Hi,
On 6/21/2023 12:15 AM, Alexei Starovoitov wrote:
> On Mon, Jun 19, 2023 at 7:00 AM Hou Tao <houtao@xxxxxxxxxxxxxxx> wrote:
>> +static void bpf_rcu_gp_acc_work(struct callback_head *head)
>> +{
>> + struct bpf_rcu_gp_acc_ctx *ctx = container_of(head, struct bpf_rcu_gp_acc_ctx, work);
>> +
>> + local_irq_disable();
>> + rcu_momentary_dyntick_idle();
>> + local_irq_enable();
> We discussed this with Paul off-line and decided to go a different route.
"A different route" means the method used to reduce the memory footprint
is different or the method to do reuse-after-rcu-gp is different ?
> Paul prepared a patch for us to expose rcu_request_urgent_qs_task().
> I'll be sending the series later this week.
Do you plan to take over the reuse-after-rcu-gp patchset ?
I did a quick test, it showed that the memory footprint is smaller and
the performance is similar when using rcu_request_urgent_qs_task()
instead of rcu_momentary_dyntick_idle().
overwrite:
Summary: per-prod-op 27.25 ± 0.08k/s, memory usage 29.52 ±
1.60MiB, peak memory usage 37.84MiB
batch_add_batch_del:
Summary: per-prod-op 45.84 ± 0.29k/s, memory usage 24.18 ±
1.44MiB, peak memory usage 30.38MiB
add_del_on_diff_cpu:
Summary: per-prod-op 13.09 ± 0.33k/s, memory usage 32.48 ±
3.51MiB, peak memory usage 53.54MiB
