mm_update_next_owner() checks the children and siblings to avoid
the "global" for_each_process() loop. Not sure this makes any sense,
but certainly it makes sense to check our sub-threads before anything
else.
Signed-off-by: Oleg Nesterov <oleg@xxxxxxxxxx>
---
kernel/exit.c | 8 +++++++-
1 files changed, 7 insertions(+), 1 deletions(-)
diff --git a/kernel/exit.c b/kernel/exit.c
index 1d1810d..b1f7135 100644
--- a/kernel/exit.c
+++ b/kernel/exit.c
@@ -340,6 +340,12 @@ void mm_update_next_owner(struct mm_struct *mm)
read_lock(&tasklist_lock);
/*
+ * Search in the sub-threads
+ */
+ if (assign_new_owner(mm, p))
+ goto done;
+
+ /*
* Search in the children
*/
list_for_each_entry(g, &p->children, sibling) {
@@ -359,7 +365,7 @@ void mm_update_next_owner(struct mm_struct *mm)
* Search through everything else, we should not get here often.
*/
for_each_process(g) {
- if (g->flags & PF_KTHREAD)
+ if (g == p || g->flags & PF_KTHREAD)
continue;
if (assign_new_owner(mm, g))
goto done;
--
1.5.5.1
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@xxxxxxxxx. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@xxxxxxxxx";> email@xxxxxxxxx </a>