Re: [RFC PATCH v4 03/14] rootns: Track init lifecycle via copy_process() and pid_namespace

From: Alexander Mikhalitsyn

Date: Fri Nov 20 2026 - 08:18:59 EST


On Fri, 2026-11-20 at 00:43 +0200, Jarkko Sakkinen wrote:
> A root namespace needs to record its init thread group's PID when
> bootstrapped and capture its exit status once all processes in the
> root
> namespace have terminated, without serializing normal fork() and
> exit()
> paths inside the container.
>
> Commit the bootstrap child reaper, bind pid_ns->rootns, and wake
> waiters
> in copy_rootns() after attach_pid(PIDTYPE_PID), and invoke
> exit_rootns()
> from zap_pid_ns_processes() after the root namespace's PID namespace
> has
> killed and reaped all member tasks.
>
> Co-developed-by: David Howells <dhowells@xxxxxxxxxx>
> Signed-off-by: David Howells <dhowells@xxxxxxxxxx>
> Signed-off-by: Jarkko Sakkinen <jarkko@xxxxxxxxxx>
> ---
>  include/linux/pid_namespace.h |  4 ++++
>  include/linux/rootns.h        | 15 ++++++++++++++
>  kernel/fork.c                 |  3 +++
>  kernel/pid_namespace.c        |  2 ++
>  kernel/rootns.c               | 38
> +++++++++++++++++++++++++++++++++++
>  5 files changed, 62 insertions(+)
>
> diff --git a/include/linux/pid_namespace.h
> b/include/linux/pid_namespace.h
> index b20baaa7e62b..6fd959f4d0d4 100644
> --- a/include/linux/pid_namespace.h
> +++ b/include/linux/pid_namespace.h
> @@ -15,6 +15,7 @@
>  #define MAX_PID_NS_LEVEL 32
>  
>  struct fs_pin;
> +struct rootns;
>  
>  #if defined(CONFIG_SYSCTL) && defined(CONFIG_MEMFD_CREATE)
>  /* modes for vm.memfd_noexec sysctl */
> @@ -45,6 +46,9 @@ struct pid_namespace {
>   struct user_namespace *user_ns;
>   struct ucounts *ucounts;
>   int reboot; /* group exit code if this pidns was
> rebooted */
> +#ifdef CONFIG_ROOTNS
> + struct rootns *rootns;
> +#endif
>   struct ns_common ns;
>   struct work_struct work;
>  } __randomize_layout;
> diff --git a/include/linux/rootns.h b/include/linux/rootns.h
> index 3f59415efadb..0557ce75195d 100644
> --- a/include/linux/rootns.h
> +++ b/include/linux/rootns.h
> @@ -13,6 +13,8 @@
>  struct file;
>  struct nsproxy;
>  struct pid;
> +struct pid_namespace;
> +struct task_struct;
>  struct user_namespace;
>  
>  enum rootns_state {
> @@ -41,6 +43,9 @@ struct rootns {
>  struct rootns *get_rootns(struct rootns *c);
>  void put_rootns(struct rootns *c);
>  bool is_rootns_file(struct file *file);
> +void copy_rootns(struct task_struct *tsk, struct pid *pid,
> + struct rootns *rootns);
> +void exit_rootns(struct pid_namespace *pid_ns, struct task_struct
> *tsk);
>  #else
>  static inline struct rootns *get_rootns(struct rootns *c)
>  {
> @@ -55,6 +60,16 @@ static inline bool is_rootns_file(struct file
> *file)
>  {
>   return false;
>  }
> +
> +static inline void copy_rootns(struct task_struct *tsk, struct pid
> *pid,
> +        struct rootns *rootns)
> +{
> +}
> +
> +static inline void exit_rootns(struct pid_namespace *pid_ns,
> +        struct task_struct *tsk)
> +{
> +}
>  #endif /* CONFIG_ROOTNS */
>  
>  #endif /* _LINUX_ROOTNS_H */
> diff --git a/kernel/fork.c b/kernel/fork.c
> index 5efceb8d58a7..08fc28ad1eed 100644
> --- a/kernel/fork.c
> +++ b/kernel/fork.c
> @@ -83,6 +83,7 @@
>  #include <linux/fs_struct.h>
>  #include <linux/magic.h>
>  #include <linux/perf_event.h>
> +#include <linux/rootns.h>
>  #include <linux/posix-timers.h>
>  #include <linux/user-return-notifier.h>
>  #include <linux/oom.h>
> @@ -2576,6 +2577,8 @@ __latent_entropy struct task_struct
> *copy_process(
>     &p->signal->thread_head);
>   }
>   attach_pid(p, PIDTYPE_PID);
> + if (thread_group_leader(p))
> + copy_rootns(p, pid, NULL);
>   nr_threads++;
>   }
>   total_forks++;
> diff --git a/kernel/pid_namespace.c b/kernel/pid_namespace.c
> index 127a5fd23e70..081c27aa85bd 100644
> --- a/kernel/pid_namespace.c
> +++ b/kernel/pid_namespace.c
> @@ -24,6 +24,7 @@
>  #include <linux/sched/signal.h>
>  #include <linux/idr.h>
>  #include <linux/nstree.h>
> +#include <linux/rootns.h>
>  #include <uapi/linux/wait.h>
>  #include "pid_sysctl.h"
>  
> @@ -280,6 +281,7 @@ void zap_pid_ns_processes(struct pid_namespace
> *pid_ns)
>   current->signal->group_exit_code = pid_ns->reboot;
>  

Dear Jarkko,

>   acct_exit_ns(pid_ns);
> + exit_rootns(pid_ns, me);

We have a very similar mechanism around pidfd's to track exit
information for a task. As far as I can see we already have (struct
rootns *)->init_pid reference, so we can get exit information for free
(look for PIDFD_INFO_EXIT).

Kind regards,
Alex

>   return;
>  }
>  
> diff --git a/kernel/rootns.c b/kernel/rootns.c
> index 3bbf7677080e..bde7cf4baebe 100644
> --- a/kernel/rootns.c
> +++ b/kernel/rootns.c
> @@ -183,3 +183,41 @@ bool is_rootns_file(struct file *file)
>  {
>   return file->f_op == &rootns_fops;
>  }
> +
> +void copy_rootns(struct task_struct *tsk, struct pid *pid, struct
> rootns *c)
> +{
> + if (likely(!c || !is_child_reaper(pid)))
> + return;
> +
> + scoped_guard(spinlock, &c->lock) {
> + if (c->state != ROOTNS_FORKING)
> + return;
> + ns_of_pid(pid)->rootns = get_rootns(c);
> + WRITE_ONCE(c->init_pid, get_pid(pid));
> + WRITE_ONCE(c->state, ROOTNS_RUNNING);
> + }
> + wake_up_all(&c->waitq);
> +}
> +
> +void exit_rootns(struct pid_namespace *pid_ns, struct task_struct
> *tsk)
> +{
> + struct rootns *c = pid_ns->rootns;
> + struct task_struct *leader;
> + int status;
> +
> + if (!c)
> + return;
> +
> + leader = tsk->group_leader;
> + status = (tsk->signal->flags & SIGNAL_GROUP_EXIT) ?
> + tsk->signal->group_exit_code : leader->exit_code;
> +
> + scoped_guard(spinlock, &c->lock) {
> + c->exit_code = status;
> + WRITE_ONCE(c->state, ROOTNS_DEAD);
> + }
> +
> + pid_ns->rootns = NULL;
> + wake_up_poll(&c->waitq, EPOLLHUP);
> + put_rootns(c);
> +}