[PATCH] kfence: unconditionally use unbound work queue

[PATCH] kfence: unconditionally use unbound work queue

[Marco Elver]

Unconditionally use unbound work queue, and not just if
wq_power_efficient is true. Because if the system is idle, KFENCE may
wait, and by being run on the unbound work queue, we permit the
scheduler to make better scheduling decisions and not require pinning
KFENCE to the same CPU upon waking up.

Fixes: 36f0b35d0894 ("kfence: use power-efficient work queue to run delayed work")
Reported-by: Hillf Danton <hdanton@sina.com>
Signed-off-by: Marco Elver <elver@google.com>
---
 mm/kfence/core.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/mm/kfence/core.c b/mm/kfence/core.c
index 4d21ac44d5d3..d7666ace9d2e 100644
--- a/mm/kfence/core.c
+++ b/mm/kfence/core.c
@@ -636,7 +636,7 @@ static void toggle_allocation_gate(struct work_struct *work)
 	/* Disable static key and reset timer. */
 	static_branch_disable(&kfence_allocation_key);
 #endif
-	queue_delayed_work(system_power_efficient_wq, &kfence_timer,
+	queue_delayed_work(system_unbound_wq, &kfence_timer,
 			   msecs_to_jiffies(kfence_sample_interval));
 }
 static DECLARE_DELAYED_WORK(kfence_timer, toggle_allocation_gate);
@@ -666,7 +666,7 @@ void __init kfence_init(void)
 	}
 
 	WRITE_ONCE(kfence_enabled, true);
-	queue_delayed_work(system_power_efficient_wq, &kfence_timer, 0);
+	queue_delayed_work(system_unbound_wq, &kfence_timer, 0);
 	pr_info("initialized - using %lu bytes for %d objects at 0x%p-0x%p\n", KFENCE_POOL_SIZE,
 		CONFIG_KFENCE_NUM_OBJECTS, (void *)__kfence_pool,
 		(void *)(__kfence_pool + KFENCE_POOL_SIZE));
-- 
2.31.1.818.g46aad6cb9e-goog

Re: [PATCH] kfence: unconditionally use unbound work queue

[Alexander Potapenko]

On Fri, May 21, 2021 at 1:16 PM Marco Elver <elver@google.com> wrote:
>
> Unconditionally use unbound work queue, and not just if
> wq_power_efficient is true. Because if the system is idle, KFENCE may
> wait, and by being run on the unbound work queue, we permit the
> scheduler to make better scheduling decisions and not require pinning
> KFENCE to the same CPU upon waking up.
>
> Fixes: 36f0b35d0894 ("kfence: use power-efficient work queue to run delayed work")
> Reported-by: Hillf Danton <hdanton@sina.com>
> Signed-off-by: Marco Elver <elver@google.com>
Reviewed-by: Alexander Potapenko <glider@google.com>

> ---
>  mm/kfence/core.c | 4 ++--
>  1 file changed, 2 insertions(+), 2 deletions(-)
>
> diff --git a/mm/kfence/core.c b/mm/kfence/core.c
> index 4d21ac44d5d3..d7666ace9d2e 100644
> --- a/mm/kfence/core.c
> +++ b/mm/kfence/core.c
> @@ -636,7 +636,7 @@ static void toggle_allocation_gate(struct work_struct *work)
>         /* Disable static key and reset timer. */
>         static_branch_disable(&kfence_allocation_key);
>  #endif
> -       queue_delayed_work(system_power_efficient_wq, &kfence_timer,
> +       queue_delayed_work(system_unbound_wq, &kfence_timer,
>                            msecs_to_jiffies(kfence_sample_interval));
>  }
>  static DECLARE_DELAYED_WORK(kfence_timer, toggle_allocation_gate);
> @@ -666,7 +666,7 @@ void __init kfence_init(void)
>         }
>
>         WRITE_ONCE(kfence_enabled, true);
> -       queue_delayed_work(system_power_efficient_wq, &kfence_timer, 0);
> +       queue_delayed_work(system_unbound_wq, &kfence_timer, 0);
>         pr_info("initialized - using %lu bytes for %d objects at 0x%p-0x%p\n", KFENCE_POOL_SIZE,
>                 CONFIG_KFENCE_NUM_OBJECTS, (void *)__kfence_pool,
>                 (void *)(__kfence_pool + KFENCE_POOL_SIZE));
> --
> 2.31.1.818.g46aad6cb9e-goog
>


-- 
Alexander Potapenko
Software Engineer

Google Germany GmbH
Erika-Mann-Straße, 33
80636 München

Geschäftsführer: Paul Manicle, Halimah DeLaine Prado
Registergericht und -nummer: Hamburg, HRB 86891
Sitz der Gesellschaft: Hamburg