提交 c976a67b 编写于 作者: L Linus Torvalds

Merge branch 'for-3.19-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/tj/cgroup

Pull cgroup fix from Tejun Heo:
 "The lifetime rules of cgroup hierarchies always have been somewhat
  counter-intuitive and cgroup core tried to enforce that hierarchies
  w/o userland-visible usages must die in finite amount of time so that
  the controllers can be reused for other hierarchies; unfortunately,
  this can't be implemented reasonably for the memory controller - the
  kmemcg part doesn't have any way to forcefully drain the existing
  usages, leading to an interruptible hang if a following mount attempts
  to use the controller in any way.

  So, it seems like we're stuck with "hierarchies live on till they die
  whenever that may be" at least for now.  This pretty much confines
  attaching controllers to hierarchies to before the hierarchies are
  actively used by making dynamic configurations post active usages
  unreliable.  This has never been reliable and should be fine in
  practice given how cgroups are used.

  After the patch, hierarchies aren't killed if it isn't already
  drained.  A following mount attempt of the same mount options will
  reuse the existing hierarchy.  Mount attempts with differing options
  will fail w/ -EBUSY"

* 'for-3.19-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/tj/cgroup:
  cgroup: prevent mount hang due to memory controller lifetime
......@@ -1909,7 +1909,7 @@ static void cgroup_kill_sb(struct super_block *sb)
*
* And don't kill the default root.
*/
if (css_has_online_children(&root->cgrp.self) ||
if (!list_empty(&root->cgrp.self.children) ||
root == &cgrp_dfl_root)
cgroup_put(&root->cgrp);
else
......
Markdown is supported
0% .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑!
想要评论请 注册