pid: reorder fields in pid_namespace to reduce false sharing

[PATCH] pid: reorder fields in pid_namespace to reduce false sharing

Posted by Mateusz Guzik 2 weeks, 3 days ago

alloc_pid() loads pid_cachep, level and pid_max prior to taking the
lock.

It dirties idr and pid_allocated with the lock.

Some of these fields share the cacheline as is, split them up.

No change in the size of the struct.

Signed-off-by: Mateusz Guzik <mjguzik@gmail.com>
---

this is independent of other patches

i got an inconsistent win in terms of throughput rate, but relative
contention between pidmap lock and the rest went down

 include/linux/pid_namespace.h | 14 +++++++-------
 1 file changed, 7 insertions(+), 7 deletions(-)

diff --git a/include/linux/pid_namespace.h b/include/linux/pid_namespace.h
index 0e7ae12c96d2..b20baaa7e62b 100644
--- a/include/linux/pid_namespace.h
+++ b/include/linux/pid_namespace.h
@@ -27,6 +27,13 @@ struct pid_namespace {
 	struct idr idr;
 	struct rcu_head rcu;
 	unsigned int pid_allocated;
+#ifdef CONFIG_SYSCTL
+#if defined(CONFIG_MEMFD_CREATE)
+	int memfd_noexec_scope;
+#endif
+	struct ctl_table_set	set;
+	struct ctl_table_header *sysctls;
+#endif
 	struct task_struct *child_reaper;
 	struct kmem_cache *pid_cachep;
 	unsigned int level;
@@ -40,13 +47,6 @@ struct pid_namespace {
 	int reboot;	/* group exit code if this pidns was rebooted */
 	struct ns_common ns;
 	struct work_struct	work;
-#ifdef CONFIG_SYSCTL
-	struct ctl_table_set	set;
-	struct ctl_table_header *sysctls;
-#if defined(CONFIG_MEMFD_CREATE)
-	int memfd_noexec_scope;
-#endif
-#endif
 } __randomize_layout;
 
 extern struct pid_namespace init_pid_ns;
-- 
2.48.1

Re: [PATCH] pid: reorder fields in pid_namespace to reduce false sharing

Posted by Christian Brauner 2 weeks, 1 day ago

On Tue, 20 Jan 2026 21:48:20 +0100, Mateusz Guzik wrote:
> alloc_pid() loads pid_cachep, level and pid_max prior to taking the
> lock.
> 
> It dirties idr and pid_allocated with the lock.
> 
> Some of these fields share the cacheline as is, split them up.
> 
> [...]

Applied to the kernel-7.0.misc branch of the vfs/vfs.git tree.
Patches in the kernel-7.0.misc branch should appear in linux-next soon.

Please report any outstanding bugs that were missed during review in a
new review to the original patch series allowing us to drop it.

It's encouraged to provide Acked-bys and Reviewed-bys even though the
patch has now been applied. If possible patch trailers will be updated.

Note that commit hashes shown below are subject to change due to rebase,
trailer updates or similar. If in doubt, please check the listed branch.

tree:   https://git.kernel.org/pub/scm/linux/kernel/git/vfs/vfs.git
branch: kernel-7.0.misc

[1/1] pid: reorder fields in pid_namespace to reduce false sharing
      https://git.kernel.org/vfs/vfs/c/e7d0463629d1

Re: [PATCH] pid: reorder fields in pid_namespace to reduce false sharing

Posted by Jan Kara 2 weeks, 3 days ago

On Tue 20-01-26 21:48:20, Mateusz Guzik wrote:
> alloc_pid() loads pid_cachep, level and pid_max prior to taking the
> lock.
> 
> It dirties idr and pid_allocated with the lock.
> 
> Some of these fields share the cacheline as is, split them up.
> 
> No change in the size of the struct.
> 
> Signed-off-by: Mateusz Guzik <mjguzik@gmail.com>

Looks ok. Feel free to add:

Reviewed-by: Jan Kara <jack@suse.cz>

								Honza

> ---
> 
> this is independent of other patches
> 
> i got an inconsistent win in terms of throughput rate, but relative
> contention between pidmap lock and the rest went down
> 
>  include/linux/pid_namespace.h | 14 +++++++-------
>  1 file changed, 7 insertions(+), 7 deletions(-)
> 
> diff --git a/include/linux/pid_namespace.h b/include/linux/pid_namespace.h
> index 0e7ae12c96d2..b20baaa7e62b 100644
> --- a/include/linux/pid_namespace.h
> +++ b/include/linux/pid_namespace.h
> @@ -27,6 +27,13 @@ struct pid_namespace {
>  	struct idr idr;
>  	struct rcu_head rcu;
>  	unsigned int pid_allocated;
> +#ifdef CONFIG_SYSCTL
> +#if defined(CONFIG_MEMFD_CREATE)
> +	int memfd_noexec_scope;
> +#endif
> +	struct ctl_table_set	set;
> +	struct ctl_table_header *sysctls;
> +#endif
>  	struct task_struct *child_reaper;
>  	struct kmem_cache *pid_cachep;
>  	unsigned int level;
> @@ -40,13 +47,6 @@ struct pid_namespace {
>  	int reboot;	/* group exit code if this pidns was rebooted */
>  	struct ns_common ns;
>  	struct work_struct	work;
> -#ifdef CONFIG_SYSCTL
> -	struct ctl_table_set	set;
> -	struct ctl_table_header *sysctls;
> -#if defined(CONFIG_MEMFD_CREATE)
> -	int memfd_noexec_scope;
> -#endif
> -#endif
>  } __randomize_layout;
>  
>  extern struct pid_namespace init_pid_ns;
> -- 
> 2.48.1
> 
-- 
Jan Kara <jack@suse.com>
SUSE Labs, CR