Threads also have an exit code on their own, so report it in
TASKSTATS_CMD_ATTR_PID.

For TASKSTATS_CMD_ATTR_TGID, instead of relying only on the exit code of the
leader, we use task->signal->group_exit_code if not null as suggested by
Oleg Nesterov.

Also, document that as of this patch, fill_threadgroup() must be called after
add_tsk() as it may overwrite some stats.

Signed-off-by: Guillaume Chazarain <[EMAIL PROTECTED]>
Cc: Balbir Singh <[EMAIL PROTECTED]>
Cc: Jay Lan <[EMAIL PROTECTED]>
Cc: Jonathan Lim <[EMAIL PROTECTED]>
Cc: Oleg Nesterov <[EMAIL PROTECTED]>
---

 kernel/taskstats.c |    3 +++
 kernel/tsacct.c    |   12 +++++++-----
 2 files changed, 10 insertions(+), 5 deletions(-)

diff --git a/kernel/taskstats.c b/kernel/taskstats.c
index 42d3110..24d7f62 100644
--- a/kernel/taskstats.c
+++ b/kernel/taskstats.c
@@ -181,6 +181,9 @@ static void send_cpu_listeners(struct sk_buff *skb,
  *     memory usage), so are taken from the group leader.
  * XXX_threadgroup() methods deal with the first type while XXX_add_tsk() with
  * the second.
+ *
+ * fill_threadgroup() may overwrite stats from add_tsk(), so it must be called
+ * after add_tsk().
  */
 static void fill_threadgroup(struct taskstats *stats, struct task_struct *task,
                             bool tg_stats)
diff --git a/kernel/tsacct.c b/kernel/tsacct.c
index 24056aa..526b134 100644
--- a/kernel/tsacct.c
+++ b/kernel/tsacct.c
@@ -44,6 +44,8 @@ static void fill_wall_times(struct taskstats *stats, struct 
task_struct *tsk)
 void bacct_fill_threadgroup(struct taskstats *stats, struct task_struct *task,
                            bool tg_stats)
 {
+       int group_exit_code;
+
        BUILD_BUG_ON(TS_COMM_LEN < TASK_COMM_LEN);
 
        rcu_read_lock();
@@ -53,11 +55,11 @@ void bacct_fill_threadgroup(struct taskstats *stats, struct 
task_struct *task,
 
        fill_wall_times(stats, task);
 
-       if (thread_group_leader(task)) {
-               stats->ac_exitcode = task->exit_code;
-               if (task->flags & PF_FORKNOEXEC)
-                       stats->ac_flag |= AFORK;
-       }
+       if (thread_group_leader(task) && (task->flags & PF_FORKNOEXEC))
+               stats->ac_flag |= AFORK;
+
+       group_exit_code = tg_stats ? task->signal->group_exit_code : 0;
+       stats->ac_exitcode      = group_exit_code ? : task->exit_code;
 
        stats->ac_nice          = task_nice(task);
        stats->ac_sched         = task->policy;

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

Reply via email to