While refactoring cgroup creation, a5bca2152036 ("cgroup: factor out
cgroup_create() out of cgroup_mkdir()") incorrectly onlined subsystems
before the new cgroup is associated with it kernfs_node.  This is fine
for cgroup proper but cgroup_name/path() depend on the associated
kernfs_node and if a subsystem makes the new cgroup_subsys_state
visible, which they're allowed to after onlining, it can lead to NULL
dereference.

The current code performs cgroup creation and subsystem onlining in
cgroup_create() and cgroup_mkdir() makes the cgroup and subsystems
visible afterwards.  There's no reason to online the subsystems early
and we can simply drop cgroup_apply_control_enable() call from
cgroup_create() so that the subsystems are onlined and made visible at
the same time.

Signed-off-by: Tejun Heo <tj@kernel.org>
Reported-by: Konstantin Khlebnikov <khlebnikov@yandex-team.ru>
Fixes: a5bca2152036 ("cgroup: factor out cgroup_create() out of cgroup_mkdir()") 
Cc: stable@vger.kernel.org # v4.6+

1 files changed, 6 insertions, 0 deletions

diff --git a/man/ifpps.8 b/man/ifpps.8
index a0abba7..46088dd 100644
--- a/man/ifpps.8
+++ b/man/ifpps.8
@@ -77,6 +77,12 @@ On 10Gbit/s cards or higher, receive and transmit statistics are usually
 accumulated each > 1sec. Thus, it might be advised to alter the timing
 option to a higher accumulation interval for such cards.
 
+.SH BUGS
+Systems with a failry high number of cores (> 32) are currently not
+supported. This should however not be a big deal to fix that. The only
+challenge would be to display the presented information in a sane way,
+probably by selectively hiding uninteresting statistics.
+
 .SH LEGAL
 ifpps is licensed under the GNU GPL version 2.0.