ptrace: permit ptracing of /sbin/init
Oleg Nesterov [Wed, 30 Apr 2008 07:53:15 +0000 (00:53 -0700)]
Afaics, currently there are no kernel problems with ptracing init, it can't
lose SIGNAL_UNKILLABLE flag and be killed/stopped by accident.

The ability to strace/debug init can be very useful if you try to figure out
why it does not work as expected.

However, admin should know what he does, "gdb /sbin/init 1" stops init, it
can't reap orphaned zombies or take care of /etc/inittab until continued.  It
is even possible to crash init (and thus the whole system) if you wish,
ptracer has full control.

See also the long discussion: http://marc.info/?t=120628018600001

Signed-off-by: Oleg Nesterov <oleg@tv-sign.ru>
Acked-by: Roland McGrath <roland@redhat.com>
Acked-by: Pavel Emelyanov <xemul@openvz.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

kernel/ptrace.c

index 5f8d452..dcc199c 100644 (file)
@@ -168,8 +168,6 @@ int ptrace_attach(struct task_struct *task)
        audit_ptrace(task);
 
        retval = -EPERM;
-       if (task->pid <= 1)
-               goto out;
        if (same_thread_group(task, current))
                goto out;
 
@@ -521,12 +519,6 @@ struct task_struct *ptrace_get_task_struct(pid_t pid)
 {
        struct task_struct *child;
 
-       /*
-        * Tracing init is not allowed.
-        */
-       if (pid == 1)
-               return ERR_PTR(-EPERM);
-
        read_lock(&tasklist_lock);
        child = find_task_by_vpid(pid);
        if (child)