OSDN Git Service

cgroup: move cgroup->sibling unlinking to cgroup_put()
authorTejun Heo <tj@kernel.org>
Wed, 14 May 2014 13:15:01 +0000 (09:15 -0400)
committerTejun Heo <tj@kernel.org>
Wed, 14 May 2014 13:15:01 +0000 (09:15 -0400)
commit4e4e28472365f8c7a7c55f6b5706f68bc40c5b13
tree537d6c32f129fd90347463f5e645c6dfe51f5860
parent9e4173e1f24fa3bd562f13b92ee34c7dfb1db7c9
cgroup: move cgroup->sibling unlinking to cgroup_put()

Move cgroup->sibling unlinking from cgroup_destroy_css_killed() to
cgroup_put().  This is later but still before the RCU grace period, so
it doesn't break css_next_child() although there now is a larger
window in which a dead cgroup is visible during css iteration.  As css
iteration always could have included offline csses, this doesn't
affect correctness; however, it does make css_next_child() fall back
to reiterting mode more often.  This also makes cgroup_put() directly
take cgroup_mutex, which limits where it can be called from.  These
are not immediately problematic and will be dealt with later.

This change enables simplification of cgroup destruction path.

Signed-off-by: Tejun Heo <tj@kernel.org>
Acked-by: Li Zefan <lizefan@huawei.com>
kernel/cgroup.c