pidns: make pid->level and pid_ns->level unsigned

These values represent the nesting level of a namespace and pids living in it,
and it's always non-negative.

Turning this from int to unsigned int saves some space in pid.c (11 bytes on
x86 and 64 on ia64) by letting the compiler optimize the pid_nr_ns a bit.
E.g.  on ia64 this removes the sign extension calls, which compiler adds to
optimize access to pid->nubers[ns->level].
Signed-off-by: NPavel Emelyanov <xemul@openvz.org>
Cc: "Eric W. Biederman" <ebiederm@xmission.com>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

include/linux/pid.h

@@ -60,7 +60,7 @@ struct pid
 	/* lists of tasks that use this pid */
 	struct hlist_head tasks[PIDTYPE_MAX];
 	struct rcu_head rcu;
-	int level;
+	unsigned int level;
 	struct upid numbers[1];
 };
 

include/linux/pid_namespace.h

@@ -20,7 +20,7 @@ struct pid_namespace {
 	int last_pid;
 	struct task_struct *child_reaper;
 	struct kmem_cache *pid_cachep;
-	int level;
+	unsigned int level;
 	struct pid_namespace *parent;
 #ifdef CONFIG_PROC_FS
 	struct vfsmount *proc_mnt;

kernel/pid_namespace.c

@@ -66,7 +66,7 @@ static struct kmem_cache *create_pid_cachep(int nr_ids)
 	return NULL;
 }
 
-static struct pid_namespace *create_pid_namespace(int level)
+static struct pid_namespace *create_pid_namespace(unsigned int level)
 {
 	struct pid_namespace *ns;
 	int i;