In a prior patch series we tried to cleanly differentiate between:
(1) The task has already been reaped.
(2) The caller requested a pidfd for a thread-group leader but the pid
actually references a struct pid that isn't used as a thread-group
leader.
as this was causing issues for non-threaded workloads.
But there's cases where the current simple logic is wrong. Specifically,
if the pid was a leader pid and the check races with __unhash_process().
Stabilize this by using the pidfd waitqueue lock.
Signed-off-by: Christian Brauner <brauner@xxxxxxxxxx>
---
Christian Brauner (2):
exit: move wake_up_all() pidfd waiters into __unhash_process()
pidfs: ensure consistent ENOENT/ESRCH reporting
kernel/exit.c | 5 +++++
kernel/fork.c | 31 +++++++++++++------------------
kernel/pid.c | 5 -----
3 files changed, 18 insertions(+), 23 deletions(-)
---
base-commit: 1e940fff94374d04b6c34f896ed9fbad3d2fb706
change-id: 20250411-work-pidfs-enoent-747579160c8c
