Messages in this thread Patch in this message | | | Date | Thu, 14 Jul 2016 16:58:44 +0200 | From | Oleg Nesterov <> | Subject | Re: Severe performance regression w/ 4.4+ on Android due to cgroup locking changes |
| |
On 07/14, Peter Zijlstra wrote: > > OK, not too horrible if I say so myself :-) > > The below is a compile tested only first draft so far. I'll go give it > some runtime next.
Yes, thanks.
But note that we do not need RCU_NONE. All we need is the trivial change below. Damn, I am trying to find my old rcu-sync patches which I didn't send, but can't... OK, this almost off-topic right now, just this "enter" is ugly and we can't switch the slow/fast modes dynamically.
The rest of you patch is "optimize the slow path" and we already discussed it before, I personally like it. Perhaps you can redo it without RCU_NONE part?
Of course, this leads to another question: do we really need rcu-sync at all, or should we change percpu-rwsem to always work in the "slow" mode which is not that slow with your change... I'd like to keep it ;)
What do you think?
Oleg.
--- a/kernel/cgroup.c +++ b/kernel/cgroup.c @@ -5605,6 +5605,8 @@ int __init cgroup_init(void) BUG_ON(cgroup_init_cftypes(NULL, cgroup_dfl_base_files)); BUG_ON(cgroup_init_cftypes(NULL, cgroup_legacy_base_files)); + rcu_sync_enter(&cgroup_threadgroup_rwsem.rss); + get_user_ns(init_cgroup_ns.user_ns); mutex_lock(&cgroup_mutex);
| |