The patch titled
sched_exit: move the callsite to do_exit()
has been added to the -mm tree. Its filename is
sched_exit-move-the-callsite-to-do_exit.patch
See http://www.zip.com.au/~akpm/linux/patches/stuff/added-to-mm.txt to find
out what to do about this
------------------------------------------------------
Subject: sched_exit: move the callsite to do_exit()
From: Oleg Nesterov <oleg@xxxxxxxxxx>
sched_exit() is called by release_task(). If the task auto-reaps itself
this call happens a bit too early (the task still uses cpu after that). If
the task goes to TASK_ZOMBIE this call is unpredictably delayed.
I think it is better to do sched_exit() right before the last schedule().
We can use read_lock_rcu() instead of write_lock(tasklist). In that case
it is possible that sched_exit() changes ->time_slice/->sleep_avg of the
already dead ->parent, but I think this is tolerable.
Signed-off-by: Oleg Nesterov <oleg@xxxxxxxxxx>
Cc: Ingo Molnar <mingo@xxxxxxx>
Cc: Nick Piggin <nickpiggin@xxxxxxxxxxxx>
Cc: Con Kolivas <kernel@xxxxxxxxxxx>
Cc: Peter Williams <pwil3058@xxxxxxxxxxxxxx>
Cc: "Paul E. McKenney" <paulmck@xxxxxxxxxx>
Signed-off-by: Andrew Morton <akpm@xxxxxxxx>
---
kernel/exit.c | 3 ++-
kernel/sched.c | 6 +++++-
2 files changed, 7 insertions(+), 2 deletions(-)
diff -puN kernel/exit.c~sched_exit-move-the-callsite-to-do_exit kernel/exit.c
--- a/kernel/exit.c~sched_exit-move-the-callsite-to-do_exit
+++ a/kernel/exit.c
@@ -171,7 +171,6 @@ repeat:
zap_leader = (leader->exit_signal == -1);
}
- sched_exit(p);
write_unlock_irq(&tasklist_lock);
spin_unlock(&p->proc_lock);
proc_pid_flush(proc_dentry);
@@ -950,6 +949,8 @@ fastcall NORET_TYPE void do_exit(long co
if (tsk->splice_pipe)
__free_pipe_info(tsk->splice_pipe);
+ sched_exit(tsk);
+
/* PF_DEAD causes final put_task_struct after we schedule. */
preempt_disable();
BUG_ON(tsk->flags & PF_DEAD);
diff -puN kernel/sched.c~sched_exit-move-the-callsite-to-do_exit kernel/sched.c
--- a/kernel/sched.c~sched_exit-move-the-callsite-to-do_exit
+++ a/kernel/sched.c
@@ -1606,10 +1606,13 @@ void fastcall wake_up_new_task(task_t *p
*/
void fastcall sched_exit(task_t *p)
{
- task_t *parent = p->parent;
+ task_t *parent;
unsigned long flags;
runqueue_t *rq;
+ rcu_read_lock();
+ parent = p->real_parent;
+
/*
* If the child was a (relative-) CPU hog then decrease
* the sleep_avg of the parent as well.
@@ -1625,6 +1628,7 @@ void fastcall sched_exit(task_t *p)
(EXIT_WEIGHT + 1) * EXIT_WEIGHT + p->sleep_avg /
(EXIT_WEIGHT + 1);
task_rq_unlock(rq, &flags);
+ rcu_read_unlock();
}
/**
_
Patches currently in -mm which might be from oleg@xxxxxxxxxx are
avoid-tasklist_lock-at-getrusage-for-multithreaded-case-too.patch
ptrace-document-the-locking-rules.patch
list-introduce-list_replace-helper.patch
list-use-list_replace_init-instead-of-list_splice_init.patch
when-config_base_samll=1-the-kernel-261611-cascade-in-kernel-timerc-may-enter-the-infinite-loop.patch
when-config_base_samll=1-the-kernel-261611-cascade-in-kernel-timerc-may-enter-the-infinite-loop-use-list_replace_init.patch
sched_exit-fix-parent-time_slice-calculation.patch
sched_exit-move-the-callsite-to-do_exit.patch
proc-remove-tasklist_lock-from-proc_pid_readdir-simply-fix-first_tgid.patch
proc-dont-lock-task_structs-indefinitely.patch
proc-dont-lock-task_structs-indefinitely-task_mmu-small-fixes.patch
simplify-fix-first_tid.patch
cleanup-next_tid.patch
de_thread-fix-lockless-do_each_thread.patch
coredump-optimize-mm-users-traversal.patch
coredump-speedup-sigkill-sending.patch
coredump-kill-ptrace-related-stuff.patch
coredump-kill-ptrace-related-stuff-fix.patch
coredump-dont-take-tasklist_lock.patch
coredump-some-code-relocations.patch
coredump-shutdown-current-process-first.patch
coredump-copy_process-dont-check-signal_group_exit.patch
pidhash-temporary-debug-checks.patch
-
To unsubscribe from this list: send the line "unsubscribe mm-commits" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
