On Wed, Jan 23, 2019 at 05:31:44PM -0500, Chris Down wrote: > memory.stat and other files already consider subtrees in their output, > and we should too in order to not present an inconsistent interface. > > The current situation is fairly confusing, because people interacting > with cgroups expect hierarchical behaviour in the vein of memory.stat, > cgroup.events, and other files. For example, this causes confusion when > debugging reclaim events under low, as currently these always read "0" > at non-leaf memcg nodes, which frequently causes people to misdiagnose > breach behaviour. The same confusion applies to other counters in this > file when debugging issues. > > Aggregation is done at write time instead of at read-time since these > counters aren't hot (unlike memory.stat which is per-page, so it does it > at read time), and it makes sense to bundle this with the file > notifications. ... > Signed-off-by: Chris Down <chris@xxxxxxxxxxxxxx> > Acked-by: Johannes Weiner <hannes@xxxxxxxxxxx> Acked-by: Tejun Heo <tj@xxxxxxxxxx> Michal has a valid counterpoint that this is a change in userland visible behavior but to me this patch seems to be more of a bug fix than anything else in that it's addressing an obvious inconsistency in the interface. Thanks. -- tejun
