start the global /sbin/init with 0,0 special pids
Oleg Nesterov [Fri, 8 Feb 2008 12:19:11 +0000 (04:19 -0800)]
As Eric pointed out, there is no problem with init starting with sid == pgid
== 0, and this was historical linux behavior changed in 2.6.18.

Remove kernel_init()->__set_special_pids(), this is unneeded and complicates
the rules for sys_setsid().

This change and the previous change in daemonize() mean that /sbin/init does
not need the special "session != 1" hack in sys_setsid() any longer. We can't
remove this check yet, we should cleanup copy_process(CLONE_NEWPID) first, so
update the comment only.

Signed-off-by: Oleg Nesterov <oleg@tv-sign.ru>
Acked-by: "Eric W. Biederman" <ebiederm@xmission.com>
Cc: Pavel Emelyanov <xemul@openvz.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

init/main.c
kernel/sys.c

index 9a5b18c..ed9747f 100644 (file)
@@ -833,7 +833,6 @@ static int __init kernel_init(void * unused)
         */
        init_pid_ns.child_reaper = current;
 
-       __set_special_pids(task_pid(current));
        cad_pid = task_pid(current);
 
        smp_prepare_cpus(setup_max_cpus);
index c326d6d..7de9c98 100644 (file)
@@ -1054,12 +1054,11 @@ asmlinkage long sys_setsid(void)
        if (group_leader->signal->leader)
                goto out;
 
-       /* Fail if a process group id already exists that equals the proposed
-        * session id.
+       /* Fail if a process group id already exists that equals the
+        * proposed session id.
         *
-        * Don't check if session == 1 because kernel threads and CLONE_NEWPID
-        * tasks use this session id and so the check will always fail and make
-        * it so init cannot successfully call setsid.
+        * Don't check if session == 1, clone(CLONE_NEWPID) creates
+        * this group/session beforehand.
         */
        if (session != 1 && pid_task(sid, PIDTYPE_PGID))
                goto out;