krcp variable 's lock , will happen deadlock.
Or in single CPU scenario.

> +             bnode = get_cached_bnode(krcp);
> +             raw_spin_unlock_irqrestore(&krcp->lock, flags);
> +             if (!bnode)
> +                     break;
> +             free_page((unsigned long)bnode);
> +     }
> +
> +     return i;
> +}
>Also i forgot to add in my previous comment to this path. Can we >access
>to page cache once and then do the drain work? I mean if we had >100 objects
>in the cache we would need to access to a krcp->lock 100 times.
>
>What about something like below:
>
><snip>
>static int free_krc_page_cache(struct kfree_rcu_cpu *krcp)
>{
>        struct llist_node *page_list, *pos, *n;
>        int freed = 0;
>
>        raw_spin_lock(&krcp->lock);
>        page_list = llist_del_all(&krcp->bkvcache);
>        krcp->nr_bkv_objs = 0;
>        raw_spin_unlock(&krcp->lock);
>
>        llist_for_each_safe(pos, n, page_list) {
>                free_page((unsigned long) pos);
>                freed++;
>        }
>
>        return freed;
>}
><snip>

  this change looks better.
  Thanks 
  Qiang
> +
>  static unsigned long
>  kfree_rcu_shrink_count(struct shrinker *shrink, struct shrink_control *sc)
>  {
>       int cpu;
>       unsigned long count = 0;
> +     unsigned long flags;
>
>       /* Snapshot count of all CPUs */
>       for_each_possible_cpu(cpu) {
>               struct kfree_rcu_cpu *krcp = per_cpu_ptr(&krc, cpu);
>
>               count += READ_ONCE(krcp->count);
> +
> +             raw_spin_lock_irqsave(&krcp->lock, flags);
> +             count += krcp->nr_bkv_objs;
> +             raw_spin_unlock_irqrestore(&krcp->lock, flags);
>Should we disable irqs?

>
>       return count;
> @@ -3598,6 +3621,8 @@ kfree_rcu_shrink_scan(struct shrinker *shrink, struct shrink_control *sc)
>               struct kfree_rcu_cpu *krcp = per_cpu_ptr(&krc, cpu);
>
>               count = krcp->count;
> +             count += free_krc_page_cache(krcp);
> +
>               raw_spin_lock_irqsave(&krcp->lock, flags);
>               if (krcp->monitor_todo)
>                       kfree_rcu_drain_unlock(krcp, flags);
> --
> 2.17.1

Thanks!

--
Vlad Rezki