cpuset_rwsem is a static variable. It's initialized at build time and so
there's no need for explicit runtime init leaking one percpu int.
Signed-off-by: Daniel Vacek <[email protected]>
---
kernel/cgroup/cpuset.c | 2 --
1 file changed, 2 deletions(-)
diff --git a/kernel/cgroup/cpuset.c b/kernel/cgroup/cpuset.c
index a29c0b13706bb..87fe410361b3d 100644
--- a/kernel/cgroup/cpuset.c
+++ b/kernel/cgroup/cpuset.c
@@ -3281,8 +3281,6 @@ struct cgroup_subsys cpuset_cgrp_subsys = {
int __init cpuset_init(void)
{
- BUG_ON(percpu_init_rwsem(&cpuset_rwsem));
-
BUG_ON(!alloc_cpumask_var(&top_cpuset.cpus_allowed, GFP_KERNEL));
BUG_ON(!alloc_cpumask_var(&top_cpuset.effective_cpus, GFP_KERNEL));
BUG_ON(!zalloc_cpumask_var(&top_cpuset.subparts_cpus, GFP_KERNEL));
--
2.38.1
On 12/20/22 10:14, Daniel Vacek wrote:
> cpuset_rwsem is a static variable. It's initialized at build time and so
> there's no need for explicit runtime init leaking one percpu int.
>
> Signed-off-by: Daniel Vacek <[email protected]>
> ---
> kernel/cgroup/cpuset.c | 2 --
> 1 file changed, 2 deletions(-)
>
> diff --git a/kernel/cgroup/cpuset.c b/kernel/cgroup/cpuset.c
> index a29c0b13706bb..87fe410361b3d 100644
> --- a/kernel/cgroup/cpuset.c
> +++ b/kernel/cgroup/cpuset.c
> @@ -3281,8 +3281,6 @@ struct cgroup_subsys cpuset_cgrp_subsys = {
>
> int __init cpuset_init(void)
> {
> - BUG_ON(percpu_init_rwsem(&cpuset_rwsem));
> -
> BUG_ON(!alloc_cpumask_var(&top_cpuset.cpus_allowed, GFP_KERNEL));
> BUG_ON(!alloc_cpumask_var(&top_cpuset.effective_cpus, GFP_KERNEL));
> BUG_ON(!zalloc_cpumask_var(&top_cpuset.subparts_cpus, GFP_KERNEL));
It will be clearer if you mention that DEFINE_STATIC_PERCPU_RWSEM() is
used to set up cpuset_rwsem at build time. Other than that, the patch
looks good to me.
Cheers,
Longman
On Tue 2022-12-20 16:14 +0100, Daniel Vacek wrote:
> cpuset_rwsem is a static variable. It's initialized at build time and so
> there's no need for explicit runtime init leaking one percpu int.
>
> Signed-off-by: Daniel Vacek <[email protected]>
> ---
> kernel/cgroup/cpuset.c | 2 --
> 1 file changed, 2 deletions(-)
>
> diff --git a/kernel/cgroup/cpuset.c b/kernel/cgroup/cpuset.c
> index a29c0b13706bb..87fe410361b3d 100644
> --- a/kernel/cgroup/cpuset.c
> +++ b/kernel/cgroup/cpuset.c
> @@ -3281,8 +3281,6 @@ struct cgroup_subsys cpuset_cgrp_subsys = {
>
> int __init cpuset_init(void)
> {
> - BUG_ON(percpu_init_rwsem(&cpuset_rwsem));
> -
> BUG_ON(!alloc_cpumask_var(&top_cpuset.cpus_allowed, GFP_KERNEL));
> BUG_ON(!alloc_cpumask_var(&top_cpuset.effective_cpus, GFP_KERNEL));
> BUG_ON(!zalloc_cpumask_var(&top_cpuset.subparts_cpus, GFP_KERNEL));
> --
> 2.38.1
Good catch.
Reviewed-by: Aaron Tomlin <[email protected]>
--
Aaron Tomlin
Hi,
On 12/20/2022 8:44 PM, Daniel Vacek wrote:
> cpuset_rwsem is a static variable. It's initialized at build time and so
> there's no need for explicit runtime init leaking one percpu int.
>
> Signed-off-by: Daniel Vacek <[email protected]>
> ---
> kernel/cgroup/cpuset.c | 2 --
> 1 file changed, 2 deletions(-)
>
> diff --git a/kernel/cgroup/cpuset.c b/kernel/cgroup/cpuset.c
> index a29c0b13706bb..87fe410361b3d 100644
> --- a/kernel/cgroup/cpuset.c
> +++ b/kernel/cgroup/cpuset.c
> @@ -3281,8 +3281,6 @@ struct cgroup_subsys cpuset_cgrp_subsys = {
>
> int __init cpuset_init(void)
> {
> - BUG_ON(percpu_init_rwsem(&cpuset_rwsem));
> -
> BUG_ON(!alloc_cpumask_var(&top_cpuset.cpus_allowed, GFP_KERNEL));
> BUG_ON(!alloc_cpumask_var(&top_cpuset.effective_cpus, GFP_KERNEL));
> BUG_ON(!zalloc_cpumask_var(&top_cpuset.subparts_cpus, GFP_KERNEL));
Acked-by: Mukesh Ojha <[email protected]>
-Mukesh
On Tue, Dec 20, 2022 at 5:59 PM Waiman Long <[email protected]> wrote:
>
> On 12/20/22 10:14, Daniel Vacek wrote:
> > cpuset_rwsem is a static variable. It's initialized at build time and so
> > there's no need for explicit runtime init leaking one percpu int.
>
> It will be clearer if you mention that DEFINE_STATIC_PERCPU_RWSEM() is
> used to set up cpuset_rwsem at build time. Other than that, the patch
> looks good to me.
That's true. I only figured later.
Whoever is going to apply it, feel free to amend the message if you like.
--nX
> Cheers,
> Longman
>
On Tue, Dec 20, 2022 at 04:14:15PM +0100, Daniel Vacek wrote:
> cpuset_rwsem is a static variable. It's initialized at build time and so
> there's no need for explicit runtime init leaking one percpu int.
>
> Signed-off-by: Daniel Vacek <[email protected]>
Applied to cgroup/for-6.2-fixes with the following description:
cgroup/cpuset: no need to explicitly init a global static variable
cpuset_rwsem is a static variable defined with DEFINE_STATIC_PERCPU_RWSEM().
It's initialized at build time and so there's no need for explicit runtime
init leaking one percpu int.
Signed-off-by: Daniel Vacek <[email protected]>
Reviewed-by: Aaron Tomlin <[email protected]>
Acked-by: Mukesh Ojha <[email protected]>
Signed-off-by: Tejun Heo <[email protected]>
Thanks.
--
tejun