On Fri, 26 Feb 2016 11:59:40 -0800
"Shi, Yang" <yang.shi@xxxxxxxxxx> wrote:
BTW, it sounds this is not the only point where kernfs_node could be
updated, __kernfs_remove should need synchronize_sched too.
Question is, can the kernfs of a cgroup be removed while the cgroup is
still active?
I don't see the kernfs_rename_lock being held anywhere in that
remove. If it can be an issue there with this patch, then it's an issue
today, because the kernfs_mutex is not held by the tracepoint.
-- Steve