cgroup: protect cgroup->nr_(dying_)descendants by css_set_lock
authorRoman Gushchin <guro@fb.com>
Fri, 19 Apr 2019 17:03:03 +0000 (10:03 -0700)
committerTejun Heo <tj@kernel.org>
Fri, 19 Apr 2019 18:26:48 +0000 (11:26 -0700)
commit4dcabece4c3a9f9522127be12cc12cc120399b2f
tree0d985b3d1232ff76e67a0ac07213e4c8e4ab4a7b
parentaade7f9efba098859681f8e88d81a5b44ad09b12
cgroup: protect cgroup->nr_(dying_)descendants by css_set_lock

The number of descendant cgroups and the number of dying
descendant cgroups are currently synchronized using the cgroup_mutex.

The number of descendant cgroups will be required by the cgroup v2
freezer, which will use it to determine if a cgroup is frozen
(depending on total number of descendants and number of frozen
descendants). It's not always acceptable to grab the cgroup_mutex,
especially from quite hot paths (e.g. exit()).

To avoid this, let's additionally synchronize these counters using
the css_set_lock.

So, it's safe to read these counters with either cgroup_mutex or
css_set_lock locked, and for changing both locks should be acquired.

Signed-off-by: Roman Gushchin <guro@fb.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
Cc: kernel-team@fb.com
include/linux/cgroup-defs.h
kernel/cgroup/cgroup.c