pidfs: provide information after task has been reaped

Hey,

Various tools need access to information about a process/task even after
it has already been reaped. For example, systemd's journal logs and uses
such information as the cgroup id and exit status to deal with processes
that have been sent via SCM_PIDFD or SCM_PEERPIDFD. By the time the
pidfd is received the process might have already been reaped.

This series aims to provide information by extending the PIDFD_GET_INFO
ioctl to retrieve the exit code and cgroup id. There might be other
stuff that we would want in the future.

Pidfd polling allows waiting on either task exit or for a task to have
been reaped. The contract for PIDFD_INFO_EXIT is simply that EPOLLHUP
must be observed before exit information can be retrieved, i.e., exit
information is only provided once the task has been reaped.

Note, that if a thread-group leader exits before other threads in the
thread-group then exit information will only be available once the
thread-group is empty. This aligns with wait() as well, where reaping of
a thread-group leader that exited before the thread-group was empty is
delayed until the thread-group is empty.

With PIDFD_INFO_EXIT autoreaping might actually become usable because it
means a parent can ignore SIGCHLD or set SA_NOCLDWAIT and simply use
pidfd polling and PIDFD_INFO_EXIT to get get status information for its
children. The kernel will autocleanup right away instead of delaying.

To: Oleg Nesterov <oleg@redhat.com>
Cc: linux-fsdevel@vger.kernel.org
Cc: Jeff Layton <jlayton@kernel.org>
Cc: Lennart Poettering <lennart@poettering.net>
Cc: Daan De Meyer <daan.j.demeyer@gmail.com>
Cc: Mike Yuan <me@yhndnzj.com>
Signed-off-by: Christian Brauner <brauner@kernel.org>

---
Changes in v3:
- Fix various minor issues.
- Link to v2: https://lore.kernel.org/r/20250304-work-pidfs-kill_on_last_close-v2-0-44fdacfaa7b7@kernel.org

Changes in v2:
- Call pidfs_exit() from release_task().
- Don't provide exit information once the task has exited but once the
  task has been reaped. This makes for simpler semantics. Thus, call
  pidfs_exit() from release_task().
- Link to v1: https://lore.kernel.org/r/20250228-work-pidfs-kill_on_last_close-v1-0-5bd7e6bb428e@kernel.org

--- b4-submit-tracking ---
# This section is used internally by b4 prep for tracking purposes.
{
  "series": {
    "revision": 3,
    "change-id": "20250227-work-pidfs-kill_on_last_close-a23ddf21db47",
    "prefixes": [],
    "base-branch": "vfs-6.15.pidfs",
    "history": {
      "v1": [
        "20250228-work-pidfs-kill_on_last_close-v1-0-5bd7e6bb428e@kernel.org"
      ],
      "v2": [
        "20250304-work-pidfs-kill_on_last_close-v2-0-44fdacfaa7b7@kernel.org"
      ]
    }
  }
}