[Devel] [PATCH rh7 1/2] mm/memcg: Don't enable interrupts too soon.

[Andrey Ryabinin]

When mem_cgroup_move_parent() moves huge page it disables interrupt:

	if (nr_pages > 1) {
		VM_BUG_ON_PAGE(!PageTransHuge(page), page);
		flags = compound_lock_irqsave(page);
	}

and calls:
	ret = mem_cgroup_move_account(page, nr_pages, ...

which does the following:

	local_irq_disable();
	mem_cgroup_charge_statistics(to, page, nr_pages);
	...
	local_irq_enable();

So the last local_irq_enable() enables irq too early, which may lead
to the deadlock. mem_cgroup_move_account() should use local_irq_save()/
local_irq_restore() primitives instead.

Found while investigating https://jira.sw.ru/browse/PSBM-76011
but unrelated.

Signed-off-by: Andrey Ryabinin <aryabinin at virtuozzo.com>
---
 mm/memcontrol.c | 5 ++---
 1 file changed, 2 insertions(+), 3 deletions(-)

diff --git a/mm/memcontrol.c b/mm/memcontrol.c
index 239fbca70b59..efc455d8ca81 100644
--- a/mm/memcontrol.c
+++ b/mm/memcontrol.c
@@ -3597,15 +3597,14 @@ static int mem_cgroup_move_account(struct page *page,
 
 	/* caller should have done css_get */
 	pc->mem_cgroup = to;
-	move_unlock_mem_cgroup(from, &flags);
+	spin_unlock(&from->move_lock);
 	ret = 0;
 
-	local_irq_disable();
 	mem_cgroup_charge_statistics(to, page, nr_pages);
 	memcg_check_events(to, page);
 	mem_cgroup_charge_statistics(from, page, -nr_pages);
 	memcg_check_events(from, page);
-	local_irq_enable();
+	local_irq_restore(flags);
 out_unlock:
 	unlock_page(page);
 out:
-- 
2.13.6