From: Oleg Nesterov Date: Wed, 15 Oct 2025 12:36:13 +0000 (+0200) Subject: Revert "pid: make __task_pid_nr_ns(ns => NULL) safe for zombie callers" X-Git-Url: http://git.ipfire.org/cgi-bin/gitweb.cgi?a=commitdiff_plain;h=1cf2e88e0651dbd5fb7f5651c32d617de759b855;p=thirdparty%2Flinux.git Revert "pid: make __task_pid_nr_ns(ns => NULL) safe for zombie callers" This reverts commit abdfd4948e45c51b19162cf8b3f5003f8f53c9b9. The changelog in this commit explains why it is not easy to avoid ns == NULL when the caller is exiting, but pid_vnr() is equally unsafe in this case. However, commit 006568ab4c5c ("pid: Add a judgment for ns null in pid_nr_ns") already added the ns != NULL check in pid_nr_ns(), so we can remove the same check from __task_pid_nr_ns(). Signed-off-by: Oleg Nesterov Link: https://patch.msgid.link/20251015123613.GA9456@redhat.com Signed-off-by: Christian Brauner --- diff --git a/kernel/pid.c b/kernel/pid.c index 72c9372b84b8..3b96571d0fe6 100644 --- a/kernel/pid.c +++ b/kernel/pid.c @@ -558,8 +558,7 @@ pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type, rcu_read_lock(); if (!ns) ns = task_active_pid_ns(current); - if (ns) - nr = pid_nr_ns(rcu_dereference(*task_pid_ptr(task, type)), ns); + nr = pid_nr_ns(rcu_dereference(*task_pid_ptr(task, type)), ns); rcu_read_unlock(); return nr;