Re: Fix pacct bug in multithreading case.

[Roland McGrath]

That change looks correct to me, and I see that it's gone in.

It made me think of another issue affecting the accounting records in
related circumstances when the old group leader died after a non-leader exec.
I think this patch addresses that, and along with your patch, gives more
sensible times in all cases.


Thanks,
Roland


[PATCH] Take original leader's start_time in non-leader exec.

The only record we have of the real-time age of a process, regardless of
execs it's done, is start_time.  When a non-leader thread exec, the
original start_time of the process is lost.  Things looking at the
real-time age of the process are fooled, for example the process
accounting record when the process finally dies.  This change makes the
oldest start_time stick around with the process after a non-leader exec.
This way the association between PID and start_time is kept constant,
which seems correct to me.

Signed-off-by: Roland McGrath <roland@xxxxxxxxxx>

---

 fs/exec.c |   12 ++++++++++++
 1 files changed, 12 insertions(+), 0 deletions(-)

7cda52efb6ff969f049bc2ab3742b0341e45184a
diff --git a/fs/exec.c b/fs/exec.c
index 0291a68..a45f712 100644
--- a/fs/exec.c
+++ b/fs/exec.c
@@ -678,6 +678,18 @@ static int de_thread(struct task_struct 
 		while (leader->exit_state != EXIT_ZOMBIE)
 			yield();
 
+		/*
+		 * The only record we have of the real-time age of a
+		 * process, regardless of execs it's done, is start_time.
+		 * All the past CPU time is accumulated in signal_struct
+		 * from sister threads now dead.  But in this non-leader
+		 * exec, nothing survives from the original leader thread,
+		 * whose birth marks the true age of this process now.
+		 * When we take on its identity by switching to its PID, we
+		 * also take its birthdate (always earlier than our own).
+		 */
+		current->start_time = leader->start_time;
+
 		spin_lock(&leader->proc_lock);
 		spin_lock(&current->proc_lock);
 		proc_dentry1 = proc_pid_unhash(current);
-- 
1.2.4

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/