2021-12-28 08:05:05

by Zqiang

Subject: [PATCH] rcu: per-cpu rcuc kthread are created only when rcutree.use_softirq=0

In non-RT kernel, if the RCU_BOOST is enabled, the per-cpu rcuc
kthread will be created, however under the rcutree.use_softirq=1,
the RCU core processing only in softirq context, the rcuc kthread
doesn't do anything, so remove RCU_BOOST interference.

Signed-off-by: Zqiang <[email protected]>
---
kernel/rcu/tree.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/kernel/rcu/tree.c b/kernel/rcu/tree.c
index 9b58bae0527a..ed64b4fcb8a0 100644
--- a/kernel/rcu/tree.c
+++ b/kernel/rcu/tree.c
@@ -2902,7 +2902,7 @@ static int __init rcu_spawn_core_kthreads(void)

for_each_possible_cpu(cpu)
per_cpu(rcu_data.rcu_cpu_has_work, cpu) = 0;
- if (!IS_ENABLED(CONFIG_RCU_BOOST) && use_softirq)
+ if (use_softirq)
return 0;
WARN_ONCE(smpboot_register_percpu_thread(&rcu_cpu_thread_spec),
"%s: Could not start rcuc kthread, OOM is now expected behavior\n", __func__);
--
2.25.1

2022-01-01 16:44:32

by Paul E. McKenney

Subject: Re: [PATCH] rcu: per-cpu rcuc kthread are created only when rcutree.use_softirq=0

On Wed, Dec 29, 2021 at 12:05:10AM +0800, Zqiang wrote:
> In non-RT kernel, if the RCU_BOOST is enabled, the per-cpu rcuc
> kthread will be created, however under the rcutree.use_softirq=1,
> the RCU core processing only in softirq context, the rcuc kthread
> doesn't do anything, so remove RCU_BOOST interference.
>
> Signed-off-by: Zqiang <[email protected]>

Looks sane to me, but adding Sebastian on CC for his thoughts.

Thanx, Paul

> ---
> kernel/rcu/tree.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/kernel/rcu/tree.c b/kernel/rcu/tree.c
> index 9b58bae0527a..ed64b4fcb8a0 100644
> --- a/kernel/rcu/tree.c
> +++ b/kernel/rcu/tree.c
> @@ -2902,7 +2902,7 @@ static int __init rcu_spawn_core_kthreads(void)
>
> for_each_possible_cpu(cpu)
> per_cpu(rcu_data.rcu_cpu_has_work, cpu) = 0;
> - if (!IS_ENABLED(CONFIG_RCU_BOOST) && use_softirq)
> + if (use_softirq)
> return 0;
> WARN_ONCE(smpboot_register_percpu_thread(&rcu_cpu_thread_spec),
> "%s: Could not start rcuc kthread, OOM is now expected behavior\n", __func__);
> --
> 2.25.1
>

2022-01-03 13:17:16

by Sebastian Andrzej Siewior

Subject: Re: [PATCH] rcu: per-cpu rcuc kthread are created only when rcutree.use_softirq=0

On 2022-01-01 08:44:26 [-0800], Paul E. McKenney wrote:
> On Wed, Dec 29, 2021 at 12:05:10AM +0800, Zqiang wrote:
> > In non-RT kernel, if the RCU_BOOST is enabled, the per-cpu rcuc
> > kthread will be created, however under the rcutree.use_softirq=1,
> > the RCU core processing only in softirq context, the rcuc kthread
> > doesn't do anything, so remove RCU_BOOST interference.
> >
> > Signed-off-by: Zqiang <[email protected]>
>
> Looks sane to me, but adding Sebastian on CC for his thoughts.

Yes, it makes sense. invoke_rcu_core_kthread() is only invoked for
!use_softirq so it makes to create the threads based on this condition.

Acked-by: Sebastian Andrzej Siewior <[email protected]>

> Thanx, Paul

Sebastian

2022-01-03 14:36:37

by Paul E. McKenney

Subject: Re: [PATCH] rcu: per-cpu rcuc kthread are created only when rcutree.use_softirq=0

On Mon, Jan 03, 2022 at 02:17:11PM +0100, Sebastian Andrzej Siewior wrote:
> On 2022-01-01 08:44:26 [-0800], Paul E. McKenney wrote:
> > On Wed, Dec 29, 2021 at 12:05:10AM +0800, Zqiang wrote:
> > > In non-RT kernel, if the RCU_BOOST is enabled, the per-cpu rcuc
> > > kthread will be created, however under the rcutree.use_softirq=1,
> > > the RCU core processing only in softirq context, the rcuc kthread
> > > doesn't do anything, so remove RCU_BOOST interference.
> > >
> > > Signed-off-by: Zqiang <[email protected]>
> >
> > Looks sane to me, but adding Sebastian on CC for his thoughts.
>
> Yes, it makes sense. invoke_rcu_core_kthread() is only invoked for
> !use_softirq so it makes to create the threads based on this condition.
>
> Acked-by: Sebastian Andrzej Siewior <[email protected]>

Thank you both! As usual, I could not resist a bit of wordsmithing,
please see below.

Thanx, Paul

------------------------------------------------------------------------

commit 6e68b781388cfaca95a07493a060c4a6e4ee5d0f
Author: Zqiang <[email protected]>
Date: Wed Dec 29 00:05:10 2021 +0800

rcu: Create per-cpu rcuc kthreads only when rcutree.use_softirq=0

The per-CPU "rcuc" kthreads are used only by kernels booted with
rcutree.use_softirq=0, but they are nevertheless unconditionally created
by kernels built with CONFIG_RCU_BOOST=y. This results in "rcuc"
kthreads being created that are never actually used. This commit
therefore refrains from creating these kthreads unless the kernel
is actually booted with rcutree.use_softirq=0.

Acked-by: Sebastian Andrzej Siewior <[email protected]>
Signed-off-by: Zqiang <[email protected]>
Signed-off-by: Paul E. McKenney <[email protected]>

diff --git a/kernel/rcu/tree.c b/kernel/rcu/tree.c
index aeea5487632f7..5c0ffee7a43a0 100644
--- a/kernel/rcu/tree.c
+++ b/kernel/rcu/tree.c
@@ -2903,7 +2903,7 @@ static int __init rcu_spawn_core_kthreads(void)

for_each_possible_cpu(cpu)
per_cpu(rcu_data.rcu_cpu_has_work, cpu) = 0;
- if (!IS_ENABLED(CONFIG_RCU_BOOST) && use_softirq)
+ if (use_softirq)
return 0;
WARN_ONCE(smpboot_register_percpu_thread(&rcu_cpu_thread_spec),
"%s: Could not start rcuc kthread, OOM is now expected behavior\n", __func__);

2022-01-04 08:39:08

by Zqiang

Subject: RE: [PATCH] rcu: per-cpu rcuc kthread are created only when rcutree.use_softirq=0

On Mon, Jan 03, 2022 at 02:17:11PM +0100, Sebastian Andrzej Siewior wrote:
> On 2022-01-01 08:44:26 [-0800], Paul E. McKenney wrote:
> > On Wed, Dec 29, 2021 at 12:05:10AM +0800, Zqiang wrote:
> > > In non-RT kernel, if the RCU_BOOST is enabled, the per-cpu rcuc
> > > kthread will be created, however under the rcutree.use_softirq=1,
> > > the RCU core processing only in softirq context, the rcuc kthread
> > > doesn't do anything, so remove RCU_BOOST interference.
> > >
> > > Signed-off-by: Zqiang <[email protected]>
> >
> > Looks sane to me, but adding Sebastian on CC for his thoughts.
>
> Yes, it makes sense. invoke_rcu_core_kthread() is only invoked for
> !use_softirq so it makes to create the threads based on this condition.
>
> Acked-by: Sebastian Andrzej Siewior <[email protected]>

Thanks Sebastian review.

>Thank you both! As usual, I could not resist a bit of wordsmithing, please see below.

Hi Paul , thank you for your modification, your description is clearer.

Thanks,
Zqiang

>
> Thanx, Paul
>
>------------------------------------------------------------------------
>
>commit 6e68b781388cfaca95a07493a060c4a6e4ee5d0f
>Author: Zqiang <[email protected]>
>Date: Wed Dec 29 00:05:10 2021 +0800
>
> rcu: Create per-cpu rcuc kthreads only when rcutree.use_softirq=0
>
> The per-CPU "rcuc" kthreads are used only by kernels booted with
> rcutree.use_softirq=0, but they are nevertheless unconditionally created
> by kernels built with CONFIG_RCU_BOOST=y. This results in "rcuc"
> kthreads being created that are never actually used. This commit
> therefore refrains from creating these kthreads unless the kernel
> is actually booted with rcutree.use_softirq=0.

> Acked-by: Sebastian Andrzej Siewior <[email protected]>
> Signed-off-by: Zqiang <[email protected]>
> Signed-off-by: Paul E. McKenney <[email protected]>
>
>diff --git a/kernel/rcu/tree.c b/kernel/rcu/tree.c index aeea5487632f7..5c0ffee7a43a0 100644
>--- a/kernel/rcu/tree.c
>+++ b/kernel/rcu/tree.c
>@@ -2903,7 +2903,7 @@ static int __init rcu_spawn_core_kthreads(void)
>
> for_each_possible_cpu(cpu)
> per_cpu(rcu_data.rcu_cpu_has_work, cpu) = 0;
>- if (!IS_ENABLED(CONFIG_RCU_BOOST) && use_softirq)
>+ if (use_softirq)
> return 0;
> WARN_ONCE(smpboot_register_percpu_thread(&rcu_cpu_thread_spec),
> "%s: Could not start rcuc kthread, OOM is now expected behavior\n", __func__);