tracing, sched, vfs: Fix 'old_pid' usage in trace_sched_process_exec()

1. TRACE_EVENT(sched_process_exec) forgets to actually use the
   old pid argument, it sets ->old_pid = p->pid.

2. search_binary_handler() uses the wrong pid number. tracepoint
   needs the global pid_t from the root namespace, while old_pid
   is the virtual pid number as it seen by the tracer/parent.

With this patch we have two pid_t's in search_binary_handler(),
not really nice. Perhaps we should switch to "struct pid*", but
in this case it would be better to cleanup the current code
first and move the "depth == 0" code outside.

Signed-off-by: Oleg Nesterov <oleg@redhat.com>
Cc: David Smith <dsmith@redhat.com>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Denys Vlasenko <dvlasenk@redhat.com>
Link: http://lkml.kernel.org/r/20120330162636.GA4857@redhat.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
diff --git a/fs/exec.c b/fs/exec.c
index 23559c2..644f6c4 100644
--- a/fs/exec.c
+++ b/fs/exec.c
@@ -1370,7 +1370,7 @@
 	unsigned int depth = bprm->recursion_depth;
 	int try,retval;
 	struct linux_binfmt *fmt;
-	pid_t old_pid;
+	pid_t old_pid, old_vpid;
 
 	retval = security_bprm_check(bprm);
 	if (retval)
@@ -1381,8 +1381,9 @@
 		return retval;
 
 	/* Need to fetch pid before load_binary changes it */
+	old_pid = current->pid;
 	rcu_read_lock();
-	old_pid = task_pid_nr_ns(current, task_active_pid_ns(current->parent));
+	old_vpid = task_pid_nr_ns(current, task_active_pid_ns(current->parent));
 	rcu_read_unlock();
 
 	retval = -ENOENT;
@@ -1405,7 +1406,7 @@
 			if (retval >= 0) {
 				if (depth == 0) {
 					trace_sched_process_exec(current, old_pid, bprm);
-					ptrace_event(PTRACE_EVENT_EXEC, old_pid);
+					ptrace_event(PTRACE_EVENT_EXEC, old_vpid);
 				}
 				put_binfmt(fmt);
 				allow_write_access(bprm->file);